ORIGINAL PAPER

Export behavior of German SMEs in the Eurozone

Kenan Bagci

Published online: 8 December 2012# Springer-Verlag Berlin Heidelberg 2012

Abstract The paper investigates the export behavior of German small and medium-sized enterprises (SMEs) in response to the introduction of Euro and the subsequentfall in trade costs. Based on a testable prediction derived from Helpman et al.(American Economic Review 94:300–316, 2004) and Helpman (Journal of EconomicLiterature 44:589–630, 2006), the paper argues that the likelihood of SMEs tobecome an exporter increases after the monetary union. By using an empirical probitestimation technique, it is found that SMEs, and especially medium-sized enterprises,indeed benefit from cost reductions in entering into export markets. What is more,firms operating in furniture, electrical equipment, plastics, medical instruments,machinery and textiles are more likely to benefit from monetary union. These resultsare robust to using exchange rate volatility as a proxy for the cost reductions due tomonetary union.

Keywords MonetaryUnion . Export Behavior . Small andMedium-sized Enterprises

JEL Codes F15 . F33 . D21 . F23

1 Introduction

Monetary unions extensively improve the economic environment in which firmsoperate, mainly through elimination of transaction costs and exchange rate uncer-tainty, increase in price transparency and possibility to exploit economies of scale at alarger market. Additionally, firms save on administrative costs, costs arising from

Int Econ Econ Policy (2013) 10:613–629DOI 10.1007/s10368-012-0227-2

Much of the work for this paper was done during my PhD study at the University of St. Gallen. Specialthanks go to Simon Evenett for his valuable orientation and two anonymous referees for their usefulcomments. I also thank to Sandra Gottschalk from ZEW for the gracious provision of the data.

K. Bagci (*)SESRIC, Kudüs Cd. No:9, 06450 Ankara, Turkeye-mail: kbagci@sesric.org

technical regulations being different, costs to obtain information (information costs),and costs from fragile financial conditions.1 Greater nominal exchange rate stability,lower transaction costs, and price transparency2 reduce information costs and therebyenhance competition and increase international competitiveness of enterprises.

Managing financial flows at a lower cost is particularly important for small andmedium-sized enterprises that are less able to benefit from economies of scale. Inmanaging exchange rate risks, smaller firms are less able to hedge than larger firmsbecause of three main reasons. The use of derivative markets is more costly forsmaller transactions. They are also less likely to be diversified in terms of thecurrencies in which they transact. Lastly, small firms are less likely to have thefinancial resources to absorb adverse currency movements. Exchange rate uncertaintydiscourages firms from selling in foreign markets due to a lack of price transparency aswell. Although the price will be known at the point of transaction when paid forimmediately, the actual price in terms of domestic currency will remain unknown. Thiswill apply to many transactions between firms, where payment is often made a month ormore after delivery of the goods. Therefore, monetary union is especially beneficial forsmall firms’ trade as it improves the operational capability by removing currency risk,increasing transparency, and reducing transaction costs within the union.

In this context, this paper analyzes the export behavior of firms in response to theintroduction of single currency in the case of German SMEs. As explained in the nextsection, the scope of the extensive works on trade impacts of monetary unionremained remarkably narrow. This paper takes another step to go beyond the studiesof how big is the magic and investigates the impact on the export behavior of smalland medium-sized enterprises, in the case of the German SMEs in the Eurozone.Following the theoretical approach developed by Helpman (2006) and Helpman et al.(2004), theoretical prediction on the likelihood of small and medium sized enterprisesto become an exporter is derived in section 3. The prediction is then tested by usingMannheim Innovation Panel (MIP) dataset provided by ZEW. The data descriptionand the methodology are explained in section 4. Before concluding, section 5presents the estimation results.

2 Related literature

Forming a monetary union between countries having certain monetary independenceis practically complicated and also controversial as it entails both benefits and costs.Though the topic may be linked to many research questions, its association with tradeflows became one of the major research subjects of the last decade. The scope of theextensive works on trade impacts of monetary union, however, remained remarkablynarrow. The previous studies focused typically on the volume effect of monetary

1 Even the anticipation of entry into EMU has produced a remarkable shift in the interest rate structure insome countries by substantially narrowing the risk premia vis-a-vis German government securities. This initself would reduce the interest burden on government debt and serve as stimulus to investment by theprivate sector.2 Increased price transparency may have far-reaching effects, on the capability, willingness, and desire ofenterprises to change their production and operation processes. It may reduce the number of distributionand storage facilities, or even may change their operating bases for multinational enterprises.

614 K. Bagci

union on trade at aggregate level, without paying attention to the particulardynamics that may lead to higher trade volumes. Only recently there appearedsome works completed by utilizing micro data and some sectoral analyses havebeen conducted.

The empirical literature on trade effects of monetary union concentratingtypically on the volume effect of trade generally exploit gravity equations invarious forms to estimate the magnitude. Baldwin (2006) provides an outstandingreview of literature and discusses some of the important mistakes made in the previouspapers. The results found in majority of the papers reveal clear evidence in favor ofhigher trade volumes betweenmember countries. It is, however, hardly possible to find adiscussion why and how should that occur except a few papers (e.g., Alesina and Barro2002).

Empirical studies analyzing trade effects of monetary union specific for theEuro area usually present positive effects. Micco et al. (2003) find 5–10 %increase in bilateral trade among member countries, and even more cautious studyby Bun and Klaassen (2004) suggests a 3 % increase in trade by considering the trendeffect as well. There are, however, controversial views in explaining the increase intrade. Some authors, including Nitsch (2002) and Berger and Nitsch (2005), arguethat introduction of the Euro simply coincided with an accelerating process ofEuropean-wide political and institutional reforms favoring trade. Others, includingBaldwin (2006), argue that the effect pertains more closely to the Euro itself and a risein the number of exporting firms in a given country is likely to be the key to explaintrade creation.

In the same fashion, Fontagné and Freudenberg (1999) find that the elimination ofexchange rate variability has fostered product differentiation in European trade: i.e.,intra-industry trade is occurring more in horizontally differentiated goods (two-waytrade in varieties) than in vertically differentiated goods (two-way trade in qualities).Nguyen et al. (2007) find that the introduction of the Euro is associated with areduction in both the number of firms that have significant foreign exchange exposureand the magnitude of exposure. Baldwin and Taglioni (2004) show that exchange ratevolatility naturally hinders exporting by small firms, so reduced volatility tends toespecially promote exports from small firms. These results provide the first support-ing evidence for the prediction.

In addition to the analysis conducted at aggregate level, there are fewempirical studies conducted at sectoral level. De Nardis et al. (2008) conductan empirical analysis to study the effect of the single currency across industries foreuro area members and find that the euro effect was not uniformly distributed amongsectors: only in 11 industrial sectors out of 25 there was a positive and significantimpact of the euro on export flows. Flam and Nordström (2006) estimate currencyunion effects at different stages of processing and for different industries, findingevidence of a positive effect for semi-finished and finished products and for industriescharacterized by highly processed products, such as pharmaceuticals and machinery.

Berthou and Fontagné (2008) use French firm level data to analyze the microeffects of the euro. They find that the euro has a positive effect on the extensivemargin, but no effect on the intensive margin. Moreover, the descriptive analysis ofthe French data reveals that the effect on the extensive margin is a result of theincrease in the number of products exported by a firm rather than the effect of an

Export behavior of German SMEs in the Eurozone 615

increase in the number of exporting firms. The most closely related study to thispaper, on the other hand, belongs to Esteve-Pérez et al. (2011). They analyze theimpact of the euro on the relationship between firm size and exports using arepresentative sample of Spanish manufacturing firms. Their results indicate thatthe introduction of the euro has remarkably weakened the role of firm size in thedecision to export to the Eurozone, suggesting that the euro has reduced the thresholdsize in order to export to Eurozone countries.

With respect to the mechanism driving export behavior of firms, heteroge-neous firm literature suggests that declining trade costs force low-productivityplants to exit the market either because of increased import competition fromforeign varieties (Bernard et al. 2003) or because of increased probability ofdeath at all levels of productivity after an increase in imports, while the deathof low-productivity plants is actually driven by the entry into exporting ofother domestic firms (Melitz 2003). As is especially clear in Melitz’s approach,decrease in trade costs reduces the cut-off level above which firms survive in exportmarkets and reduced cut-off level eases the access to foreign markets. A naturalimplication of lower trade costs is higher trade volumes, as verified by empiricalstudies.

3 Derivation of the prediction

Whatever the magnitude of reduction in trade costs, some firms will find theirway to export markets due to lower trade costs and other non-negligible trade-related cost reductions following monetary union. While the impact is unevenlyspread over firms with different sizes, SMEs are predicted to benefit over-proportionately. Below we first briefly discuss the importance of size in man-aging transactions in multiple currencies and then describe the testable predic-tion derived from Helpman (2006) and Helpman et al. (2004).

3.1 Transactions in multiple currencies

In order to illustrate the significance of foreign exchange rate volatility on firmperformance, we consider a simple cost function for managing foreign curren-cy transactions. Trade costs due to transactions in multiple foreign currenciescan be formulated for firm j as the sum of some fixed cost of foreign exchangerisk management (q), which includes personnel and administrative costs, andvariable costs of managing that risk depending on the size of transactions inforeign currency (k), volatility of exchange rate (σ), and rate of transaction costs (t),and other costs (ω) related to transactions in foreign currency, including informationcosts. Suppose the trade costs of a firm due to transactions in foreign currencies takethe form

V ðjÞ ¼ qþ f k;σ; t;wð Þ ð1ÞThe components of variable costs depend completely on the size of trans-

actions. When it comes to average costs, they have, therefore, no specialimplication for firms of different sizes. The total fixed cost of managing foreign

616 K. Bagci

currency transactions (q) is, however, independent of the firm’s output, but the sizeof firms matters for average fixed costs: the larger the size of transaction (and size offirm), the lower is the average fixed cost. Therefore, the fixed cost in a linear costfunction gives rise to economies of scale, because the larger the firm’s output, the lessis the fixed cost per unit.

The cost function in (1) makes it possible to study the behaviors of firms ofdifferent sizes. As discussed previously, SMEs are disadvantaged with respect to theirability to manage international transactions and relatively higher level of averagefixed costs inhibits profitable international operations for SMEs. Assuming monop-olistic competition, profit-maximizing firms have to meet the criteria that the mar-ginal revenue is equal to marginal cost plus marginal cost of transactions in foreigncurrencies (MR ¼ MC þ f 0 k;σ; t;wð Þ ) in the export market in order to sell profit-ably.3 After the elimination of the cost of transactions with multiple currencies, onlythe firms with MR� f 0 k;σ; t;wð Þ but with MR ≥ MC will be able to profitablyexport. Since P = AC, the cost reductions reduce the price level settled by SMEs morethan big enterprises due to over-proportional decrease in average costs and increasetheir competitiveness by generating price advantage for SMEs.

3.2 Derivation of the prediction on export behavior

After the short representation of the role of foreign currency transactions in exportbehavior of firms of different sizes, we turn to derivation of the prediction. Derivationof the prediction directly follows from the theoretical approach used in Helpman(2006) and Helpman et al. (2004). Following Helpman (2006), consider an industrysupplying a differentiated product with continuum of firms manufacturing differentbrands. The demand function for brand of firm j is xðjÞ ¼ ApðjÞ�" , with x indicatingthe quantity, p the price, A a measure of the demand level, and " ¼ 1= 1� að Þ thedemand elasticity, which is assumed to be constant with 0 < α < 1. Productivity levelθ(j) for firm j is discoverable only after the firm enters the industry. c/θ(j) is thevariable production cost per unit of output and cft is the fixed cost, with c measuringthe cost of resources and ft a measure of fixed production costs in terms of resources.Then, profit-maximizing strategy for the firm entering the market requires it to charge

pðjÞ ¼ c=aθðjÞ , yielding the operating profits of ðjÞ ¼ θðjÞ"�1Z � cft , where Z �1� að ÞA c=að Þ1�" .It follows that the profit function of a domestic firm is pd yð Þ ¼ yðjÞZ � cfd and

the profit function of an exporting firm px yð Þ ¼ t1�"yðjÞZx � cfx , where yðjÞ ¼θ"�1ðjÞ measures the productivity level of firm j and t measures the melting icebergtrading costs such that t>1 units have to be shipped for one unit to arrive. Figure 1depicts both profit functions for the case of t"�1fx > fd . When the two demand levelsare the same πd is steeper than πx as a result of the trading costs and assumption onthe relative size of the trading costs. It follows that firms with productivity levely < yd have to exit market, while firms with productivity above yd can profit fromserving the domestic market, but only firms with productivity y > yx can profitadditionally from exporting.

3 MC represents here all marginal costs except the costs due to transactions in multiple currencies.

Export behavior of German SMEs in the Eurozone 617

In this setting, trading costs, t, typically include transport costs, insurance,fees, transaction and hedging costs (when currencies differ), duties, and othercosts that may stem from the differences in language, legal system and the like.After monetary union, the transaction and hedging costs naturally disappear. Byassumption, this will lead to reductions in both fixed (cfx) and variable costs (t)parameters and result in a change in the pattern to become an exporter. If productivityadjusted fixed costs (cufx/y) are smaller for some firms than the updated trade costinclusive parameter ( t1�"

u A ), then these firms will subsequently become exporters.This is illustrated in Fig. 1. ya

x represents the initial productivity cut-off level forall firms. Reductions in variable trade costs rotate the profit line for exportersand yb

x becomes the new cut-off level. With reductions in fixed costs, the profit lineshifts upwards and yc

x becomes the final cut-off level. Rotation of the profit line hasno particular implication for firm size. However, the shift from yb

x to ycx improves

the likelihood of SMEs becoming exporters. Why this applies especially for SMEsis because of the fixed cost parameter (q) in Eq. (1). Average cost of doingbusiness with multiple currencies will be higher for SMEs than big companiesand SMEs will initially not be able to enter into export market due to higheraverage trade costs. As costs due to multiple currency transactions are eliminated,these firms will over-proportionately benefit from cost reductions. This leads toan increase in the likelihood of SMEs to become exporters (E) after monetaryunion (m), P E ¼ 1 Xj ;m ¼ 1ð Þ > P E ¼ 1 Xj ;m ¼ 0ð Þ .

For given productivity level for each firm independent of size, the numberof exporters as well as the probability to become an exporter will increaseafter productivity cut-off level decreases. Additionally, the reduction in aver-age costs will be higher for SMEs and these firms will get an additional costadvantage in export market participation. Since firms know their productivityonly after they enter into market, firms enjoying a higher reduction inaverage costs in addition to already lower productivity cut-off will over-proportionately increase their chance to survive in export market, implyingthat the probability of SMEs will increase compared to their previous status with

Fig. 1 Increase in the likelihoodof SMEs to become exporter

618 K. Bagci

higher productivity cut-off level and compared to other firms’ status with lowerproductivity cut-off level.4

Reductions in trade costs increase the profits that existing exporters can earn inforeign markets and reduce the export productivity cutoff above which firms export.There are two different mechanisms that may bring about an increase in the numberof exporting firms. The first one is the direct mechanism: if firms sell or want to sellabroad, their trade-cost-inclusive marginal costs will decrease and they will bettercompete with incumbent firms in foreign markets. The second one is an indirectmechanism and concerns the firms that do (or must) import some intermediary inputsfrom foreign markets to become more productive. For these firms, marginal costs willfall and they will automatically update to exporter status, provided that they areproductive enough. For other firms serving only the domestic market and not dealingwith foreign trade, the impact of the single currency on their marginal costs will beprobably limited. At the end, the share of non-exporters will decrease in totaleconomic activity.

4 Empirical approach

4.1 Data

Analysis of whether the likelihood of high productive SMEs to become an exporterincreases after monetary union requires a very detailed dataset broken down by firmsize. Since there is only limited data available for such an analysis, we use firm leveldata from a representative survey of the German manufacturing sector, the MannheimInnovation Panel (MIP) published by ZEW, to detect the exporter status of Germanfirms after monetary union. Data is available from 1994 to 2004 and is stratified by

4 From the same theoretical application of Helpman et al. (2004) and Helpman (2006), another testableprediction can be derived for the location decision of multinational enterprises (MNEs) and that can be adirection for further research. The logic goes as follows. Reductions in trade costs may also promote thetrade in raw materials and intermediary products. One outcome of this result is that the companies do notneed to be closely located to the places where they acquire, and probably process, these goods. Multina-tional enterprises usually have complex enterprise structure with facilities in more than one location andthey confront with a hard decision of being closer to local markets or making use of increasing returns toscale, generating a proximity-concentration trade-off.

Given a firm’s profit from FDI is pfdi yð Þ ¼ yðjÞZ � cffdi , by using the same notation, it isstraightforward to argue that whenever the condition px yð Þ > pfdi yð Þ or 1� t1�eð ÞyZ > cffdi � cfxapplies for MNEs, then they prefer concentrating spatially and serving remote markets from a centrallocation. After the elimination of certain trade costs with the introduction of single currency, if the gainfrom the change in trade costs overweighs the cost advantage of serving markets locally at givenproductivity level, Δt1�" yZð Þ þΔc fxð Þ > pfdi yð Þ � px yð Þ , then firms prefer concentration over proxim-ity. Though practically controversial, even a tiny change in trade costs can have great impact on theproductivity cut-off level needed in order to benefit from serving markets through foreign direct investment.

As in the prediction on the export behavior of SMEs, firms discover their productivity only after theyenter the market. The overall probability of random draw for high productivity firms—that prefer proximityto concentration—will decrease after the elimination of some export costs. Firms that otherwise wouldprefer FDI to export at a given probability level may, therefore, prefer concentrating instead of proximity,implying an increase in their overall probability of concentrating. Admittedly, in real economic life, theexpected change in the firms’ decision to serve foreign markets through FDI or export will not happenovernight. For some firms that will take perhaps years to adapt to new economic environment.

Export behavior of German SMEs in the Eurozone 619

firm size, therefore useful in analyzing the behavior of small and medium sizedenterprises. Table 1 provides the descriptive statics for exporters and non-exporters.As it is evident, exporters are more productive, more innovative, larger, and locatedmostly in the West.

The empirical literature finds evidence in favor of recent international tradetheories with heterogeneous firms by proving a robust positive correlation betweenproductivity at the firm level and exporting. Micro-evidence on this issue is availablefor a number of the Eurozone countries: for Spain (Delgado et al. 2002), for Italy(Castellani 2002), for Germany (Arnold and Hussinger 2005) and the German state ofLower Saxony (Bernard and Wagner 2001), and for France (Eaton et al. 2008). Thelimitation on the availability of longitudinal data at the firm level is one of the mostimportant challenges for researchers. Therefore, previous studies also worked com-monly with regional or country-level data, without having the possibility to workwith cross-country data.

4.2 Estimation methodology

When the dependent variable is dichotomous rather than continuous, ordinary leastsquares (OLS) becomes an inefficient estimation technique, and the underlying linearprobability model (LPM) that is being estimated represents a poor choice of modelspecification. The common solution to the deficiencies of the LPM model as esti-mated via OLS is to adopt a different model specification. The Probit model con-strains the estimated probabilities to be between 0 and 1, and relaxes the constraintthat the effect of independent variables is constant across different predicted values ofthe dependent variable.

The probit model assumes that while we only observe the values of 0 and 1 for theexport status (EXP), there is a latent, unobserved continuous variable EXP* thatdetermines the value of EXP. We assume that EXP* can be specified as: EXP*

i ¼a0 þ a1x1i þ a2x2i þ . . .þ akxki þ ui and that EXPi ¼ 1 if EXP*

i > 0 and EXPi ¼0 otherwise; where x1; x2; . . . ; xk represent vectors of random variables and u

Table 1 Descriptive Statistics(Mean) for Exporters andNon-exporters

Variable Exporters Non-exporters

Productivity 0.276 0.210

R&D Intensity 0.023 0.008

Wage 5.566 5.897

Non-innovator 0.223 0.532

Size 2.020 1.406

Export Intensity 0.291 0

Investment Intensity 0.068 0.0864

Sales (million €) 130.137 30.802

East Germany 0.276 0.494

Number of Employees 375.350 125.065

Number of Observations 5601 1939

620 K. Bagci

represents a random disturbance term. It is straightforward that Pr EXPi ¼ 1ð Þ ¼Pr a0 þ a1x1i þ a2x2i þ . . .þ akxki þ ui > 0ð Þ and Pr EXPi ¼ 1ð Þ ¼ 1� Pr ui < �ða0 þ a1x1i þ a2x2i þ . . .þ akxkið ÞÞ a n d Pr EXPi ¼ 1ð Þ ¼ 1� F � a0 þ a1x1iþðða2x2i þ . . .þ akxkiÞÞ where F is the cumulative density function of the variable u.If we make the usual assumption that u is normal distributed, we obtain PrðEXPi ¼1Þ ¼ 1�Φð�ða0 þ a1x1i þ a2x2i þ . . .þ akxkiÞÞ ¼ 1� Φð�XiaÞ ¼ ΦðXiaÞ , whereΦ represent the cumulative normal distribution function. Using maximum like-lihood techniques we can compute estimates of the coefficients (αs) and theircorresponding standard errors that are asymptotically efficient.

In order to estimate the probability of being exporter after monetary union forSMEs, we follow Arnold and Hussinger (2005) in selecting the appropriate covariatesand estimate an empirical probit model in which export behavior depends on a varietyof observed, firm-specific characteristics:

P EXPi;t ¼ 1� � ¼ Φ LPi;t�1;RDi;t�1;Filteri;t�1; Skilli;t�1; SMEi;t�1;Easti;t�1;MUt�1;Dt

� �

where Φ is a normal cumulative density function, LP is labor productivity, RD is theratio of the expenditures in research and development to turnover, Filter separates outthe non-innovator firms, Skill is the average wages and used as a proxy for averagelevel of skill in each firm, East is a dummy variable if a firm is located in EastGermany, MU is a dummy for the monetary union, and SME is a dummy for smalland medium sized enterprises where SME is defined as firms with employees lessthan 250. More explicitly, firms with less than 50 employees are considered to besmall enterprises and firms with more than 50 but less than 250 employees areregarded as medium-sized enterprises. All covariates are lagged one period as inArnold and Hussinger (2005). Finally, in addition to industry dummies to control forindustry-specific effects, we add year dummies to capture time-specific effects notspecific to an individual firm and not attributable to monetary union.

4.3 Identifying the Euro impact

There are both measurable and immeasurable benefits of greater economic integrationwith monetary union. Elimination of exchange rate volatility and subsequent hedgingcosts, elimination of transaction costs and some other information costs are the maincomponents of measurable part of the benefits. The huge literature on the impact ofexchange rate volatility on trade provides no clear evidence and the findings aremostly mixed (see, e.g., Tenreyro 2007). On the other hand, perception of living inlarger, more transparent, and expectedly more stable market bestow the economicagents with increased buoyancy and resilience that invigorate the enthusiasm to meetnew challenges. As a supporting premise to that idea, the story postulated by the“endogeneity of optimum currency area (OCA)” hypothesis has the implication that amonetary union may turn into an optimum currency area, even if it was not before(Frankel and Rose 1998). The basic intuition behind this hypothesis is that monetaryintegration reduces trading costs beyond the elimination of the costs from exchangerate volatility.

In estimating the impact of monetary union, we in general use dummy or interac-tion variables, which help to identify discrete changes in the variables of interest.

Export behavior of German SMEs in the Eurozone 621

Whether the identified impact is completely attributable to monetary union is,however, ambiguous. In order to deal with this concern, we use nominal exchangerate volatility as a proxy for the reductions in trade costs after monetary union andconduct alternative estimations to verify the initial findings. Higher volatility is likelyto depress the volume of trade as higher exchange rate risk lowers the expectedrevenue from exporting5 and monetary union eliminates this risk and validates the useof volatility as a proxy for the reductions in trade costs due to monetary union. Wemeasure exchange rate volatility against ECU/Euro as

σi ¼ std d log sið Þð Þ½ �;where si is the nominal exchange rate of country i against ECU/Euro. Explicitly,volatility is the standard deviation of the changes in the logarithm of bilateralexchange rates (as commonly defined in the literature, e.g., Gros and Thygesen1998) and constructed using monthly data over 1.1976 to 12.2008 (Fig. 2).6 Thismeasure has the property of being zero in the presence of an exchange rate thatfollows a constant trend as after monetary union, therefore it is a plausible proxy forreductions in trade costs after monetary union.

As depicted in Fig. 2, volatility in the Eurozone countries was small but non-negligible before the introduction of the Euro. On the other hand, in two formermembers of ECU with relatively higher volatility, nominal exchange rate volatilitydisappeared in Ireland after monetary union but stayed at remarkably high levels inUnited Kingdom. This clearly depicts the gains from monetary union in terms ofreductions in exchange rate volatility. It also validates the use of exchange ratevolatility as a proxy for the reductions in trade costs after monetary union. The areabetween the vertical lines shows the volatility for the period that is considered in theempirical analysis.

5 Findings

Hypothesizing that SMEs are more likely to become exporters after monetary unionis equivalent to say that they are less likely to export under given market conditionsdue to inherent boundaries. Starting with the relevance of size in probability of beingan exporter, Fig. 3 demonstrates the fact that big enterprises are more likely tobecome exporters at every level of productivity. At higher productivity levels, theprobability of being exporter for medium-sized enterprises is approaching to that forbig enterprises. Besides, firms located in East Germany are disadvantaged comparedto their counterparts in West Germany and their likelihood of becoming exporter is onaverage lower.

After presenting this crucial evidence on the relevance of size on being exporter,the main findings on the export behavior of SMEs are presented in Table 2, withstandard probit estimation results in (col. I, col. III and col. V) and marginal effects at

5 Risk-averse firms may reduce their foreign activity and relocate production toward domestic markets.6 In order to observe the changes in the volatility of exchange rates over a longer period, the range of data inthe figures is kept intentionally long. When using it as a proxy in empirical part, the span of the data fits tothe original dataset.

622 K. Bagci

the means in (col. II, col. IV, and col. VI). The probit regression coefficients give thechange in the probit index, also called a z-score or standard normal scores and can betransformed into predicted probabilities using tables of standard normal distribution.Z-score interpretation is useful when seeking to compare the relative standings of theunits with different means or standard deviations. The coefficients reported in termsof z-scores may be hard to interpret. The marginal effects at the means are relativelyeasier to interpret, which report the change in the probability for an infinitesimalchange in each independent, continuous variable and the discrete change in theprobability for dummy variables.

The results in col. I indicate that a one-unit increase in the productivity levelresults in a 1.55 standard deviation increase in the predicted probit index. Thecoefficient for size is interpreted to mean that the change from 0 to 1 reducesthe predicted probit index by 0.8 standard deviations. And the coefficient formonetary union indicates that after monetary union the predicted probit index

a Volatility in the Former Euro Area b Volatility in Ireland and UK0

.51

1.5

2

1976 1981 1986 1991 1996 2001 2006Year

05

1015

1976 1981 1986 1991 1996 2001 2006

Ireland United Kingdom

Vol

atili

ty

Vol

atili

ty

Fig. 2 Exchange rate volatility. a Volatility in the Former Euro Area. b Volatility in Ireland and UK.Former Euro Area countries include Austria, Belgium-Luxembourg, Finland, France, Germany,Italy,Netherlands, Portugal, and Spain

.4.6

.81

0 .2 .4 .6 0 .2 .4 .6

West Germany East Germany

Small Ent. Medium-sized Ent. Big Ent.

Productivity

Pre

dict

ed P

roba

bilit

y

Fig. 3 Predicted probability of exporting vs. firm size

Export behavior of German SMEs in the Eurozone 623

increases 0.24 standard deviations. In terms of marginal effects, the probabilityof SMEs is in general 18.4 % lower than other firms (col. II). However after1999, the likelihood of these firms entering foreign markets increases by 6.6 %.The findings additionally confirm that more productive firms are more likely tobecome exporters. Firms with higher skills (lower personnel cost per sales) andhigher R&D intensity are again more likely to enter into export markets.Likelihood of exporter status decreases by 12.3 % when firms are located inEast Germany and by 15.8 % when they are not innovators.

Columns IV and VI separate out the impact of small and medium-sized enterprisesrespectively. Higher productivity is associated with higher likelihood of becoming anexporter for both small and medium-sized enterprises. It is, however, small enter-prises that are particularly disadvantaged due to their size. After monetary union, theirlikelihood of becoming an exporter does not change statistically significantly (col.IV). The impact of single currency is stronger for medium-sized enterprises (col. VI),which indicates around 6 % increase in the probability of becoming an exporter. Sincemedium-sized enterprises are already at the boundary in terms of exporting potential,this result should not be surprising.

Table 2 Export Behavior of SMEs

Small & Medium-sized E. Small Enterprises Medium-sized Ent.

I II III IV V VI

Productivity 1.554+ 0.435+ 1.309+ 0.363+ 1.848+ 0.530+

(0.231) (0.064) (0.234) (0.064) (0.234) (0.066)

Size −0.804+ −0.184+ −0.786+ −0.231+ 0.322+ 0.089+

(0.086) (0.015) (0.059) (0.018) (0.059) (0.016)

Size x Mon. Union 0.241** 0.066** −0.004 −0.001 0.229*** 0.062***

(0.118) (0.031) (0.081) (0.023) (0.085) (0.021)

Skill −0.072+ −0.020+ −0.091+ −0.025+ −0.066*** −0.019***(0.020) (0.006) (0.020) (0.006) (0.020) (0.006)

R&D Intensity 6.382+ 1.784+ 7.037+ 1.952+ 6.457+ 1.851+

(0.905) (0.249) (0.914) (0.249) (0.934) (0.263)

Filter (Non-innovator) −0.524+ −0.158+ −0.443+ −0.131+ −0.580+ −0.179+(0.046) (0.015) (0.047) (0.015) (0.046) (0.015)

East −0.419+ −0.123+ −0.478+ −0.141+ −0.498+ −0.151+(0.044) (0.014) (0.045) (0.014) (0.044) (0.014)

Constant 1.570*** 1.618*** 0.601

(0.505) (0.512) (0.501)

Pseudo R2 0.214 0.214 0.246 0.246 0.209 0.209

Table presents the probit estimation results for SMEs. Columns I, III, and V present standard probitestimation results and columns II, IV and VI present the marginal effects obtained from standard probitestimation. All estimators include year and industry dummy variables. Small enterprises are the firms withless than 50 employees and medium-sized enterprises are the firms with more than 50 but less than250 employees. Robust standard errors are in parentheses. Statistical significance are denoted by* p<0.10, ** p<0.05, *** p<0.01, + p<0.001

624 K. Bagci

The impact of monetary union on exporter status of firms is depicted inFig. 4. In the left panel, the predicted probability of small-sized enterprises does notchange both in West and East Germany. For medium-sized enterprises, on the otherhand, the probability to export after the monetary union clearly increases both in Westand East Germany (right panel). It is also important to recognize that, in general, asthe level of productivity increases, the probability of exporting rises more rapidly forfirms in East Germany compared to those in West Germany.

Having identified the impacts at aggregate level, it is essential to study theresponses at sectoral level, as each sector may have a different response toreductions in trade costs due to monetary union. Table 3 presents the relativemarginal impact of cost reductions for each sector compared to all other remainingsectors. Estimations have been conducted for each industry separately with standardcovariates, but only the coefficients for the industries are reported. The resultsindicate that firms operating in mining, chemicals, food & tobacco, glass& ceramics, wood & paper industries become less likely to enter into foreignmarkets. Firms operating in electrical equipment, plastics, medical & otherinstruments, machinery, furniture and textiles industries become more likelyto enter into export markets. Finally firms in transport equipment and metalindustries experience no significant impact on export behavior. The mostdisadvantaged firms are in mining, food & tobacco and wood & paperindustries, for which the likelihood to export decreases by 18–26 %. Thefirms benefiting most are in the furniture industry, with an increase inprobability by around 14 %. These sectoral level results are in line withthe previous findings of Flam and Nordström (2006) and De Nardis et al.(2008).

In order to gain a better grasp of the factors that account for export behavior offirms, the export behavior is further investigated for firms with different productivitylevels. As reported in Table 4, productivity matters only for firms whose productivityis lower than average. That is to say, if firms are on average less productive, anincrease in the productivity will have strong impact on their export behavior. Thisimpact is stronger for SMEs. However, these impacts are vanishes for firms with toolow productivity levels, as reported in col. III and VII. For firms with already highproductivity, there are no significant changes in export behavior due to furtherincrease in productivity.

.6.7

.8.9

0 .2 .4 .6 0 .2 .4 .6

West Germany East Germany

Before Monetary Union After Monetary Union

Productivity

.51

0 .2 .4 .6 0 .2 .4 .6

West Germany East Germany

Before Monetary Union After Monetary Union

Productivity

Pre

dict

ed P

roba

bilit

y

Pre

dict

ed P

roba

bilit

y

Fig. 4 Predicted probability of exporting for SMEs before and after Monetary Union

Export behavior of German SMEs in the Eurozone 625

Size of firms remains to play a significant role in affecting export behavior.When all enterprises are considered together, increase in the size of firmsincreases their probability of exporting as well (c. I–IV). SMEs are, however,disadvantaged at all productivity levels compared to larger firms (col. V-VIII).This implies that, in terms of export behavior, growing from medium size tobig size matters more than growing from small size to medium size. Aftermonetary union, the relative significance of being larger in size decreases byabout 4 % for firms with productivity level above the average (col. I). Nofurther impact of monetary union is detected for other firms with differentproductivity levels. Furthermore, being an innovator always significantly improvesthe probability of exporting. R&D intensity appears to be especially important for firmswith productivity below the average (col. II, III, VI, and VII). Being located in the Eastdoes not affect the firms with significant dispersion from the average productivity levels(col. III, IV, VII, and VIII).

Up to this point, we provided important and very crucial evidence on the exportbehavior of firms of different sizes at aggregate level, sectoral level and differentproductivity levels. It may, however, be argued that these results are not necessarilyoutcomes of monetary union. Even though we include dummy variables for each yearto control for time specific impacts not attributable to monetary union, one may claimthat the predicted impact may include factors other than monetary union. In order toprovide supporting evidence, we use volatility of German Mark (DM) against Euro/ECU as a proxy for the elimination of trade costs due to monetary union.

Table 5 provides the results when exchange rate volatility is used as a proxy for theelimination of exchange rate volatility following monetary union. The marginalimpacts of all covariates are very close to previous estimation results. Exchange ratevolatility turns out to reduce the overall probability of exporting for SMEs. Compared

Table 3 Export Behavior of SMEs Operating in Different Industries

Industries Coefficient Industries Coefficient Industries Coefficient

1. Mining −0.261+ 6. Glass, Ceramics −0.058* 10. Machinery 0.080+

(0.000) (0.095) (0.000)

2. Chemicals −0.058* 7. Wood, Paper −0.177+ 11. Metals −0.001(0.094) (0.000) (0.957)

3. ElectricalEquipment

0.080+ 8. Plastics 0.090+ 12. Furniture 0.138+

(0.000) (0.000) (0.000)

4. Food,Tobacco

−0.206+ 9. Medical and Otherinstruments

0.074*** 13. Textiles 0.086+

(0.000) (0.001) (0.000)

5. TransportEquipment

0.000

(0.990)

Table presents the marginal effects obtained from standard probit estimation for different industries. Allestimators include year dummy variables. P values are in parentheses. Statistical significance are denotedby * p<0.10, ** p<0.05, *** p<0.01, + p<0.001

626 K. Bagci

to other firms, small enterprises do not benefit from the elimination of thevolatility, but medium-sized enterprises significantly gain from the eliminationof exchange rate volatility. These results confirm the previous findings andprovide supporting evidence that the impact is indeed due to introduction ofsingle currency.

6 Concluding remarks

There has been much research in attempting to quantify the monetary union’simpact on trade volume during the last decade. It is clear that monetary unionnot only decreases the costs of doing business, but also extends the opportu-nities and improves conditions in which firms do businesses. As for SMEs, themanagement of exchange rate risk is a particularly disadvantageous task forthem, as they do not have the critical size which gives them access to the mostmodern hedging instruments. They also lack qualified staff to discern exchangerate risks accurately. They are sometimes obliged to use the services ofintermediaries and that increases their costs even further. The introduction of

Table 4 Export Behavior of Firms with Different Productivity Levels – Marginal Effects

All Enterprises Small and Medium-sized Enterprises

I II III IV V VI VII VIII

Prod. Level y > y y < y y < y � σ y > y þ σ y > y y < y y < y � σ y > y þ σ

Productivity −0.042 0.822+ 0.319 0.056 −0.021 1.082+ 0.407 0.054

(0.511) (0.000) (0.617) (0.673) (0.757) (0.000) (0.516) (0.699)

Size 0.114+ 0.189+ 0.157** 0.097+ −0.132+ −0.218+ −0.055 −0.099***(0.000) (0.000) (0.013) (0.000) (0.000) (0.000) (0.773) (0.002)

Size x Mon.Union

−0.039** 0.012 0.063 0.011 0.049 0.080 0.149 −0.042

(0.028) (0.670) (0.504) (0.732) (0.101) (0.135) (0.541) (0.490)

Skill −0.029+ −0.012 0.024 −0.013 −0.027+ −0.003 0.038* −0.011(0.000) (0.124) (0.261) (0.334) (0.000) (0.724) (0.070) (0.417)

R&DIntensity

0.401 2.732+ 3.359+ 0.324 0.340 2.538+ 3.049+ 0.415

(0.209) (0.000) (0.000) (0.585) (0.314) (0.000) (0.000) (0.484)

Filter (Non-innovator)

−0.072+ −0.139+ −0.096* −0.077** −0.094+ −0.177+ −0.125** −0.092***

(0.000) (0.000) (0.085) (0.011) (0.000) (0.000) (0.021) (0.003)

East −0.047** −0.163+ 0.008 −0.027 −0.047** −0.145+ 0.034 −0.040(0.017) (0.000) (0.892) (0.440) (0.019) (0.000) (0.564) (0.271)

Table reports the marginal effects of probit estimation on the export behavior of firms withdifferent productivity levels. y denotes the productivity level y denotes the average productivity leveland σ denotes the standard deviation. All estimators include dummy variables for year and industry. Pvalues are in parentheses. Statistical significance are denoted by * p<0.10, ** p<0.05, *** p<0.01, +p<0.001

Export behavior of German SMEs in the Eurozone 627

the euro effectively removes this disadvantage and makes them more competitive ininternational markets.7

In line with the prediction derived from Helpman (2006), estimation resultsindicate that the likelihood of small and medium-sized enterprises to become anexporter increases. Medium-sized corporations benefit more than small corporations.This implies that size matters. What is more, firms operating in furniture, electricalequipment, plastics, medical instruments, machinery and textiles are more likely tobenefit from monetary union. These sectoral level results are in line with the previousfindings of Flam and Nordström (2006) and De Nardis et al. (2008). The results arerobust to using exchange rate volatility as a proxy for the cost reductions.

Furthermore, if firms are on average less productive, an increase in the productiv-ity will have strong impact on their export behavior. This impact is stronger forSMEs. After monetary union, the relative significance of being larger in sizedecreases by about 4 % for firms with productivity level above the average. Amongthe other results, being an innovator always significantly improves the probability ofexporting. R&D intensity appears to be especially important for firms with produc-tivity below the average. Being located in the East does not affect the firms with toohigh or too low productivity levels compared to the average productivities.

7 Introduction of monetary union is associated with reduction in trade costs with potential impact to changethe export and production behavior of firms. In fact, cost reductions may be quite trivial, especially in thecase of euro area where trade costs have been falling for decades, and it may be naïve to expect quitesignificant changes due to reductions in trade costs in such scales. It is, however, fair to argue that highereconomic integration gives investors and consumers more confidence and perception of living in a greatermarket and all of these may have more solid impacts than cost reductions due to monetary union. Therehave been many studies conducted on relationships between exchange rate volatility and trade and most ofthem have found either no or insignificant effect of exchange rate uncertainty on trade. Therefore, eventhough the arguments are built on the reductions in trade costs, we believe that the driving force ofrestructuring should be more than mere trade cost reductions.

Table 5 Export Behavior of SMEs – Exchange Rate Volatility as an Instrument for MU

Small & Medium-sized E. Small Enterprises Medium-sized Ent.

Marg.Eff.

St. Err. Pvalue

Marg.Eff.

St. Err. Pvalue

Marg. Eff. St. Err. Pvalue

Productivity 0.431+ (0.064) (0.000) 0.360+ (0.064) (0.000) 0.526+ (0.066) (0.000)

Size −0.138+ (0.017) (0.000) −0.230+ (0.017) (0.000) 0.148+ (0.014) (0.000)

Size x Volatility −0.124** (0.061) (0.042) −0.007 (0.042) (0.876) −0.132*** (0.043) (0.002)

Skill −0.020+ (0.006) (0.000) −0.025+ (0.006) (0.000) −0.019*** (0.006) (0.001)

R&D Intensity 1.780+ (0.249) (0.000) 1.950+ (0.250) (0.000) 1.837+ (0.263) (0.000)

Filter (non-innovator

−0.156+ (0.015) (0.000) −0.130+ (0.015) (0.000) −0.178+ (0.015) (0.000)

East −0.125+ (0.014) (0.000) −0.142+ (0.014) (0.000) −0.152+ (0.014) (0.000)

Table provides probit estimation results by using nominal exchange rate volatility as an instrumentfor the reduction in trade costs due to monetary union. In each category, first column provides themarginal effect, second column provides the standard errors and third column provides the p values.All estimators include year and industry dummy variables. Statistical significance are denoted by* p<0.10, ** p<0.05, *** p<0.01, + p<0.001

628 K. Bagci

Finally, it is worth noting that the approach is partial equilibrium analysis. That somefirms export more or new firms enter into the export market in no way mean that theeconomy is better off with a monetary union. However, from the existing theoreticaland empirical analyses, it is straightforward to expect that higher exposure tothe competition makes the small and medium size enterprises more competitiveand productive, with direct contribution to the overall competitiveness of the economy.

References

Alesina A, Barro RJ (2002) Currency unions. Q J Econ 117:409–436Arnold JM, Hussinger K (2005) Export behavior and firm productivity in German manufacturing: a firm-

level analysis. Rev World Econ 141:219–243Baldwin RE (2006) The Euro’s Trade Effects. ECB Working Paper, No: 594Baldwin R, Taglioni D (2004) Positive OCA criteria: microfoundations for the rose effect. Graduate

Institute of International Studies, GenevaBerger H, Nitsch V (2005) Zooming out: the trade effect of the euro in historical perspective. Center for

Economic Studies and the Ifo Institute. Cesifo Working Paper Series, No: 1435Bernard AB, Wagner J (2001) Export entry and exit by German firms. Weltwirtschaftliches Archiv

137:105–123Bernard AB, Eaton J, Jensen JB, Kortum S (2003) Plants and productivity in international trade. Am Econ

Rev 93:1268–1290Berthou A, Fontagné L (2008) The euro and the intensive and extensive margins of trade: evidence from

French firm level data, Working Paper 2008–06 CEPII, ParisBun MJ, Klaassen FJ (2004) The Euro Effect on Trade Is Not as Large as Commonly Thought. University

of Amsterdam Discussion Paper 2004/03Castellani D (2002) Export behavior and productivity growth: evidence from Italian manufacturing firms.

Weltwirtschaftliches Archiv 138:605–628De Nardis S, De Santis R, Vicarelli C (2008) The single currency’s effects on Eurozone sectoral trade:

winners and losers? Economics: The Open-Access, Open-Assessment E-Journal 2(17):1–34Delgado MA, Fariñas JC, Ruano S (2002) Firm productivity and export markets: a non-parametric

approach. J Int Econ 57:397–422Eaton J, Kortum SS, Kramarz F (2008) An Anatomy of International Trade: Evidence from French Firms.

NBER Working Paper No: 14610Esteve-Pérez S, Gil-Parejay S, Llorca-Vivero R, Martinez-Serrano JA (2011) The impact of the euro on

firm export behaviour: does firm size matter? Oxf Econ Pap 63:355–374Flam H, Nordström H (2006) Euro Effects on the Intensive and Extensive Margins of Trade. CESifo

Working Paper No. 1881Fontagné L, Freudenberg M (1999) Endogenous symmetry of shocks in a monetary union. Open Econ Rev

10:263–287Frankel JA, Rose AK (1998) The endogeneity of the optimum currency area criteria. Econ J 108:1009–

1025Gros D, Thygesen N (1998) European monetary integration. Longman, Harlow, EssexHelpman E (2006) Trade, FDI, and the organization of firms. J Econ Lit 44:589–630Helpman E, Melitz MJ, Yeaple SR (2004) Export versus FDI with heterogeneous firms. Am Econ Rev

94:300–316Melitz MJ (2003) The impact of trade on intra-industry reallocations and aggregate industry productivity.

Econometrica 71:1695–1725Micco A, Stein E, Ordonez G (2003) The currency union effect on trade: early evidence from EMU. Econ

Pol 18:316–356Nguyen H, Faff R, Marshall A (2007) Exchange rate exposure, foreign currency derivatives and the

introduction of the euro: French evidence. Int Rev Econ Finance 16:563–577Nitsch V (2002) Honey, I shrunk the currency union effect on trade. World Econ 25:457–474Tenreyro S (2007) On the trade impact of nominal exchange rate volatility. J Dev Econ 82:485–

508

Export behavior of German SMEs in the Eurozone 629