Bureaucratic Discretion, Business Investment, and...

Bureaucratic Discretion, Business Investment,and Uncertainty

Quintin H. Beazer Florida State University

What determines whether policy environments attract or deter investment? Scholars worried about thevulnerability of market-supporting institutions to political manipulation have identified delegation to independentactors as way to increase policy environments’ predictability. Extant arguments, however, risk overgeneralizingfrom the experience of developed democracies. I argue that investors’ response to bureaucratic discretion—agents’leeway to make decisions and act independently of political bodies—depends upon the broader institutionalcontext. Where robust political institutions are lacking, bureaucratic discretion acts as a source of unpredictabilitythat deters investors; conversely, political institutions that share the cost of monitoring help to mitigate uncertaintyabout how bureaucrats will use discretion in applying regulatory rules. Using survey data from over 600 enterprisesin Russia, I find that perceptions of bureaucratic discretion are negatively associated with firm managers’willingness to invest; this effect is particularly pronounced in regions where the institutional environmentdiscourages political competition.

What determines whether policy environ-ments attract or deter investment? Forinvestors who seek predictable environ-

ments, the state is a double-edged sword. On one hand,states’ involvement in economic affairs is commonlyjustified as a way to increase the predictability ofmarkets and market activity. Businesses and consumerslook to the state to help correct market failures, supplynecessary infrastructure, and reduce the transactioncosts of measurement and enforcement. On the otherhand, the state is controlled by leaders who respondto political incentives, which can make it hardfor economic actors to predict changes that mightaffect their business interests. Consequently, politicalscientists and economists have looked to bureau-cratic discretion—bureaucrats’ leeway to make de-cisions and act independently of political bodies—asa potential mechanism for stabilizing expectationsabout the regulatory environment and encouraginginvestment. In a variety of circumstances, however,we might question whether greater independencefrom politicians will make regulation more predict-able.

The literature on the state’s role in economicdevelopment identifies bureaucracy as a factor that

encourages growth, emphasizing bureaucracy’s tech-nocratic role in nurturing new markets and unitingbusiness groups together with government in thepursuit of common economic goals (Brown, Searle,and Gehlbach 2009; Kohli 2010; Rauch and Evans1999). One prominent mechanism for enhancingcredibility relies upon delegating the details of poli-cies’ implementation to independent actors, partic-ularly autonomous bureaucrats, with preferencesthat remain distinct from political leaders’ (Bendor,Glazer, and Hammond 2001). Implicitly, such argu-ments rely upon bureaucratic agents who performtheir delegated duties in a manner that investors cananticipate. Predictable bureaucracy, however, isscarce in some locations. Surprisingly, the improb-ability that the predictability assumption holdsfor the wide variety of developing and transitioneconomies has prompted little investigation intoanother question: how does the broader institutionalcontext shape bureaucratic discretion’s effect oninvestment?

To preview the argument, this article identifiesbureaucratic discretion as a primary source of uncer-tainty that deters long-term investors by underminingthe predictability with which regulatory policies are

The Journal of Politics, Vol. 74, No. 3, July 2012, Pp. 637–652 doi:10.1017/S0022381612000205

� Southern Political Science Association, 2012 ISSN 0022-3816

637

interpreted and applied.1 Where bureaucratic agentsexercise greater discretion in interpreting and applyinglaws, investors shy away from the resulting increaseduncertainty over how those applied regulatory policieswill affect their business interests. Particularly inlocations where surrounding political and legal insti-tutions do not provide decentralized mechanisms formonitoring discretionary bureaucrats’ behavior orsupply effective channels for resolving disputes withagencies, uncertainty about policy application has adirect negative effect on investment.

The next section examines prominent argumentsin the broader literature about the economic effectsof bureaucracy, identifying where underappreciationof developing world’s institutional challenges has ledscholars to suspend healthy skepticism about theindependent agents charged with managing regulatorypolicies. The article then details the logic connectingregulatory bureaucrats’ policy discretion, the broaderpolitical environment, and economic actors’ invest-ment decisions. The empirical section that followsemploys survey data from over 600 enterprises inRussia to test the argument’s microlevel predictions.The data show that discretionary bureaucratic behav-ior has a robust negative relationship with firmmanagers’ willingness to invest; additional analysesshow that this association is particularly strong forfirms in politically uncompetitive regions. The articleconcludes by discussing the research’s implicationsand contributions.

Bureaucracy and Investment

The term investor encompasses actors such as firms,venture capital groups, and private individuals withresources that could potentially be plowed intoentrepreneurial ventures. Examples of investmentinclude expanding or starting new operations, con-ducting research and development, and buying newmachinery—long-term projects that require that in-vestors pay extensive costs initially but then wait forreturns to materialize several time periods in thefuture. Because they need to make costly decisionsbased upon forecasts, investors seek predictable envi-ronments and avoid situations characterized by highuncertainty (Aizenmann and Marion 1999). In this

regard, politicians’ control over the creation and im-plementation of economic policy can present a prob-lem because frequent turnover and fickle electoratesmake it harder for investors to anticipate changesin how politicians choose to regulate the economy.Indeed, empirical studies showing that investmentresponds negatively to volatility and the threat ofunpredictable changes emphasize the premium thatinvestors place on predictable policy environments(Aizenmann and Marion 1993; Nooruddin 2011).Bureaucratic discretion has been suggested by existingresearch as one viable solution to this problem; byinsulating the state apparatus from political decisions,bureaucratic discretion may help stabilize policy ex-pectations and encourage investment.

Various theoretical arguments associate bureau-cratic discretion with improved economic outcomes.For some, discretion is a way to harness technocraticexpertise. Proponents claim that independence allowsagencies to recruit and retain career-minded profes-sionals, ostensibly improving the economic climatethrough higher-quality governance and fewer officialsthat use their authority irresponsibly (Rauch and Evans1999). Arguments of this sort cast bureaucracy as thestate’s ‘‘neutral competence,’’ portraying bureaucraticdiscretion as a way to prevent elected officials fromderailing effective public policy in their pursuit ofparticularistic goals (Rauch 1995).

In addition to harnessing expertise, broad delega-tion of authority to bureaucrats might make the policyenvironment more predictable by providing continu-ity and insulating government programs against majorpolitical changes (Lewis 2003; Miller 2000). This viewinforms the ‘‘Bureaucratic Quality’’ measure of theInternational Country Risk Guide (ICRG), a set ofcross-national risk ratings that is prominent in theeconomic development literature.2 Consider theICRG’s reasoning: ‘‘Bureaucracy is another shockabsorber that tends to minimize revisions of policywhen governments change . . . .[High-quality] bu-reaucracy has the strength and expertise to governwithout drastic changes in policy or interruptions ingovernment services. In these low-risk countries, thebureaucracy tends to be somewhat autonomous frompolitical pressure’’ (ICRG 2010). According to thislogic, discretionary bureaucrats have the ability touncouple policy implementation from the instabil-ities associated with political competition.

1An online appendix with supplementary material for this articlewill be available at http://journals.cambridge.org/jop. Upon pub-lication, data and supporting materials necessary to reproducenumerical results will be made available at http://dvn.iq.harvard.edu/dvn/dv/ghb.

2Studies use this measure both explicitly, beginning with Knackand Keefer (1995) as well as implicitly, since the World Bank usesICRG data to build its own governance indicators (Kaufmann,Kraay, and Mastruzzi 2009).

638 quintin h. beazer

Finally, delegation to independent bureaucratshas been proposed to solve commitment problems thatcan deter investors. Recognizing politicians’ incentiveto make opportunistic changes to business-friendlypolicies once firms have sunk immobile capital, busi-nesses may curtail their investment plans unless polit-ical leaders can credibly commit to their decisions bymaking policies difficult to reverse (Jensen 2006; Levyand Spiller 1994). Commitment at the policy imple-mentation stage ceases to be a problem, the logicgoes, if political leaders can ‘‘lock-in’’ policies byhanding the management of policies to autonomousbureaucratic actors who have different preferences.Delegation-as-commitment arguments appear regu-larly in political economy, such as ceding control ofmonetary policy to independent central bankers(Cukierman, Web, and Neyapti 1992; Rogoff 1985),committing to property rights by establishing inde-pendent judiciaries (Levy and Spiller 1994), orremoving the oversight of regulatory rules from thepolitical arena to independent agencies (Bertelli andWhitford 2009). Again, independence from politi-cians provides the key to credibility—granting dis-cretion to bureaucratic agencies helps to stabilizeinvestors’ expectations about the policy environment.

Such explanations contribute greatly to ourunderstanding of the political dynamics that shapeinvestment, but applying them too broadly can skewour picture of how investors react to policy environ-ments. Specifically, arguments that link bureaucraticdiscretion to beneficial economic outcomes riskovergeneralizing from countries with robust politicaland administrative institutions. Like the institutionaldesign literature, these theories rely heavily on theexperience of developed countries (Levitsky andMurillo 2009).3 Without evidence that bureaucraciesin other institutional environments behave similarly,however, we should exercise extreme caution ingeneralizing conclusions about the economic benefitsof bureaucratic discretion to the developing world.

Figure 1 emphasizes the uniqueness of the OECD’sregulatory environment relative to other countries.Plotting the within-country standard deviations to aglobal survey on the percentage of firm managers’ timespent on dealing with officials and meeting regulatoryrequirements, Figure 1 makes two points. First,

within many countries of the developing world, thereis surprisingly little uniformity in how entrepreneursexperience their regulatory environment. Second,company managers in developing economies appearto face greater uncertainty than their counterparts inthe OECD in doing what it takes to meet regulatoryrequirements. To the extent that the within-countryvariation reflects officials’ discretion in the interpret-ing and applying the very same legal codes, the figuressuggests that scholarly discussions about bureaucraticdiscretion within developed democratic countries maynot be very informative for the rest of the world.4 Lateron, the analysis addresses potential explanations forthe group differences; for the time being, Figure 1underscores the point that relying too heavily on theexperience of developed countries could bias ourinferences about the economic consequences ofbureaucratic discretion.

We gain valuable insights into the relationshipbetween investment and bureaucratic discretion bylooking beyond developed democracies’ well-structuredand predictable regulatory environments. For example,touting insulated bureaucracies as a solution to commit-ment problems obscures the fact that delegation doesnot eliminate moral hazard so much as relocate it frompoliticians to bureaucrats. Bureaucratic control onlyaids investors as long as they can anticipate theoutcome from bureaucrats’ involvement; where thisis not the case—and that may be frequently in somelocations—discretion merely increases uncertaintyabout the application and interpretation of laws atthe hands of unelected officials. Similarly, becausemany governments struggle to recruit and traintechnocrats of the ‘‘Weberian’’ sort, greater bureau-cratic independence can backfire when agencies do notpossess sufficient human or financial resources toimplement policies effectively (Huber and McCarty2004). Accounting for the challenges to governmentsand businesses in developing economies reminds usthat, as a strategy for improving the investmentclimate, substituting bureaucratic control for politicalcontrol depends heavily upon the bureaucrats them-selves. Borrowing from the principal-agent framework,the theoretical analysis in the next section explains thepotential pitfalls of bureaucratic discretion.

3With its championing of bureaucratic autonomy as a keycontributor to the economic success of developmental states inEast Asia, the literature on state-led development stands out as anexception (Amsden 1989; Johnson 1982).

4The full World Bank sample of OECD countries appears in thefigure: Ireland, Germany, Greece, Portugal, and Spain. As someof the lowest-performing economies in Europe, the inclusion ofthe latter three countries make the comparisons even more starkthan if the sample focused only on the economic leaders of theEuropean Union. A full list of countries is available in theappendix.

bureaucratic discretion and business investment 639

Bureaucratic Discretion andUncertainty

A principal-agent approach to bureaucracy stresses boththe practical necessity that gives rise to delegation as wellas the pitfalls that accompany it. Delegation to bureau-crats by lawmakers is ubiquitous because governmentleaders cannot hope to enforce and implement lawsthemselves (Shipan 2004). As politicians’ agents, bu-reaucrats have informational advantages over theirpolitical principals (Weingast and Moran 1983). Theygenerally stay in their offices longer than politicians,they have specialized knowledge and training that pol-iticians lack, and they have hidden information abouthow they have fulfilled their assigned duties. Theseinformational advantages create opportunities for bu-reaucrats to act contrary to politicians’ expectations,such as providing more zealous enforcement or, alter-natively, applying rules too laxly (McNollgast 1987).

Imperfect control within hierarchical relationshipscreates a gap between principals’ directives and agents’execution. Politicians recognize that granting discre-tion only increases the risk of divergent outcomes, buteliminating discretion completely would require theimpossible task of assigning procedures for everycontingency and specifying the appropriate behaviorfor every imaginable circumstance (Huber and Shipan2002). Even knowing that it may weaken their control,

sometimes government leaders may expand bureau-crats’ discretion because it allows bureaucrats toexecute their duties in complex cases and encouragesspecialization (Bawn 1997; Gailmard and Party 2007).

Scholars rarely consider the consequences ofprincipal-agent dynamics for individuals outside thehierarchical relationship, but leaders’ imperfect controlover their bureaucratic agents has important implica-tions for entrepreneurial activity. Considering that everylaw must go through their hands before it goes intoeffect, bureaucrats play a critical part in determining howlaws govern both society and economy (Lipsky 1980). Asauthorized agents of the state, inspectors and regulatorsrefer to some guiding principles or overarching policy,then make subjective decisions about how businesses’current circumstances map onto those criteria (Brehmand Gates 1997). The trouble is that for business actorscontemplating long-term investments, any uncertaintyabout how regulatory rules will affect their businessposes an obstacle.

By easing constraints on bureaucrats, greaterdiscretion only exacerbates the problem that agents’implementation can differ significantly from politi-cians’ legislated policy.5 Increased discretion widens

FIGURE 1 Firm Managers in Developing Countries Have Less Uniform Experiences with Regulators

5To the extent that insufficiently articulated legislative policies them-selves may frustrate investors in the developing world, the predict-ability with which bureaucrats implement those laws is likely to playan even more pivotal role for economic actors in those locations.


the range of potential outcomes that bureaucrats mayproduce, making the policy environment even lesspredictable. Thus, the same information asymmetrythat troubles the principal-agent relationship alsoimpedes investors’ ability to forecast regulatory con-ditions, and this uncertainty increases with regulatoryagents’ discretion. This observation provides thetestable empirical prediction that bureaucratic dis-cretion should depress investment:

d ‘‘Dangerous Discretion’’ Hypothesis: Holding otherfactors constant, bureaucratic discretion should benegatively correlated with investment.

In contrast, existing research predicts that any neg-ative side-effects that discretion has on the regulatoryenvironment are overshadowed by the economicbenefits to investors from increased bureaucraticexpertise or more insulation against political control:

d ‘‘Benevolent Bureaucrat’’ Hypothesis: Holding otherfactors constant, bureaucratic discretion should bepositively correlated with investment.

Institutional Constraints onRegulatory Uncertainty

The claim that bureaucratic discretion deters invest-ment rests upon two arguments. First, studies ofinvestment unambiguously point to uncertainty as adeterrent for investors. Second, the principal-agentframework implies that uncertainty about policy im-plementation should increase with the amount ofdiscretion delegated to agents.6 Yet, predictions thatdiscretion deters investment fit better for developingeconomies than they do with scholarly accounts ofbureaucratic discretion in the developed world. Thisincongruity suggests that we need further theorizing tounderstand why investors’ response to bureaucraticdiscretion might differ across institutional settings.

Several plausible mechanisms could make un-certainty about bureaucratic discretion higher indeveloping economies. At first blush, differences inbureaucratic corruption present one possible explan-ation. Observers of developing countries regularlyequate greater discretion with increased opportuni-ties for bureaucrats to take or extort bribes. Unfortu-

nately, a mixed empirical record makes it difficult todraw conclusions about whether corruption hurts orhelps investment (Treisman 2007).7 In fact, severalpromising studies suggest that corruption’s predict-ability matters more to investors than its costs(Malesky and Samphantharak 2008; Wei 1997),implying that those who wish to use corruption topredict differences across the two groups must alsosupply a theoretical argument for why discretionshould lead to corruption that is more predictablein developed countries than in developing countries.8

Variation in state capacity represents anotherplausible alternative, positing that bureaucratic man-agement produces unanticipated outcomes whenagencies lack sufficient resources to carry out theirassignments. Such an explanation highlights thetension in many developing countries between states’ambitious economic policies and the challenges oftranslating them into practice. In locations such assub-Saharan Africa, India, and the postcommuniststates of Eastern Europe, scholars have identifiedinadequate local cadres and tight resource constraintsas main contributors to the inconsistent implemen-tation of national policies (Kohli 2010; Varese 2001).If greater discretion widens the range of outcomesthat bureaucrats may produce while attempting tomanage burdensome obligations, the combinationof resource-poor agencies and high discretion maypresent an especially unattractive option for economicactors who are seeking predictable environments.

This study concentrates on a third explanation,that developed democracies’ supporting institutionsprovide additional oversight of government agents,thereby reducing investors’ uncertainty about bu-reaucrats’ discretionary behavior. The principal-agentliterature shows that, although discretionary agentsare hardly predictable automatons, politicians canstill take costly actions to influence how bureaucratsinterpret and apply delegated policies. In particular,politicians increase the likelihood that bureaucratswill implement policies in a predictable manner bymonitoring agents closely, but the resource-intensive

6This claim finds support in empirical evidence from both globaland country-specific enterprise surveys. For analyses showingthat bureaucratic discretion correlates with greater unpredict-ability in the policy environment, please see the appendix.

7Scholars have taken both sides of the theoretical debate. Bribepayments may act as ‘‘unofficial’’ taxes that allocate resourcesinefficiently and crowd out beneficial economic institutions(Mauro 1995); alternatively, corruption may make some invest-ments possible by helping firms bypass regulatory demands orobtain preferential treatment (Leff 1964; Slinko, Yakovlev, andZhuravskaya 2005).

8The question of how bureaucratic discretion relates to thepredictability of corruption is theoretically interesting, at leastpartly because predictions are not obvious a priori. Addressingthis question in a rigorous theoretical and empirical manner,however, falls outside the scope of this article.


nature of monitoring often makes it unattractive toleaders. Conveniently, lawmakers can keep bureau-crats in check without directly paying all the coststhemselves by allowing interested constituents andother institutional actors to help in the monitoringprocess.

The literature on ‘‘fire-alarm’’ oversight providesa specific example of how such decentralized mon-itoring of bureaucrats can take place. In studies ofAmerican bureaucracy, fire-alarm oversight refers toa system of rules and administrative processes thatcreate opportunities for the public to examine admin-istrative practices, question agency decisions, and iden-tify actions that violate legislative goals (McCubbinsand Schwartz 1984). Such arrangements allow politi-cians to reduce their own monitoring and rely insteadon self-interested constituents to report troublingbureaucratic behavior. In addition, by providing dis-satisfied actors with clear, formalized channels forseeking recourse, politicians can further shift enforce-ment costs off onto supporting institutions, such ascourts (McNollgast 1987). Generalizing the logic ofdiffuse monitoring beyond the specific example of fire-alarms and the American Congress, we see thatbroader political institutional arrangements wield thissame mechanism and often do so in a way that makesthe regulatory environment more predictable forinvestors.

In this article, I call attention to the beneficialrole of political competition in mitigating the un-certainty that businesses associate with bureaucraticdiscretion. Institutions that encourage political com-petition make policy implementation more predictablein at least two ways. First, competitive environmentsincrease the number of politically active groups andenhance their access to the policy-making process. Bygiving economic actors direct and indirect opportu-nities to influence regulatory policy or complain aboutadministrative practices, political competition helps tospread the costs of monitoring bureaucrats acrossactors and stabilize expectations about how bureau-crats will use their discretion to interpret and applyregulations. In a related manner, politically compet-itive environments give rise to a wide array of institu-tional watchdogs, such as political parties andindependent courts, that can provide formal, predict-able channels for settling disputes that arise fromagencies’ actions. Second, competitive politics help tokeep politicians attentive, effectively raising the costs toprincipals associated with not monitoring their agents.Politicians worried about losing office have a pressingelectoral incentive to monitor the bureaucracy inorder to avoid scandal or prevent political opposition

from capitalizing on constituents’ dissatisfaction withinconsistent legal requirements or aggravating treat-ment by state officials. Likewise, electorally sensitiveleaders should be more responsive in handlingregulatory disputes associated with agents’ discretion.

With their robust institutions and vibrant politicalcompetition, developed democracies are in the bestposition to foster an effective, polity-wide realizationof fire-alarm oversight. In developed democracies, theinstitutional environment actively encourages organ-ized political opposition, free media, and politicallysavvy interest groups to help monitor the behavior ofstate agents. Moreover, from courts to political parties,a wide array of effective institutional watchdogs standready in these countries to help settle disputes that arisefrom agencies’ actions. It is hardly surprising, then,that studies of investment based upon developeddemocracies have a sanguine attitude about bureau-cratic discretion. In such environments, the surround-ing institutional context plays a critical role in reducingthe uncertainty that investors associate with bureau-cratic discretion.

In marked contrast to developed democracies,regulatory oversight in many developing countries istypically characterized by limited political competi-tion and inaccessible political institutions. Givenunstable and toothless institutions in many parts ofthe world (Levitsky and Murillo 2009), political andlegal institutions in developing countries often can-not play the supporting role required to makebureaucratic discretion more predictable. For exam-ple, the diffuse monitoring of state agents cannotwork well if institutions restrict citizens’ ability toparticipate in political life or shut organized societalgroups out of the policy process. Furthermore, with-out intense political competition, leaders face littleelectoral pressure to respond to constituents’ com-plaints. Thus, in addition to reducing the channelsfor the effective monitoring of discretionary bureau-crats, institutions that discourage political competi-tion also weaken politicians’ incentives to respond toconcerns when they are raised. In contrast to thedynamics within politically competitive environ-ments, low-competition settings do little to alleviatebusinesses’ uncertainty about the way in whichdiscretionary bureaucrats choose to implement pol-icy, nor do they provide economic actors with manyviable options to handle disputes should they arise.

This discussion helps us understand why bureau-cratic discretion in some regulatory environmentspresents a particularly sharp problem for economicactors. The political institutional context of economicregulation affects the level of uncertainty that


bureaucratic discretion creates for investors. Businessesshould more readily invest where political institutionsspread monitoring costs across multiple nongovern-ment and institutional actors and then allow those sameparties to be involved in the policy process. Under suchcircumstances, investors should have clearer expect-ations about what regulators want and how agencies’discretion will affect their business interests. But, wherepolitical institutions restrict political participation orgive leaders’ few incentives to check regulatory agencies,bureaucratic discretion remains a stumbling block forinvestors since they cannot rely on institutional con-straints to enhance the predictability of the regulatoryenvironment. Thus, the following empirical predictionthat institutional context conditions the relationshipbetween bureaucratic discretion and investment:

d The ‘‘Conditional Constraints’’ Hypothesis: Holdingother factors constant, the relationship betweenbureaucratic discretion and investment should bemore negative in regions characterized by lowpolitical competition than in regions with highamounts of political competition.

As political scientists, we should consider more care-fully how the discretionary actions of bureaucraticagents determine policy environments’ predictabilityand, consequently, their attractiveness to business.When bureaucrats apply statutes consistently andpredictably, business actors can treat regulation as aknown parameter in their investment formula. Asbureaucrats’ discretion grows, investors experiencegreater uncertainty about how regulation will affecttheir business interests now and in the future. Thisuncertainty is especially problematic within regulatoryenvironments that lack diffuse monitoring mecha-nisms; without supporting institutions to encouragefire-alarm oversight and stabilize investors’ expect-ations, investors shrink from committing their assetsto unpredictable rule enforcement.

Empirical Analysis: Evidencefrom Russia

I test the empirical predictions generated above in thecontext of the regions of the Russian Federation.9

Several factors make postcommunist Russia an ex-cellent location for examining investors’ response tobureaucratic discretion. For starters, bureaucratic

authority in Russia’s regional economies has consid-erable scope. Regional bureaucratic agencies interpretand apply an array of statutes, including registrationlaws, zoning ordinances, tax codes, and safety regu-lations. Given the impact of these policy areas onexpected profitability, how regional bureaucrats carryout these duties matters greatly for investors. Fur-thermore, scholars have noted that vague laws withextensive delegation to executive agencies are commonin Russian legislation, and evidence suggests thatagents exercise this discretion regularly (Solomon2008). Remington elaborates: ‘‘Many legislative actsin Russia have been so vague that the bureaucracyhas been able to eviscerate, reinterpret, or ignorethem freely’’ (2006, 278). Cross-regional studies ofregulatory reforms indicate that regional bureaucratsguard this autonomy, providing evidence that regu-lators in many regions have resisted enforcing federalreforms that would standardize small business regu-lations (Yakovlev and Zhuravskaya 2008). Methodo-logically, Russia’s federal structure also helps tomitigate some of the challenges posed by studyingbureaucracy and regulation in a cross-national set-ting, where measurement problems and unobservedheterogeneity in countries’ legal or administrativefactors are often difficult to address. In this regard,regions’ shared administrative legacies and the over-arching sovereignty of Russia’s federal law help tohold constant many institutional and policy factorsthat might vary cross-nationally.10

Modeling Businesses’ Response toBureaucratic Discretion

I conduct statistical analyses using data from a survey ofRussian enterprise managers conducted in 2005 byTimothy Frye (see Fyre 2006). Given growing skepti-cism about the validity and biases of popular cross-national governance indicators (Kurtz and Shrank2007), enterprise surveys offer an effective alternativefor studying how specific regulatory factors affectindividuals’ behavior. Commissioned specifically forstudying how firms respond to their institutionalenvironment, this survey features items on firm invest-ment and, importantly, about managers’ perceptions ofinstitutional actors, including regulatory bureaucrats.

9Rather than refer to Russia’s 83 subnational units by their formal(and unwieldy) constitutional title, ‘‘Subjects of the Federation,’’I use the more colloquial term—regions.

10Admittedly, the decision to use subnational data is not withoutcosts. In this case, however, any trade-off in terms of results’generalizability are likely to be small relative to the associatedgains in precision.


Thus, the individual-level data on how firms perceivetheir regulatory environment make a natural fit fortesting the argument’s microeconomic predictionsabout individual actors’ subjective investment decisions.

The survey was conducted by a well-respectedMoscow polling firm, the Levada Center, polling 666firm managers in 11 different regions.11 In regards tosize and sector, the sample distribution of firmsapproximates the national population of enter-prises.12 With only 12% majority state-owned and5% foreign-owned, the modal firm is a domestic,private enterprise.13

The dependent variable is investment by firms.Following Frye (2006), I operationalize it using firmmanagers’ response to the following item: ‘‘Do youplan to make any large investment in the next twelvemonths for the development of your firm (i.e.,construction, reconstruction, capital renovation ofthe building or surroundings, equipment updates,etc.)?’’ I collapse answers on the question’s original4-point scale (‘‘yes,’’ ‘‘likely yes,’’ ‘‘likely no,’’ and‘‘no’’) into a dichotomous variable that gives a valueof 1 for the two affirmative answers and assigns a 0 tothe remaining, negative answers.14 Firms in thesample lean towards no investment plans, with 60%of respondents answering negatively.15

To test the prediction that firms’ investmentdecisions correlate negatively with their perceptionsof bureaucratic discretion, I operationalize the inde-pendent variable using an item about the independenceof regional bureaucrats’ decision making: ‘‘To whatdegree is independent decision making, separate fromother government bodies, characteristic of bureaucrats,

administrators, and various inspectors in your re-gion?’’16 Respondents’ answers range on a 4-point scalethat evaluates regional bureaucrats’ autonomy as char-acteristic ‘‘to a high degree,’’ ‘‘essentially to a highdegree,’’ ‘‘to a lesser degree,’’ or ‘‘completely unchar-acteristic.’’17 For ease of interpretation, I invert theoriginal scale so that higher scores reflect perceptions ofgreater bureaucratic discretion.18

Contrary to prevailing explanations, my argu-ment predicts that this variable should correlatenegatively with firms’ plans to invest, and a basiccomparison of means in the raw data shows that itdoes. Among respondents that see little independencefor regional regulatory bureaucrats, 44% indicate thatthey will invest in their firms soon; in contrast,among managers that identify regulatory bureaucratsas independent in their decision making, just 32% offirms plan to invest (p 5 0.015). To ensure that thisresult, which contradicts conventional expectations,is not spurious, I add a battery of control variablesidentified in existing research.

Controlling for certain regulatory-related varia-bles is of particular interest in order to ensure that mymeasure of bureaucratic discretion does not tap otherpolicy concerns beyond investors’ worries about howregulations are interpreted and applied. Other fea-tures of the policy environment may also driveinvestment, such as the actual policies of regulatoryregimes or the frequency with which laws change. Interms of policy content, scholars have paid specialattention to the idea that high tax rates discourageinvestment (Jensen 2006; Li 2006). Accordingly, Iinclude respondents’ evaluations about the degree towhich high tax rates hinder their firm’s development.Frequent legislative changes could deter investorswith uncertainty about the regulatory environment

11For more detail on the strategy behind the stratified samplingdesign, see Frye (2006). Including at least one region from each ofthe seven federal districts, survey sampled firm in the followingcities: Ekaterinburg, Khabarovsk, Moscow, Nizhniy Novgorod,Novgorod, Omsk, Rostov, Smolensk, Tula, Voronezh, and Ufa.

12Like most enterprise surveys, the sample is truncated because itcontains no data for potential firms who were deterred from initialinvestment or those short-timers who entered the market but thenclosed their doors. This survivor bias implies that respondents arefirms who have higher tolerance for risk or have managed to copewith existing barriers. Any bias of this type works against myhypotheses, making it harder to find a negative relationshipbetween bureaucratic discretion and firm investment.

13The online appendix provides descriptive statistics on the basiccharacteristics of the sample.

14The results do not depend on this coding choice; all therelationships hold with the original coding.

15The online appendix provides descriptive statistics for thisvariable and all others included in the analyses.

16Because Russia’s regions have different titles, such as oblast,territory, republic, and district, the item’s wording changesslightly to fit the appropriate regional designation, but thecharacter trait and actors in question do not change.

17Because this key question relies upon respondents’ perceptionsof bureaucratic discretion, I provide evidence in the appendixsupporting this measure’s internal and external validity. Aextensive set of additional analyses show that the article’s findingsare robust to controlling for a host of outside factors that mightshape respondents’ answer to the prompt, such as respondents’political knowledge, optimism, experience interacting with thegovernment, familiarity with the region, or sectoral/legal charac-teristics. Furthermore, empirical analyses demonstrate that, inaggregate, respondents’ perceptions correlate as expected withexternal measures of bureaucratic discretion and private invest-ment at the region level.

18Recoding this measure into an dichotomous indicator of lowversus high bureaucratic discretion creates no substantive changefor the results.


much in the same way that bureaucratic discretiondoes over the application of those policies. To controlfor policy volatility’s potential relationship to bothbureaucratic discretion and firm investment decisions,I include managers’ response regarding the extent towhich frequent changes in legislative and statutory actsare an obstacle to their firms. Conventional argumentswould anticipate that both these variables shouldassociate negatively with the dependent variable. As acheck for whether my measure merely reflects invest-ors’ antipathy towards corruption, I include in onemodel a dummy indicator for company managers’perceptions that regional bureaucrats are corrupt. Asnoted earlier, the literature disagrees on this variable’sexpected sign, but if corruption truly represents theheart of investors’ concerns about discretion, then weshould expect its inclusion into the model to over-shadow the bureaucratic discretion variable.

Since models rely on respondents’ subjectiveperceptions of bureaucratic discretion, we might worrythat such perceptions may be colored heavily bygeneral attitudes towards the region or its governmentin a way that also correlates with firms’ dispositiontowards future investment. As a precaution, I includevariables that measure separately firms’ assessments ofthe regional political institutions to help account forregional ‘‘halo effects’’ that might vary with bothperceptions of bureaucratic discretion and plans toinvest. These controls capture respondents ratings ofthe regional administration, regional arbitrationcourts, and governor; we would expect more favorableassessments of these institutions to correlate positivelywith firm investment. I also add controls for keyeconomic factors that might similarly influence firms’decisions to invest. I control for difficulties in gettingaccess to finance with a measure of firms’ problemsobtaining credit. High levels of economic competitionmay pressure firms to invest in innovative processesand products when they might not otherwise want todo so; accordingly, I control for respondents’ viewsabout competition as a problem for their businessgrowth. Similarly, managers may refrain from invest-ment projects that they think will be hobbled by apoor labor pool. Thus, I include a measure formanagers’ view about the shortage of skilled labor asa hindrance to business.

In addition to the perception-based measures, Icontrol also for firm-specific characteristics that mightaffect both firms’ investment plans as well as theirregulatory experience. Declining sales might stall in-vestment plans, so I include a control variable forwhether firm sales have increased, decreased, or stayedthe same over the past three years. Small firms might

have less need or ability to invest; as such, I control forfirm size with the (logged) number of employees.Finally, to control for potentially higher investmentamong private and start-up enterprises, I includedummy variables for private firms (as opposed tostate-owned) as well as a dummy for firms that wereprivatized following the collapse of communism.

I estimate the model predicting intentions to investusing logistic regression. To account for unobservedregion-specific factors that may influence the relation-ship between firms’ perceptions of bureaucraticdiscretion and their investment plans, I estimaterandom-intercept, random-coefficient models thatallow coefficient estimates on the intercept and bu-reaucratic discretion variable to vary with a region-specific error.19 Results from these analyses appear inTable 1.

Results

To demonstrate that estimated relationships holdacross various model specifications, Table 1 reportsthe results from multiple models. Model 1 estimates abaseline specification that includes only the inde-pendent variable and a few controls for the invest-ment environment and firm-specific factors. Next,Model 2 tests the results’ robustness in a full set ofcontrols, followed by Model 3, which examinesdiscretion’s relationship with investment once wecontrol for firm managers’ perceptions of bureau-cratic corruption. Model 4 adds a control for recentpast investment as well as dummies for firms’ sectorand legal form in order to control for a wider range offirm-level characteristics that might determine firms’contact with regulatory agencies as well as shape theirinvestment plans.

From a substantive standpoint, the statistics inthe first two rows display the most important empiricalresult: perceived decision-making independence forregulatory bureaucrats correlates with a lower proba-bility that firms plan to make any large investment overthe coming year. Whether in sparse models or con-trolling for a large number of potentially confoundingfactors, the coefficient estimates on the bureaucraticdiscretion variable in all models are negative andstatistically significant from zero at conventional levels.This leads us to reject the null hypothesis that perceived

19Results do not change substantively if models drop the randomeffects and use standard errors that are clustered by regioninstead. Similarly, results hold using a standard logit model.Results available in the appendix.


bureaucratic autonomy has no relationship to firms’investment plans. Substantively, the relative size of thisestimated effect is large as well. Using the estimatesfrom Model 2 and holding other variables at theirsample median, a firm manager who perceives thatregulatory agents have very high independence indecision making has a predicted probability of invest-ing that is 27% lower than a identical respondent whoreports that regional bureaucrats have a moderatelylow degree of independence. Not only do these findings

support the claim that increased bureaucratic discre-tion is associated with reduced incentives for firms toinvest for future returns, but they also suggest that,averaged across the sampled regions, the negativeestimated effects are potentially quite substantial.

Among the other variables directly related to thepolicy environment, firm problems with high tax ratesare also a strong predictor of firm investment deci-sions. Supporting arguments that stress the economicincentives of specific policy content, the coefficients on

TABLE 1 Discretion Associated with Lower Probability of Investment

Firm Investment (1) (2) (3) (4)

Bureaucratic Discretion -0.458***(0.145)

-0.612***(0.161)

-0.573***(0.174)

-0.606***(0.191)1 5 no discretion, 4 5 high discretion

Frequent Changes to Laws 0.101(0.112)

0.066(0.118)

0.139(0.125)1 5 no obstacle, 5 5 very serious obstacle

High Tax Rates -0.392***(0.123)

-0.437***(0.131)

-0.404***(0.147)1 5 no obstacle, 5 5 very serious obstacle

Regional Administration 0.509***(0.195)

0.447**(0.210)

0.477**(0.229)1 5 poor job, 5 5 excellent job

Regional Courts -0.150(0.147)

-0.181(0.152)

-0.041(0.173)1 5 poor job, 5 5 excellent job

Regional Governor 0.037(0.106)

-0.324*(0.184)

-0.284(0.197)

-0.437**(0.222)1 5 poor job, 5 5 excellent job

Access to Finance -0.087(0.067)

-0.034(0.082)

-0.026(0.086)


Labor Shortages -0.028(0.086)

0.001(0.091)


Competitive Pressures 0.107(0.085)

0.127(0.089)


Privatized Firm 0.043(0.273)

0.044(0.282)

0.209(0.346)dummy, 1 5 privatized, former SOE

Annual Sales 0.476***(0.166)

0.442**(0.182)

0.330*(0.190)

0.258(0.217)-1 5 decreasing, 1 5 increasing

Firm Size 0.335***(0.073)

0.298***(0.084)

0.284***(0.088)

0.337***(0.117)number of employees (logged)

Private Firm 0.721**(0.337)

0.694(0.439)

0.542(0.443)dummy, 1 5 private ownership

Bureaucratic Corruption -0.015(0.254)dummy, 1 5 perceived as corrupt

Past Investment 2.086***(0.294)dummy, 1 5 invested in last 3 yrs.

Constant -1.940***(0.674)

-0.594(1.031)

-0.023(1.151)

-2.523(1.712)

Dummies for Sector & Legal Form No No No YesLog-likelihood -290.198 -240.182 -220.298 -201.094AIC 600.396 514.363 476.595 472.188No. of Cases 470 403 3402

Note: Survey data from Frye (2006). Firm investment is a dummy variable where 1 indicates the firm plans to invest during the comingyear. Coefficients represent estimates from multilevel logistic regressions with a random coefficient for the bureaucratic discretionvariable and random intercepts at the region level; standard errors in parentheses. Out of space concerns, region-specific effects andsector/legal-form variables not reported. * p , 0.10, ** p , 0.05, *** p , 0.01


all tax rate variables are statistically significant andnegative, suggesting that high tax rates provide dis-incentives to managers to sink capital into new invest-ments. Together, the strong performance of both thebureaucratic discretion and tax rate variables suggeststhat policy environments can affect investment viamultiple channels and that investors’ concerns aboutcontent and application might reinforce each other. Incontrast, neither frequent changes to laws nor bureau-cratic corruption (Model 3) bear statistically signifi-cant relationships to firm investment in thesedata, countering the expectations of conventionalexplanations.

The models reveal a nuanced relationship be-tween government institutions and the investmentclimate. The probability of firms’ intent to invest hasa significant, positive association with expressedsupport for the regional administration, while Mod-els 2 and 4 suggest that, ceteris paribus, supportiveviews of the regional governor and investment planscorrelate negatively. Such findings could arise iffirms’ noninvestment relates to frustrations that apoor-quality regional bureaucracy is hampering thegovernor’s good policies.20 In regards to economiccourts, however, the analyses reveal no significantassociation between respondents’ assessments of re-gional economic courts and decisions to invest.

The economic variables have mixed success inpredicting investment. Problems with finance, com-petitive pressures, labor shortages, and status as aprivatized firm have no statistically significant rela-tionship with investment in any model. Privateownership and increasing sales trends, on the otherhand, do have significant, positive relationships withinvestment in earlier models, although uncertaintyabout this relationship grows as more controls areadded. Of the main economic variables, only largerfirm size is consistently estimated to have a signifi-cant, positive association with firms’ plans to invest,reflecting perhaps smaller firms’ difficulties withfunding the expansion of their business operations.

To test the sensitivity of these findings to model-ing assumptions and specifications, I conduct anumber of robustness checks. Investigating whetherthe estimated relationship between perceived discre-tion and firms’ intent to invest is spurious due tounobserved sector- or firm-specific factors, Model 4

controls for economic sector or firms’ legal form.Model 4 also adds a dichotomous indicator for pastinvestment out of concerns that past investment givesoccasion for encounters with regional bureaucracyand affects investment strategies for the future. Theinclusion of these controls do not change the resultsmeaningfully. Results are also robust to alternatemodeling strategies, such as basic logistic regression,as well as to the inclusion of additional substantivecontrols. These additional robustness checks arereported in the appendix.

A Conditional Model of BureaucraticDiscretion & Investment

In support of my hypothesis, the previous analyseshave demonstrated that a robust association betweenfirm managers’ perception of regulatory bureaucrats’discretion and lower probabilities of firm investment.Since this finding within a developing country suchas Russia appears to contradict existing research thathas drawn heavily from the experience of developedeconomies, it is worth investigating why these resultsdepart from prominent arguments. To what extentdoes the relationship between discretion and invest-ment depend on the broader institutional context? Inmaking the case that uncertainty over the applicationof laws deters investors who seek predictable environ-ments, I have claimed that uncertainty surroundingbureaucratic discretion is a particularly acute prob-lem for economic actors in environments wheresupporting institutions do not provide effective fire-alarm oversight.

In this section, I investigate the proposition thatthe absence of certain institutional checks makesbureaucratic discretion especially problematic forinvestors. Using a multilevel modeling strategy, Icombine the previous firm-level survey data withinformation on the 11 sampled regions to examinehow the relationship between bureaucratic discretionand firm investment is affected by the degree towhich regional institutions allow for open and activepolitical competition.21 Critically, variation in thecompetitiveness of regional politics presents a directtest of the delegation-as-commitment logic. Conven-tional credible commitment arguments suggest that

20Additional analyses support this interpretation. Regressing thedependent variable on a measure for the difference between tworatings shows that higher ratings for the governors relative to theregional administration correlate negatively with firms’ invest-ment plans (p 5 0.06).

21Multilevel models provide flexibility in modeling hierarchicalrelationships, helping us to make inferences about variables thatwe observe at different levels of analysis (Gelman and Hill 2007).The nested relationship of firms within regions makes this anatural modeling choice here.


political leaders’ moral hazard problem is high wherepoliticians face few constraints on ex-post behavior(Jensen 2006; Miller 2000). This logic implies that, byproviding an autonomous counterbalance to rela-tively unconstrained leaders, delegation to independ-ent agents should encourage firm investment in thoseplaces where political leaders lack organized politicalopposition or have little threat of losing office. Incontrast, my argument predicts that, because closedpolitical processes and unaccountable governmentinhibit the effectiveness of diffuse monitoring of stateagents, more discretion to regulatory bureaucrats willdeter investors by increasing policy uncertainty.

To measure political competition in firms’ regions,I use a regional democracy index created for use in theMoscow Carnegie Center’s Regional Monitoringproject. The original measures expert assessment ofRussia’s regions along 10 different dimensions at fourtime periods between the years 1991–2006. For thisanalysis, I use the time period immediately prior to thesurvey (2000–2004), constructing an additive index ofregional political competition that captures threedimensions of the political environment: representa-tive elections (existence of free and fair elections, fewlimitations on political rights), the openness of polit-ical life (the extent of transparency and public involve-ment in the political sphere), and pluralism(participation by stable parties or legislative factionsbefore and after elections). Each of these three scorescan range from 1 to 5 (ordered worst to best); whencombined and mean-centered for ease of interpretingresults, the lowest score is a -3 (Khabarovsk Krai andthe Republic of Bashkortostan) and the highest is 5(Sverdlovsk Oblast). Summary statistics appear in theappendix.22

Results

Is there evidence that bureaucratic discretion presentsa particular problem for economic actors in settingswith less democratic institutions? Even before intro-ducing statistical controls, a basic comparison of firm

investment in high- versus low-competition regionsindicates that this is, in fact, the case. In highlycompetitive regions, there is no statistically significantdifference between managers with differing percep-tions of bureaucratic discretion (p 5 0.658). Inregions with little political competition, however,only 25.8% of firm managers that perceive bureau-crats to have high discretion report to have invest-ment plans for next year, as compared to 42.8% ofmanagers who see bureaucrats as having low dis-cretion (x2(1) 5 8.281, p 5 0.004). Although firms inhigh-competition regions seem less concerned aboutbureaucrats’ discretion over policy, firms’ percep-tions of regional bureaucrats’ discretion in applyingand interpreting laws appear to be strongly related toinvestment decisions in low-competition regions.

The models in Table 2 extend the analysis to amultilevel logistic regression model of firm invest-ment. The new interactive model adopts the samerandom-intercept, random-coefficient specification asearlier and includes all variables as specified in Table 1,plus three additional region-level variables: the measureof regional political competition, its cross-level in-teraction with managers’ perceptions of bureaucraticdiscretion, and a (logged) variable for average re-gional GDP per capita (2002–2004) to help controlfor the level of regional economic development.

Beginning first with the control variables, thenegative coefficient estimates on high tax rates andgovernor approval in this new model are statisticallysignificant, as are the positive coefficients on firms’rating of the regional administration and firm size.Sales trends, private ownership, and regional GDPper capita also display positive relationships withfirms’ intent to invest, although the statistical un-certainty around these estimates increases as morecontrols enter the analyses.

Figure 2 uses a predicted probability graph todiscuss the estimated relationship between perceivedregulatory discretion, institutional constraints, andfirms’ willingness to invest. Holding all variables attheir sample medians, I manipulate whether a hypo-thetical firm perceives that regional bureaucrats havehigh or low discretion and then predict the probabilityof firm investment for different levels of regionalpolitical competition. In the sample’s least politicallycompetitive environment, a hypothetical firm managerwho perceives regional bureaucrats to have high dis-cretion has a predicted probability of investment that ismore than 50 percentage points lower than an identicalcounterpart who reports regional bureaucrats as havingno discretion. As the level of regional political competi-tion increases, however, this gap diminishes as firms’

22The original Petrov and Titkov measure also includes compo-nent scores for economic liberalization, corruption, civil society,elite recruitment and coordination, municipal governance, mediafreedom, and ‘‘regional political structure.’’ To maintain con-ceptual focus on electoral accountability and political participa-tion, I leave these components out of the analysis. Results arerobust, however, to using the full set of measures. For an English-language discussion of data collection and basic descriptions,albeit regarding earlier versions of the dataset, see McMann andPetrov (2000).


TABLE 2 Political Context Conditions Investors’ Response to Discretion

Firm Investment (5) (6) (7) (8)

Bureaucratic Discretion -0.474***(0.135)

-0.645***(0.157)

-0.614***(0.168)

-0.672***(0.184)1 5 no discretion, 4 5 high discretion

Frequent Changes to Laws 0.115(0.112)

0.082(0.118)


High Tax Rates -0.385***(0.122)

-0.417***(0.130)

-0.379***(0.144)1 5 no obstacle, 5 5 very serious obstacle

Regional Administration 0.508***(0.196)

0.433**(0.210)

0.450*(0.230)1 5 poor job, 5 5 excellent job

Regional Courts -0.149(0.148)

-0.169(0.152)

-0.022(0.174)1 5 poor job, 5 5 excellent job

Regional Governor 0.003(0.104)

-0.372**(0.185)

-0.328*(0.198)

-0.441**(0.222)1 5 poor job, 5 5 excellent job

Access to Finance -0.087(0.067)

-0.027(0.083)

-0.023(0.086)


Labor Shortages -0.046(0.086)

-0.007(0.091)


Competitive Pressures 0.089(0.085)

0.109(0.089)


Privatized Firm 0.078(0.275)

0.081(0.283)

0.206(0.346)dummy, 1 5 privatized, former SOE

Annual Sales 0.469***(0.165)

0.459**(0.182)

0.357*(0.191)

0.275(0.219)-1 5 decreasing, 1 5 increasing

Firm Size 0.338***(0.072)

0.306***(0.083)

0.294***(0.087)

0.333***(0.115)number of employees (logged)

Private Firm 0.766**(0.336)

0.774*(0.441)

0.620(0.444)dummy, 1 5 private ownership

Bureaucratic Corruption -0.071(0.252)dummy, 1 5 perceived as corrupt

Past Investment 2.049***(0.295)dummy, 1 5 invested in last 3 yrs.

Constant -3.081***(1.155)

-2.149(1.436)

-1.577(1.565)

-3.920*(2.133)

GDP per capita 0.330(0.245)

0.440*(0.252)

0.421(0.268)

0.445(0.308)in constant 2000 rubles per 1000 persons (logged)

Regional Political Competition -0.139(0.111)

-0.185(0.127)

-0.148(0.130)

-0.188(0.149)mean-centered index, -3 5 uncompetitive,

5 5 highly competitivePolitical Competition 3 Bureaucratic Discretion 0.124**

(0.059)0.144**

(0.066)0.124*

(0.066)0.172**

(0.080)interaction

Dummies for Sector & Legal Form No No No YesLog-likelihood -286.738 -235.432 -216.731 -195.699AIC 599.477 510.864 475.463 467.399No. of Cases 470 403 365 402

Note: Survey data from Frye (2006). Firm investment is a dummy variable where 1 indicates the firm plans to invest during the comingyear. Models also include region-level political data from the Moscow Carnegie Center and economic data from Rosstat. Coefficientsrepresent estimates from multilevel logistic regressions with a random coefficient for the bureaucratic discretion variable and randomintercepts at the region level; standard errors in parentheses. Out of space concerns, region-specific effects and sector/legal-formvariables not reported. * p , 0.10, ** p , 0.05, *** p , 0.01


predicted probability of investing converges steadilytowards that of the low-discretion profile. In fact, whereregional institutions foster open and competitive poli-tics, the overlapping confidence intervals suggest thedifference between a high-discretion (the solid line) andlow-discretion scenario (the dotted line) is negligiblefrom a statistical viewpoint.

These results support the empirical predictions ofmy argument, but run directly counter to conventionallogic that, given unconstrained political leaders, invest-ors should welcome greater independence for executiveagents as a type of substitute for political constraints onleaders’ ex-post behavior. Quite the opposite appearstrue in these data—the good that substitutes for desirablepolitical institutions is less discretion rather than more.This finding puts scholars’ seemingly sanguine attitudetowards bureaucratic discretion into perspective. In thedeveloped democracies that inform a majority ofstudies, high-quality institutions attenuate investors’uncertainty about independent bureaucrats’ behavior.On the spectrum’s other end where much of theworld lives, economic actors in the low-democracy,high-discretion scenario find themselves in the worst

of all worlds—high uncertainty over the application ofregulatory rules coupled with low-quality regionalinstitutions that offer limited political channels forresolving investors’ difficulties.

The results in Table 2 are robust to a variety ofestimation strategies and model specifications. Addi-tional analyses show that using alternate variablesthat should be related to political competition, suchas measures of civil society’s strength or proportionalrepresentation rules that increase the voice of minoritygroups, produce qualitatively similar results. Usingthe index’s three components separately instead of theentire index creates no substantive changes in theoutcomes, although the estimate on the interactionterm is associated with higher statistical uncertainty inmodels using the index’s election component only.Following Brown, Searle, and Gehlbach (2009), Icontrol for regional bureaucrats per capita to checkthe possibility that bureaucracy size may influenceinvestors’ opinion of both bureaucratic discretionand regional investment climate, but neither thisvariable nor additional region-level controls, such aspopulation or regional infrastructure (measured byrailway density) change the results. The empiricalrelationships do not depend on the chosen modelingstrategy, either, producing similar results in bothnonmultilevel logistic regression with region covariatesor, alternatively, multilevel analysis via Bayesian esti-mation with diffuse priors. Results for robustnesschecks appear in the appendix.

Conclusion

This article argues that bureaucratic discretion con-strains many localities’ ability to attract investment.As policymakers’ agents, bureaucrats interpret andapply regulations, but not always in a way that isconsistent with politicians’ legislative intent. Therefore,even knowing the content of a particular policy,investors may have little information about howdiscretionary bureaucrats will choose to implementand enforce that policy, either today or in the future. Ininstitutional environments where the absence of diffusemonitoring makes this uncertainty acute, private mar-ket actors will invest less because they cannot formstable forecasts over their long-term investments.

Using survey responses from over 600 enterprisesoperating across Russia, I have shown that firms areless likely to invest in fixed capital assets when theyperceive that regulatory bureaucrats in their region takedecisions independently of other government bodies.Furthermore, I find this negative relationship to be

FIGURE 2 Investors’ Response to BureaucraticDiscretion Shaped by InstitutionalContext (Predicted Probabilities)


especially pronounced for firms in regions with limitedpolitical competition, suggesting that greater independ-ence for agents in such circumstances only heightensinvestors’ uncertainty about the policy environment.

This article makes several useful contributions.First, to the general political economy literature, itreintroduces a group of oft-overlooked governmentactors—rank-and-file bureaucrats—and highlights po-tential benefits to considering more frequently theirrole in shaping economic outcomes. For instance, in aconventional treatment of noncredible commitments,the theoretical emphasis lies on politicians’ time-inconsistent preferences and their inability to commitcredibly not to reverse policy after investors have sunktheir quasi-irreversible investments. Rather than focuson politicians’ intertemporal dilemma, my argumentmakes a separate point rooted in the division betweenthe negotiating parties and those charged with imple-menting the deal: even without the time-inconsistencyproblem, political promises may not yield predictableresults if principals have weak control over the agentscharged with carrying out those promises. To theextent such commitments are also ‘‘noncredible,’’ itsuggests that there are gains to be had from conceptu-alizing a broader family of credibility problems thatwould include a role for agency issues.

Second, for research programs where bureauc-racy already plays a central role, this article addsnuance to our understanding of how bureaucracyinfluences economic behavior. I offer post-Sovietstudies of bureaucracy as an example. Within theliterature on postcommunist development, bureau-cratic institutions have been criticized as ‘‘grabbinghands’’ that drive entrepreneurs into the unofficialeconomy (Fyre and Zhuravskaya 2000) and forcebusinesses to pool resources in common defense(Duvanova 2007). At the same time, others havedefended bureaucrats as ‘‘helping hands’’ that pro-vide infrastructure and support to companies makingtheir way in a new economic system (Brown, Searle,and Gehlbach 2009). This article highlights a unifyingframework for the opposing camps—bureaucrats’ability to shape the predictability of firms’ regulatoryenvironment—and encourages us look for variationin the broader institutional context that will helpexplain when bureaucratic discretion should mitigateor magnify economic uncertainty.

Finally, in finding that the negative relationshipbetween discretion and investment is strongest inregions where leaders are least constrained by polit-ical competition, this research produces a result thatis somewhat counterintuitive from the standpoint ofconventional arguments: delegation to independent

agents may be counterproductive in precisely thoseplaces where the literature expects it to help most.Rather, the findings suggest the benefits of extensivedelegation may only begin to outweigh the negativeswhere surrounding institutions can help reduce in-vestors’ uncertainty about how regulatory bureau-crats will use their discretion. While this study hasfocused on the diffuse monitoring encouraged bypolitical competition, future research can build uponthese findings by elaborating additional mechanismsby which supporting political institutions can affectthe predictability of policy implementation.

Acknowledgments

I thank Daniel Blake, Noah Buckley, Ethan Bueno deMesquita, Dinissa Duvanova, Timothy Frye, ScottGehlbach, Marcus Kurtz, Irfan Nooruddin, WilliamMinozzi, Scott Powell, Graeme Robertson, JoshuaTucker, and Craig Volden for helpful suggestions anddiscussion. Special thanks go to Timothy Frye forsharing his data, three anonymous referees at theJournal of Politics, and the Department of PoliticalScience at The Ohio State University. Any remainingerrors are my own.

References

Aizenman, Joshua, and Nancy Marion. 1993. ‘‘MacroeconomicUncertainty and Private Investment.’’ Economic Letters 41 (2):207–10.

Aizenman, Joshua, and Nancy Marion. 1999. ‘‘Volatility and theInvertment Response.’’ Economica 66: 157–79.

Amsden, Alice. 1989. Asia’s Next Giant: South Korea and LateIndustrialization. Oxford: Oxford University Press.

Bawn, Kathleen. 1997. ‘‘Choosing Strategies to Control theBureaucracy: Statutory Constraints, Oversight, and the Com-mittee System.’’ Journal of Law, Economics, and Organizations13 (1): 101–26.

Bendor, Jonathan, Amihai Glazer, and Thomas Hammond. 2001.‘‘Theories of Delegation.’’ Annual Review of Political Science 4:235–69.

Bertelli, Anthony, and Andrew Whitford. 2009. ‘‘PerceivingCredible Commitments: How Independent Regulators ShapeElite Perceptions of Regulatory Quality.’’ British Journal ofPolitical Science 39 (3): 517–37.

Brehm, John, and Scott Gates. 1997. Working, Shirking, andSabotage: Bureaucratic Response to a Democratic Public. AnnArbor: University of Michigan Press.

Brown, David, John Searle, and Scott Gehlbach. 2009. ‘‘HelpingHand or Grabbing Hand? State Bureaucracy and PrivatizationEffectiveness.’’ American Political Science Review 103 (2):264–83.

Cukierman, Alex, Steven Web, and Bilin Neyapti. 1992. ‘‘Measur-ing the independence of central banks and its effect on policyoutcomes.’’ The World Bank Economic Review 6 (3): 353–98.


Duvanova, Dinissa. 2007. ‘‘Bureaucratic Consuption and Collec-tive Action: Business Associations in the Post-CommunistTransition.’’ Comparative Politics 37 (4): 441–61.

Frye, Timothy. 2006. ‘‘Original Sin, Good Works, and PropertyRights in Russia.’’ World Politics 58 (8): 479–504.

Frye, Timothy, and Ekaterina Zhuravskaya. 2000. ‘‘Rackets,Regulation, and the Rule of Law.’’ Journal of Law, Economics,and Organization 16 (2): 478–502.

Gailmard, Sean, and John W. Patty. 2007. ‘‘Slackers and Zealots:Civil Service, Policy Discretion and Bureaucratic Expertise.’’American Political Science Review 51 (4): 873–89.

Gelman, Andrew, and Jennifer Hill. 2007. Data Analysis UsingRegression and Multilevel/Hierarchical Models. Cambridge:Cambridge University Press.

Huber, John D., and Charles R. Shipan. 2002. DeliberateDiscretion? The Institutional Foundations of BureaucraticAutonomy. Cambridge: Cambridge University Press.

Huber, John D., and Nolan McCarty. 2004. ‘‘BureaucraticCapacity, Delegation, and Political Reform.’’ American Polit-ical Science Review 98 (3): 481–94.

ICRG. 2010. ‘‘International Country Risk Guide.’’ Accessed July2010: http://www.prsgroup.com/ICRG.aspx.

Jensen, Nathan M. 2006. Nation-States and the MultinationalCorporation: A Political Economy of Foreign Direct Investment.Princeton, NJ: Princeton University Press.

Johnson, Chalmers A. 1982. MITI and the Japanese Miracle:the Growth of Industrial Policy, 1925-1975. Palo Alto, CA:Stanford University Press.

Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2009.‘‘Governance matters VIII: Aggregate and Individual Gover-nance Indicators, 1996-2008.’’ Policy Research Working Paper4978. Washington, DC: The World Bank.

Knack, Stephen, and Philip Keefer. 1995. ‘‘Institutions andEconomic Performance: Cross-Country Tests Using AlternativeInstitutional Measures.’’ Economics & Politics 7 (3): 207–27.

Kohli, Atul. 2010. ‘‘State Capacity for Development.’’ Global EventWorking Paper. United Nations Development Programme.

Kurtz, Marcus, and Andrew Shrank. 2007. ‘‘Growth and Gover-nance: Models, Measures, and Mechanism.’’ Journal of Politics69 (2): 538–54.

Leff, Nathaniel H. 1964. ‘‘Economic Development Through Bureau-cratic Corruption.’’ American Behavioral Scientist 8 (3): 8–14.

Levitsky, Steven, and Maria Victoria Murillo. 2009. ‘‘Variation inInstitutional Strength.’’ Annual Review of Political Science 12:115–33.

Levy, Brian, and Pablo Spiller. 1994. ‘‘International trade andinternational investment in the product cycle.’’ Journal ofLaw, Economics, and Organizations 10 (2): 201–46.

Lewis, David E. 2003. Presidents and the Politics of Agency Design:Political Insulation in the United States Bureaucracy,1946-1997. Palo Alto, CA: Stanford University Press.

Li, Quan. 2006. ‘‘Democracy, Autocracy, and Tax Incentives toForeign Direct Investors: A Cross-National Analysis.’’ Journalof Politics 68 (1): 62–74.

Lipsky, Michael. 1980. Street-Level Bureaucracy. New York:Russell Sage Foundation.

Malesky, Edmund, and Krislert Samphantharak. 2008. ‘‘Predict-able Corruption and Firm Investment: Evidence from aNatural Experiment and Survey of Cambodian Entrepre-neurs.’’ Quarterly Journal of Political Science 3 (3): 227–67.

Mauro, Paolo. 1995. ‘‘Corruption and Growth.’’ QuarterlyJournal of Economics 110 (3): 681–712.

McCubbins, Matthew, Roger Noll, and Barry Weingast. 1987.‘‘Administrative Procedures as Instruments of Political Con-trol.’’ Journal of Law, Economics, & Organization 3 (2): 243–77.

McCubbins, Matthew, and Thomas Schwartz. 1984. ‘‘Congres-sional Oversight Overlooked: Police Patrols Versus FireAlarms.’’ American Journal of Political Science 28 (1): 165–79.

McMann, Kelly, and Nikolai Petrov. 2000. ‘‘A Survey of Democ-racy in Russia’s Regions.’’ Post-Soviet Geography and Econom-ics 41 (3): 155–82.

Miller, Gary J. 2000. ‘‘Above Politics: Credible Commitment andEfficiency in the Design of Public Agencies.’’ Journal of PublicAdministration Research and Theory 10 (2): 289–328.

Nooruddin, Irfan. 2011. Coalition Politics and Economic Develop-ment: Credibility and the Strength of Weak Governments.Cambridge: Cambridge University Press.

Rauch, James. 1995. ‘‘Bureaucracy, Infrastructure, and EconomicGrowth: Evidence from U.S. Cities During the ProgressiveEra.’’ American Economic Review 85 (4): 968–79.

Rauch, James, and Peter Evans. 1999. ‘‘Bureaucracy and Growth:A Cross-National Analysis of the Effects of ‘‘Weberian’’ StateStructures on Economic Growth.’’ American SociologicalReview 64 (5): 748–65.

Remington, Thomas F. 2006. Democratization, Separation ofPowers, and State Capacity. In The State After Communism:Governance in the New Russia, ed. Timothy J. Colton andStephen Holmes. Rowman & Littlefield Publishers, Inc.,261–98.

Rogoff, Kenneth. 1985. ‘‘The Optimal Degree of Commitment toan Intermediate Monetary Target.’’ The Quarterly Journal ofEconomics 100 (4): 1169–89.

Shipan, Charles R. 2004. ‘‘Regulatory Regimes, Agency Actions,and the Conditional Nature of Congressional Influence.’’American Political Science Review 98 (3): 467–80.

Slinko, Irina, Evgeny Yakovlev, and Ekaterina Zhuravskaya. 2005.‘‘Laws for Sale: Evidence from Russia.’’ American Law andEconomics Review 7 (1): 284–318.

Solomon, Peter. 2008. ‘‘Law in Public Administration: HowRussia Differs.’’ Journal of Communist Studies and TransitionPolitics 24 (1): 115–35.

Treisman, Daniel. 2007. ‘‘What Have We Learned About theCauses of Corruption From Ten Years of Cross-NationalEmpirical Research?’’ Annual Review of Political Science 10:211–44.

Varese, Federico. 2001. The Russian Mafia. Oxford: OxfordUniversity Press.

Wei, Shang-Jin. 1997. ‘‘Why is Corruption So Much MoreTaxing than Tax? Arbitrariness Kills.’’ Working Paper 6255.National Bureau of Economic Research.

Weingast, Barry, and Mark Moran. 1983. ‘‘Bureaucratic Discre-tion or Congressional Control? Regulatory Policymaking bythe Federal Trade Commission.’’ Journal of Political Economy91: 765–800.

Yakovlev, Evgeny, and Ekaterina V. Zhuravskaya. 2008. ‘‘Reformsin Business Regulation: Evidence from Russia.’’ Available atSSRN: http://ssrn.com/abstract5965838.

Quintin H. Beazer is Postdoctoral Associate atYale University, New Haven, Connecticut 06511,and Assistant Professor at Florida State University,Tallahassee, Florida 32306.


Bureaucratic Discretion, Business Investment, and...

Documents

Transcript of Bureaucratic Discretion, Business Investment, and...