INSTITUTIONAL DETERMINANTS OF THE INVESTMENT …

UNIWERSYTET WARSZAWSKI

WYDZIA L NAUK EKONOMICZNYCH

INSTITUTIONAL DETERMINANTS

OF THE INVESTMENT PROCESSES

IN POLAND

Essays on selected issues

A Ph.D thesis submitted by Joanna TyrowiczPromoted by Professor Jerzy Wilkin

June 2005

INSTITUTIONAL DETERMINANTS OF THE

INVESTMENT PROCESSES IN POLAND

Joanna Tyrowicz

Version: June 5th, 2005

CONTENTS

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2. Financial institutions and economic development - a general framework . . . . . . 102.1 The theoretical foundations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.2 The empirics of finance and growth . . . . . . . . . . . . . . . . . . . . . . . . 15

3. The market structure in the banking sector and the investment processes . . . . . 213.1 The competition in the Polish banking sector . . . . . . . . . . . . . . . . . . 233.2 Model I - asymmetry of response to shock . . . . . . . . . . . . . . . . . . . . 24

3.2.1 Econometric obstacles . . . . . . . . . . . . . . . . . . . . . . . . . . . 263.2.2 Data and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.3 Model II - credit-deposit nexus . . . . . . . . . . . . . . . . . . . . . . . . . . 283.3.1 Data and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.4 Policy implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4. The Open Pension Funds as Capital Providers . . . . . . . . . . . . . . . . . . . . 314.1 The architecture of the new social security system . . . . . . . . . . . . . . . 324.2 The systemic shortcomings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344.3 The impact on the financial markets . . . . . . . . . . . . . . . . . . . . . . . 35

4.3.1 The strength of the link with the capital market . . . . . . . . . . . . 364.3.2 The efficiency of the pension funds . . . . . . . . . . . . . . . . . . . . 384.3.3 The costs of OPFs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394.3.4 The efficiency of savings management . . . . . . . . . . . . . . . . . . 40

4.4 Possible remedies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

5. The Venture Capital In Poland . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465.1 The VC industry in Poland . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485.2 How to run a VC fund in Poland . . . . . . . . . . . . . . . . . . . . . . . . . 505.3 But what is a good business opportunity? . . . . . . . . . . . . . . . . . . . . 535.4 The perspective for the future . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

6. The Efficiency of The Investment Decisions . . . . . . . . . . . . . . . . . . . . . . 606.1 Theoretical foundations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

6.1.1 Q-Theory of investment . . . . . . . . . . . . . . . . . . . . . . . . . . 616.1.2 Marginal versus average Tobin’s q . . . . . . . . . . . . . . . . . . . . 64

6.2 Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

Contents ii

6.2.1 The hypothesis of managerial discretion . . . . . . . . . . . . . . . . . 666.2.2 Some controversies over this approach . . . . . . . . . . . . . . . . . . 696.2.3 Other approaches to cash flow sensitivity of investment . . . . . . . . 70

6.3 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 726.3.1 Methodology and data . . . . . . . . . . . . . . . . . . . . . . . . . . . 736.3.2 The Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

6.4 Possible refinements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

8. Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Appendix 80

A. Endogenous growth model with the financial institutions . . . . . . . . . . . . . . . 81

B. The factors controlled for by Leitinger and Schofer . . . . . . . . . . . . . . . . . . 87

1. INTRODUCTION

Without investment there is neither economic growth nor social development. This state-ment is common knowledge and as such seems to require no further elaboration. Observingthe GDP growth rates of the Polish economy one might think that despite the temporaryperiods of slowdown our economy experiences no problems in this domain. On the otherhand, however, complaints about low levels of investment are very frequent, while invest-ments themselves are rarely believed to be of self-sustainable nature. Entrepreneurs point tolow capital availability and politicians emphasise insufficient increase in employment offers.In addition, the central bank is criticised as well for ’too high’ interest rates and hence ’tooexpensive’ credit. Similarly, the economic policy of the government towards the entrepreneursis found to be lacking development strategy and entreprise targeted programmes. Can onethus state that the economic development happens ’despite’?

Evidently, when talking about investment, one tends to combine many aspects: local andglobal availability of capital, supply of investment projects, monetary policy, fiscal policy,development strategies of the local and central governments, etc. This thesis has much morehumble scope of interest. Its main purpose is to analyse the institutional entourage of theinvestment processes in the search to analyse the efficiency of the financial system design.To a certain extent, institutions provide the fundamentals for the ceteris paribus - shouldthey fail to efficiently manage the processes of investment, resources are not allocated to themaximum general benefit regardless of the capital abundance and growth outlooks.

The fifteen years of transformation have yielded a number of formal and informal insti-tutional designs which influence directly the processes of investment. These include the Acton Open Pension Funds from 1999 one one hand, but also unofficial internal bank blueprintsabout the criteria of assigning credits. As these are mostly strategies of the players, they arerarely observable and one usually addresses them with the competition-enhancing-efficiencyargument. Even if this argument was always true, perfect competition is rather seldom, whilstinformation asymmetry as well as moral hazard contribute permanently to this landscape.

This thesis aims at identifying whether de facto institutions tend to have the pro-investmentand thus pro-development character. Should we find that a particular financial institutionaldesign hinders the process of investment, such an institution would be claimed inefficient.Furthermore, we attempt to identify the fundamental sources of these inefficiencies and thusprovide some indications with regards to regulation and future policies implications. Themain research question approached in this thesis may be summarised to the following: withrespect to the financial institutions that emerged during the transition, can weconsider them to embed pro-development characteristics. Unfortunately, a negativeanswer to this questions seems to find confirmation in the empirical analysis. Although un-

1. Introduction 2

doubtedly not all investment capital provider categories are analysed, the evidence providedpoints to the necessity of changing, adjusting or enforcing the institutional design in all ofthe analysed domains.

There is a number of reasons why the financial sector and its activity may influence therate of economic growth which follows directly from the role the financial institutions play inthe economy. As Merton and Bodie (1995) suggest, the principal role of the financial systemin the economy is to smooth the process of the ”allocation of resources, across space and time,in uncertain environment”. Specifically, following the synthesis of Levine (1997), financialintermediaries channel the resources to the most profitable sectors of an economy. They alsomonitor managers and exert corporate control ameliorating moral hazard risk. In particular,by providing liquidity, financial institutions permit risk averse savers to hold deposits ratherthan liquid but unproductive assets (like money). Moreover, this mobilisation of savingsallows to increase the amount of resources available to entrepreneurs.

So much for the theory. Unfortunately, one tends to mistake most profitable from abank point of view for the most efficient from an economy point of view as well as expectsmore information under the term ”amelioration of moral hazard”. In reality, however, agentsexhibit goals other than maximisation of profits (rentability, market share, turnover, valueto the shareholders, cost minimisation, etc.). Furthermore, product differentiation as well asmarketing techniques often blur this picture even further. For example, the banking sectormay be very competitive on the deposit market where information is available easily, butat the same time behave differently at the investment credit market. Finally, marketingtechniques (like branding, loyalty inducing etc.) seriously undermine the trust in financialmarkets efficiency, especially in a relatively young market economy.

In Poland there are de facto two types of capital providers: the banks and the privateinvestors1 This division is of institutional nature in a sense that the law and the followingprivate institutions shape the capital allocation processes differently. As far as banks areconcerned, it seems crucial to determine first the objective function and then proceed to thequestion of to what extent their nature of typical ’utility’ maximising agents influences theirinvestment credit policies. Is it possible that these innate characteristics may either hinderor effectively anihilate investment? If so, what are the possible remedies?

For the purpose of this thesis, the category of private investors is narrowed to OpenPension Funds2 and Private Equity/Venture Capital environment. In the case of the latter,despite their marginal share in GDP, they may be perceived as an important barometer ofthe economic agents’ behaviour, thus justifying their presence among the analysed institu-tional designs. One should expect them to constitute the influential source of capital for theentrepreneurs, while their interests should be focused around high return/risk ratio of theundertaken investments. However, they are also regular entreprises with internal problem,like the the principal-agent problem for example. Do we have any mechanisms forcing themto behave the expected way? Is the invisible hand of the competition strong enough to ensurethe desired general equilibrium outcome? And over the short run? If not, what are the tools

1 For the purposes of this thesis this category is limited to the institutionalised investors.2 The second pillar of the social security system.

1. Introduction 3

at our disposal?The issues indicated above focused on the efficiency of the capital provision processes

should the investment initiative arise. For the time being, however, the very investmentdecision remained unadressed. This is dealt with in the final part of the thesis where we aimat assessing the investment behavioural patterns. Due to little information on the quality ofprojects ex ante and the noisiness of the ex post data in most cases it is rather difficult toevaluate the investment decisions. In addition, long term strategies of the agents should alsobe controlled for in this kind of scrutiny, whilst access to this type of data is even more limited.On the other hand, however, should one dispose a fairly objective valuation of the companiesprospects, one might be able to seeks these answers. We thus employ a agency theorymodel to Polish stock-listed companies, aiming to assess the correctness of the investmentdecisions by their managers. This kind of query allows also to specify whether shareholdersenjoy sufficient protection of their rights, hence identifying the managerial discretion problemamong the Polish stock-listed companies.

One may thus specify the four main domains of interest addressed in this thesis:

• The impact of the market structure in the banking sector on the investment creditpolicies implemented.

• The efficiency of investment funds - in particular the pension funds - in transformingthe economy savings into investments.

• The role of the PE/VC industry in Poland.

• The extent to which the investment decisions themselves are taken correctly.

Since the above listed institutional designs profoundly differ in functionalities, for each ofthis domains we employ a different methodology, the main scope remaining the developmentorientation of an analysed setting. In particular, we first develop an endogenous growthmodel justifying the analysis of the financial institutions in the development context. Despitethe efforts, we have not been able to provide sound direct empirical evidence between theanalysed institutional settings and the growth rates of the Polish economy. Nevertheless,having placed the financial infrastructure in the landscape of economic growth discussionsin a general framework and providing evidence to chosen model specification, we providegrounds to further analyses and move to inquiring specifically the Polish environment.

The first element of the institutional setting analysed in this thesis concerns the bankingsector. One clearly observes a wide spread between the credit and the deposit rates in thecommercial banks. This wedge is traditionally attributed to the risk factor, but may alsofollow from the market structure and competition patterns, while little research is observed inthis area. Equally, whether this spread is widening over time and whether one can really claimit to be a consequence of relatively uncompetitive market structure has not been analysed inthe literature to our best knowledge. Theoretical fundamentals are provided in an industrialorganisation model borrowed from the oil industry and adapted to the financial institutionsneeds. The testable hypotheses derived from this model are confronted with the data reported

1. Introduction 4

by commercial banking institutions operating in Poland, thus providing sound justificationto the claim of a particular type of ’credit rationing’.

We later move to other capital providers, first analysing the pension funds as a repre-sentative in a wider category of investment funds. Due to intensive information disclosureobligation, open pension funds (OPFs) seem very fruitful from the researcher point of view.On the other hand, due to the scale of their operations, they are and will be by far the mostimportant players among the investment funds. However, analysing the incentives structurebuilt into the institutional design of this market one may hypothesise that pension fundsare only marginally encouraged to actively participate in the financial markets, which rarelyinvites them to efficiently transform the economy savings into investment. We develop aquantitative method allowing to assess the degree of OPFs involvement in the financial mar-kets. This model is supported by theoretical considerations providing justification to such astrategy.

The fourth chapter of this thesis is devoted to another form of capital providers, namelythe venture capital funds. Although their role in financing investment in Poland is ratherlimited - or rather precisely because of that - one is tempted to analyse their mode of op-erating as well as try to identify the potential reasons for the status quo. Unfortunately inthis domain data is significantly underprovided, which forces us to use the available resultsof quantitative research by PSIK3, its mother organisation EVCA4 as well as qualitativeanalyses based on surveys among the VC fund managers. Although this part of the thesisfocuses rather on hypothesising about the link between the observed behavioural patternsand the underlying strategies and incentives, an analysis of the legal environment from theorganisational efficiency point of view is also present.

Finally, to assess the investment processes one needs to analyse the investment decisionsby the companies themselves. As Q-theory of investment argues, in the case of companiesfor which the objective valuation is available, one may suggest the method to evaluate theefficiency of the undertaken investment decisions. Basing on the data of the Polish stock-listed companies5 we pursue an empirical analysis based on theoretical framework that can bebest fit into principal-agent literature. Thus, not only do we provide suggestions concerningthe efficiency of undertaken investment but also attempt to assess the quality of investorsprotection under Polish law and in the Polish conduct. This fifth chapter closes the thesis.

As emphasised by Levine (1996, 2004) when studying the dynamics of financial systems,it is best to adopt an analytical framework that treats functions rather than institutions asthe conceptual anchors6. In this analytical framework the functions are exogenous, and theinstitutional forms are endogenously determined (cfr. Merton and Bodie, 2004). The pointof departure to these considerations is the neoclassical paradigm of rational agents operating

3 Polish Association of Capital Investors4 European Venture Capital and Private Equity Association5 Although one might undermine the efficiency of Warsaw Stock Exchange as such, stock listed companies

might be considered to enjoy a fairly impartial valuation of their potential.6 See Merton (1993) on the functional perspective. The functional analytical framework presented here is

developed in Crane et al. (1995). Financial functions for financial institutions are also used in a different

analytic framework that originates from the important work of Diamond and Dybvig (1986).

1. Introduction 5

opportunistically in an environment of frictionless markets.If existing prices and allocations fail to conform to the neoclassical paradigm, it is helpful

to inquire why. The possible causes of such a failure might be of three kinds. Firstly, theexisting institutional rigidities, in which case we might consider applying institutional designtechniques to circumvent their unintended and dysfunctional aspects or abolish them directly,if the institutional sources are no longer needed. Secondly, the technological inadequacies,which may disappear over time as a result of innovation. Finally, the dysfunctional behavioralpatterns that cannot be offset directly by institutional changes.

Even when individuals behave in ways that are irrational, institutions may evolve to offsetthis behavior and produce a net result that is ”as if” the individuals were behaving rationally.This is a version of Adam Smith’s ”invisible hand”. Structural models that include transac-tions costs, irrational behavior or other ”imperfections” may give distorted predictions whenframed in a neoclassical ”minimalist” institutional setting of atomistic agents interacting di-rectly in markets. It is therefore essential to include the endogenous institutional response.The resulting institutional design, if not already in place, can be seen as providing either aprediction about the dynamics of future institutional change or as a normative prescriptionfor innovation.

As regards the institutions themselves, in this thesis we follow the game theoretic perspec-tive. After Aoki (2001) an institution is a summary representation of the invariant and salientfeatures of one out of possible many (Nash) equilibrium paths, internalized as shared beliefsof all the players regarding ways by which the game is being repeatedly played. This is theshared beliefs mechanism that allows for internalising the equilibria via Bayesian-type of up-grading our expectations. In addition, this is the Nash equilibrium that works as a horseof self-enforceability. One should note however that Nash equilibrium does not necessitatePareto optimality. In addition, in the above context beliefs have no normative notion.

This concept seems particularly useful for a number of reasons. Primarily, it leaves thenotion of equilibrium unspecified beyond the best-response Nash setting. Thus, it does notrule out any adaptation pattern (Bayesian upgrading of beliefs), thus allowing for changing theequilibria accordingly to changes in beliefs and their endogenisation7. Neither does it imposeany requirements on the behaviour of the agents. Bounded rationality as well as altruism fitthis approach just as well as full information utility maximising or effort minimising for thatmatter.

Secondly, in general there can be multiple Nash equilibria for a game offering a reasonabledegree of freedom of choice. The attempts of game-theorists to refine the concept of equilib-rium in order to reduce the possibility of equilibrium to one have proven to be unsuccessful.But this should be regarded rather as a blessing in disguise for institutional economists be-cause it indicates that a variety of institutions may be possible for the same environment.In fact, Aoki has made the multiplicity of equilibrium a precondition for an institution, asSugden (1989) did for his conceptualization of a convention.

Finally, the game theoretic approach allows to avoid the formal-informal institution di-7 This perception of change dates back to Schumpeter (1898) while providing a potentially powerful ana-

lytical tool

1. Introduction 6

chotomy. Let us consider here a very naive example of speed limitations. It seems ratherplausible to believe that most people (here: agents) realise both the dangers of speeding aswell as consequences of being caught by the police. Still, some of them continue to violatethe norms. In the formal-informal institution approach one should find a criterion to claimwhether an institution has been internalised or not. If 50% of drivers obey the limit in Swedenis it the same level of internalisation as 90% in Ouzbekistan? And what if back in Swedenit is no longer 50%, but the drivers exceed the limit by less kilometers per hour on average?Any such claim would constitute an easy target for a critique, while providing relatively littleinsights into the process of institutions formation.

The game theoretic approach of Aoki offers a completely different perspective. Peoplerepresent agents with independent and possibly highly complex objective functions. Thesefunctions lead to formulating some strategies, while shared beliefs allocate among them thepotential payoffs. As a consequence, each and every one of the drivers on a road enforces theindependent strategy of going above the speed limit or respecting it (to keep the considerationsat the discretionary level). As long as one may not demonstrate they do not realise thepresence of the limit, the potential fine from the police officer as well as an increased odds ofan accident, one is faced with an institution as defined by Aoki. If any design fails to satisfythis very basic condition, one may hardly name it an institution despite the potentiallyformal character of this uninternalised design, as would have been the classification in theWilliamson-North framework. This is best demonstrated on the scheme below.

1. Introduction 7

The summary representations of salient features of Nash equilibria - as Masahiko Aokinames the institutions - represent also other characteristic traits. Namely, they coexist nat-urally in four different environments: the social, the political and the organisational domainas well as the pure economic exchange. Strategies chosen by players may overpass the bound-aries of one single category. This may be again illustrated on the speeding example. First,anyone that exceeds the limits faces the risk of being forced to bear the fine costs. Secondly,there is certainly a social dimension to increasing the accident odds as well as loosing thereputation in the eyes of other drivers and potentially also the accompanying people. Fur-thermore, the political dimension may be sought in the preferences for one or other partycompeting for the votes in the closest elections. Finally, the organisational domain is presentin the way traffic in general is managed as well as the spontaneous phenomena such as thepeak hours and congestion.

In the case one or more players chose strategies from more than one domains in a coordina-tive (consistent) manner, one may apply the notion of a linked game. Because of the possibleexternalities (social surpluses) created by such linkages, a choice pattern that is unsustainablein a single domain in isolation may become sustainable in a broader set of domains. Thisleads to a formulation of a tool crucial for analysing the institutional change, namely theinstitutional complementarities. What is particularly interesting is that in linked games eachagent or a particular agent coordinates his/her own strategic choices across domains and gen-erate a single institution (equilibrium). Alternatively, we can conceive of the possibility that,even if agents may not strategically coordinate their own choices across domains, they regardan institution in another domain as a parameter, hence accordingly choose strategies in owndomain, and vice versa. In such situations, institutions evolving in each of those domains maybecome interdependent and mutually reinforcing. This intuition can be game-theoreticallywarranted8

This is a powerful and useful analytical tool for institutional analysis. First, as justmentioned, it explains why there can be a variety of over-all institutional arrangements, evenif economies face the same types of domain characteristics (such as technologies, commonmarkets connecting them), as well as why a sub-optimal over-all institutional arrangementcan persist. Second, institutional complementarities are not necessarily conditional on aconsensus among agents in a domain (i.e. it is not required that in terms of pay-offs onealternative strictly dominates the other). Only a weaker agreement in the direction of changein their ”relative” preferences associated with parametric changes matters (Aoki 2001, p. 15).Thus the emergence and sustenance of an over-all institutional arrangement may becomestable even if there is a conflict of interests among agents about the absolute preference fora component institution in isolation.

Aoki (2001) suggests that, contrary to the conventional view, an equilibrium view of aninstitution is not necessarily inconsistent with the evolutionary approach in the traditionof Schumpeter. On the contrary, the game-theoretic equilibrium view of an institution canapply the essential point that Schumpeter made regarding the nature of innovation, that is,

8 For a proof, see Aoki (2001).

1. Introduction 8

the creative destruction, to the study of institutions9.To the extent to which the game-theoretic approach is employed in this thesis, it allows

to embody the entire approach to the Polish financial institutions argued here. Firstly, someof the analysed phenomena can definitely be categorised as consistently chosen strategiespreferred by the players despite the potentially higher level equilibria - be it profits, turnover,utility, growth or any other reference criterion. Thus, we suggest that some of the Nashoutcomes observed in this sphere are rather unlikely to self-improve and might require aninstitutional intervention at the level of at least forceful influence over the beliefs (commit-ment).

To be more explicit on this issue, let us assume for the sake of the argument that players onsome market can choose between Cournot and Bertrand type of competition. The outcomeof the latter yields zero profits and a socially optimal price level. The former allows theproducers to keep the price level above the marginal costs in the equilibrium thus allowingfor economic profits. Let us further assume, that the activity companies perform on thisparticular market is strictly licensed, which definitely limits the free entry and eliminatesthe n → ∞ type of analysis. Thirdly, let us assume that one of the players is wronglybut consistently perceived by consumers to be a definitely safer supplier than any others,whatever the reason. Finally, for some non fully rational reason but in a consistent manner,consumers prefer bigger producers to smaller ones, while being forced to accept relativelyhigh switching costs imposed by the producers.

From even this rough sketch, one sees that the evident market outcome is of Stackelber-gian type with quantities as choice variables and consumers highly concentrated among fewbiggest players. Furthermore, producers are likely to make efforts to further decrease pricesensitivity of consumers by introducing other arguments in favour of further concentration.Similarly, consumers understanding the rules of the game, are not likely to be attracted byspecial programmes and undercutting, expecting it to be a short-term activity only aim-ing at convincing them to ’shop around’ the current offers, bearing every time considerableswitching costs. Evidently, rendering this market a perfectly competitive one requires muchmore than levying the entry limitations. Obviously, this example was designed to refer to theconsumer banking in Poland.

Secondly, none of the framework demonstrated in this thesis could be clearly identified asa formal or informal institution in the North-Williamson line of thinking. On the other hand,the summary representation about how the game is repeatedly played is exactly what one mayobserve when conducting interviews with the participants to this market. The persistenceargument is surprisingly often dominating the pure economic rationale, while the conflicts ofinterest, as is argued in the thesis, lead to clearly suboptimal but feasible from the players’points of view outcomes.

To refer once again to the example of Polish consumer banking sector, except for entryregulation, all of the described institutions would have been labeled informal. However, whatis really crucial is that they are mostly based on beliefs and on conduct type of fundamentals.

9 He also proposes three potential frameworks of modeling the process of institutional change. Cfr. Aoki

(2001).

1. Introduction 9

Rational or not, consumers are not likely to shop around searching the cheapest bank orhighest deposit rates every month or even quarter. With more and more operations performedby banks automatically, effort involved in changing an account number is considerable. Thus,although consumers are aware that a current solution is far from being optimal, they are rarelywilling to change it allowing the systemic inefficiencies to persist.

Thirdly, in the sphere of potential political implications, the approach initiated by Aokiallows for seeking solutions on a much wider scale than the legal change or necessitating moreeffective law enforcement. Altering the incentives in the systems plays a role of introducinga shock to the current status of beliefs. Although the endogenisation of the updating processis not automatic, one seems to enjoy more tools for inducing the institutional change.

This is to say that perhaps on some instances it is not necessary to change an equilibriumsolution to affect the strategies of players, as it might suffice to change the alternatives. In thebanking and final consumers example, the revolution introduced by dedicated retail banks10

followed by the onset of internet banking11 and has caused many clients to finally abandontheir expensive and traditional banks with much narrower ’menu’. The response to thisleakage of clients was the horizontal integration of products - being in a bank, a client canhave life insurance and pension account in the same group. This is equivalent to imposinghigher switching costs and decreasing the propensity to look for cheaper banking productselsewhere.

Lastly, there is an additional benefit in the sphere of the improvement perspectives. North-Williamson framework shares the common characteristic of the persistence of institutions.Political economy literature makes the economists even less optimistic about the possiblesituation improvement. In the game theoretic framework not only is the rapid change possible(updating beliefs, introducing incentives), but also the sustainability matter seems to be farmore promising.

10 That is Handlobank and Millenium.11 Among others Inteligo and VolksvagenBank.

2. FINANCIAL INSTITUTIONS AND ECONOMIC DEVELOPMENT - A

GENERAL FRAMEWORK

The analysis of the institutional design of the financial markets in this thesis is justifiedby the silent assumption that they have the impact on the processes of investment thusinfluencing the development of an economy. However this link is not as evident as one mightthink. Although as early as in the 19th century Bagehot (1873) stressed the role of financialsystem in commencing and sustaining the industrial revolution in England, some suggest theopposite. Namely, Robinson (1952) claims that this was just a response to the needs of theeconomy. Similarly, Lucas (1988) in his Nobel lecture went as far as to stating that ”theimportance of financial matters is badly over-stressed”.

Empirical research in this domain, on the other hand, is obviously troubled by the en-dogeneity problem. The solution given by the Instrumental Variables, although useful to alarge extent, cannot unequivocally grant the financial institutions a role in fostering economicgrowth. Although many of the findings seem to be robust to the endogeneity problem, thecausality is still uncertain.

Furthermore, the financial systems themselves pose a challenge for a comparative study, asthe differ substantially between countries. The most rough division runs along the Atlanticseparating the continental banking system and the American one, more market oriented(with the UK bearing more of the American than the European traits). The two majorjustifications for the very presence of the financial markets, i.e. the information asymmetryand the moral hazard are tackled differently in both these solutions making any comparisonsrather difficult. In the European system, financially constrained companies run the risk ofbeing subject to credit rationing by banks (whatever the reason). On the other hand, however,specialisation among banks helps to alleviate the moral hazard problem - costs of mimickingraise substantially, lowering the costs of obtaining the funds even for small projects. TheAmerican solution is exactly the opposite. The initial barriers to participate are rather highdue to significant information costs. Conversely, once a company is able to participate inthe stock market, it enjoys relatively easier access to capital as the information asymmetryproblems are solved mutually by the investors and the specialised institutions.

It seems worthy to emphasise that in this reasoning both banks and the stock marketsare assumed to play their role efficiently. Efficiently not in the understanding that they areable to make money on their core activities, but in the understanding that they actuallyplay the expected roles either because they constitute their objectives or because their objec-tives and the social objectives coincide. This assumption is a very strong one. As forcefullydemonstrated by Graham, Harvey and Rajgopal (2004), the very details of information dis-closure rules in the US significantly affect the behaviour of companies, often to the loss of

2. Financial institutions and economic development - a general framework 11

the shareholders as well as stakeholders. Specifically, vast majority of firms regard earnings(and EPS) as a key indicator to the outsiders. Consequently, they are willing to sacrificethe economic value at the altar of smooth earnings. Furthermore, they would even forgoprojects of highly positive NPV and IRR if this implied falling short of the current quarter(sic! ) declared earnings. Dramatically, from the purely theoretic point of view, not only isthis an obvious inefficiency, but also an irrationality of agents. Stunningly for an economist,in the real world, this type of behaviour is simply an implied and induced by the structure ofincentives. The statistical as well as economic significance of Graham, Harvey and Rajgopal(2004) results are even more alarming if one considers that they only analyse the disclosurerules.

Summarising, on the aggregate level, the finance-growth link raises serious methodologicaldoubts. The aim of this thesis is much more humble. We aim at pondering the particularitiesin the relations between the financial system and the entrepreneurs trying to verify whetherone can state that the institutions pertaining on this market do not hinder the processesof investment. Importantly, this approach has its roots in the new growth theory. In theneo-classical growth model as developed by Solow (1956) and Swan (1956) the influence offinancial system on economic growth is confined only to the short-run. On the steady-statelong-run equilibrium path stock markets, banks, and other financial institutions have noinfluence over the growth rate. Yet, the endogenous approach to growth allows to model thefinancial system in the development process via facilitating capital mobilisation and reducingthe costs of acquiring it. Followingly, the returns to innovation are increased and the growthrate raises.

2.1 The theoretical foundations

The first to raise this channel of transmission was Schumpeter (1912). Evaluation of newprojects and monitoring supposedly allows for channelling private capital to the highest valueuse and ameliorating the moral hazard and adverse selection problems. Since in such situationthe most prospective entrepreneurs are provided with financial support, the economy withwell-developed financial system tends to grow faster and in a more stable manner. Evidently,in perfectly competitive world of general equilibrium inhabited by identical agents (i.e. thebasic classical assumptions) there is no room for the Schumpeterian notion of ideas and”creative destruction”. On the other hand, it seems that Schumpeter is much closer to realityto the extent that intellectual property rights, markets troubled by imperfect competition andagency problems with incomplete contracts tend to be observed more often. Consequently, theboom of the ”new” growth theory in the end of 1980s completed in this respect the revolutiontriggered by emergence of the new institutional economics and industrial organisation.

The financial markets were first modeled in the endogenous growth framework by Kingand Levine (1993), who allow the financial system to direct the investment to activities thatenhance productivity most. Using the Shumpeterian framework of ”creative destruction”they develop a model similar to the one proposed by Aghion and Howitt (1992) as well asGrossman and Helpman (1991). They assume two channels of transmission from the finan-


cial markets to the economic growth. Primarily, the very presence of financial markets offersthe investors an evaluation of the prospective entrepreneurs, thus alleviating the informa-tion asymmetry problem. Secondly, by actually allocating the investments, banks and otherfinancial institutions generate certain positive externalities due to the coordination of theR&D.

King and Levine (1993) reasoning is based on the fact that financial intermediaries arespecialised in solving the asymmetric information problems. Thus, the costs of researching,evaluating and monitoring the entrepreneurs are lower in case of the financial intermediariesthan in the case of investors. Importantly, they also emphasise that distortions on the financialmarkets may actually hinder the growth process.

However, as pointed out by Van Cayseele (2002), one should be very careful in specifyingthe costly factors of monitoring the loans. Specialisation in alleviating the information asym-metry problems means that the firms one could a priori classify as potentially prone to creditrationing (e.g. small firms) de facto are not constrained because they can address an inter-mediary specialised in solving this particular type of either adverse selection or moral hazardproblem. As suggested by Audretsch and Elston (2002) this precisely underlies their empir-ical findings for Germany - smaller firms have relatively fewer liquidity constraints becausethey benefit form the structure of specialised institutions (Spaarkase). Similarly, Mulkay,Hall and Mairesse (2000) find that smaller French companies face less financial constraintsthan their bigger American counterparts. This is accredited to a different structure of thefinancial markets. The continental banking system (in its French edition) seems to providebetter financial products for small companies than the American stock market system. Call-ing on these examples, Van Cayseele (2002) suggest that one should think about modelsendogenising the specialisation in the banking industry.

Similar line of reasoning seems to be in line with cross-country results in Rajan andZingales (1998), who find that financial dependence is not only firm specific, but also sectorspecific. Research, evaluation and monitoring in some sectors (e.g. biotechnology) is naturallymuch more costly than in others (e.g diary). This approach has been followed by Suda(2004), who extends the endogenous growth model of Aghion and Howitt (1992) by stressingthe channel through which the rate of innovations and technological progress are explicitlyinfluenced by the financial system. The crucial assumption rests upon the presence of thecredit market imperfections. As a result, the lower cost of acquiring external funding shouldinduce more innovations and faster growth, especially in sectors that are most reliant onexternal financing. Also, the overall increase in productivity due to the more careful choiceof innovation projects is identified.

Specifically, if we assume the Cobb-Douglas production function (Y = AH1−αKα, whereA denotes the state of ”environment”, H represents the human and K the physical capital),we can divide the economy into three levels: the R&D sector, the intermediate good sectorand the final good one. With the standard neoclassical assumptions concerning K and H

(decreasing marginal productivity and constant returns to scale), the state of environment, A,comprises various policy and control variables that are likely to influence the rate of growth.

Producers of the final good use a constant number of N varieties of intermediate inputs.


The assumption of constant N is held to eliminate for analytical reasons one of the potentialand exogenous sources of economic growth. In the ”new” growth theory innovation happens atthe intermediate stage of production. Each type of intermediate good has a ”quality ladder”along which the improvements can occur. These improvements build on the currently besttechnology and take the form of a sequential increase in the productivity level by the factor q.The physical capital, K, consists of an aggregate of different varieties of the quality-adjusted

capital goods, which can be represented by: K ={∑N

j=1 Xαj,kj

}1/α=

{∑Nj=1(qkjXj)α

}1/α.

Therefore, the parameter kj describes how many improvements in quality have occurred insector j. Once again for the sake of the argument, we assume that only the best existingquality of intermediate good j (with quality level qkj ) is available currently for production.

The crucial element concerns the property rights over the innovation. A successful inno-vator retains the exclusive right over the use of their improved intermediate good. In otherwords, once the intermediate good is discovered, its innovator enjoys monopoly power inthe production of their particular variety forever by being granted a patent. This securesthe funds for financing the R&D activity. If market was assumed perfectly competitive (i.e.there would be no advantage of the inventor over the followers) there would be no incentivesto engage into R&D as the free-riding would effectively eliminate all possible revenues. Inaddition, this assumption seems to be rather intuitive, to the extent to which one operates inenvironment with sound law enforcement. Finally, although patent protection is eternal, thedemand for this variety lasts only till the better quality good is invented. This last assumptionis evidently necessary to induce constant and self-sustainable innovative activities.

Whether we actually need the patent protection in the real world, however, remains anopen question. As Scherer (2004) convincingly demonstrates, there seems to be a numberof cases where creativity and intellectual rights property were not accompanying each other.Namely, there was no positive impact whatsoever of the copyright introduction on the musiccomposition in 18th and 19th century. Composers were ”doing what their hearts told them”just as now people often engage into business activities led by the ”entrepreneurial spirit”.Similarly, on-line communities are organised for many non-purely economic motivations, whilemany outsider consumers free-ride on their acquis. Nevertheless, in the modeled world, patentprotection is the only mean to introduce incentives for innovative activity and for this reasonthey are incorporated in the model.

The costs of entry are equal to the value of the resources devoted to the innovation process.In this model we assume that most of the firms cannot finance the R&D expenditures fromtheir contemporaneous activities. Hence, in order to engage in innovation activity they haveto acquire some externally funded resources, which is obviously costly. Thus a wedge iscreated between the cost of internal (but no longer available) and external resources costs1.

As suggested above, we assume that the extent to which firms need external financingdepends on the sector. Specifically, the external dependence takes form of the mark-up on

1 The literature on the financial market imperfections has offered several theories basing on the asymmetry

of information between insiders and outsiders that can explain this assumption. We cover these problems later

in the thesis.


R&D expenditures, that is:

Z∗j,kj= Zj,kj

· (1 + fj),

where Zj,kjequals the effective expenditures on the research process and fj is the parameter

of the external funding dependence of sector j. The higher fj , the more resources mustbe devoted to keep the effective expenses constant across sectors. Costs associated withrelaying on external funding are assumed to be proportional to the total expenditures onR&D. No assumptions concerning the functional form of fj are necessary. More specifically,it is sufficient to state that some time-varying industry-specific amount of resources must bedevoted to cover external funds costs. We further postulate that financial system developmentwill reduce the wedge between the costs of internal and external funds thus fostering thegrowth rate of the economy.

For brevity, the model, as in Suda (2004) is explicitly developed in Appendix A. The finalresult of derivations produces the equilibrium economy growth rate as given by the followingequation2:

γ =(qα/(1−α) − 1) ·

((1

ζ )HA1/(1−α)(

1−αα

)α2/(1−α)

∑Nj=1 ψ(j)(1 + fj)− ρ

)1 + θ · (qα/(1−α) − 1)

This equation indicates that changes in fj affect the GDP growth rate. It implies that costsassociated with external financing negatively influence the rate of economy growth. It isalso interesting to note that the model developed in this section should be considered as anextension of standard ”creative destruction” approach. This extension implicitly introducesfinancial markets and allows us to explain the relationship between finance and growth in avery simple setting.

The model also predicts that industries that depend heavily on external finance growdisproportionately faster in the countries with better financial system. In terms of model’sparameters, the lower value for fj results in higher effective expenditures, Zj , and greaterprobability of success in innovation activity. Therefore, the model indicates that the compo-sition of industrial structure is influenced by the development of financial system.

Finally, but most importantly, the bottom line conclusion from the model helps to specifythe efficiency criterion for the financial institutions. Namely, efficiency follows from alleviatingthe wedge between costs the internal and external of capital. Whether a financial institution -be it a bank, an investment fund or a stock exchange - is profitable or not is solely a problemof its owners. Economically speaking, should they be unable to provide for the operationcosts, they might be considered obsolete on one hand or public goods on the other. From theeconomy point of view, their major role is to minimise the effective cost of external capitalto the borrowers. Thus, we arrive at an approach to analysing the efficiency of the financialinstitutions. A simple policy implication one may easily derive even at this stage is thatthe institutional design should incorporate incentives inducing profit maximising agents toactually realise these aims. This is to say, verifying whether they are in their optimisation

2 Where δ = [qα/(1−α) − 1] and β = ( 1ζ) ·HA1/(1−α)

(1−α

α

)· α2/(1−α).


problem and placing them there by the force of law and its execution, should the economyinterest and the private benefit diverge in any domain.

2.2 The empirics of finance and growth

Despite the methodological obstacles and interpretational difficulties, a large body ofrecent research work suggests that well-functioning financial institutions promote economicgrowth. These conclusions emerge from cross-country comparisons, firm-level studies, time-series research and econometric investigations that use panel techniques.

In their historical research, North (1990), Levine (2002), Neal (1990), and Rousseau andSylla (2003) have all concluded that the regions - be they cities, countries, or states — whohave developed the relatively more sophisticated and better functioning financial systemswere the ones that were the subsequent leaders in economic development of their times. Anintegrated picture of these findings suggests that in the absence of a financial system that canprovide the means for transforming technical innovation into broad enough implementation,technological progress will not have a statistically significant and - in real terms - substantialimpact on the economic development and growth of the economy.

More specifically, the model as developed in the previous section has been tested by Suda(2004) with the results satisfying both the statistical and economic standards. Furthermore,although it is very difficult to measure whether a country’s financial system is comparativelyadept at reducing information acquisition costs firm level studies provide insights into the roleplayed by financial intermediaries in easing information asymmetries (Schiantarelli, 1995).

However, one might wonder whether financial structure is something characteristic to acountry, or does it change as countries develop. Goldsmith (1969) pioneered the cross-countrywork in this area. He traced the relationship between the mix of financial intermediariesand economic development for 35 countries over the period 1860-1963. The World Bank(1989) and Demirguc-Kunt and Levine (1996) extended Goldsmith’s work by examining theassociation between the mix of financial intermediaries, markets, and economic developmentfor approximately 50 countries over the period 1970-1993. These works find that financialstructure differs importantly across countries and changes as countries develop economically.However, the causality remains undetermined.

Four basic findings emerge from these studies. As countries get richer over time or asone shifts from poor to richer countries, financial intermediaries get larger as measured bythe total assets or liabilities of financial intermediaries relative to GDP. Secondly, banksgrow relative to the central bank in allocating credit. Thirdly, as far as non-banks - such asinsurance companies, investment banks, finance companies, and private pension funds - areconcerned, they also grow in importance with development. Finally, stock markets becomelarger (as measured by market capitalization relative to GDP) and more liquid (as measuredby trading relative to GDP, market capitalization, and stock price variability). While these”patterns” pose a challenge to financial theorists, they must be treated cautiously because thedata suffer from numerous problems3. Thus, while there is a general trend involving financial

3 For example, it is difficult to distinguish private from public banks and development banks from commercial


structure and the level of GDP per capita, there are important exceptions and differenceswithin income groups.

Another strand of research went in the direction of efficient savings mobilisation. Theresults in Bencivenga and Smith (1991) also rely upon the predictions of ”new” growth theo-rym but focus their analysis of the financial system on the hypothesis that ”savings matter”for growth.4 The role of financial intermediaries with special attention to banks is taken fromthe Diamond-Dybvig (1983) model. Banks, by providing liquidity to the depositors, increasethe savings and thus foster economic growth. Moreover, the very appearance of banks isdetermined in the model.

Greenwood and Smith (1997) in turn present two models with endogenous market for-mation. In the first model, they extend Bencivenga and Smith (1991) model and analyse therole of banks and stock market in allocation of funds to the highest value use in the economicsystem. In their model banks play the role of providing liquidity to consumers, thus raisingtheir propensity to save and eventually increasing the savings rate. In the second model,they investigate the link between increasing specialisation, market development, and growth.They postulate that growth should lead to an increase in market activity, which in turn maystimulate further economic growth.

Also, there exists considerable debate, with sparse evidence and insufficient theory, aboutthe relationship between financial structure and economic growth. The classic controversyinvolves the comparison between Germany and the United Kingdom. Starting early in thiscentury, economists argued that differences in the financial structure of Germany and Englandhelp to explain Germany’s more rapid economic growth rate during the latter half of the 19thcentury and the first decade of the 20th century (Gerschenkron, 1962). The premise is asfollows. Germany’s bank-based financial system, where banks have close ties to industry,reduces the costs of acquiring information about firms. This has the positive effect not onlyin terms of reducing the research, evaluation and monitoring costs but also facilitates savingsmobilisation. The British financial system, being more securities oriented, is deprived of thischaracteristic.

Indeed, quite a bit of evidence suggests that German banks were more closely tied to indus-try than British bankers. Unlike England, nearly all German bankers started as merchants.The evolution from entrepreneur to banker may explain the comparatively close bonds be-tween bankers and industrialists. For example, German bankers frequently ”... mapped outa firm’s paths of growth, conceived farsighted plans, decided on major technological innova-tions, and arranged for mergers and capital increases” (Gerschenkron 1968, p. 137). PrivateGerman bankers also organized and promoted an impressive array of major manufacturing

banks in many countries. Similarly, the definition of a bank and of a non-bank are not always consistent

across countries. Furthermore, there is nothing causal about these relationships. These patterns alone do

not suggest that poor countries can accelerate their growth rates by changing the structure of their financial

systems. Finally, many differences exist across countries at similar stages of economic development (World

Bank, 1989).4 It seems useful to remind that in the neo-classical approach nothing but the population growth rate and

the rate of exogenous technological progress determine the long-run growth. The savings rate, among others,

has no impact on the rate of growth once an economy stabilises on its steady-state path.


companies during the mid-19th century (Tilly 1967, p. 179). Besides this entrepreneurialrole, some evidence suggests that German bankers tended to be more committed to the long-term funding of their clients than English bankers. Short term credits could be transformedinto longer-term securities more easily in Germany (Tilly 1967, p. 178-181)5.

While bank-industry relationships may have been closer in Germany, this does not neces-sarily imply that the German financial system was better at risk management, providing liq-uidity, or facilitating transaction. Furthermore, economists disagree over whether the growthdifferential between the U.K. and Germans was really very large6. Last, but not least, thediscrepancy between the development path of the UK and Germany followed largely fromthe different sectoral distribution. The so-called ”winning sector” argument (light industryin the first half of the 19th century versus heavy industry in the 20th century) seems to havemore explaining power than the financial system differences, not to mention being far moreintuitive.

The debate concerning bank-based versus market-based systems eventually expanded toinclude comparisons with the United States. The main hypothesis suggests that the Germanbank-based system may reduce information asymmetries and thereby allow banks to allocatecapital more efficiently and to exert corporate control more effectively. In contrast, the UnitedStates’ securities market-based financial system may offer advantages in terms of boostingrisk sharing opportunities (Allen and Gale, 1994). While this approach highlights the relevantissues, just as in the UK-Germany case it is not sufficient to determine the ”better” functionalsolution for the time being. Further research in this direction seems thus indispensable.

Many of the arguments involving bank-based versus securities market-based financialsystems have been used to compare Japan and the United States. For example, researchsuggests that Japanese bankers are more closely tied to industrial clients than U.S. bankers.This closer connection may mitigate information asymmetries (Hoshi et al., 1990), which mayfoster better investment and faster growth. Thus, the structure of the Japanese financialsystem is sometimes viewed as superior to the financial structure of the United States andan important factor in Japan’s faster growth rate over the last four decades. Interestingly,however, the recent banking problems and slower growth in Japan have led some to arguethat the absence of a credible takeover threat - through the efficient stock markets - hasimpeded proper corporate governance and competitiveness.

There are severe analytical problems with linking financial structure to economic per-formance. First, existing research on financial structure does not quantify the structureof financial systems or how well different financial systems function overall. For example,German bankers may have been more closely connected to industrialists than their Britishcounterparts, but less capable at providing liquidity and facilitating transactions. Similarly,while Japanese keiretsu may lower information acquisition costs between banks and firms,

5 One can easily observe the similarities between the German system as described above and the modern

Japanese keiretsu. This issue is re-addressed later.6 Although German manufacturing production grew noticeably faster than Britain’s in the six decades

before World War I, Germany’s overall per capita GNP growth rate was 1.55 while the U.K.’s was 1.35 over

the period 1850 to 1913 (Goldsmith 1969, p. 406-407).


this does not necessarily imply that the Japanese financial system provides greater risk shar-ing mechanisms or more accurately spot promising new lines of business. Furthermore, whileJapan is sometimes viewed as a bank-based system, it has one of the best developed stockmarkets in the world7! Thus, the lack of quantitative measures of financial structure and thefunctioning of financial systems make it difficult to compare financial structures.

Second, given the array of factors influencing growth in Germany, Japan, the UnitedKingdom, and the United States, it is analytically difficult - and perhaps reckless - to attributedifferences in growth rates to differences in the financial sector. Moreover, over the post WorldWar I period, the devastated Axis powers may simply have been converging to the incomelevels of the United States, such that observed growth rate differentials have little to do withfinancial structure. Thus, before linking financial structure with economic growth, researchersneed to control for other factors influencing long-run growth.

A third factor that complicates the analysis of financial structure and economic growth ismore fundamental. The current debate focuses on bank-based systems versus market-basedsystems. Some aggregate and firm level evidence, however, suggest that this dichotomy isinappropriate. The data indicate that both stock market liquidity and the level of bankingdevelopment correlate with the economic growth over subsequent decades (Levine and Zer-vos, 1996). Thus, it is not banks or stock markets - bank AND stock market developmentindicators both predict economic growth. It is the presence and the quality rather than thefunctional form.

A fourth factor involves the problem of causality. It may simply be that financial markets -be it bank-based or securities-based - evolve faster in faster growing countries simply preciselybecause of their faster growth. There are business opportunities to realise and as rationalagents they try to keep up with the economy’s pace. To a certain extent we have no means toidentify the potential growth rates, i.e. the one that would pertain ceteris paribus should thefinancial markets be perfectly efficient in researching, evaluating and monitoring the projectsas well as mobilising the savings in a particular economy. Researchers have been observingthe countries with relatively similar income levels (US, Japan, Germany and the UK inearly research and comparative studies or approximately 50 most industrialised countriesin the latest studies). Furthermore, despite all cultural differences, all these countries havedeveloped within a fairly similar institutional setting with respect to the issues of trust,human and social capital as well as general institutional framework. Is there any economicallysignificant role to the financial institutions is a question that requires further elaboration.

Finally, there are important interactions between stock markets and banks during eco-nomic development that have not been the focus of bank-based versus market-based com-parisons. As noted, greater stock market liquidity is associated with faster rates of capitalformation. Nonetheless, new equity sales do not finance much of this new investment (Colinand Mayer, 1988), though probably important differences exist across countries. Most newcorporate investment is financed by retained earnings and debt. This raises an importantpuzzle: stock market liquidity is positively associated with investment, but equity sales do

7 Cfr. Demirguc-Kunt and Levine (1996).


not finance much of this investment. This observation is sustained on a firm-level - in rela-tively poor countries even, enhanced stock market liquidity actually tends to boost corporatedebt-equity ratios, as forcefully demonstrated by Demirguc-Kunt and Maksimovic (1996).However, for industrialized countries, debt-equity ratios fall as stock market liquidity rises,suggesting that stock market liquidity causes a particular type of crowding out (substitutionof debt finance). Furthermore, on the corporate finance level, there are different ”schools” ofcash-flow management - some of the most prominent suggesting that debt is always betterthan equity to companies and higher equity only gives an opportunity to raise debt evenfurther8.

In addition, as well as being unable to shed adequate light on the causal relation be-tween the development of the banking system and economic growth, the measures used formeasuring the role of banks in financial development have the further shortcoming that theyessentially concentrate on the role of banks in stimulating capital accumulation. However, asforcefully demonstrated by Easterly and Levine (2001) forcefully demonstrate, factor accumu-lation cannot explain the differences in income levels and the growth rates among countries.Thus, this line of defending the role of financial systems in determining growth is not apromising one. Furthermore, as some of researchers emphasise, the specific role of banks inthe economic system is not to mitigate and intermediate savings, but rather to certify thequality of borrowers, monetizing liabilities which otherwise would fail to find purchasers inthe markets9.

This description of banking activity has been offered in the past by prestigious academicslike Wicksell, Schumpeter, von Mises, Kalecki and Keynes, and leading bankers like AlbertHahn or Luigi Lugli. For example in 1920 the German banker Hahn wrote: ”The activity ofbanks consists ..., if one ignores the legal form and considers the economic significance of theprocess as decisive, in providing guarantees, in acting as the guarantors of borrowers. Theyfurnish, so to speak, borrowers with the general trust that they lack. On this view, therefore,they are none other than intermediaries of credit in the literal sense of the expression, orin other words the intermediaries of trust” (Hahn, 1920, after: Luchetti et. al., 2000)10.Some claim that banks are essential for economic development in that they are a crucialdevice for the selection of entrepreneurs and the allocation of (first) financial and (then) realresources (Diamond, 1984). This is where they are supposed to essentially differ from otherfinancial institutions. If this is true, it is to this that the indicators of the banking system’sdevelopment should refer.

All these considerations point to the complexity of the relations between the functioningof all financial systems. Viewed from the perspective of corporate decisions and potential en-

8 Namely the MIT and Princeton business schools suggest this line of reasoning. For comparison, the main

line of the Harvard Business School cases in corporate finance stresses the benefits of resorting to equity.9 See: Minsky (1986), Moore (1988), Fama (1985), Stiglitz and Weiss (1988).

10 This assertion would most probably meet with general agreement today, with the difference that authors

working in the mainstream of the dominant theory (neoclassical or neo-Keynesian if more precise lables are

to be employed) would maintain that banks ”intermediate trust” by intermediating savings, while those who

adhere to more heterodox schools of thought (post-Keynesian, post-Kaleckian, neo-Schumpeterian) would

claim that banks perform their functions by creating money.


dogeneity problems, they call upon further research into this domain. A better understandingof the links between the stock markets, banks and corporate behavioural decision patternsseems vital for constructing a good endogenous model of development as well as providingrobust empirical results corroborating or underminig the potential link between the financialstructure and economic growth on the aggregate level.

To overcome these shortcomings, in this thesis we resort to the level of a particular”relation” or, to use the terminology suggested by Aoki (2001) domains of the game. We keepthe game as a unit of considerations, analysing the potential payoffs to the players (namely, anentrepreneur and a financial institution) aiming at asserting whether the incentive structureforces the socially desired outcomes and the Nash equilibria in imperfect world to overlap.

3. THE MARKET STRUCTURE IN THE BANKING SECTOR AND THE

INVESTMENT PROCESSES

When thinking about the financial system as such, banks constitute the first and mostprincipal pillar. As Freixas and Rochet (2002) state it, ”a bank is an institution whose currentoperations consist of granting the loans and receiving deposits from the public” (p.1), withthe key words: institution, current and public. The common knowledge states that ’they liveon lending money rather than on borrowing it’, whilst a large body of literature has beendeveloped to explain the relation between the firm and the bank.

Primarily, banks seem to positively influence the shareholders’ value. James (1987)demonstrates a positive abnormal price reaction to an announcement of bank credit an-nouncement as compared to public straight debt of privately placed debt1. Furthermore, asdemonstrated by Billett, Flannery and Garfinkel (1995) loans from high quality banks aremore likely to be viewed as positive news than loans from low quality lenders.

Similarly, there is evidence in support of the significant role of the banks in the earlystage of company’s development. Theoretically speaking, young companies are especiallyexposed to adverse selection problems, while moral hazard temptations are notably stronger.These two effects must inflate the rate of interest on credit but close relationship may beused as a mean of overcoming the problem. One one hand, if a bank cannot use the historyas an indication of the future, the multiplicity of information about the client might be usedinstead. On the other, with time and experience the rate of interest charged to the entreprisesshould decrease.

Surprisingly, this intuition finds little or no support in data. The implicit assumptionthat the lower cost of capital is passed through to the borrowers in the lower interest rateis contradicted by the observation of small and medium size entreprises both in the US(Petersen and Rajan, 1994) and in Europe (Degryse and Van Cayseele, 1998). Firms tend toconcentrate their borrowing from one source but concentration decreases as firm size increases- the smallest companies are close to one lender on average, while medium size reach the levelof about three.

However, not only the length of the relation seems to have no positive impact on the costof credit to the borrower, but also the knowledge the bank has about the potential borrower(the monthly turnover, the number of accounts, the frequency of payments by the clients)seems to be unimportant in the credit assignment decision process. Furthermore, banks seemto detest competition from other banks: if firm lends from more than one bank the cost of

1 This issue was further refined by Lummer and McConnell (1989), who differentiated between first an-

nouncements and renewals of loans - they have found a positive abnormal effect only for the renewals of

loans.

3. The market structure in the banking sector and the investment processes 22

debt increases (the interest rate rate increases with multiplicity of relationships, or, qualitydecreases if they lend at more banks)2.

All these results seem to contradict the economic intuition, but if we consider financialmarkets to be informationally intensive ones, some facts may become a little bit more clear.Although, as suggested above, some effect may be observed on the availability of the credit,virtually zero or even negative impact on the interest rate might be a consequence of thesimple premise. Despite the fact that the relationship reduces the lender’s expected cost, atthe same time it increases the informational monopoly of the capital providing institution.This may be summarised by a following statement: if any bank that has no knowledge ofcompany A will charge it certain percentage on the loan, why should I significantly lower myoffer - company A will not go anywhere else anyway. Or will it? Thus, the market structureas well as the degree of competition seem to play a crucial role in providing potential solutionsto this problem.

As far as the role of banks in the economy is concerned, a major problem is identifiedin the very process of measuring the development of the banking system3. These variablesare usually of two types. The first refers to the presence and diffusion of the banking systemshare of liquid liabilities in GDP belongs to the most frequently used4 The second group ofvariables instead measures the amount of financing intermediated by banks. Among thesevariables are the ratio between domestic credit and GDP5, the share of credit granted to theprivate sector, or the credit granted to the private sector in ratio to GDP6.

All these measures give rise to interpretative problems and they are only partially able tocapture the role performed by banks in economic development. Firstly, there is the problemof causality. The growth of the banking system and the amount of credit disbursed areclosely influenced by the level of economic development. The wide presence of banks and theimportance of bank lending in areas which grow more rapidly than others may be indicativeof a reverse causal relation between finance and economic growth. After all, banks grantcredit on demand by firms.

In order to counter this classic objection, most researchers resort to the classic argumentof post hoc ergo propter hoc: the presence of banks and the amount of credit granted aregood predictors of growth in subsequent years and can therefore be presumed to be one of itscauses. The weaknesses of this argument are equally well known. Firstly, variables may havebeen omitted which explain both financial development and growth. Secondly, and especially,the capacity of the development of the banking system to predict growth may be due to thesimple fact that production must be financed in advance. Consequently, in granting credit,the banks are only making correct predictions about the future growth of the real economy.

2 Some evidence, though weak, has been found in the support of the thesis, that better knowledge about the

borrower can positively influence the availability of the credit. Although statistically significant, the results

were of low economic importance. See: Petersen and Rajan (1987).3 Cfr. Lucchetti, Papi and Zazzaro (2000).4 See: Gertler and Rose (1994), King and Levine (1993b, 1993c). In the analysis on the regional level, the

proportion of bank branches to the resident population is also often applied (Ferri and Mattesini, 1997).5 See: Rajan and Zingales (1998)6 See: King and Levine (1993b, 1993c), Levine (1998, 1999) and Beck, Levine and Loayza (1999).


In short, as Rajan and Zingales (1998, p. 560) have pointed out, “financial development maysimply be a leading indicator rather than a causal factor”.

In this part of the thesis we will focus on the impact of the markets structure on thebehaviour of the banks in Poland, but only concerning the investment credit. This choiceis dictated by the scope of the thesis and the crucial role the banking sector plays in thecontinental system in fostering investment and thus development. This problem is approachedbasing on an industrial organisation model which links the market structure and the time ofthe response to the change in the underlying fundamentals (i.e. the cost of capital to thebanks).

This chapter is organised as follows. We first present the landscape of the banking sectorin Poland which serves the purpose of assessing the degree of competition on this market.We demonstrate a time link between the changes in the market structure and the investmentcredit conditions. We then develop a theoretical model of the investment credit to apply itto the Polish data in section three. In the last part we attempt to provide certain policyimplications concerning the demonstrated market failure.

3.1 The competition in the Polish banking sector

As demonstrated by Demirguc-Kunt, Laeven and Levine (2003) on a cross-section ofcountries, concentration is positively and significantly linked with the net interest margins,even constrolling for bank specific factors. This finding is about the first to empiricallyprove the link between the market structure and the pricing behaviour in the banking sec-tor. Previously, it has been widely believed that due to the speed of information spreadingfinancial markets, and thus financial institutions, are on average and in the long run perfectlycompetitive.

Although evidently high degree of concentration is not equivalent to lower level of compe-tition among the players, there seem to be reasonable grounds to use concentration measuresas indicators of competitiveness. Some researchers suggest, though, that concentration resultsfrom more efficient and faster growing banks taking over the less efficient and less rapidlyexpanding ones. Thus, the alternative hypothesis states that markets are more concentratedprecisely due to higher competition accompanied by constantly improving efficiency (Bergerand Hannan, 1989). This belief, however, excludes from considerations the option of collusionamong the biggest market players.

Corvosier and Gropp (2001) analysed the European countries, trying to trace the con-centration - retail interest rates nexus on the sample of 10 EU countries for which detaileddata about the credit and deposit offers by banks as well as market shares were available.They have found that higher concentration results in collusion, but not always and every-where (p. 24). Namely, moving from a moderately concentrated banking market to a highlyconcentrated one, margins on both deposits and credits were increased by approximately 0,1to 0,2 percentage points. However, in case of savings and time deposits the results werethe opposite. They suggest that differences in the switching costs as well as the availabilityof information might provide justification to these contradicting results. Nevertheless, the


collusion following from concentration seems robust across countries and across time.The availability of Polish data in this respect is incomparably poorer, making any similar

studies rather impossible. However, even making use of the available data reveals that theclaims of collusion in the banking sector in general are not deprived of grounds.

POROWNANIE ZMIENNOSCI W CZASIE HHI oraz spreadow - ZUPELNIE PRZY-PADKIEM WSPOLEGZYSTUJA.

RYSUNKI.

3.2 Model I - asymmetry of response to shock

The oil shocks in the 1970s and 1980s as well as the Persian Gulf war have channeledthe attention of the policymakers and economists to the response of retail gasoline pricesto the fluctuations in the world crude oil prices. Refineries and oil companies were accusedof speculating in the difficult times of war, while more radical politicians raised the ideasof rationing and central allocation of quantities and prices. Some observers pointed to thetime of the reaction suggesting that retailers increase the prices immediately but lower themwith a certain delay. This discussion seems very similar to what we observed in 2000 and2001 with respect to the Polish banking sector. Unfortunately, to our best knowledge littleempirical work was devoted to this problem.

In this thesis, we test for the asymmetry in the speed of retail price responses and findsupporting evidence. Such asymmetries, however, are as inconsistent with textbook monopolybehavior as they are with simple models of competition. Although such a pricing patterncould indicate market power at some level of the distribution chain, the connection is notimmediately apparent. Borenstein, Cameron and Gilbert (1992) provide a useful analyticaltool to approach this type of a problem.

Many analysts claim that increasing the cost of the investment credit immediately afterthe central bank’s decision to raise the interest rates is already a proof for uncompetitivestructure on this market. However, prices in competitive markets should reflect the opportu-nity cost of the inputs, and not the accounting cost of acquiring it, so this complaint is easyto dismiss. What is important to note is that the asymmetry in the response time (immediatein the case of increases and delayed in the case of decreases) provides a strong proof for theclaim of insufficient competition.

To estimate the rate at which prices adjust to changes in the underlying fundamentals, weassume a simple linear long run relationship between central bank interest rates and the cost ofcredit, R = φ0+φ1C. While we recognize that the adjustment is not instantaneous, we assumethat the adjustment function is time-invariant during our sample period and is independentof absolute magnitude of the change in the interest rates. Defining ∆Ct = Ct − Ct−1 and


∆Rt = Rt −Rt−1 the adjustment could be modeled as:

∆Rtt = β0∆Ct

∆Rtt+1 = β1∆Ct

∆Rtt+2 = β2∆Ct (3.1)

......

∆Rtt+n = βn∆Ct,

where the superscript on ∆R indicates that it is solely the change resulting from the periodt change and n is the number of periods it takes for retail prices to complete adjustment tothe period t change7.

Under these assumptions, the total change in retail prices in any period t will depend onthe price changes in the previous n periods.

∆Rt = ∆Rtt + ∆Rt−1

t + ∆Rt−2t + ...+ ∆Rt−n

t =n∑

i=0

βi∆Ct−i (3.2)

This equation, however, imposes symmetric responses in the case of decreases and in the caseof increases. Recognising that the process can have a different nature these two cases is thecrucial part of the model. We thus refine it to:

∆Rtt = β0∆Ct

∆Rtt+1 = β1∆Ct

......

∆Rtt+n = βn∆Ct,

if ∆Ct > 0 (3.3)

∆Rtt = γ0∆Ct

∆Rtt+1 = γ1∆Ct

......

∆Rtt+n = γn∆Ct,

if ∆Ct ≤ 0

Defining ∆C+t = max{∆Ct; 0} and ∆C−t = min{∆Ct; 0} the adjustment of retail credit

prices may be rewritten as

∆Rt =∑

(βt∆C+ + γt∆C−) (3.4)

This specification adapts Borenstein, Cameron and Gilbert (1992) to the banking sector.The main innovation of this thesis is that we distinguish among the products offered by thebanks. Evidently, a part from the investment credit, banks offer also the liquidity credits to

7 Although here we ignore the systematic drift in the interest rates in Poland, we do control for it in the

econometric specification. Here, it serves the purpose of simplifying the presentation.


the firms. Perhaps more importantly, however, they also offer products to consumers - theconsumption credit and the mortgage. Consequently, one might ask whether the potentialasymmetry in the response to the changes in monetary policy are similar among the products.To be able to account for this, we consider different bank products and attempt to distinguishboth the time of the response and as well as its strength.

3.2.1 Econometric obstacles

A number of econometric issues must be addressed before proceeding with estimation ofan equation similar to the above. The issues that we discuss here arise in the estimation ofall of the downstream price transmissions. First, the restrictions imposed on the lag responsestructure. The additive lag structure we use places few constraints on the adjustment path,allowing it even to be non-monotonic. It also allows a certain intertemporal independencethat may be non-standard. For instance, if the interest rate increases by 25 base points inmonth t and decreases by the same amount in month t+ 1, our model would not necessarilycause the direction of adjustment to reverse when the interest rate does. The retail pricecould continue to rise in week t + 18. This contrasts with a standard partial adjustmentmodel. If the long run equilibrium relationship is assumed to be R = φo + φ1C we couldestimate a partial adjustment model as:

Rt −Rt−1 = β(φ0 + φ1C −Rt−1). (3.5)

This however is not applicable in this case.Bacon (1991) tests for asymmetry in adjustment rates by including a quadratic term in

the adjustment process:

Rt −Rt−1 = β1(φ0 + φ1C −Rt−1) + β2(φ0 + φ1C −Rt−1)2, (3.6)

so that the test of β2 = 0 is the test of whether adjustment to increases and decreases occurequally quickly. The partial adjustment model, however, imposes equal proportional adjust-ments towards the new equilibrium in all periods after a shock to crude oil prices, which is aserious constraint. Furthermore, Bacon’s method for diagnosing asymmetry with a quadraticterm imposes a structure on the asymmetry, implying that the asymmetry becomes propor-tionally larger as the difference between the current retail price and the long-run equilibriumprice increases.

Secondly, one might consider the long-run relationship between the interest rates andthe credit prices. The principle advantage of the partial adjustment model over the lagadjustment model presented above is that equation (3.4) takes no account of the long runrelationship between the prices of the upstream and downstream goods, and the tendencyto revert towards that relationship. To address this, we estimate 3.4 as an error correctionmodel (ECM). The error correction term is the one-period lagged residual from the regressionRt = φ0 + φ1Ct + φ2MOt, where MOt is a time trend. The monthly regression is then:

Rt −Rt−1 = θ0 +n∑

i=0

(βi∆C+t−i + γi∆C−t−i) + θ1(Rt−1 − φ0 − φ1Ct−1 − φ2MOt) (3.7)

8 This would occur if βt > γt−1.


The constant term included accounts for the fact that the margins may have systematicallychanged during our sample period, whether due to inflation or other factors. Evidently, thisspecification is not a canonic form error correction model due to the

∑ni=0 (βi∆C+

t−i + γi∆C−t−i)term on the right hand side. However, not only is this the common practice but also theasymptotic properties of the estimates are unlikely to be affected by this deviation.

The drawback of applying this methodology is that the results obtained are difficult tobe interpreted in the economic terms. It is not possible to state by how much is the costof credit inflated as a byproduct of the uncompetitive market structure. To partly alleviatethis problem we compare the investment and the consumption credit with the underlyingreasoning as follows. Should the response pattern in these two cases be significantly different,one might conclude that banks (as firms) have certain preferences among the borrowers,effectively rationing one group at the expense of the other. Such an observation is obviouslypotentially interesting from the policy implications point of view. In addition, the long runequilibrium interpretation imposed by the ECM specification seems particularly useful, asmay be interpreted as an identification of the real underlying fundamentals.

The last potential econometric obstacle to be considered here is the problem of endogene-ity. Does the credit policy of the commercial banks influence the decisions of the monetaryauthority? To the extent the analysis of the central bank official decision criteria is concerned,this does not seem to be the case. However, one may not exclude this option over the wholesample due to the young age of this institutional design. Thus, endogeneity seems to intro-duce potentially serious problems from the technical point of view. Due to the specificationof the problem, using lagged variables as instruments may not be considered here.

3.2.2 Data and results

Data used in this section are taken from the Polish Central Bank and cover the period ofDecember 1996 to March 2005. In this period one can hardly talk about exploding inflationrates or other important sources of credit risk justifying the anti-lending approach of thebanks. Neither can one talk about poor performance of the economy, which could explainthe preference of consumption credit over the investment.

In the analysis we employed monthly observations on reported investment credit interestrates for loans of differentiated durations (one to five and more years). For each of theseseries we constructed a model as described by equation (3.7), where the cost C was proxiedby the central bank’s interest rate9. This measure, although cannot be identified directlywith the cost of capital to the banks may be a good indication of changes in these costs. Toavoid the potential criticism of inappropriate measure, we also performed the analysis whereWIBOR and the cost of deposit were used as proxies for the cost C. As will be presented,the conclusions remain essentially unaffected by the choice of proxy.

One could suggest that a direct measure of the cost of capital should be used in themodel. However, for Polish banks very little information is available in this respect. Despitesome efforts to apply the CAPM model to the banking sector (Kochaniak 2003), any analysis

9 The observations were corrected for the seasonal effects.


are mere approximations based on relatively low quality data - approximations which areburdened with the limitations following from a particular theory or econometric methodologyemployed. More importantly, equity cost or required risk of return (the measures that wereobtained so far in the literature) do not reflect the cost of mobilising deposits but the efficiencyof a bank to make profits on its current operations, which is not of any interest to us. Thus, wefollowed the initial approach of using one of the three: central bank’s interest rate, WIBORand deposit rate. All these were corrected for the obligatory reserve ratio.

Summarising the econometric considerations, equation Rt = φ0 + φ1Ct + φ2MOt reflectsthe long run equilibrium, while the ECM (3.7) describes the adjustment dynamics. Theconstant term in this equation tells what part of the deviation is corrected for within amonth. Further, the error correction term suggests whether one can effectively talk about acointegration in this respect, while β’s and γ’s assess the symmetry of the response to thechange in the underlying fundamentals. The results are presented in the table below.

DWIE TABELE Z WYNIKAMIW DLUGIM OKRESIE RELACJA JEST ZGODNA Z OCZEKIWANIAMI, ALE ECM

POKAZUJE, ZE NIE MA PRAKTYCZNIE ZADNEJ REAKCJI STOP KREDYTOWYCHNA ZMIANY KOSZTOW KAPITALU. JEZELI COKOLWIEK, TO TYLKO WZROSTY,A Z CALA PEWNOSCIA BRAK WPLYWU SPADKOW! CO NALEZALO UDOWODNIC.

3.3 Model II - credit-deposit nexus

The latter model employed in this section concerns the credit-deposit relationship. Asimplied by common sense, banks should maintain a close link between their deposits and theallocated credits. The lack of this link is not necessarily a proof for the efficient interbankingmarket, as asymmetry in one bank would immediately induce the reverse disequilibrium in theother. Thus, on the aggregate level of the entire sector, the strong link should be sustained.However, the lack of it may be interpreted in terms of insufficient competition, as any bankcan only afford the discrepancy knowing that either its structure is being mimicked by otheror no other bank will threaten him on the ”short” position. Therefore, should we find thatchanges in deposits and changes in credits cointegrate poorly or not at all, we would findsome indication of effective collusion between the Polish banks.

Monti-Klein model of monopolistic bank. Prediction: jesli bank jest konkurencyjny,udzielone kredyty i depozyty powinny sie kointegrowac. Brak tej kointegracji rownowaznyjest z niewystarczajacym poziomiem konkurencji.

3.3.1 Data and results

Dane panelowe dla polskich bankow 1996-2004 (miesieczne). Statystycznie kointegracjajest, ale na rysunku wyraznie widac, ze ciagna ten wynik cztery obserwacje - zasadniczochmura kropek nie wskazujaca na jakakolwiek kointegracje miedzy udzielonymi kredytamiaa pozyskanymi depozytami.


3.4 Policy implications

As has been recently suggested by the analysts and the newspapers, Polish banks havecommenced the process of mobilising the savings. A number of largest commercial bankshave initiated the advertisement campaigns concerning the new, higher deposit rates. Thisincreased activity, as emphasised by the analysts, is caused by the expected credit boom, asin terms of average borrowing Poland lags behind its European partners, while similar effectswere observed in Portugal, Spain and Ireland. However, this only concerns the consumercredit and the mortgage. None of the banks launches a campaign focused on entreprises andinterbanking-credit spreads pertain at an astonishingly high level of 7%. Evidently, fromsimple macroeconomic identities, more consumption credit implies more consumption, whichfurther implies higher output. However, funds available for the R&D and investment in newtechnologies will only be generated if higher corporate earnings will be accompanied by moreintensive credit activity from the side of the banking sector.

On the other hand, however, the ’common belief’ seems to suggest that credit to theentrepreneurs is so low because firms do not want to resort to this solution. They supposedlyprefer self-financing. Taking into consideration very low level of capital accumulation andlittle tradition of entrepreneurship in Poland this explanation seems highly unlikely. Never-theless, it is continuously repeated by the banking sector representatives10.

The two models presented in this chapter suggest that the collusion in the banking sectorat best diverts the funds from the entrepreneurs to the consumers making the investmentprocess far more challenging. This indirect credit rationing is a consequence of the incentivestructure. Firstly, the high supply of government, a part from causing a traditional crowdingout effect, causes the returns on investment to be sufficiently high from the foreign own-ers point of view. Among the efficiency indicators applied by the managers are the ROIs.Having government bonds at disposal banks are able to demonstrate relatively high capitalproductivity without taking too much risks. At the same time they also evade the problemof searching, evaluating and monitoring the business opportunities - this costly activity canbe avoided if little or no investment credits are allocated.

On the first glance there seems to be a variety of options available to the economy regula-tor. However, a more profound analysis shows that in reality only very radical solutions canprove efficient. One could consider for instance differentiated rates of reserves for consumptionand investment credit. Since there is little or no effect of the obligatory reserves regulationhowever, there seems to be little role for the central bank in this respect. Secondly, one couldopen the market to the new entrants thus strengthening competition11. Unfortunately, toactually address the problems indicated earlier, this new player (or the threat) should befocused on investment banking, while the Polish banking system is generally characterised byunspecialised banks, who provide a variety of products to final consumers. The new playerif anything, is likely to conform with these norms, and expecting it to act against its own

10 Among others, Jan Krzysztof Bielecki, former prime minister, currently the head of Board in the biggest

Polish bank, PKO S.A. See: Polityka, Dec. 9-16, 200411 A credible threat of doing so should have the same effect


interest (chose the difficult path instead of the easy one) tends to be a wishful thinking typeof regulation.

4. THE OPEN PENSION FUNDS AS CAPITAL PROVIDERS

The discussion on whether to create nation-wide social security systems or whether stateshould in any way intervene in the inter-temporal allocation of resources is a long and engagingone. Starting from Milton Friedman’s Capitalism and Freedom and Nozick-alike libertariansthrough neoliberals, social democrats and socialists, there are many stand points on this issueand each is equally worthy in one’s value system. This problem is probably particularly tensein transition countries who are in the process of moving away from the distorted form ofwelfare state. Where they stop on the way depends on many social, political and economicfactors establishing a very complex problem. In this thesis, however, we intend to touchupon a small constituent of this issue. We analyse the Polish social security system reform of1999 trying to verify whether the realisation fulfils its initial aim of deepening the financialmarkets and efficiently directing the economy savings into productive investment. Shiftingretirement savings from the Social Security Fund to private and independent pension fundswas believed to trigger the process of deepening the Polish financial markets (van Giunrken,1999).

There were three fundamental prerequisites for the reform in Poland. On one hand, as inmost European economies, the state pay-as-you-go (PAYG) system seized being capable ofserving future generations for demographic reasons. On the other hand, the national savingrate was believed to be too low. In addition, reform was expected to ameliorate the problem ofcapital market shallowness and increase the efficiency as well as the transparency of nationalsavings management.

The analysis of the nature of the reform should comprise two major constituents: the studyof the social security as well as the efficiency scrutiny into the financial market solutions.However, due to the institutional design of the reform, as specified below, essentially thesocial security component can barely be addressed. For this reason, Polish experience mayand should be perceived only as an attempt to answer the fundamental question of how todesign a financial services market capable to serve as a private pillar of the social securitysystems.

The major hypothesis we try to support is that from the users’ point of view the currentsystem is in its idea equivalent to the obligatory bank savings, although significantly moreexpensive. Moreover, the system does not facilitate raising the economy’s savings rate, as itsimpact on consumption patterns is rather negligible. Lastly, albeit via market mechanism,majority of resources is allocated to state bonds helping to finance budget deficit. Thus, sinceproper competition mechanisms are not built into the system, funds are effectively divertedfrom entrepreneurs. Hence, the observed effect of the reform on the capital formation andallocation is minuscule.

4. The Open Pension Funds as Capital Providers 32

4.1 The architecture of the new social security system

In 1999 government introduced four general reforms: health care, education, administra-tion and social security system. All of them required creating a new institutional architectureas well as introducing new rules concerning contributions and benefits. Five principles can beidentified as having guided the changes in the case of social security (Gomo lka and Styczen,2000). These have laid the foundation for the architecture of the new system.

• The diversification principle: the necessity of enhancing the security and efficiencyof the pension system should be obtained by diversifying the system between state andprivate funds as well as between obligatory and voluntary contributions. The system ofretirement pensions consists of three pillars: (I) state PAYG, (II) private open pensionfunds and (III) other pension funds. Pillars II and III are private, while only pillars Iand II receive obligatory contributions (see: scheme below).

• The distribution principle: maintain the PAYG rule in the public part of the newsystem, while making it less redistributive and more transparent in order to immune thesystem to temporary political pressures. Obligatory contributions remain proportionalto earnings (a form of a payroll tax, 19.52% of gross wage), but are also subject to acut-off point equal to 2.5 times the average wage. The employer is obliged to contributethe same amount as the employee and the payment is made upfront each month. Theestablishment of personal accounts to which all the obligatory contributions of partic-ipants in the new system are assigned is the key innovation. The collected personalcontributions, if not paid out in the form of pensions, may be inherited.

• The capital-funding principle: make it capital funded as well as induce adequateregulation of the private part. The private pension funds which constitute pillar IIreceive a specified part of all the obligatory contributions (7.3% of gross wage, theremainder is left in the PAYG state pillar I). The key issue of ensuring efficiency ofthe pension funds is recognised in two ways: contributors are freely allowed to changethe fund (twice a life without any costs) and a floor was imposed on the fund rates ofreturn (at first biannual, triennial after 2004 amendment).

• The savings principle: savings measures introduced must cover the cost of reform.The reform clearly implies the deterioration of the public finance in the initial phase,as part of the obligatory contributions is diverted to pillar II. To contain this deteriora-tion, supplementary reforms were introduced, namely: tighter criteria for handicappedbenefits, raising the effective retirement age, reducing the scope for early retirementand reducing the growth rate of benefits.

• The gradual phasing-in principle: the phasing-in of the new system must be spreadover a prolonged period of time, and should not involve people near retirement. Agradual phasing-in of the reform is intended, above all, to protect the rights of olderworkers, as they remain in the old system. On the other hand this helps to limit thecost of reform implementation.


Graph: the new architecture of the pension system

As the authors of the reform admit, the necessary process of shifting away from theinefficient and unsustainable PAYG system was prolonged over at least 20 years. In addition,even after this period, PAYG remains one of the pension sources, providing 30% of the benefit.

According to the rules outlined above, each citizen born after 1969 is obliged to participatein both I and II pillars (the choice of II pillar is left to the worker). Citizens born between1949 and 1969 are free to choose, whether they join the II pillar at all or remain in thePAYG system solely (if not, their pension is calculated according to new, more transparentprinciples). Finally, citizens born before 1949 do not have the choice, as they are obligatorymembers of the PAYG system. The government is the lender of last resort for the secondpillar, which effectively neglects the risk of loosing savings in result of a bankruptcy.

Everybody, regardless of age, is entitled to join any pension fund within the scheme ofthe III pillar. However, contribution is limited to 7% of wage and is not tax deductible. Inaddition, the solution of Individual Pension Accounts was introduced in 2004, allowing todeposit voluntary contributions in banks and financial institutions up to the level of tax oninterest redemption (the equivalent of approximately 800-850 euro on the annual basis).


4.2 The systemic shortcomings

Although the system seems fully consistent with the set of five principles, major lack oflegal regulation may be observed in the second pillar1. Primarily, according to the act of 1997,Open Pension Funds (OPFs) are not responsible for paying out the benefits - they collectthe contributions and invest them indirectly2 till the retirement on personalised accounts.Furthermore, no pension plans are offered by OPFs and fund participants have no knowledgeas to the algorithm according to which their benefits will be calculated. In other words, theyonly accumulate the contributions, while in principle separate institutions are to manage theprocesses of funds investment and were to distribute benefits.

In addition, the regulation of 30% of benefit following from the PAYG system will be verydifficult to implement. Since the remaining 70% is provided by private pension institutions,depending on particular contributions and realised rate of returns on investment, the PAYGpart would have to be fully personalised imposing a intense logistical constraint on the PAYGinformation system. So far the system is hardly capable of addressing the contributionscorrectly with constantly pertaining high rate of mistakes (up to 12%)3. Furthermore, asstated above, there exists no scheme according to which the benefits are calculated by the2nd pillar institutions, which makes it impossible to predict the future expenses of the 1stpillar.

The first problem arises with the fact that the act of 1997 specifies that for the purposeof paying out benefits independent entities should be created. Open Pension Associations(OPAs) may be founded by the same financial group as OPFs, but this is not an exclusiverequirement. By definition people are free to chose their pension association. OPAs are toreceive accumulated contributions of each participant at the moment of retiring. The decisionof moving to retirement is irreversible as well as the choice of association. However, the act of1997 does not irrevocably specify that these are the OPAs who will take the role of benefitsdistributors. Neither does it provide legal grounds for their foundation. In addition, OPAs- if they are organised at all - will only offer the payment scheme when the beneficiary isalready dependent upon them, so the decision to pension off is rather dictated by the age andsocially acceptable pattern and not a strategic and financially-wise one. Consequently, whencompared to the previous state-controlled system, the situation is only slightly changed fromthe contributors-beneficiaries point of view.

Secondly, in the literary understanding of the pillar metaphor, one could state that 2ndpillar undoubtedly lacks its second leg. Till present, there is no legal act in force specifyingthe responsible for benefits allotment. So far, all claims occurring due to inheritance ordivorce were settled by court. Because one could only join 2nd pillar if one had at least 15

1 For a detailed analysis of the reform shortcomings, see: Mularczyk and Tyrowicz (2004, 2005)2 Formally, for this purpose Polish Pension Associations are founded and they are also free to offer 3rd pillar

savings. Practically, they are obligatorily run by the same financial group.3 Approximately one out of eight contributions is misallocated among the pension funds (the toll reported

after the public statement on a press conference by the representatives of the Superintendence of Pension

Funds, March 2003) The problems with the Social Security Fund information system are widely known to the

public opinion. Apart from logistical issues, there were several corruption claims involved.


years to retirement left, first considerable group of beneficiaries will only appear in 2009.Thirdly, Open Pension Funds are not de facto pension funds, participant and the OPF do

not agree on the paying scheme. Thus, there is no risk sharing mechanism in force, which isa crucial conclusion from the institutional point of view. One could also state, that OPFs arenot pension funds even de iure, since they were not required to employ actuarial specialistsand perform any actuarial calculations.

Fourth, even though the contributions are addressed to participants’ accounts, propertyrights are not well specified. The law is not strict on who actually owns the money from thecontribution: the participant, the open funds, the social fund or the state. This uncertaintyabout the property rights leaves the re-nationalisation alternative available, which violatesthe necessary condition of irrevocability, though such a decision would be evidently riskyfrom the political point of view.

Summarising, the main part of the reform remains incomplete, while both participantsand market players lack any kind of indication of the future shape of the system. In addition,some stakeholders favour the alternative of organising only one distributing identity, namelya nation wide pension association. There are neither defined contribution nor defined benefitplans and prospective beneficiaries are not provided with any calculation algorithm. Theyare, of course, tempted by the marketing visions of retirement in Bahamas but they are onaverage poorly informed about the absence of OPAs’ regulations4

4.3 The impact on the financial markets

With the wind of 1989 and the gradual development of the banking system, some changesoccurred. Despite initially strong impact of inflation, people eagerly opened accounts in new,commercial banks. The popularity of a vista accounts and short term deposits is believed tobe the key factor of success in the stage of forming financial markets as well as in the initialinvestment process (Jurkowski, 2001).

However, as stated above, Polish economy suffers from a capital gap, which necessitatesimplementing a strategy aiming at introducing capital accumulating institutions and creat-ing investment friendly environments. Forming Open Pension Funds was expected to playa significant role in implementing this strategy. Unfortunately, OPFs are rather inefficientin transforming the participants’ savings into investment capital. They avoid direct capitalinvolvement (via for instance capital funds) and are reluctant as stock exchange players, thelatter being partly justifiable by the shallowness of the stock market. More importantly,

4 One should also state that this situation contributes to the low popularity of the 3rd pillar products. For

six months of an intense marketing campaign 100 000 customers have opened an Individual Pension Account

(150 000 within a year). For the purposes of comparison: in one of the private commercial banks over a

quarter more than 160 000 clients opened an internet account, despite other comparable options already

available on the market. The reform is identified with merely shifting the obligation of benefit distribution

from negatively assessed Social Security Funds to private financial institutions - the awareness of the actual

mechanisms remains on the very low level. For individuals doubting the entire social security system another

product offered by essentially the same provider is equally unworthy. Similarly, if one is convinced about

secure and wealthy future, it is optimal not to engage into independent saving activity.


government and communal bonds continuously contribute to almost 70% of their assets (see:figure below). Furthermore, these bonds are rarely purchased on the secondary market real-ising rather the buy-and-hold strategy.

Graph: bonds in portfolios

In other words, whereas in the previous system subsidisation of the Social Security systemoccurred directly, after the reform there are at least two intermediaries in this process. Inthe pay-as-you-go system participants used to provide Social Security Fund with means tofinance current benefits. With the new system participants provide their contributions whichare then taken to the capital market and used as a mean to finance government budget deficitwhich partly follows from the current Social Security Fund obligations.

4.3.1 The strength of the link with the capital market

It is likely that the process of deepening the financial markets is much more time-consuming and the effects cannot be observed after only six years of the reform implementa-tion. Another method to verify the extent to which OPFs participate in the financial marketis to run a CAPM on their returns, treating the stock exchange index as a market indicator.A detailed study with respect to every fund based on daily data from May 1999 to June 2002can be found in Czerwinska (2003). The obtained β-coefficients depending on the fund andyear range from -0.014 to 0.1 (with the R2 of approximately 0.15 on average). It is importantto emphasise that the estimated risk free interest rate of these models ranges from 8.8% to18.5% for particular funds. Her conclusions point to risk averse profile as well as very lowdiversification of investment.

We decided to use panel data technique for the entire sample of OPFs. We grouped thedata by fund and obtained similar results on the period of August 2001 to August 2004. To


measure the risk free interest rate we applied the treasury bills yields. For the OPFs weused daily rates of return, although the biannual ones are specified by the regulator as thereference level. This was motivated by the nature of the market data set - obtaining biannualrates of return for treasury bills would require imposing an implicit and strong assumptionabout the unchanging conditions of public offerings. To measure the market risk we used thedifferential between the rate of return to the main Warsaw Stock Exchange index (WIG) andthe yield of the treasury bills. The results are presented below.

Table 4.1. Dependent variable: OPFs biannual daily rates of return

Independent variables Coefficients

Treasury bills (biannual daily yields) 0.24***WIG-TBills (biannual and daily) 0.13***Constant 0.18***

No. of observations 2801 (21 groups)R2 within 0.09R2 between 0.88χ2 statistic 6454.6***

Notes: Panel data random effects GLS estimation.

*** denotes statistical significance at 1% level. Fund dummies not reported.

The estimation is highly significant and the results are consistent with the study byCzerwinska (2003), despite different time span and data characteristics. The values of theestimators (significantly different from each other at 1% significance level) suggest that bothrisk free returns and stock market yields are useful indicators of the OPFs biannual rates ofreturn. Nevertheless, the estimated β-coefficient remains at a low 0.13 level, while a radicaldifference between both R2 obtained suggests that the explanatory power of the model israther low although OPFs tend to behave similarly. This last conclusion is particularlyworth stressing, as with such a variety of funds one would expect at least some differentiationin the investment strategies.

Summarising, CAPM model is useful for analysing the determinants of the OPFs returnsonly in statistical terms. Insofar as the economic interpretation is concerned, model haslittle explaining power whereas the high and significant value of the constant term suggeststhat OPFs returns are a result of rather passive investment strategy focused on safe, longrun opportunities. This is confirmed by the observation that the standard deviation of theOPFs returns over the analysed period was relatively low and significantly lower than theone for treasury bills. Hence, one might state that fund managers avoid active speculationeven on this market. The graph below demonstrates the average composition of the OPFsportfolios, where the first number denotes the share averaged over time, while the latterindicates the volatility ratio across funds. The extremely low volatility in the case of bondsand National Investment Fund shares again suggest that pension funds consistently chooseto mimic strategies of investing in the government securities.

In other words, although all OPFs investment occurs exclusively via the market mecha-nism, this group of investors is not particularly active on the markets - neither risk less nor


involving significant uncertainty. The results suggest also that this is a consistent patternacross the funds implying low incentives for investment strategies diversification.

Graph: average composition of portfolio

One needs to admit that in the previous system, current contributions were neither accu-mulated nor invested. If the PAYG system liabilities were below contemporaneous contribu-tions, the surplus was taken over by the budget. In the case of Social Security Fund liquidityproblems, the government covered the deficit. Currently, contributions constitute savingsand although they still finance the budget deficit, it occurs via the market mechanism. Thisevidently contributes to increasing the transparency of the system as well as deepening thefinancial markets. Unfortunately, it only concerns a very particular segment of this marketand happens at a considerable cost. The cost-benefit efficiency of this solution is analysed inthe next section.

4.3.2 The efficiency of the pension funds

As presented in the previous section, the reform has not achieved its initial aims. Onelacks the ground to state that it significantly affected the savings patterns, while it onlymarginally contributed to deepening the financial markets. Above all, with the current status,pension funds reform did not introduce the risk sharing mechanisms leaving the problemof benefit payouts essentially unaddressed. OPFs are not pension funds in the canonicalunderstanding of the word, as they only play a role of contribution collectors and savingsmanagers.

Despite all these shortcomings the new system involving Open Pension Funds might


provide benefits in terms of value-for-money analysis. To be able to verify this hypothesiswe need to specify a benchmark case. For simplicity we assume a slightly peculiar exampleof virtually no system - instead of introducing the institutional design of pension funds thesystem would impose an obligation to deposit monthly exactly the same contributions ona bank account without the right to retrieve these holdings till a certain age. We furtherassume that there is no cost associated to this saving strategy, while in return it brings theaverage long term commercial bank deposit rate.

The choice of the benchmark was dictated by the observation emphasised in the previoussections that OPFs are only responsible for collecting and managing the contributions butthey bear no actuarial risk. Thus, in purely economic terms, according to the act of 1997 apension fund is equivalent to specialised money box. The main advantage the new systemenjoys over other solutions in this respect concern the capital adequacy rules and investmentprudence. Thus, although a bank account would not be as secure without additional regu-lations and the change of financial services provider would be a much more complex issuein this setting, we treat that only as a benchmark. We do not claim it to be a comparablysuitable systemic solution, underlying only the fact that in predictable conditions the newsystem is not functionally very different from the chosen benchmark.

4.3.3 The costs of OPFs

Open Pension Funds are privately funded institutional investors as well as legal entitiesobliged by the act of 1997 to perform certain reporting activities. Consequently, they face twotypes of costs - those associated with organising and operating a financial institution (benefitscollection, funds management, etc.) and those following from regulator decision (essentially:reporting to the Pension Funds Superintendence as well as to the participants). They mayseek two sources to cover these costs: own capital and fees charged to the beneficiaries. Whilethe first option is rarely applied for obvious reasons, the latter is strongly regulated.

First, OPFs are entitled to a certain part of each first contribution. This rate is decreasingwith the participation in the fund. The regulator assumes that being a member of a fund fora longer period of time entitles the participant to a special treatment, which is reflected inthe fee charged. The details are given in the table below.

Table 4.2. The average management fee (% of each first contribution)

Year 1 2 3 4 5 6-10 11-15 16-20 21-25 26 27 28...

Fee 8,94 8,88 6,43 6,27 6,23 6,22 6,18 5,26 5,12 5,03 5,01 4,97Source: Own calculations based on the Superintendence of Pension Funds Quarterly Bulletins

Secondly, OPFs are also allowed to charge a management fee of approximately 0.02%to 0.05% of the accumulated capital, monthly at the end of each month. The size of thefee is inversely related to the size of the fund, which follows from an amendment of 2004.Previously it used to be a constant percentage of 0.05% monthly regardless of the number ofparticipants. The details of the current regulation are outlined in the table below.


Table 4.3. Structure of monthly management fees depending on OPFs size (as % of accu-mulated contributions collected)

Size (in bln PLN) 0-8000 8000-20.000 20.000-35.000 35.000-65.000 65.000-...

Max. monthly fee 0,0540 0,0504 0,0452 0,0371 0,0181Source: Superintendence of Pension Funds Quarterly Bulletins

While this new regulation allows the participants to internalise obvious returns to scalepreviously enjoyed exclusively by the fund, one can find little support to justify the verypresence of this fee. The efficiency argument is often raised here, namely that funds will thushave an incentive to actively manage the participants’ savings invested as they will have theirshare in the high returns earned for the members. However, funds are allowed to charge thisfee irrespectively of their performance, whilst similarly regardless of their performance withthe accumulation of members and their contributions this fee is in the perspective of 10-15years strictly increasing in time.

On the other hand there are two sides to every coin. The new fees - id est after theamendment of 2004 - are significantly lower than initially. Furthermore, the criteria accord-ing to which the respective levels were specified are not evident. Although clearly a badregulation should be corrected as soon as possible, from the point of view of the pensionfunds changing such an important indicator, changes the profitability outlooks of the entireenterprise. Changing the rules of the game when it has already started decreases the trustin the solidity of institutionalisation of this solution with the obvious consequences.

4.3.4 The efficiency of savings management

To insure the minimum well being of participants regardless of the fund they choose someregulations were introduced. Firstly, a floor for the rates of returns has been introduced toensure participants protection. It takes the value of either half of the average rate of return or4 percentage points below the average for the industry, whichever is lower5. Each fund thatdoes not meet this requirement needs to compensate its participants from owners’ capital.The time horizon for the rates of return was changed from biannual to three year ones 6,which obviously affects the investment strategies .

Most obviously, funds also need to obey certain risk standards. By the act of 1997 theymust not bear more risk than the balanced investment funds, based on the results observedfor the previous period. These standards are executed on a monthly basis by Pension FundsSuperintendence. As is easily observed, OPFs have little incentives to outperform the sector.There is no bonus to beating the average rate of return, except for highly doubtful monthlymanagement fee on capital described in the previous section. On the other hand, however,neither competition regulator nor legal mechanisms preventing collusion are in place. Inaddition, there is also no control over excessive consolidation.

To asses the efficiency of OPFs as capital accumulators vis-a-vis a standard financialinstitution that is a bank we compare the observed risk-less interest rates with OPFs results

5 Mathematically: R = min0.5 · R; R− 4%6 Amendment of 2004.


and - in a very simplified way - estimate the differential necessary to ensure OPFs outperformthe bank account strategy.

Consider a very basic financing model to compare two different types of medium and longrun investment strategies (OPF versus an alternative strategy). Assume:

• Ki- monthly accumulated amount transfer to OFE or an alternative system;

• αi - part of Ki, OPFs management fee, which they are allowed to deduce from thecontributions;

• rOFE =∑

iweighti ·K1,t−K0,t

K0,t-average weighted OPFs rates of return (weights based

on the market shares) with K0,t - initial capital at time 0 and K1,t - capital after aspecified period (usually monthly);

• rFREE - risk free interest rate identified with the return from annual T-bills (rTBILLSFREE )

or commercial bank deposits (rDEPOSITFREE ), depending on the scenario analysed.

We consider future values (FV) of a 480 monthly investment strategy: from 25 till 65years of age. Since any assumptions on the interest rates patterns over such a long time spanis a disputable issue, we avoid this problem by comparing the cash flows directly at time t.

We further assume that such a long run investment strategy is essentially costless underthe bank strategy. This assumption has two grounds. First, most citizens already have anindependent bank account. Secondly, there are many bank alternatives where opening andmaintaining an account actually is costless (e.g. internet banking). We assume that thereturn in case of such a strategy is determined by the interest on treasury bills (the proxy forrisk-less interest rate). Based on the above specification, the cash-flows per person are givenby:

FV (rFREE) =480∑t=1

Ki(1 + rFREE)(480−t)

FV (rOFE , αi) =480∑t=1

Ki[(1− αi)(1 + rFREE)](480−t) (4.1)

Comparing these two amounts allows to formulate the following hypothesis:

FV (rOFE , αi)− FV (rFREE) > 0. (4.2)

It is thus sufficient to compare [(1−αi)t(1+rFREE)]t and (1+rFREE)t under the assumptionof strict stochastic domination. Taking appropriate roots we obtain a system of equationsfor each t (and averaging over i):

rOFE ≥(1 + rFREE)

(1− αi)− 1, where 1− αi ∈ (0, 1). (4.3)

Of course the above inequality is a stochastic one and we do not know the processes followedby rt,OFE , rt,FREE , neither do we know Ki. What we have at our disposal is the completeinformation about the realisation of these processes up to now. We know rt,OFE and rt,FREE


in the period 1999 to 2005 as well as the pattern of αi evolution (constructed from Tables 2and 3). Basing on this data set and an additional assumption of scaling Ki to 1 we did twosimulations.

Simulation IWe compute the gap between the accumulated returns to two types of investments: OPFs

and our artificial one controlling for the costs associated to OPFs. The artificial benchmark isconstructed basing on the average long term deposit rate in the commercial banks as reportedby the Central Bank (monthly data on three year or more deposits, annual rates). From thesetime series we conclude that the artificial investment strategy was working better than OPFs.This outcome can be attributed to both the structure of OPFs rates of return and significantcosts (α factor in our model). We then ask the data by how much on average must OPFsoutperform simple bank deposits over the next nearly thirty five years to ensure that the newsystem is efficient from the value-for-money point of view.

For the reasons specified above it seems important to analyse the accumulated increasein value and not only contemporaneous rates of return. We compare the observed risk-lessinterest rates with OFEs results and - in a very simplified way - estimate the differentialnecessary ensure OFEs outperform the bank account strategy. The following formula isapplied:

68∑t=1

[(1− α) ·

i∏m=1

[1 + rm,OFE ]−i∏

m=1

[1 + rm,FREE ]]−

480∑l=69

[(1− α)

[1 + rOFE ]l− 1

[1 + rFREE ]l

]> 0, (4.4)

where rm denotes observed rates of return, while estimates of the expression in second bracketson the RHS of the above equation is only considered in a form of a differential. These simplecalculations demonstrate that OPFs would need to maintain on average returns 0,4 - 0,5percentage points higher than the commercial banks deposit rates.

Table 4.4. Simulation results - OFE versus a bank deposit

Amount invested monthly 1 PLNOFE cost structure As in: Tables 4.2 and 4.3(∗)

No. of monthly observations of returns 68No. of monthly contributions till retirement 480 (40 years)Earned in OFE 76,9 PLNEarned on a bank deposit 77,28 PLN

Sensitivity analysisDifferential for rFREE

t = 0.00 0,49%Differential for rFREE


t = 0.08 0,38%Source: own calculations. OPFs data from the Superintendence of Pension Funds. Data on commercial

banks from the Central Bank of Poland. The annual fee calculated as the average weighted by the

market shares.

For the derivations a constant differential between the two rates of return was considered.


The results show, that values for α matter significantly in comparing the completelyrisk-free investment strategy and participating in the pension fund. It is only natural torequire OPFs to realise the effective returns above the bank deposit. With the interest ratesdecreasing in Poland on its accession to European Monetary Union, OPFs are rather likelyto achieve the rates of return required in this simulation. However, one should be onlymoderately optimistic about this result.

Simulation IIOpen Pension Funds are believed to play the role of crucial institutional investors on

the Polish capital market. If so, the secure investment rather than secure deposit should betreated as a benchmark. Thus we repeated the above procedure for the most profitable lowrisk investment opportunity over the relevant time span. Public offerings of the governmentbonds in the period 1999-2004 were believed to be the most attractive from the risk-returntrade-off point of view. We used the annual yields of these bonds, as indicated at the day ofoffering as a reference level (regardless of maturity, dates of public offerings and OPFs unitvalues accorded). This is equivalent to a simple buy&sell strategy.

Table 4.5. Simulation results - OFE versus a risk-free investment strategy

Amount invested monthly 1 PLNOFE cost structure As in: Tables 4.2 and 4.3(∗)

No. of monthly observations of returns 68No. of monthly contributions till retirement 480 (40 years)Earned in OFE 76,9 PLNEarned on a risk-free investment strategy 88,29 PLN

Sensitivity analysisDifferential for rFREE



t = 0.08 0,46%Source: own calculations. OPFs data from the Superintendence of Pension Funds. Data on bond returns

from Money.pl. The annual fee calculated as the average weighted by the market shares.

For the derivations a constant differential between the two rates of return was considered.

Clearly, the gap between the effective OPFs returns and the government bond yieldsis much higher than in the previous case. Furthermore, although the budged needs willeventually decrease over time, the effect of the current public offerings will last for eventhirty years in some cases. Thus, although government and communal bonds constitute thevast majority of OPFs assets, funds should not be perceived as active investors on this marketdue to the high returns differential.

Summarising, treating bank as a benchmark for analysing the efficiency of the OPFs isa rather perfidious example. We did that only in order to show that effectively OPFs ratesof return are beaten even by the essentially risk free rates. Relatively low OPFs rates in thebeginning make it difficult to obtain relatively high benefits in the end due to accumulationprocess, even if the observed rates of return at the end might suggest that OPFs strategyoutperforms the risk-free one. There is also an additional effect of the timing of joining a


fund, which concerns most of the current beneficiaries obliged to join a fund within a firstyear of the reform (25 kohorts).

However, there are sound arguments against leaving retirement savings in the banksdomain . The main argument on our side follows from the observation that from the userpoint of view (the cost perspective) the difference between the OPF and the bank accountis not that evident, especially considering that the banking sector is also fully guaranteed bythe state.

4.4 Possible remedies

Undoubtedly, finalising the reform by completing the regulation on the Open Pension Asso-ciations is vital. Without specifying the institutional set-up in this respect, one is unable toascertain the very social security dimension of the reform. On one hand, the redistributiveand social net issues remain unaddressed. On the other, the specific aspects of the regulation(e.g. the investment strategies allowed, the variety of products offered, etc.) will surely pre-determine the behavioural patterns in this sector. On the contrary to what one might think,Open Pension Funds and Open Pension Associations - should they be eventually formed -are neither victims nor winners of this system.

Obviously, as was demonstrated above, due to the conflict of interest and inadequateregulation, OPFs abuse their position towards the contributors-beneficiaries. As the previoussection suggests, OPAs are likely to follow that way too. The fact that none of the OPFshas ever brought reasonable net profit combined with the observed mergers and their poorinvestment results might indicate that these companies are troubled by some systematic andpersistent inefficiencies hazarding their very existence. However, as we try to demonstrate,OPFs are formed and managed to provide for their costs of living. Literarily from the scratchan entire sector was created with its corporate structure, technical and physical infrastructure,non-negligible employment and, consequently, considerable costs. These are organisationsof considerable size both in terms of employees and in terms of fixed capital who succeedin paying their own bills from the management and maintenance fees they charge to theirclients.

If one wanted to render them to financial institutions whose interests would coincide withthe interests of their clients one would have to introduce some benchmarking and incentivesinstead of only security regulations. One of such solutions would be externalising the floorimposed on the OPFs rates of return. Tying it to some market indicators would not increasethe risk on the contributors side while forcing the funds to compete for returns instead ofcompeting for clients. Another possible solution would be to make the benchmark systemsymmetric. With the status quo a fund can only be punished for realising too low returnswhereas no incentives are provided for significantly outperforming the sector.

Also wide scale education seems necessary, as the society seldom realises the mechanismsof the new system. For many citizens the insurance agent was the only source of information.While far from being impartial, on many occasions they themselves were misinformed. From1997 on, consecutive governments did not propose the regulations necessary to complete the


reform. Reasons might be sought, among others, in problems of forming any stable coalitionin this respect. However, there are only two years left before first beneficiaries appear in thesystem.

5. THE VENTURE CAPITAL IN POLAND

The role of Venture Capital in stimulating the economic growth has been largely analysedin the economic literature. It is widely accepted that the existence of the sophisticated VentureCapital industry is a major fact behind the America’s ability to encourage innovation andsustain technological growth (cfr. Keuschnig, 2002). There are strong empirical prerequisitessupporting this statement. Kortum and Lerner (2000) show, that VC backed firms are moreinnovative and produce more and more valuable patents. In addition, they also have higherquality of management, as demonstrated by Hellman and Puli (2000, 2001).

On the other hand, there also seem to occur macro level effects, since Venture Capitalsector is known to stimulate employment, as was shown Wasmer and Weil (2000) in a panelstudy on 20 OECD economies. And even though in Europe these effects seem to be lessspectacular than in the United States (Botazzi and Da Rin, 2001), Venture Capital is believedto be one of the main sources of innovative investment. As forcefully demonstrated byRomain and von Pottelsberghe (2004), Venture Capital turns out highly significant in thetotal multifactor productivity function, both stand alone and as a factor intensifying the R&Dexpenses and overall business activity. This is fully recognised by the European Commission,who has drafted the Risk Capital Action Plan (EC, 2002)1 and its implementation documents.

In this thesis we attempt to focus on the case of Poland for two main reasons. Firstly, asa transition economy, Poland has really little own funds for financing technological develop-ment. Accessing the European Union provides a significant improvement in this domain, butonly a small part of EU funded money will actually be spent on the R&D activities in hightechnology. Secondly, the industrial R&D spending is largely limited by the availability ofown funds, because of the absorption limitations. Thus, in the case of Poland Venture Capitalfunds emerge as one the of main sources of R&D investment and play especially significantrole in realising the aims underlying the Lisbon Strategy.

On the other hand, Venture Capital along with the funds in the R&D sector providethe business know-how, which signifies further their importance. In the case of a formallycentrally planned economy, with average economic awareness very limited in the society, themanagement techniques and the operational experience are particularly required.

Understanding the mechanisms encouraging start-up entrepreneurship and innovation ismainly important because the main policy interest in a viable VC sector focuses on its roleas an engine of innovation driven growth and job creation in new industries . The analysisof a sample of firms located in Silicon valley by Hellmann and Puri (2000, 2002) showsthat Venture Capital vitally enhances the professionalisation and commercial orientation of

1 The European Commission (1998, 1999 and 2002) gives a detailed outline on the content and the imple-

mentation of the action plan for risk capital in the European Union.

5. The Venture Capital In Poland 47

young firms . Venture Capital backed firms introduce more radical innovations and pursuemore aggressive market strategies compared to other start-ups. For example, once a VentureCapital joins the firm and provides finance, the probability of introducing the new productjumps up by a factor of more than three! Rapid market introduction is strategically importantbecause the first firm enjoys a first mover advantage.

On a more aggregate level, Kortum and Lerner (2000) show that one dollar of R&Dspending in Venture Capital backed firms creates more patents and more radical innovationsthan the same expenditure in other firms. They calculate that Venture Capital financed R&Daccounts for roughly 14 percent of U.S. industrial innovation in 1998 although it amounts toonly about 3 percent of all R&D funds. This empirical evidence shows that VC significantlypromotes innovation and business growth. Evidently, one is troubled with the selection biasin this type of research, as VC ’s are specialising in seeking and financing R&D, while in manycorporations research departments have large teams and consistently growing budgets whichis probably only partly justified by the productivity growth. Thus, the empirical evidencesuggests rather higher efficiency in financing R&D due to lower, so to say, fixed costs andnot actually the differences in the R&D spending productivity across sectors and forms offinancing.

The real effects of Venture Capital, i.e. the ability to locate and select promising projectsand to add value in terms of strategic business advice, depend not only on the Venture Capitalfunds own managerial qualifications and investment know-how but also on their incentivesto be actively engaged in portfolio companies. The supply of experienced financiers withuseful business contacts and knowledge of the industry is a scarce resource that is not easilyaccumulated in short term and may become a considerable bottleneck in the development ofa healthy Venture Capital sector. An equilibrium analysis of the Venture Capital industryshould pay due attention to the slow entry of experienced financiers.

The other precondition for the development of an active Venture Capital sector and ahigh rate of business creation is the supply of entrepreneurs with innovative ideas. In fact,representatives of the industry often complain about the lack of high quality entrepreneurs.However, high quality entrepreneurs refers rather to renown specialists in a domain whoseposition can strengthen the process of product dissemination in the launch sphere. Highlyskilled managers and engineers are available and finding proper people - although undoubtedlyan art - does not pose an obstacle.

It seems that the creation of highly innovative firms requires an active research environ-ment. It is no accident that the Venture Capital industry is usually geographically concen-trated in the neighbourhood of publicly and privately funded centres of basic research thathost numerous researchers who might consider to turn their research ideas into a businessstart-up2. The active role of the government in this respect cannot be overestimated, asstimulating the process of concentrating research spending in areas with a potential for com-mercial applications and a correspondingly high probability of spin-offs definitely allows to

2 In the most classical example, Silicon Valley is located close to Stanford and San Francisco, both highly

recognised in the academic and entrepreneurial world, with San Francisco being the second most popular US

city among immigrants from Europe and South-East Asia.


increase the efficiency of allocation. However, such an intervention cannot be direct in termsof specifying physical localisations and economy sectors. Fostering basic research spendingshould wisely raise the probability of researchers coming up with innovative business ideas.

Although, as common proverb states, the nature hates the vacuum, Venture Capital sectoris significantly underdeveloped in Poland with its share in the GDP of only 0.098%, whereasin France and Great Britain this number reached 0.37% and 0.85% respectively3. Accordingto Schofer and Leitinger (2002), there are strong prerequisites preventing the Venture Capitalsector’s development and these lie in the sphere of infrastructure and operation (insufficienttechnical infrastructure, too little roads, phone lines, low level of education etc.).

However, as follows from the recent report by European Private Equity and VentureCapital Association (EVCA), success stories do happen in Eastern Europe, while out of 25described cases as much as 7 originate in Poland. The main thesis of our paper is that thelimitations on the Venture Capital development in Poland come from the institutional side,whereas the infrastructure shortcomings play only secondary role. Legal limitations on theoperation of the Venture Capital funds along with some behavioural patterns on the side ofthe funds themselves provide the main reasons for sector’s relative underdevelopment.

5.1 The VC industry in Poland

Polish Venture Capital sector is in its 14th year of development with approximatelythirty funds operating currently in Poland in spite of which, it still lags behind the rest ofthe world in terms of recognition as the reliable business partner. Only recently it begins tobe recognized as very important source of knowledge and capital.

In terms of numbers, 90% of all transactions on the Polish VC/PE market concern PrivateEquity investments (buy-outs, expansion, etc.), only the remaining 10% concern financingof early stage investments4. Furthermore, this percentage is declining over the past yearssuggesting the marginalisation of seed and start-up investment which is to a large extentequivalent to diverting funds from technology and R&D sectors. This declining amount ofVenture Capital investments increases the capital gap on the Polish market. This is especiallyharmful to the economy, as Venture Capital focuses by definition on small an medium size.This is were the highest demand for capital lies and VC/PE funds knowledge and experienceis necessary.

According to the 2002 survey carried out by EVCA, the volume of total private equityinvestment in the Polish market grew significantly in comparison to 2001, reaching PLN529 million (EUR 137 million). Both Polish-based and foreign-based private equity investorscontributed to this growth. Polish-based private equity firms invested PLN 422 million (EUR109 million) in while at the same time foreign-based private equity managers invested PLN107 million (EUR 28 million) in Poland in 2002 - up nearly 120% on the PLN 49 million (EUR13 million) they invested in 2001. Poland was clearly an attractive destination for private

3 Data for 2003, after PSIK 2003 Annual Report.4 For the purposes of this thesis, the division between Private Equity and Venture Capital is kept along the

line of adding value in strategic domains rather than via consolidation and financing growth.


equity investments in 2002, when it returned to more normal levels of economic activity aftergoing through a relative low in 2001. It is worth noting that in 2002 private equity investmentin Poland was higher than in four EU countries: Ireland (EUR 129 million), Austria (EUR123 million), Portugal (EUR 93 million), and Greece (EUR 17 million).

Graph: PE/VC investment in Poland (source: EVCA)

The EVCA survey focuses on fund managers that are based in Poland and invest bothin Poland and throughout Central and Eastern Europe. The investment patterns of thesePolish-based private equity houses changed in 2002, when their focus shifted from regionalto domestic investments. Of the total of PLN 454 million invested by Polish-based fundmanagers in 2002, PLN 422 million (93%) went to Polish companies and PLN 32 million(7%) was invested in the other CEE countries. The respective allocations in 2001 were PLN337 million (61% of total) in Poland and PLN 215 million (39%) in the CEE region.

The number of companies that received investment increased significantly between 2001and 2002. Polish-based funds completed investments in 86 companies in Poland and CEE in2002, up from 57 in 2001. At the same time the proportion of Polish companies increased by51% in 2002, 74 of all companies receiving investment were Polish. In other CEE countries, 12companies attracted investment in 2002. The large increase in the number of companies thatreceived investment was driven primarily by a significant growth in ’follow-on’ investments,i.e. where companies that had received private equity capital previously were given additionalcapital by the same investor. Of the 86 companies that received funding in 2002, 63 werefollow-on investments, whereas 23 companies received private equity funding for the firsttime. In 2001, of the 57 companies receiving funding, 32 were follow-on investments and 25were first time investments.


This focus on existing companies is further seen in the amount of investment allocatedto first time and follow-on investments. Of the total PLN 454 million invested by Polish-based funds in 2002, 64% went into follow-on investments while only 36% went into first timeinvestments.

According to a historical study realised by Polish Association of Capital Investors (PSIKDATA), the characteristics of the Polish Venture Capital market differ significantly fromwhat is observed in other countries. Primarily, most of these investments are of privateequity type, which has the disadvantage that VC fund support is limited to the financing.Furthermore, it suggests that instead of stimulating the necessary R&D spendings, fundsfocus on restructuring and managerial buy-outs. Thus, VC sector has had for the time beinga much smaller influence over the development of Polish economy. Furthermore, as suggestedearlier, this sector plays a marginal role in terms of size of both investment and contributionto GDP.

There can be numerous reasons for this status quo. As pointed by Leitinger and Schofer(2002), Poland is outpaced by its neighbours in terms of development in economic, legaland social environment as well as ’entrepreneurial spirit’. On average, Poland scored 67%of the EU15 level, which was the lowest of the considered group, worse than Russia used inthe CEECs group for the comparison reasons. Following their argumentation, some of theshortcomings like legal regulations can be alleviated rather easily, whereas others - mainlyinsufficient infrastructure - require more time and effort. However, although they tried tocapture the general characteristics of legal and social environment, they did not control forthe quality of Venture Capital legal regulations and the attitude towards Venture Capitalfunds.

5.2 How to run a VC fund in Poland

In reality, risk capital investors lack a legal vehicle to operate in Poland. On the otherhand, those who decided to commence activity despite the unfavourable legal conditions arerather Private Equity funds investing mainly in low-risk large scale restructuring projects.So far many managements teams and entrepreneurs perceived funds as ”vultures” and fearedthat VC/PE objective is only to take over the control their companies. This perception wasto a large extent grounded.

In this section, we investigate the legal aspect and later turn to the behavioural aspectsof the Venture Capital presence in the Polish economy. Theoretically, VC funds do not needto limit their activity to the country where they are originated. In other words, funds maybe organised elsewhere and still operate in Poland. This solution, however, has some seriousdrawbacks. First of all, tax and ownership procedures are more complex and thus more costlyand bear more risk. Secondly, the further from the fund to the beneficiary companies, theless efficient management and narrower possibilities to provide managerial know-how and anyintangible form of support. Thirdly, the flow of information is largely limited, even despitethe recent developments in the communication technologies. Therefore, although feasible,funds avoid this form of organising their operations.


There are two general acts defining forms of organising enterprises to be considered here:the Act on Investment Funds and the Commercial Companies Code5. We briefly discuss therelevant regulations below.

Comercial Companies Code specifies six types of legal identities, of which two have nolimit on the liability of owners. The latter four may be grouped into limited liability andpartnership companies:

• Limited liability and joint-stock company.

Although they fulfill four of the crucial requirements - i.e. investors’ limited liabilityand the presence of board or revision commission to monitor the activity of managers,one can easily increase the base capital while deposits do not need to be diversified - theyare not designed as VC vehicles. Primarily, there is double taxation as investors areforced to pay both CIT (on the corporate level) and PIT (on personal level). Secondly,at the moment of investment owners do not enjoy any tools allowing for limiting theoperational costs. Thirdly, joint-stock company, if public, is obliged to disclose crucialinformation from the strategic point of view.

Regardless of all these obstacles, some of the Venture Capital funds undertake theiractivity using this legal solution6.

• Limited partnership and limited partnership joint-stock company.

These legal forms are characterised by being a legal person with limited capacity. Thepartners are either general (komplementariusz) or limited (komandytariusz), meaningthat the latter can only be held responsible to the amount of his share, while the formeris fully and personally responsible for the obligations of the company. Consequently,only general partners have the right to represent the company, while the identity oflimited partners can be kept secret. Furthermore, in this form of company sharescannot be public, neither freely traded. What is even worse, the vending of shares isstrongly limited while the new partner is only accepted if he receives the approval ofthe majority of owners.

In the case of limited partnership joint-stock company, shares are replaced by stocks,which allows for easier trading. Although this particular solution seems to be the mostsuited for VC activity, there are no funds of this form. It can be partly attributedto the uncertain tax situation (unequivocal interpretations of VAT regulations) as wellas the problem of finding skilled managers who would be willing to accept the unlim-ited liability clause. Using a legal person with limited liability as a general partneris strongly opposed by most prominent lawyers specialising in commercial companiescode7. Furthermore, it was not designed as a vehicle for VC, discouraging potentialinvestors from using this solution. Finally, no one has used it so far, which imposeshigher risk on the eventual pioneer.

5 Dz.U. 2002 No. 49 poz. 448 and Dz.U. 2000 No. 94 poz. 1037 respectively.6 E.g. MCI S.A. and KCI S.A.7 Including professor Stanis law So ltysinski, one of the Commercial Companies Code author.


Investment Funds is a legal person managed by a separate identity (the investment fundsociety, TFI). Fund participants do not take the responsibility for its liabilities, including taxobligations, whereas TFIs are fully responsible to the fund participants for mismanagementof the fund. One can become a participator by buying an investment certificate - a publiclytradable security. Investment Funds are generally of three types:

• Closed-end investment funds

Except for proper tax treatment, investment funds are not a good vehicle for VentureCapital. First of all, they are very costly to set up as there are many formal requirementsconcerning securities emission. Secondly, they are obliged to disclose extensive andfrequent information on their activities (like detailed asset evaluation every quarter).They also have diversification requirements (no more than 10% in one activity), whichmakes it almost impossible to start a Venture Capital fund (one would need at least 10projects at the start). Furthermore, all funds must be gathered at the beginning, whichvirtually prevents the Venture Capital fund from development. On the top of this all,participants have little control over the investment activities (investors have insufficientinstruments to influence the TFI).

• Specialised closed-end investment funds

They are in general quite similar in their construction. However, investment certificatesare not obligatorily subject to public emission regulations. Thus, disclosure regulationsare much less restrictive, while fund participators enjoy comparably stronger controlover the fund managers, as investors have better tools to supervise the fund. In ad-dition, diversification principle was relaxed to the maximum of 20% in one activity.Moreover, investors have better tools to supervise TFI. On the other hand, base capitalrequirements are more demanding imposing a very low number of investors with highparticipation rates.

There are no closed-end investment funds in the Venture Capital sector. Despite morefavourable regulation, neither are there specialised investment Venture Capital funds. Asunderstandable as it is in the first case, in the latter it follows from insufficient provision onfunds. This type of fund organisation requires large institutional investors and only with thechanges in the regulation on pension funds (from August 2004 onwards they are be allowedto invest in high-risk activities) this type of players could become interested in participatingin the VC sector. Thus, the situation should gradually change over the next years.

Summarising, there were no actual legal regulations enabling the origination an efficientVenture Capital fund in Poland. In addition, despite efforts of the Polish Capital InvestorsSociety (PSIK) there are no prerequisites to believe that the situation should change in thenearest future. Despite the low score in the ’entrepreneurial spirit’ category in the Leitingerand Schofer (2002) Venture Capital funds do exist in Poland, some of them using the existinglegal structure and originating a formal investment fund locally. The structure of this multipleidentity is presented on the graph below. As one can observe, the partnership joint-stock


companies seem to enjoy some advantages over other solutions, but their application is againstthe spirit of Commercial Companies Code as suggested by its authors.

Graph: the best VC solution.

Surprisingly, investors do not point to the legal limitations when giving their views onthe constraints on VC sector development. In the survey study by Kornasiewicz (2003)some of the obstacles as identified by funds managers themselves were suggested. Theyconsistently point to little capital and low number of good business opportunities. We proceedto addressing this issues in the next section.

5.3 But what is a good business opportunity?

Macroeconomically speaking, Poland has incredible potential for growth as it is a largecountry, where only the most simple economic reserves were so far explored in the pastdecade. The society is enriching very fast. Needs for advanced technologies and value addingservices are expanding, especially in large and medium size cities. Moreover, because of theshortages always present in the previous system, people demand novelties much stronger thanin other societies. For instance, despite the general economic slowdown, demand for new cars,television sets and home equipment is still increasing (although dynamics have dropped) andsignificantly over 15 million people owns a cellular phone (nearly 40% of population). Thisindicates, that there will be room for successful investments.

On the other hand, as raised by many researchers and analysts, so far only the most


simple reserves of the Polish economy have been explored. The successful restructuring of thecentrally planned economy into a market one is only the first step and there is still large roomfor economic development both in technology and services and in the traditional productionsectors. The structural funds from the European Union combined with the administrationalknow-how transfer have the potential to overcome the inertia of the current institutionalsystem and improve the overall quality of governance.

Thus, joining the EU should turn out to be another positive shock to the Polish economy,whose size may only be compared to the 1989 transformation. This gives grounds to believein the large potential of Poland in the perspective of next two-three decades.

According to a study by Kornasiewicz (2004), however, managers of Venture Capitalfunds in Poland perceive the lack of good projects to be the main reason hampering thedevelopment of VC sector. More explicitly8:

• 55% suggests that quality of the managers falls short of their requirements,

• 41% point to the underdevelopment of financial markets, while additionally 18% raisethe problem of insufficient capital supply,

• 32% of answers gave the low number of good investment opportunities as the mainreason,

• 27% gave poor VC PR as an important obstacle, whilst 10% of respondents (25% amongmiddle and biggest funds) complaint about insufficient information and difficult accessto entrepreneurs

In addition, 13% (almost 30% among the smallest funds) point to problems of enforcing theexit strategy.

These responses are somewhat surprising if one considers the nature of Venture Capitalactivity. First, VC ’s are generally only interested in projects giving the expected annualreturn on investment of approximately 30% (at the least, doubling the invested capital withinthe three years). Secondly, the investors allocate resources to funds so that they couldidentify projects satisfying 30% per annum requirement (and occasionally their sectoral orrisk preferences). Finally, VC funds are by definition specialised in reaching investors andgathering capital necessary to finance the positively evaluated projects.

The study by Kornasiewicz (2003) reveals also the preferred investment types as wellas gives some indication about the evaluation criteria9. Polish VC funds avoid investmentbelow $1mln with the average size of $3-7mln. They rather enter projects with at the least50% of participation share with strong preference on state privatised companies. As regardsthe stage of development they prefer the expansion and managerial buy-outs and buy-ins.Finally, the sectors of interest include predominantly FMCG, pharmaceutical, medical care,retail wholesale and telecom (with this order of undertaken projects frequency). What theyare willing to offer includes capital as well as financial services intermediation. This is arather stunning picture of the VC sector in Poland.

8 The percentages do not add up to 100% as surveyed could choose more than one option from the list.9 The views presented here are the ones of the author and not the one to be found in Kornasiewicz (2003).


As regards the selection and evaluation process, funds assign highest attention to thesector (70% of respondents) and the prospective profitability of the project (85%) as well ashigh quality of managers (85%). As standard as these criteria may seem, they are followedby current profitability (68%) and a good expansion strategy (78%). In other words, VCfunds want a ready and highly profitable project in an expanding sector that only needsfinancing and whose up-to-date performance may be carefully examined (little partner risk).The expectations of VC funds only concern two aspects: growth of the shareholders valueand easy and feasible exit strategy. To find such projects, funds resort to information fromfinancial institutions (70% of funds).

Summarising this picture, Polish VC sector is interested in hands-off management of bigstate companies subject to favourable restructuring and with easily vendable facilities, oper-ating in expanding sectors and managed by highly skilled entrepreneurs. More importantly,these are not their Christmas wishes but the evaluation criteria applied in assessing arrivingprojects. A small size project suggested by biotech specialists with potentially expandingbut specialised clients and partners but without management experience fails to fulfil any ofthe requirements. Furthermore, since banks are unlikely going to be interested in this typeof project due to high uncertainty, funds will rarely be informed about the very existence ofsuch projects.

Evidently, in an economy restructuring for fifteen years it is not easy to find such projectsany more. Furthermore, as regards the PR of funds, these are all domains that only the VCsector itself can control. This is true, that venture capitalists are perceived as ’vultures’, com-ing to the company, extracting the resources and selling it (preferably to a foreign competitorwho is likely to close the facility)10. But it is also true that these are the funds themselveswho earned this reputation as a result of some of their engagements. Low level of knowl-edge about VC sector is equally a consequence of insufficient efforts by Polish Association ofRisk Capital Investors, who publish nothing but relatively short pamphlets and descriptivestatistics on the sector. Out of approximately twenty five VC funds only as little as 6 have awebpage, while press releases about engagements and exits are extremely rare. In most cases,consumers do not realise that a company has a VC partner. In a survey conducted amongthe students of 10 main universities in Poland 85% answered that joint venture and venturecapital are equivalent terms (65% among economics and business studies students)11.

Moreover, as far as availability of capital and exit options are concerned, VC fundsare financial intermediaries specialised in gathering capital wherever they are able to findinterested investors. In terms of assigning roles in the economy, it is their job to find andconvince both local and international investors about the business opportunities they areable to find and monitor. It is true that a relatively shallow stock exchange makes thisexit strategy rather unfeasible, but without exceptions all of the IPOs on the Warsaw StockExchange within the past three years have been extremely successful, whilst it also seemsthat investors were attracted by the moments of IPOs leaving the market afterwards.

10 See: Report On Small And Medium Entreprises In Poland (1999-2004) published annually by Polish

Agency For Entrepreneurial Development.11 A survey by Rzeczpospolita, April 25th, 2003.


Finally, VC sector is much more than just capital provider to high risk investments. Theydispose of the highly skilled specialists trained in drafting strategies and project management.Usually, this allows for the so-called hands-on management with the additional positive ex-ternality of transferring know-how and soft skills. This typically value adding activity is thesource not only of higher company value growth but also networking within and betweendifferentiated sectors of the economy. However, it seems that in Poland VC sector prefersto limit its input to providing capital, while expecting entrepreneurs to already posses allskills and background indispensable in this type of projects. The additional inefficiency ofthis approach is that vertical and horizontal integrating opportunities might be forgone dueto insufficient information sharing and the lack of holistic approach.

Concluding, the Venture Capital sector in Poland plays only marginal role in the econ-omy. Its representatives complain about the environment, while there seems to exist no legalvehicle enabling venture capitalists to openly run their activities. On the other hand, it alsoseems true that little competition among the funds as well as initially extremely favourableconditions with numerous attractive state companies privatised have resulted in a peculiarreaction of the fund managers: although they prefer to maintain control over companies, theyresort to hands-off management and actually wait for the entrepreneurs to approach them.So far they have mostly invested in large projects of restructuring and privatising promissingstate owned entreprises with strong focus on traditional sectors of FMCG and pharmaceu-tical. They rarely engage into seed or start-up activity in technology or IT sectors. As themain source of information other financial institutions are usually given, while funds ”en-joy” bad publicity of vampiring the chosen companies. In addition, a standard strategy inother developed countries is to shop around the funds looking for a better offer to the initialproject owner. Interviews with Polish businessmen trying to receive financial support fromVC fund, talking to more than one fund is perceived as violation of trust and makes anyfurther negotiations virtually impossible.

5.4 The perspective for the future

The Lisbon Strategy has emerged from the initially virtuous assumption that Europe needsto intensify activities in the domain of new technologies or else it will lag behind even further.Regardless of how we judge its feasibility and actual realisation from the today perspective,this initial motivation was not wrong. Europe does need to intensify technological develop-ment. And for this purpose Europe does need Venture Capital. Nonexclusively but inevitably.

The above statement, as crude as it seems, conveys the main motivation for dealingwith this subject. Poland in the fourteenth year of transformation cannot afford sufficientfinancing for science, industrial R&D, infrastructure development and necessary education.This is partly because it is a relatively large country with large needs and a considerablecapital gap. On the other hand, however, infrastructure investment and industrial R&D, ifrealised wisely, are extremely profitable at the current stage of development.

Therefore, there seems to be no reason why state should intervene in the domains naturallylying in the interest of Venture Capital funds. However, Venture Capital sector is rather


underdeveloped, when compared to EU member states but also to other countries of theregion. There are some important reasons for them. Unlike Leitinger and Schofer (2002) weclaim that infrastructure shortcomings and some social and general legal characteristics ofthe economic environment are of only secondary importance. The activity of Venture Capitalfunds is largely limited by the unfavourable legal setting regarding risk capital. Nevertheless,we identify the main barrier to development in the funds themselves.

We suggest that the VC sector is currently in a transitory period. The initial conditionsof extremely favourable business outlooks are already extracted and the Polish VC ’s will beforced to adapt to the new reality and change the modus operandi to the one more alikethe European neighbours. From this perspective, two main conclusions should be derivedwith regards to the behaviour of the Venture Capital funds in Poland. First, it is fullyunderstandable why they have been behaving in the described way. Secondly, the changingenvironment closes the past chances opening the new ones at the same time. Bad publicityis undoubtedly a serious problem to be addressed as soon as possible.

Furthermore, strategic approach seems to be the key change to adopt. When analysinga potential enterprise decisive weight should be put to seeking where value is added withina project. Improving cost efficiency and intensifying distributional efforts is currently notsufficient for success. Looking for technological innovation is equally unsatisfactory, as it isthe growth potential that is crucial as well as self-sustainability.

So far, as evidenced above the largest Venture Capital fund successes in Poland came fromefficient restructuring. This is still possible in agriculture and sometimes in manufacturing,but on an economy scale, restructuring needs to be replaced by strategy in approach. It seemsinformative that up to date no large success came from strategic Venture Capital investment- there were little attempts and mostly unsuccessful.

Some of the most spectacular failures in the Venture Capital enterprises resulted from thelack of partnership and team work. High rotation in the investment teams and steering com-mittees led to inappropriate decisions and decreased trust between companies and VentureCapital fund. Thus it seems necessary to insure stable project managers and unchangeableteams. On the other hand, companies often lack support from the Venture Capital funds,who limit their activity to providing finance and controlling advancements. Companies shouldtreat Venture Capital fund representatives are their partners, as only this way mutual com-mitment and efficient communication can be ensured. Provision of know-how and interveningeven at the operational level is not only needed but also essential for the success.

Another aspect of crucial importance is building corporate culture. Widely discussedgovernance in the public administration has its counterpart in the corporate world. Com-panies rarely recognize the importance of building teams, under investing in human capitaland allowing for high rotation of qualified employees. Business ethics is also on a very lowlevel (corruption), which further deteriorates corporate culture. Furthermore, this lack ofcorporate culture has a direct transmission to the results of Venture Capital funds - at leastthree large projects failed solely on this ground. Building corporate culture is crucial forassuring sustainability of their growth, which will certainly be priced at the exit. On theother hand, efforts in this direction create a positive atmosphere around the fund.


Finally, Venture Capital funds are rarely widely known as a potential source of capitalfor entrepreneurs. Not only do they seem mysterious to most companies, but they are alsosurrounded by the bad fame of buyout and (hostile) takeovers. Another characteristic ofVenture Capital funds sector is that there is practically no competition for projects betweenfunds. This is one of the reasons for such a slow development in this sector as well as low paceof changes. There are relatively few channels through which VC funds could inform abouttheir activity. Companies simply do not know, that there are venture capital funds and thatthey are willing to provide financing to good projects. The best channel for obtaining goodPR is maintaining partner relations with financed companies and making efforts to increasepublicity.

Is there any role for the state in this process? Evidently, legal environment needs to bechanged to allow for open operations of VC ’s. There are many solutions in differentiatedEuropean legal systems and surely they can be adapted to the Polish one12.

Secondly, as Schertler (2003) suggests, labour market rigidities are an important factorstimulating the activity of VC ’s. Interestingly, they seem to stimulate the productivity ofthe sector, but this effect should rather be attributed to the differences in capital-labourratios between the countries of different labour market rigidities. This undoubtedly calls forstimulating the capital accumulation in the Polish economy.

Intriguingly, Schertler (2002a, 2003) suggests also that deep determinants of capital mar-ket emergence (with the particular attention on VC ) are the focus on value adding activities(both to the final consumer and in the production and management process). Venture Capitalmarkets emerge only if the value-added by venture capitalists’ active involvement in form ofmanagement support is high compared to the costs of this management support. Further-more, the demand for Venture Capital by young, high-technology companies must exceed acritical level. This critical level is determined by the indivisibility of innovative ideas and bythe specialisation of venture capitalists on particular fields of technologies. In other wordshands-on management and sectoral specialisation are indispensible for the development ofthis sector, both of which require the Polish VC funds to change their behaviour.

In addition, there are some advantages of offering VC ’s a choice between the loan schemesand the public equity. Both schemes can be used to promote VC investment in high-technology investments. Although, under both schemes experienced investors save on man-agement support, offering them a choice can lead to a positive self-selection, where inexperi-enced VC ’s tend to choose public equity where the incentives to enter the market are higher,while more experienced focus on public loans where incentives to save on management sup-port is lower. This has been demonstrated and empirically verified in Schertler (2002b) andpoints to the role for the state in subsidising R&D activity even in the private sector.

Furthermore, venture capitalists may be attracted to investing in start-up entreprisesby a wise subsidising policy. Schertler (2000) demonstrates that venture capitalist onlyfinances start-up enterprises if he has sufficient expertise to make high-risk investments in newtechnology profitable in terms of their expected value. It is shown that a venture capitalist

12 See: Europe Private Equity Special Paper, EVCA, 1997


who already has sufficient expertise reduces his management support in the start-up entrepriseunder a public subsidy. Moreover, venture capitalists who do not have sufficient expertisemay finance start-up entreprises if future losses of the start-up investment are partly coveredby the government.

As Romain and van Pottelsberghe (2004) demonstrate, VC contributes to growth throughtwo main channels: new product and processes on the markets as well as developing the ab-sorptive capacity of the knowledge generated by both private and public research institutions.Venture Capital improves the ”crystallisation” of knowledge into new products and new pro-cesses with large potential spillovers, feedback and externalities. The numerous projects intheir seed and start-up phase would definitely benefit from such policies in Poland. TheLisbon Strategy and Structural Funds definitely offer means of financing such policies, buteither the milieu of risk capital investors and the policymakers need to find an incentive andcompromise in drafting and enforcing such regulations.

6. THE EFFICIENCY OF THE INVESTMENT DECISIONS

It has been often raised in pubic debates that Polish economy suffers from a capital gap.The problem is perceived as important to the extent that it limits the investment processescrucial for the economy growth as well as the catching up process. High trade deficit and theresulting foreign investment partly alleviated the problem. Nevertheless insufficient fundingis still believed to be the major obstacle for Polish entreprises.

On the other hand, one is rarely able to quote any research providing either the estimatedsize of the gap or the measure of its consequences. Furthermore, the argument seems to beabused by some of the firms as well as government representatives. In this paper we intendto approach this problem applying the agency theory model to the data of Polish stock-listedcompanies.

Theory of firm states that if financial markets satisfy efficiency hypothesis there is nodifference to the company whether it uses its own capital or borrows it. Thus, there isequivalence in financing and there should be no constraint on investment resulting fromownership. This seminal statement by Miller and Modigliani induced high confusion amongthe empirical economists. Namely, the postulated orthogonality of investment and cash flowsdid not find the support in data as early as in 1950s (cfr. Meyer and Kuh, 1957).

Two types of theoretical explanations where provided to explain this phenomenon. One,building on the information asymmetry and adverse selection arguments suggests that firmswith positive NPV investment opportunities will forgo profitable investment to avoid theexcessive cost of external financing, as insiders are much better informed about the quality ofthe projets than external capital providers. This hypothesis by Myers and Majluf (1984) hasbeen empirically verified by Fazzari, Hubbard and Petersen (1988) as well as Himmelberg andPetersen (1994) proving that liquidity constraints determine the cash flows to be importantdeterminants of investment decisions. This hypothesis may be interpreted positively in thesense that it is the inefficiency of the financial markets and not the firms that yields theresults strongly contradicting the Miller and Modigliani theory.

On the other hand, however, Jensen (1986) has raised the alternative explanation sug-gesting that agency problems might play an important role here. In particular, since themanagers’ utility may follow from other aspects that just remuneration (e.g. corporate jets,cars, new headquarters or prestigious entreprises), one risks over-investment instead of payingout dividends to shareholders.

In this thesis we apply the methodology proposed by Vogt (1994) allowing to distinguishbetween these two explanations for the observed correlations between cash flows and invest-ment spending. The major drawback of this methodology ensues from the fact, that it canonly be applied to companies for which objective, market valuation is available. Thus, the

6. The Efficiency of The Investment Decisions 61

scope of research in this research is limited to stock-listed companies.

6.1 Theoretical foundations

Discussing the issue of investment one immediately faces the problem of defining the costof capital. The first and the most rough approach takes the users’ approach specifying theactual cost of capital to:

rK(t) = [rfree(t) + σ − pK(t)pK(t)

] · p(t),

where pK(t) denotes the market price of capital (also capital goods), σ signifies the depreci-ation, while dot corresponds to relative change over time. If we want to control for the taxshield (or the tax advantage) the above formula changes to:

rK(t) = [rfree(t) + σ − pK(t)pK(t)

] · (1− tax rate) · p(t).

This formula, however, has some obvious shortcomings following from implicit assump-tions necessary to derive it. Firstly, there are no limits on debt/equity ratio in this model. Toput it more explicitly, companies could borrow the entire capital which is rather unfeasible inthe real world. Secondly, there is no room for uncertainty of future returns as it is an instan-taneous model. Lastly, and most importantly, this formula assumes costless and immediatejumps in capital neglecting the adjustment aspect.

One should consider two sources of typical adjustment costs: internal and external. Whilethe first follows from intuitively obvious causes like changing capital stock, training workersand installing new machines etc., the latter requires some consideration. One can expectcapital markets to adjust immediately and essentially without any cost to new information -these are the foundations of financial market efficiency hypotheses. However, the price of thecapital goods can change on the market (pK(t)/pK(t)) relative to other goods. Consequently,firms need to adjust their investment decisions accordingly which obviously bears cost. Inorder to immune model implications to these two types of problems one needs to incorporatethem.

6.1.1 Q-Theory of investment

Common sense considerations suggest that for a pricing model of the capital cost tobe complete it needs to incorporate the following elements: risk-free interest rate, capitalproductivity and depreciation, adjustment costs and the characteristics of the productionfunction. All these may be found in the Q-Theory of investment, which is derived in bothcontinuous and in discrete time versions under the following set of assumptions1:

• We consider one industry, where there are N firms operating;1 The model presented below is an extended version of Romer (2001)


• Adjustment costs are convex in the rate of change in capital stock k(t). They areincreasing in the scale of adjustment (c(0) = 0, c′(0) = 0, c′′(k) > 0) and it is alwayscostly to adjust capital;

• Price of capital is given by pK(t) with the changes in the price of capital goods expressedby pK(t);

• Change in the stock of capital is given by standard relation of:

k(t) = kt+1 − kt + pK(t) · It − σ · kt +pK(t)pK(t)

· kt (6.1)

= kt+1 − (1− σ +pK(t)pK(t)

) · kt + pK(t) · It.

We consider capital depreciation (σ) as well as the volatility of the capital goods prices(pK(t)) since both these elements affect the company’s investment decisions;

• Profits are proportionally increasing in company’s stock k(t). They are also decreasingin the industry-wide capital stock K(t). Thus, they take the form of Π(K(t)) · k(t),where Π′(K(t)) < 0.

Consequently, firms profits take the form of Πt(Kt) ·kt−pK(t) · It− c(It). Thus we obtainan optimization problem:

Π =∑

t

1(1 + rt)t

· [Π(Kt) · kt − pK(t) · It − c(It)] (6.2)

subject to : kt+1 = kt+1 − (1− σ +pK(t)pK(t)

) · kt + pK(t) · It

This condition follows from the contention that in the optimum there is no necessity to investany more. Since for each period kt+1 = kt + It − δkt, there is infinitely many constraints,which transforms the optimization problem to:

L =∞∑

t=0

1(1 + rt)t

· [Π(Kt) · kt − pK(t) · It − c(It)] (6.3)

+∞∑

t=0

λt(kt+1 − kt + pK(t) · It − σkt),

where λt represents the shadow value of capital at each moment in time or, put differently,the marginal value of relaxing the constraint.

If we introduce qt defined as:

qt = (1 + rt)t · λt

we obtain a variable of particular interest. From the definition it follows that qt captures thevalue of an additional unit of investment at time t+1 perceived at time t. On the other hand,it is also valuable from the investor’s point of view as it defines the value of the additional


unit of investment if managed by this particular company. Incorporating this variable intothe optimisation problem we obtain:

L =∞∑

t=0

1(1 + rt)t

·[kt ·

(Π(Kt)− qt(1 + σ)

)+ qt · kt+1 − It ·

(pK(t) · (1− qt)

)− c(It)

].(6.4)

We need to impose an additional constraint to avoid exploding equilibria - the limitsolution where it is still optimal to invest would introduce serious interpretational problemsnot to mention formal complications. The transversality condition can be written down as:

limt→∞

1(1 + r)t

· qt · kt = 0

Maximising equation (6.4) one obtains the optimum investment rates and consequentlythe optimum stock of capital as well. More precisely this yields the following first orderconditions:

FOC (1) :δL

δIt=

1(1 + rt)t

·(− pK(t)− c′(It) + qtpK(t)

)(6.5)

FOC (2) :δL

δkt=

1(1 + rt)t

·(

Π(Kt)− qt(1 + σ))− 1

(1 + rt−1)t−1· qt−1. (6.6)

In addition, δ2cδI2

t> 0 is the sufficient condition to satisfy the requirement of non-positive

second order derivatives. Thus, equations (6.5)-(6.6) yield the solutions:

qt = 1 +c′(It)pK(t)

(6.7)

Π(Kt) = (1 + rt) · (qt −∆qt)− (1− σ)qt. (6.8)

Please, note that these solutions are equivalent to Romer (2001) with the additional assump-tions of pK(t) normalised to unity and total depreciation (σ = 1).

One may easily observe that (6.7) reads as a sum of marginal future revenues of capitaldiscounted to today. Thus, Tobin’s Q sumarises all the information about the future thatshould be relevant to firms decision. More specifically, that is the present value of profitsgenerated by an additional unit of capital and the market value of this investment. Similarly,(6.7) gives the marginal replacement cost. Alternatively one can state it captures the marketevaluation of a unit of capital if managed by this particular company.

Thus one can specify a rule of thumb for a ’good investment opportunity’. If Tobin’s Qexceeds unity, investment should be implemented, whereas in all other cases despite possibleprofitability of this opportunity, investment should be forgone and the remaining cash flowsshould be ’returned’ to the owner in the form of dividend. Albeit a strong statement, it isimplied directly by the theory and signifies the nature of the investment decision in the lightof property rights literature.

The principal-agent model describes well a situation, where principal’s revenue rests inthe hands of agent. More specifically, their combined income crucially depends on the effortby agent. Although this model is useful to describe the nature of the relation between theowners (principal) and managers (agents), one seems to be missing an important part of the


picture. If the managerial discretion problem was a simple agency theoretic case, it wouldsuffice to offer managers the property rights (reward them with shares).

However, when property rights and decision rights are separated many more complexproblems come to play. As pointed by organisational economists (cfr. Baker, Gibbons andMurphy 2004), designing a contract in which owner’s rights are well protected is extremelydifficult if not impossible. This problem has far reaching consequences, as forcefully demon-strated by Malmendier and Tate (2004) - overconfident managers tend to overestimate thereturns to their investment projects, while asserting external financing as excessively costly.

It is the direct implication of the Q theory of investment that if managers are unableto find good investment projects they are obliged to distribute the cash flows among theowners. A good investment is specified as a one, for which Tobin’s q exceeds 1. Therefore,a project might be good for one company at the same time being not good enough foranother. However, any so-called long-term strategic decisions of retaining the funds for futureinvestment opportunities, low added value diversification investment or prestigious spendingincreasing the reputation of a company should and must be perceived as actions against theshareholder, that is the owner.

Within corporations, the investment decisions are usually based on IRR or NPV analyses.As pointed in any corporate finance textbook both techniques should yield the same orderingof projects evaluation (cfr. Breadley and Myers 1994, pp. 284). However, a crucial elementin each analysis is the rate of return applied in the evaluation process. The owner orientedCAPM would yield different outcomes than the firm oriented WACC. With all the theoreticaland empirical limitations on using the first, the latter largely underestimates the owner’s costof capital.

6.1.2 Marginal versus average Tobin’s q

The Q theory of investment focuses on marginal values, while these are usually unob-tainable in reality. Therefore, average replacement costs are used, although this solution hascertain consequences. Namely, in general average Q is higher than marginal. Assuming thatthe adjustment costs are only a function of investment we implicitly supposed decreasingreturns to scale in the adjustment process. Hayashi (1982) has shown that if constant returnsto scale are imposed, average and marginal Q are equivalent. However, downward slopingdemand curve for the firm’s final products inevitably causes δΠ2

δ2K < 0, thus leading to theresult indicated above.

On the other hand, also the opposite is possible, i.e average Q smaller than the marginalone. One can expect this result in the cases where a significant amount of capital becomesoutmoded due to a technical advancement for example. Although the average replacementcost is relatively high, the marginal productivity growth may be even higher as each additionalunit of new capital significantly raises the future profits.

Apart from the problems concerning the relation between marginal and average Q thereis also another drawback of using the latter, namely that the average Q no longer has its rule-of-thumb interpretation. Thus, one cannot look at the stock-listed companies, calculate their


market-to-book-value ratios and unequivocally determine which of these companies shouldinvest at all.

6.2 Literature review

The claim of Miller and Modigliani can be falsified on real-world data by simply correlat-ing cash flows and investment spending across companies - obtaining a significant correlationcoefficient suggests that for reasons neglected in their model these two variables are interde-pendent.

However, neither cash flow nor investment spending capture the future outlooks of acompany, thus leaving aside the aspect suggested by the Q theory. On the other hand, as in-dicated above, one cannot apply the rule-of-thumb implication in reality due to unavailabilityof marginal Q.

James Tobin (1969) introduced the ratio of the market value of a firm to the replacementcost of its capital stock—a ratio that he called Q — to measure the incentive to investin capital. Tobin’s Q, as it has become known, is the empirical implementation of Keynes’s(1936) notion that capital investment becomes more attractive as the value of capital increasesrelative to the cost of acquiring the capital. However, neither Keynes nor Tobin provided aformal theoretic analysis underlying the Q theory of investment. Lucas and Prescott (1971)developed a rigorous analysis of the capital investment decision in the presence of convex costsof adjustment, and observed that the market value of capital can be an important element ofthe capital investment decision, though they did not explicitly make the link to Tobin’s Q.

The link between convex costs of adjustment and the Q theory of investment was madeexplicitly by Mussa (1977) in a deterministic framework and by Abel (1983) in a stochasticframework, though the papers based on convex adjustment costs focused on marginal Q—theratio of the value of an additional unit of capital to its acquisition cost—rather than theconcept of average Q introduced by Tobin. Hayashi (1982) bridged the gap between theconcept of marginal Q dictated by the models based on convex adjustment costs and theconcept of average Q, which is readily observable, by providing conditions, in a deterministicframework, under which marginal Q and average Q are equal. Abel and Eberly (1994)extended Hayashi’s analysis to the stochastic case and also analyzed the relationship betweenaverage Q and marginal Q in some special situations in which these two variables are notequal.

On the side of costs of external financing, Stiglitz, Weiss Stiglitz and Weiss (1981) andMyers and Majluf (1984) developed formal models of moral hazard problems in debt andequity markets. Namely, costly state verification models were constructed on the fundamentof information asymmetry and adverse selection literature. The agency costs between theshareholders and the R&D management, i.e. risk-adverse R&D managers will lead to under-investment in risky R&D projects and managers tend to spend on activities that benefitthem. These costs can be avoided by leveraging the firm. However, the costs of the externalfunds to finance the R&D projects will be higher (Jensen and Meckling, 1976).

Fazzarri, Hubbard and Petersen (1988) suggest a way to pursue empirical research in this


direction. Specifically, they propose the following empirical specification:

(I/K)i,t = f(X)i,t + g(CF/K)i,t + ui,t,

where f(X)i,t signifies the firm and time specific control variables, like size, age, dividendhistory as well as average Q. Applying the split sample technique and selecting few types ofcompanies potentially financially constrained they have reached the following conclusions:

• young firms embody significant unknown information, inducing high costs of verifica-tion, therefore these companies tend to be more financially constrained,

• firms owing tangible assets enjoy easier access to capital than the R&D intensive ones,

• for high dividend companies cash flow coefficient is two times lower than for low dividendcompanies (0,23 US cents versus 0,46 US cents per USD invested); this difference ishighly statistically significant.

They have interpreted this result as a confirmation of the importance of the liquidity con-straints in the investment processes.

In a similar research Himmelberg and Petersen (1994) analysed the process of R&Dinvestment among the US enterprises. They show both empirically and formally that theexcess cost of external finance causes some firms to be liquidity-constrained. This yieldsthe obvious conclusion that cash flows become important determinants of the investmentprocesses.

6.2.1 The hypothesis of managerial discretion

The seminal paper by Jensen (1986) suggested that the results obtained applying themethodology outlined above might suffer from the omitted variable problem, namely over-looking the importance of the corporate governance within the company. Vogt (1994) pro-vided the theoretical foundations for the empirical specification of a model allowing to controlfor both financial constraints and agency problems. His proposed specification includes aninteraction term and can be represented by:

(I/K)i,t = f(X)i,t + g(CF/K)i,t + h(CF/K ·Q) + ui,t,

The rationale behind this specification is the following. The financial constraints resultin under-investment, while corporate governance has the effect of over-investing managers.Therefore, for high Q companies investment should be high, and thus highly dependent uponthe funds available. Consequently, a positive coefficient of CF/K ·Q is expected in supportof the financial constraint hypothesis. Alternatively, for low Q companies, there is no reasonto maintain high investment spending. Hence, larger dependence on the cash flows wouldsuggest managerial discretion leading to a negative sign on the CF/K ·Q coefficient.

To put it differently, a positive sign of its coefficient implies that firms with a higherTobin’s Q embody a higher cash-flow coefficient. This compares with higher liquidity con-straints which is in line with the asymmetric-information problem. A negative coefficient


is in line with managerial discretion, as the cash-flow-coefficient for lower Q-firms becomeshigher.

Other studies have replicated and extended this approach. For example, Devereux andSchiantarelli (1990) test for a set of UK firms whether different cash-flow-investment sensi-tivities are found in subsamples based on proxies for agency costs of external capital. Theproxies are firm size (capital stock and employees), the number of years since initial quota-tion, and the industry (growing or declining). The investments of large firms, newly listedfirms and firms in growth sectors exhibit higher cash flow sensitivities.

Oliner and Rudebusch (1992) interact the cash flow coefficient in an investment regressionmodel with proxies for information asymmetry (firm age, listing at exchange, and stock tradesby insiders), agency costs (insider shareholding and ownership concentration) and transactioncosts (firm size). The authors also include the dividend yield for comparison with Fazzari,Hubbard and Petersen (1988). Although for the set of US firms the individual interactionterms are insignificant, a compound measure of information asymmetry is significant andyields the predicted positive effect. The authors conclude that information problems worsenfinancial constraints.

For Canadian firms, Chirinko and Schaller (1995) define subsamples based on age (yearsof inclusion in a financial database), concentration of ownership, industry (manufacturingand other), and group or independent. The cash flow constraints are most relevant for youngfirms, firms with dispersed ownership, independent firms and manufacturers. Gilchrist andHimmelberg (1995) investigate US firms and define subsamples on the basis of size, dividendpayout ratio and the availability of rating for bonds and commercial papers.

Hoshi, Kashyap and Scharfstein (1991) investigate the cash flow sensitivities for a sam-ple of Japanese firms, which is divided into group and non-group firms. The latter ones,characterized by relatively weak ties with banks, have a higher cash flow coefficient. Hoshi,Kashyap and Scharfstein (1991) spot the importance of overinvestment through a differentialimpact of cash flow for firms with good and bad prospects. The latter distinction is madeby focussing on firms with a Tobin’s Q above and below median respectively. They find noevidence for over-investment.

Hubbard, Kashyap and Whited (1995) also suggest that observed links between invest-ment spending and internal funds may reflect managers’ over-investment. Following Jensen(1986), over-investment is expected to be relevant in mature industries. Based on profitabil-ity, 39 four-digit S.I.C. industries are defined as mature industries. Using Euler equations,it does not appear that agency costs are important for business fixed investment in thesemature US firms.

Hadlock (1998) studies the impact of insider ownership on the cash flow sensitivity ofinvestment based on both free cash flow problems and asymmetric-information problems. Aninteraction term of cash flow and insider ownership is found to be positive for insider owner-ship below 5% and negative above this threshold. Hadlock (1998) concludes that the findingsare inconsistent with the free cash flow theory and consistent with asymmetric informationproblems.

Although, Anglo-Saxon countries remain in the centre of interest for their capital market


orientation, continental system with banks as main capital providers has been analysed inthis framework as well. The European continent offers an interesting setting to considerdeterminants of investment under alternative structures. Kadapakkam, Kumar and Riddick(1998) study six OECD countries including France and Germany, next to the US, UK, Canadaand Japan. Subsamples based on size show that the cash flow investment sensitivity ishighest in the sample of large firms. This difference is most pronounced in the US andUK. France, Germany and Canada also show significant differences between the subsamplesin most analyses. For Japan, the difference is insignificant in several analyses. Firm sizehowever is only one a priori criterion that may be important in explaining cash flow investmentsensitivity.

Gugler (1998) analyses Austrian investment spending and corporate governance. He em-pirically investigates whether the validity of the asymmetric information problem and man-agerial discretion problem depends on the ownership structure of the firms. His findingssuggest that investment of bank controlled firms is not positively related to cash flow. Asym-metric information problems prevail in family-owned firms, while over-investment is moreprominent in state controlled firms and pyramidal groups.

Haid and Weigand (1998) focus on investment spending and corporate governance inGermany. Using sample splits, they report that liquidity positively affects investments inowner-controlled firms, while management controlled firms show no cash flow investmentdependency.

Audretsch and Elston (2002) investigate the relation between firm size and liquidity con-straints in German listed firms. They find that medium sized firms appear to be moreliquidity constrained than either the smallest or largest firms. An interesting aspect of thisstudy is that the authors get around one of the criticisms of Kaplan and Zingales (1997). Inparticular, the subsamples are not based on liquidity constraints, but plainly on size in orderto investigate constraints for large, medium-sized and small firms.

Van Ees and Garretsen (1994) study a sample of Dutch firms over the period 1984-1990.The authors define subsamples based on the dividend payout ratio, the year of the initialpublic listing, size (fixed assets) and interlocking directorates with banks. An interlockingdirectorate with a bank occurs when a managerial board member of an industrial firm holdsa position on the managerial or supervisory board of a bank or when a managerial boardmember of a bank holds a position on the managerial or supervisory board of an industrialfirm. They find that the cash-flow-investment sensitivity is significantly positive in Dutchfirms. No significant differences are found between subsamples based on dividends, yearslisted and size. Interlocks with banks are found to reduce the cash flow constraints. Firmswith ties to banks have a significantly lower impact of cash-flow on investment. Van Ees andGarretsen (1994) conclude that bank relations reduce the asymmetric information problemin Dutch firms.

Finally, de Jong and Degryse (2002) approach the Dutch data on the time span 1993-1998. Interestingly, they are unable to use Vogt (1994) methodology as the interaction termhe suggested turns out insignificant in all specifications. Thus, they proceed with samplesplit methodology. Although they do not confirm the Van Ees and Garretsen (1994) article


about the benefits of bank involvement, they find higher cash flow sensitivity of investment infirms with low investment opportunities, which they interpret as an indication of managerialdiscretion problems. Specific to the Netherlands, firms with low shareholder influence posit ahigher cash flow investment sensitivity. The relevance of asymmetric information is confirmedas smaller firms and firms from information sensitive industries show a larger cash flowinvestment sensitivity.

6.2.2 Some controversies over this approach

The approach that was initiated by Fazzari, Hubbard and Petersen (1988) and laterdeveloped by Vogt (1994) and that is also used in the above mentioned studies has beencriticized by Kaplan and Zingales (1997). They argue that, when examining in greater detailthe data used by Fazzari, Hubbard and Petersen (1988), their results do not support thepresence of liquidity constraints. Kaplan and Zingales (1997) argue that the apparentlyfinancially constrained firms could have augmented their use of cash and lines of credit at aparticular moment in time.

The discussion on the usefulness of cash flow investment sensitivities is continued in Faz-zari, Hubbard and Petersen (2000) and Kaplan and Zingales (2000a, 2000b). Kaplan andZingales (1997, 2000) show that the approach of defining subsamples based on an approxi-mation of liquidity constraints, as advocated by Fazzari, Hubbard and Petersen, should beinterpreted with caution. An additional reason for caution is discussed in recent studies byErickson and Whited (2000) and Gomes (2001). The authors use simulation techniques andargue that measurement error in Q influences the cash flow investment sensitivity.

In particular, Kaplan and Zingales (1997) argue that ”theoretical and empirical evidenceof greater sensitivity of investment to cash flows is not a reliable measure of the differentialbetween internal and external financing cost”. To support this statement they have developeda one-period theoretical model where they prove that what really measures the wedge betweenexternal and internal financing cost is the relation of third order derivatives of return oninvestment and cost of financing functions with respect to wedge and internal resources.However, in their terminology, what the empirical and theoretical research does is the focuson the differences in the second order derivatives. More precisely, the indications of financialconstraints are not a sufficient ground to claim the wedge between the cost of internal andexternal financing.

The reply of Fazzarri, Hubbard and Petersen (2000) reapproached the issue of the degreeof financial constraints - as much as they agree with Kaplan and Zingales (1997), they claimthat they were more interested in a 0-1 distinction between financially constrained and un-constrained companies than the exact distribution of this variable. Hence, although Kaplanand Zingales (1997) might be right in suggesting that in general the approach used is notsufficient in economic terms, empirically this methodology is rather efficient. Both counter-parties admit that theoretically one is unable to judge on this matter. Thus the subsequentempirical polemic focused on the empirical results.

Kaplan and Zingales (1997) analysed the Fazzari, Hubbard and Petersen (1988) companies


on the same time span and attempted to verify whether the financial constrain identificationindicated by them holds if other criteria are applied. Namely, they used managerial reportsas a source of information - the statement of being unable to undertake an investment dueto insufficient funding was interpreted as a financial constraint. Using this source they haveclassified the companies into five groups (from non-constrained to fully constrained). Onlyabout 22% of companies were found to be facing a liquidity constraint. They have repeatedthe Fazzari, Hubbard and Petersen (1997) analysis on the subsamples of companies and theresults turned out to be opposite to the their findings - low dividend companies have low cashflow sensitivity when compared to high dividend companies across the samples.

The reply from Fazzari, Hubbard and Petersen (2000) underlined the low reliability ofmanagerial reports (one only needs to report if one cannot undertake the previously acceptedinvestments). Therefore, companies classified by Kaplan and Zingales (1997) are de factofinancially distressed, which effectively rules out any investment projects. They defend alsotheir results on the intuitive grounds suggesting that even if the applied methodology is notperfect, on a large sample it allows to discriminate between the investment patterns thusproviding grounds to derived conclusions.

The immediate reply from Kaplan and Zingales (2000) focused on some examples of com-panies that Fazzari, Hubbard and Petersen (2000) methodology would classify as financiallyconstrained. These include, among others, Microsoft and Hewlett-Pacard - companies onewould rather not consider facing liquidity problems. Furthermore, Cleary (1999)finds thatKaplan and Zingales (2000) results are not susceptible to company selection - he repeats theirresearch on the entire sample of US stock-listed companies (Kaplan and Zingales only focuson 50 preselected examples) and obtains the same results.

6.2.3 Other approaches to cash flow sensitivity of investment

One should be cautious about interpreting this polemic. Kaplan and Zingales do notundermine the finding, that some companies might be suffering from the managerial discretionproblem. What they do in turn is to falsify the methodology of obtaining this findings onaggregate data. After this discussion between Kaplan and Zingales on one hand and Fazzari,Hubbard and Petersen on the other, extensive literature has emerged, where authors applydifferent techniques in the search for a proper discrimination technique.

For example, Van Cayselee (2002) uses the signalling model to analyse Belgian smalland medium enterprises. The taxing regulation in Belgium gives companies the opportunityto choose between immediately subtracting the R&D investment spending from revenuesor distributing this redemption over five consecutive years. He suggests that if a companydecides to account for the spending immediately in the same taxing year, this indicatesthat the investment is not perceived as particularly profitable by the tax payer. Otherwise,enterprise would like to retain the ability to inflate costs in the accounting understanding inthe consecutive years, thus lowering the effective tax rate on the profits from this investment.This model allows Van Cayselee to escape the Tobin’s Q approach and use the taxing choicesas a signal of good investment opportunities. He analyses a panel of 889 Belgian companies


on the time span of five years (with the average age of 2.5 years). His results largely confirmthe managerial discretion hypothesis. However, some results indicate that when the managersand shareholders interests coincide, investments are less sensitive to cash flows.

Also Cincera (2002) focuses on Belgian firms. Using GMM and ECM he attempts toanalyse the impact of financing constraints on both capital and R&D investment decisions.In particular, the extent to which these constraints differ across firms is investigated fromdifferent perspectives, e.g. industry sectors, firms’ size and age, regions, domestic firmsversus subsidiaries of foreign groups, quoted versus unquoted firms on the stock market.On the sample of 11 000 companies in the manufacturing sector over 1999-2001 he findsstrong evidence of a positive impact of cash flow effects on the firms investment decisions.Unfortunately, he is not able to disentangle this effect with respect to its causes.

Konings, Rozov and Vandebussche (2002) use the panel of more than 4,000 manufac-turing firms consisting of comparable data for Poland, the Czech Republic, Bulgaria andRomania between 1994-1999. They find firms in Poland and the Czech Republic to be creditconstrained, suggesting that capital markets are not functioning properly. In contrast firmsin Bulgaria and Romania seem far less dependent on internal financing to invest. They in-terpret this result as evidence of stronger persistence of soft budget constraints in the groupof slowly reforming transition countries. They too fail to discriminate between asymmetricinformation and managerial discretion as causes to the investment cash flow sensitivity.

Malmendier and Tate (2004) apply the tools of social psychology to analyse if the con-fidence of CEOs may have any impact on the investment strategies of companies. Theyforcefully demonstrate that overconfident managers tend to over-invest, regardless of the in-centive schemes and companies managed by them are significantly more responsive to cashflows in their investment decisions. Thus, by choosing a confident CEO shareholders shouldbe aware that this might aggravate the agency problems despite possible incentives.

Barkley, Foley and Wurgler (2004) approach this problem in the international context. Weoutline and test two mispricing-based theories of FDI. The ”cheap assets” or fire-sale theoryviews FDI inflows as the purchase of undervalued host country assets, while the ”cheapcapital” theory views FDI outflows as a natural use of the relatively low cost capital availableto overvalued firms in the source country. The empirical results support the cheap capitalview: FDI flows are unrelated to host country stock market valuations, as measured by theaggregate market-to-book-value ratio, but are strongly positively related to source countryvaluations and negatively related to future source country stock returns. The latter effectsare most pronounced in the presence of capital account restrictions, suggesting that suchrestrictions limit cross-country arbitrage and thereby increase the potential for mispricing-driven FDI. This is strong though indirect evidence in support of the managerial discretionhypothesis.

On the theoretical grounds, Abel (2002) develops a theoretical model of Q theory ofinvestment without any adjustment costs but with monopolistic power instead. He demon-strates that Tobin’s Q exceeds one, even without any adjustment costs, for a firm that earnsrents from monopoly power. Even when there are no adjustment costs and marginal Q isalways equal to one, Tobin’s Q is informative about the firm’s growth prospects. He devel-


ops a model to show that investment is positively related to Tobin’s Q (which is observableaverage Q). This effect can be quantitatively small, which has been taken as evidence of veryhigh adjustment costs in the empirical literature, but here is consistent with no adjustmentcosts at all. In addition, cash flow has a positive effect on investment, and this effect islarger for smaller, faster growing and more volatile firms, even though capital markets areperfect. These results provide a new theoretical foundation for Q theory and also cast doubton evidence of financing constraints based on cash flow effects on investment.

However, Abel (2002) findings are largely dependent upon the parameters of companiesanalysed. Although testing his model empirically would be rather difficult due to mea-surement and data availability problems, it does not seem that reality would support thisspecification.

6.3 The Model

Basing on the literature reviewed above, approaching this kind of research one shouldcontrol for age of the company, its size, the industry as well as bank involvement and po-tentially foreign ownership. However, we are not aware of any research in this respect forPoland2. Furthermore, most of the above listed control variables have limited applicabilityfor the Polish data.

Firstly, Warsaw Stock Exchange was only founded in 1994 with initially less than twentycompanies quoted. Obviously, it is impossible to use the dummy for companies younger thanten years. Using the age variable on the other hand would essentially result in a fixed effectestimation.

Secondly, three types of companies dominate on the market. There are relatively few bigenterprises contributing to the WIG20 index of twenty biggest market participants. Theirhistory on the market is highly differentiated as IPO was one of the privatisation solutions -some of them entered the market as early as in 1994, while three went public only in 2003.There is also an average sized majority, which comprises both static big privatised enterprisesand dynamic private companies entering the stock exchange in search for investment funds.Lastly, there is also a minority of these who recently arrived to the market and whose rolein trading is largely marginalised. This typology is rather stable over time (less than 5% ofcompanies on average change the group from year to year). Introducing company size variablewould have the same effect as described above with the additional drawback of scaling theresults.

Finally, there are many industries which are represented on the market by only onecompany, while only in three sectors there are more than 4 enterprises. Additionally, dataon bank involvement are not available. As regards the ownership, Polish law obliges theowners to inform publicly about the number of their shares only if they possess more than5% of stocks. Thus, ownership data are not reliable to the extent that they significantlyunder-represent the reality.

2 Hussain and Nivorozhkin (1997) provide the only analysis of the capital structure of the Polish stock

exchange we are aware of. Unfortunately, this paper dates back to 1997 and has largely descriptive character.


However, using the rationale supporting the use of each of these control variables weattempted introducing other measures and indicators that could help to address the problemof size, age and ownership. More specifically, we introduced a proxy for a size as well as foreignownership. What seemed particularly worth exploring, however, was the state participationin the stock market.

As indicated earlier, often IPO was a manner of privatisation. However, often state didnot sell all its shares, retaining either privileged stocks (more voting rights) or effectivelycontrol over the company (via concentrated ownership). The problem we approach here canbe rephrased as a question of how well can the owner protect his property rights, i.e. towhat extent can the managers abuse their information and decision advantage. In the case ofconcentrated ownership and decisive voting rights, as is often the case with the state partici-pation, one should ask whether these companies are troubled less by the agency problem. Ifthis does not prove to be the case, one should ask if this is not a persistent quandary resultingfrom the legal design. These questions laid the fundament for the presented results.

6.3.1 Methodology and data

We applied panel data technique to the data set of 181 non-financial companies listedon the Warsaw Stock Exchange in the years 1995-2003. Data were taken from the annualreports of the Warsaw Stock Exchange Commission (accounting reports confirmed by anexternal auditor). Unreported variables were calculated based on the data available. Thefollowing equation was under scrutiny:

Investmenti,t = α+ βControlV ariablesi,t

+ γCashF lowi,t + δTobinsQi,t + ζCashF lowQi,t + εi,t (6.9)

As indicated above, the time series are very short, which influences the quality of theresults. Theoretically, our preferred specification would be the one with a company as agrouping variable. However, either we obtain groups with 2.6 observations on average or weare forced to limit the scope of the research to companies that have a longer history on themarket. Any choice here would be highly arbitrary. We chose five years of history, whichsignificantly narrowed the sample - we based this choice on the prerequisite that with at theminimum of five predicting variables shorter series have relatively poor statistical properties.

One should note that data have sometimes inconsistencies, while it is also likely that thesediscrepancies are not fully idiosyncratic. Missing and inconsistent data occur persistently forsmaller companies, while for few larger they suggest purposeful action from the side of thereporting party (three of the identified nine were accused of misinforming the Warsaw StockExchange Commission and eventually were forced to pay the fine).

The companies vary largely also as regards the character of presence - the main or theparallel market. This difference is important as the requirements to enter the parallel marketare significantly lower than for the main market. Thus, typically, larger and more stablecompanies are listed on the main market. To account for this effect we incorporated adummy, which takes the value of 1 for the main market and 0 otherwise.


Investment is our dependent variable. It is defined as a ratio of each years investmentspending over total assets. Investment spending contains both fixed assets purchases and inintangibles. We used two specifications of this variable, i.e. the logarithm of the ratio as wellas a version standardised to (0,1) distribution. The latter is used in specifications where allnon-dummy variables were standardised - this allows to compare the strength of the impactamong the predicting variables.

To measure the market evaluation of the companies investment opportunities (TobinQ)we applied a standard measure of market-to-book-value ratio (MBV). We used market cap-italisation as the market value (number of shares in trading times their price at the end ofeach year). The book value was reported in annual accounting statements as the value ofassets at the end of each year.

For obvious reasons the measure of cash flows has to enter the estimated equation (CashF low).This variable too has been scaled by the total assets to eliminate the potential heterogeneitydue to the size effects. To control for the size and potentially age effects, we used a proxydummy Market taking the value of 1 if a company is quoted on the main market and 0otherwise.

We further included a variable suggested by Vogt (1994) allowing to discriminate be-tween the two potential causes for cash flow investment sensitivity. The interaction variableCashF lowQ will have a negative coefficient in the case of managerial discretion, while itshould remain positive for financially constrained companies.

Two other variables typically have explanatory power in Q theory of investment equations,first being the change in working capital (working capital investment). As for the previousvariables we scaled it by the replacement value of fixed assets (∆NWC). The motivationfor this variable is that firms may reduce their working capital (current assets minus currentliabilities) to smooth fixed investments (Fazzari and Petersen, 1993). Controlling for ∆NWC

allows to isolate the liquidity effect from the informational part of cash-flow (cfr. Haid andWeigand, 1998). In other words, a negative coefficient accompanied by an increase in thecash-flow-coefficient suggests that cash-flow does not capture investment opportunities. Notethat working-capital investment may be endogenous, as it is a decision variable of the firm.Thus, we instrument the net working capital ratio with the previous period net workingcapital scaled by the adequate total assets (2SLS as in Fazzari and Petersen, 1993).

The second variable with explanatory power in Q equations is current sales adjusted forthe size effects (the ratio of sales to total assets). This variable too has been instrumentedby its lagged value to take care of potential endogeneity problems.

To control for the presence of state represenatives among the shareholders we introduced aState dummy variable taking the value of 1 if the state - directly or indirectly - may be foundamong the shareholders (even below the 5% level). These data are available from the Ministryof Economy annual reports. An actual share of owned stocks would have provided a muchmore refined measure. It would introduce however a strong limitation of direct ownershipand underestimate the impact of privileged shares (in terms of voting or additional dividendrights).

Following the Vogt (1994) logic we have also constructed State Q and State Q CF vari-


ables in order to capture the peculiarities of state owned firms. The intuition implies thata negative coefficient associated with this variable should signify that managerial discretionproblems are more severe in the state owned companies.

Recognising the importance of environment stability, as well as changes in the strategy wedecided to include also a NewCEO dummy, as a proxy for changes within the organisation. Achange among the board members indicates that shareholders executed their right to influencethe policies adopted by their company - this is essentially the only instrument of control theyenjoy in the case of dispersed ownership.

Finally, to account for the corporate governance and corporate culture within a companywe decided to include another proxy dummy, namely the ForeignCEO. Acknowledging theyoung age of Warsaw Stock Exchange and little experience of Polish CEOs with shareholders’relations management we assumed that perhaps presence of a foreigner in the board mightsuggest more experience as well as higher consciousness as to the obligations towards theowners.

6.3.2 The Empirical Results

The panel data technique, while giving the chance to explore the properties of data setsotherwise unsuited for econometric analysis, requires specifying a grouping variable. In thecase of this research company should be a natural grouping variable - we are interested intracing the patterns of companies’ behaviour and not the changes in these patterns acrossthe stock exchange from year to year.

Unfortunately, due to data availability constraints such a grouping strategy results inhaving approximately three observations per each group (with the minimum of 1 and maxi-mum of 5), while approximately 160 groups. Such a data set has poor statistical propertiesand thus quality, irrespectively of their economic interpretation. In the below tables bothgroupings are reported and analysed.

Column (1) of both tables reports simple panel data GLS estimation with random effectssupported by the Hausmann test. Column (2) reports the same methodology, but all non-dummy series were standardised before performing the estimation. This serves the purpose ofcomparing the magnitude of influence of particular variables, as with all series standardisedto (0,1) distribution estimators can be directly compared. Columns (3) and (4) are similar tocolumns (1) and (2) with the main difference that ∆NWC has been instrumented with theuse of its lag. Thus, they contain 2SLS estimations to control for the potential endogeneityof net working capital.

As is visible in Table 6.1. there are reasonable grounds to believe in the managerialdiscretion hypothesis. Not only is CashF lows Q variable significant, but also bears a neg-ative sign. Although, as suggested above, some claim that this method is not errorless indiscriminating financially constrained companies against the affluent ones, the Vogt (1994)methodology seems to be robust to many possible data shortcomings. Furthermore, in ourstudy it is highly significant, which cannot be attributed to any other characteristic of the


data3. Also, the strength of the managerial discretion is found extremely high. This is byfar the most important determinant of the investment decisions comparable to the combinedimpact of sales and Tobin’s Q.

Table 6.1. Dependent variable: investment (year as a grouping variable)

Independent Raw Standardised Raw (2SLS) Standardised (2SLS)variables (1) (2) (3) (4)

Q 0,11 *** 0,29 *** 0,10 *** 0,31 ***Sales 0,07 *** 0,15 *** 0,09*** 0,14 ***CashF lows 0,53 *** 0,15 *** 0,48 *** 0,12 ***CashF lows Q -0,89 *** -0,42 *** -0,92 *** -0,46 ***∆NWC -0,14*** -0,06*** -0,15*** -0,05***State 0,13*** 0,39*** 0,14*** 0,41***State Q -0,11** -0,13* -0,13* -0,17*NewCEO -0,03 -0,04* -0,02* -0,02*ForeignCEO -0,12 ** -0,21 * -0,08 * -0,08*Market -0,12 ** -0,21 * -0,08 * -0,08*

R2 within 0,21 0,23 0,25 0,24No. of observations 640 640 640 640No. of groups 7 7 7 7

Note: Panel data GLS estimations with random effects. Company and year dummies not reported.

All data in columns (2) and (4) were standardised (except for dummies).

***, ** and * denote significance levels of 1%, 5% and 10% respectively.

The State and State Q variables perform remarkably well. Companies with partly stateownership tend to invest more. Unfortunately, in most cases these spendings can be classifiedas over-investment. The estimators showing the magnitude of influence seem to corroboratethe hypothesis that state owned companies are strongly troubled by the managerial discretionproblem. Unfortunately, State Q CashF lows variable has not proven significant on anyreasonable level, so we are unable to ascertain whether managerial discretion is strongeramong the state-owned companies.

Also the behaviour of our corporate governance indicator (ForeignCEO) seems to sup-port the argumentation presented earlier. Companies with foreign board members seem tobe more cautious about investment opportunities and tend to transmit dividends more oftenthan companies with solely Polish board members(a positive correlation coefficient of 0,8).There can be two types of explanations to this observation. Firstly, foreign owners tend toextract profits trying to realise returns to their investments while local owners are more con-cerned with development. Secondly, local businessmen are able to extract rents using othermethods (contracting, transfer pricing, etc.), while foreigners have limited access to theseways of realising returns due to smaller scale of operations in Poland. Statistically, we werenot able to find support to any of them.

3 For example, in De Jong and Degryse (2000) this variable was not significant which has forced them to

find another filter separating the financially constrained companies from the rest.


DO DOKONCZENIA.

Table 6.2. Dependent variable: investment (company as a grouping variable)

Independent Raw Standardised Raw (2SLS) Standardised (2SLS)variables (1) (2) (3) (4)

Q 0,18 *** 0,50 *** 0,10 *** 0,31 ***Sales 0,006 *** 0,14 *** 0,09*** 0,14 ***CashF lows 0,45 *** 0,15 *** 0,48 *** 0,12 ***CashF lows Q -0,87 *** -0,77 *** -0,92 *** 0,46 ***∆NWC -0,14*** -0,06*** -0,15*** -0,05***State 0,21*** 0,76*** 0,14*** 0,41***State Q -0,10** -0,07 -0,13* -0,17*NewCEO -0,03** -0,12** -0,02* -0,02*ForeignCEO -0,007 -0,02 -0,08 * -0,08*Market -0,12 ** -0,21 * -0,08 * -0,08*

R2 within 0,39 0,40 0,25 0,24No. of observations 640 640 640 640No. of groups 158 158 158 158

Note: Panel data GLS estimations with random effects. Company and year dummies not reported.

All data in columns (2) and (4) were standardised (except for dummies).

***, ** and * denote significance levels of 1%, 5% and 10% respectively.

DO DOKONCZENIA!Sprawdzenie z tym indeksem ’dobra spolka’.

6.4 Possible refinements

The overconfidence-based explanation for investment distortions has a number of novelpolicy implications. Traditional theories, which link investment-cash flow sensitivity to capitalmarket imperfections or misaligned incentives, propose timely disclosure of corporate accountsor high-powered incentives as potential remedies. Our findings suggest that these provisionsmay not suffice to address managerial discretion. A manager whose incentives are perfectlyaligned and who does not face any informational asymmetries may still invest suboptimallyif he is overconfident. He believes that he is acting in the best interest of shareholders.Thus, refined corporate governance structures, involving a more active board of directors orconstraints on the use of internal funds, may be necessary to achieve first best investmentlevels.

OGOLNIE RZECZ BIORAC, WNIOSKI Z TEJ CZESCI. ORAZ, ZE TAK NAPRAWDEZE STATYSTYCZNEGO PUNKTU WIDZENIA TE WYNIKI NIE MAJA SZCEGOLNEIWYSOKIEJ WARTOSCI.

7. CONCLUSIONS

8. BIBLIOGRAPHY

APPENDIX

A. ENDOGENOUS GROWTH MODEL WITH THE FINANCIAL INSTITUTIONS

There are two types of manufacturing activities: production of the final consumptiongood, the output, and production of the physical units of the capital goods of highest ingiven moment quality1. A third activity, research and development (R&D), designs betterquality for existing intermediate goods.

Some important assumptions concerning these processes are imposed. First, both manu-facturing activities - production of consumption goods and production of intermediate goods- use the same Cobb-Douglas production function. Moreover, the final good Y produced byall firms is physically identical and the aggregate of the output, Y , can be used in a perfectlysubstitutable manner for different purposes. That is to say that the output can be usedfor the production of intermediaries, Xj,kj

, for final consumption, and in the research anddevelopment activity. The last property asserts one-sector production model.

Specialised firms produce each variety of capital goods, and rent it out to the final goodproducers at a rental price Pj,kj

. The demand for each variety of ”quality-adjusted” interme-diate good, Xj,kj

, follows from the optimality condition that equates the rental rate to themarginal productivity. This condition is given by the equation:

Pj,kj=

dY

dXj= αAH(1−α)qkjαXj

α−1

Thus, the demand for the intermediate good of quality kj in sector j is derived to take thefollowing form:

Xj,kj= H(αA)1/(1−α)qkjα/(1−α)P

−1/(1−α)j,kj

The improvement of the quality of intermediate good is costly. In order to motivate anentrepreneur to engage in R&D activity, the successful innovator is granted a patent for thecapital good invented. Therefore, the leading-edge producer acts as a monopolist in thisenvironment. This is to say that they maximise profit function with respect to price levelusing the above demand function:

πj = Xj,kj(Pj,kj

− 1),

where the production cost of an intermediate good of any quality is constant and normalisedto one. Hence, the price of intermediate good takes form of constant mark-up on productioncosts:

∂πj

∂Pj,kj

= 0 ⇒ P = Pj,kj=

1α.

1 Model as in Suda (2004).

A. Endogenous growth model with the financial institutions 82

Thus, the monopoly price of intermediate good is constant over time and sectors.Moreover, we can simplify the demand of intermediate good j to:

Xj,kj= HA1/(1−α)qkjα/(1−α)α2/(1−α) (A.1)

Using these results, the overall production of intermediate inputs, (X), and total out-put, (Y ), can be determined for the whole economy. Using output, capital and productionequations as well as the pricing behaviour of the investors and summing up over the N givensectors, we obtain the following expressions:

Y = AH(1−α)

N∑j=1

qkjαxαj

= HA1/(1−α) · α2α/(1−α) ·N∑

j=1

qkjα/(1−α)

= HA1/(1−α) · α2α/(1−α) ·Q, (A.2)

X =N∑

j=1

Xj,kj= HA1/(1−α) · α2/(1−α) ·Q, (A.3)

where Q is defined as an aggregate quality in all sectors index:

Q =N∑

j=1

qkjα/(1−α) (A.4)

The innovation in the model takes form of increasing the quality of the existing capitalgood, that is of an increase in kj . The kj ’s innovator in sector j raises the quality from qkj−1

to qkj . This innovator is able to set the monopoly price and sell the quantity of intermediategoods given by market equilibrium. The flow of profit per the unit of time associated withthe quality rung kj takes the form of:

πj,kj= Xj

(1− α

α

)= HA1/(1−α) ·

(1− α

α

)· α2/(1−α) · qkjα/(1−α) (A.5)

The inventor of kj quality rung intermediate goods keeps their monopolistic position onlyfor a certain period of time. Once a higher quality intermediate good in sector j is introduced,the previous innovator does not enjoy the flow of profits given by equation (A.5) any longer.Innovations are standardly assumed to arrive randomly with a Poisson arrival rate pj,kj

, whichdepends primarily on the flow of resources to R&D in sector j. Specifically, we assume that:

pj,kj= Zj,kj

· φ(kj), (A.6)

where Zj,kjdenotes the resources effectively employed in research and development process

(R&D) in sector j, and φ(j) describes the productivity of used inputs, with the followingform:

φ(kj) =1ζ· q−(kj+1)α/(1−α), (A.7)

where ζ > 0 represents the costs of the research. Naturally, the higher ζ is the lower theprobability of success in innovation for given values of Zj,kj

and kj2 The second term on the

2 The parameter 1/ζ can also be interpreted as the efficiency of employed resources.


right-hand side of equation (A.7) indicates the negative effect of a project’s complexity on theprobability of success. The higher rung at the quality ladder, represented by kj + 1, requiresmore resources to be devoted to keep the probability, pj,kj

, on the same level.The probability of success in the innovation process determines for how long the monop-

olist enjoys the stream of profit. The present value of the profit from the kjth innovation insector j is given by:

Vj,kj= πj,kj

[1− exp (−rTj,kj)]/r (A.8)

where r denotes the interest rate and Tj,kj= tj,kj+1− tj,kj

is the time interval over which thekjth innovation is in the forefront3 Taking into account the Poisson distribution of innovation,the expected present value of profit computed at time tkj

is given by the expression:

E(Vj,kj) = πj,kj

/r ·∫ ∞

0(1− erτ ) · pj,kj

· e−pj,kjτdτ (A.9)

Having evaluated the integral and having substituted for πj,kj, E(Vj,kj

) can be presentedas follows:

E(Vj,kj) =

πj,kj

r + pj,kj

= HA1/(1−α) ·(

1− α

α

)· α2/(1−α) · qkjα/(1−α)/(r + pj,kj

) (A.10)

Equation (A.10) shows the expected reward form making the kjth innovation in sector j. Thisvalue confronted with the R&D costs allows the investor to decide whether or not engage inthe innovation process. Until now the presented model follows very closely the Barro andSala-i-Martin (1995) specification. Now, we extend their approach to allow for credit marketimperfections and to introduce the financial system.

Firms decide to engage into R&D activity if the expected profits are at least as high ascost of improving the existing intermediate good j. If the set-up costs for introducing a newquality of capital good exceed the present value of future profits then there is no economicreason to spend the resources on R&D. Similarly, if the set-up costs are smaller than theexpected profits, there are strong incentives to engage in the innovation activity.

The costs of entry are equal to the value of the resources devoted to the innovation process(as indicated earlier):

Z∗j,kj= Zj,kj

· (1 + fj). (A.11)

The expected reward per unit of time for pursuing the (kj +1)th innovation is pjkj·E(Vj,kj+1).

Hence, the expected flow of net profit from research in a sector j at quality rung kj is given by:Πj,kj

= pj,kj·E(Vj,kj+1)−Z∗j,kj

. For the free entry case, the equilibrium is achieved only if costsare equal to the reward for innovation; that is there is no ”extra” profit from inventing higherquality of capital good. In that case Πj,kj

= 0, i.e. Πj,kj= Zj,kj

·{φ(kj)·E(Vj,kj+1)−(1+fj)} =0 we obtain a condition to be satisfied for every sector j at every quality rung kj .

Using the expression for φ(j), the free-entry condition takes following form:

(1 + fj) · (r + pj) = (1ζ

) ·HA1/(1−α)

(1− α

α

)· α2/(1−α). (A.12)

3 We will show that in the equilibrium the interest rate is constant.


The right-hand side of equation (A.12) represents the expected rate of return from research.As this flow of profit last only until the next innovation, the rate of return must cover theordinary rate of return, r, plus the premium for the probability, p, per unit of time, correctedfor the extra cost of external financing. Followingly, the probability of successful innovationper unit of time is given by the following equation:

pj = (1ζ

) ·HA1/(1−α)

(1− α

α

)· α2/(1−α) · (1 + fj)−1 − r. (A.13)

The last equation completes the description of innovation process in the economy permittingthe shift towards the analysis of the growth process itself.

As demonstrated by equation (A.2), all components of this steady state equilibrium rela-tion between output (Y ) and the aggregate quality index (Q) are constant over time. This,the growth rate of the economy will be equal to the aggregate quality growth rate. Recallthat definition of Q, is as follows:

Q ≡N∑

j=1

qkjα/(1−α)

In sector j, the term qkjα/(1−α) does not change if no innovation occurs, but rises toq(kj+1)α/(1−α) in the case of research success. The proportionate change in this term due tothe successful innovation in sector j is equal to qα/(1−α) − 1, and the probability per unit oftime is pj . Since pj , which is defined in equation (A.13), varies across sectors, the expectedproportionate change in Q per unit of time takes be expressed in the following form:

E(∆Q/Q) =N∑

j=1

pj · ψ(j) · [qα/(1−α) − 1]. (A.14)

In this case ψ(j) accounts for the simple weightening mechanism with the interpretation thatthe innovation has occurred precisely in sector j 4. Since we do not know the exact distributionof research intensities across sectors, we do not assume any functional form for ψ(j).

As standard in the literature, we assume that N is large enough to treat Q as differen-tiable, with Q/Q non-stochastic. Then the Law of Large Numbers implies that the averagegrowth rate of Q measured over any finite interval of time will be equal to the right-handside of equation (A.14). Substituting for pj , the growth rate of aggregate quality, Q, is givenby following expression:

γQ =

N∑j=1

ψ(j)(ζ−1 ·HA1/(1−α)

(1− α

α

)· α2/(1−α) · (1 + fj)−1 − r

) · [qα/(1−α) − 1].

(A.15)The only remaining unknown remaining the interest rate (r). As in every general equilib-rium model these are the households who provide the capital, basing on the intertemporalconsumption-savings choice. We thus need to analyse their behaviour in order to determinethe actual economy’s interest rate.

4 With the obvious requirement that∑N

j=1 ψ(j) = 1.


The model employs the Ramsey (1928) idea of household optimisation behaviour. It isassumed that households maximise utility over an infinite horizon:

max

∫ ∞

0e−ρtu(Ct), (A.16)

where ρ > 0 denotes the rate of time preferences, u is instantaneous utility function, whilstthe rate of population growth equals zero (n=0). Household earns the rate of return (r) onassets and receives the wage rate (w) on the fixed aggregate quantity of human capital. Theutility is derived only from the consumption of final good so the felicity function is assumedto take following form:

u(ct) =Ct

1−θ − 11− θ

. (A.17)

Given the intertemporal budget constraint of a household, the optimal consumption pathand, hence, the growth rate of consumption, are given by the standard condition:

γC =Ct

Ct=

1θ

(r − ρ). (A.18)

In order to use this result to determine the rate of return (r) the expression for the con-sumption level (C ) must be derived. The economy’s overall consumption can be consideredas a residual from the resource constraint. Then, it is given by the following expression:

C = Y −X − Z∗, (A.19)

where Y is the total output, X denotes the overall spending on the intermediate goods, andZ∗ is the total spending on R&D. Recall that Y and X are given by the following equations:

Y = HA1/(1−α) · α2α/(1−α) ·N∑

j=1

qkjα/(1−α) = HA1/(1−α) · α2α/(1−α) ·Q,

X = HA1/(1−α) · α2/(1−α) ·N∑

j=1

qkjα/(1−α) = HA1/(1−α) · α2/(1−α) ·Q.

The amount of the total expenditure can be derived from equation (A.6). Hence, the amountof resources devoted to R&D in sector j is given by the following expression:

pj = Zj,kj· φ(kj) ⇒ Zj,kj

= pj/φ(kj) (A.20)

Taking the expressions for φ(j) and for pj the total expenditures in sector j can be presentedas follows:

Z∗j,kj= Zj,kj

· (1 + fj)

= q(kj+1)α/(1−α)

[HA1/(1−α)

(1− α

α

)· α2/(1−α) − rζ · (1 + fj)

]. (A.21)

Hence, the more advanced sector - that is with higher kj - the larger quantity of resourcesdevoted to R&D. However, as more effort is required to have the same probability of success


in more advanced sectors, the probability of success is independent of kj . Summing up overall N sectors we obtain the formula for the aggregate R&D spending, denoted by Z∗:

Z∗ ≡N∑

j=1

Z∗j,kj=

N∑j=1

q(kj+1)α/(1−α)

[HA1/(1−α)

(1− α

α

)α2/(1−α) − rζ(1 + fj)

]. (A.22)

We can now determine the consumption growth rate(γC). Since Y and X are constantmultiples of Q, they both grow at the same rate as Q. Although Z∗ cannot be representedin this way, the growth rate of the aggregate R&D spending is exactly the same one of totaloutput, and aggregate quality. To see this, note that except qkj all parameters on the right-hand side in the equation (A.22) are constant over time. Therefore, the changes of Z∗ willhave the same pattern as changes of

∑Nj=1 q

(kj+1)α/(1−α), or equivalently Q · qα/(1−α). Sincethe last expression grows at constant rate γQ, given in equation (A.15), we can concludethat the growth rate of the aggregate R&D is the same as Q growth rate. Consequently, thegrowth rate of consumption can be determined to be equal to the growth rates of Y , X, andZ∗:

γC = γY = γX = γZ∗ = γQ

Since the interest rate (r) is the clearing price of the financial markets, equating thesupply (households) with the demand

γ =1θ

(r − ρ)

γ =

β · N∑j=1

ψ(j)(1 + fj)−1 − r

· δ,where δ = [qα/(1−α)− 1] and β = (1

ζ ) ·HA1/(1−α)(

1−αα

)·α2/(1−α). Having solved it we obtain

the following steady-state values for r and γ:

r =ρ+ θδβ

∑Nj=1 ψ(j)(1 + fj)θδ + 1

(A.23)

γ =δ(β

∑Nj=1 ψ(j)(1 + fj)− ρ)

θδ + 1(A.24)

Putting the expressions for δ and β the economic growth rate in this model is given by thefollowing equation:

γ =(qα/(1−α) − 1) ·

((1

ζ )HA1/(1−α)(

1−αα

)α2/(1−α)

∑Nj=1 ψ(j)(1 + fj)− ρ

)1 + θ · (qα/(1−α) − 1)

(A.25)

This equation ends the derivation of theoretical model.

B. THE FACTORS CONTROLLED FOR BY LEITINGER AND SCHOFER

Basing on data from World Competitiveness Yearbook of the IMD in Lausanne for sevencountries (Czech Republic, Poland, Hungary, Russia, Slovenia, Slovakia and Estonia) theyhave formulated the measures of economic, social and legal environment as well as the ’en-trepreneurial spirit’. These were:

1. Economic environment:

• Urbanization

• Infrastructure maintenance and development

• Investment in telecommunications

• Telephone lines

• Cellular mobile telephone subscribers

• Computers per capita

• Internet users

• Availability of information technology skills

• Information technology

• Overall productivity (PPP)

• Labour productivity (PPP)

• Productivity in industry

• Productivity in services

• Compensation levels

• Working hours

• Labour relations

• Worker motivation

• Energy intensity

• Electricity costs for industrial clients

• Availability of finance skills

• International experience

• Marketing culture

• Business skills

B. The factors controlled for by Leitinger and Schofer 88

• Financial institutions transparency

• Cost of capital

• Availability of credit

• Availability of venture capital

• Ability to self-finance

• Interest rate spread

• Investment incentives

• Access to local capital markets

• Stock market capitalization

• Value traded on stock markets

• Listed domestic companies

• Insider trading

• Corporate boards

• Shareholder value

• Cross border ventures

• Access to foreign capital markets

• Foreign and domestic companies

• Foreign financial institutions

• Real GDP growth

• GDP/capita (PPP)

• Gross domestic investment/capita

• Gross domestic savings/capita

• Development of GDP

• Share of industry in GDP

• Share of services in GDP

• Exports of goods

• Trade to GDP-ratio

• Trade

• Black market economy

2. Legal environment:

• Collected total tax revenues

• Real personal taxes

• Average corporate tax rate on profits

• Legal regulation of financial institutions

B. The factors controlled for by Leitinger and Schofer 89

• Rights and responsibilities of shareholders

• Confidentiality of financial transactions

• Financial legal matters

• Competition laws

• Product and service liability

• Government price controls

• Labour regulations

3. Social environment:

• Total expenditure on R&D per capita

• Total expenditure on R&D

• Business expenditure on R&D per capita

• Total R&D personnel nationwide per capita

• Funding for technological development

• Technological cooperation

• Development and application of technology

• Nobel prizes

• Company-university cooperation

• Total public expenditure on education

• Pupil-teacher ratio (secondary education)

• Higher education achievement

• Patents granted to residents

• Number of patents in force

• Education in finance

• Qualified engineers

• Bribing and corruption

• Income distribution-lowest 20%

• Values of society

4. Entrepreneurial spirit

• Attitudes towards globalisation

• Creation of firms

• Real corporate taxes

• Bureaucracy

INSTITUTIONAL DETERMINANTS OF THE INVESTMENT …

Documents

Transcript of INSTITUTIONAL DETERMINANTS OF THE INVESTMENT …