The impact of credit shocks on the European labour market

Bodnár, Katalin et al. Article The impact of credit shocks on the European labour market Baltic Journal of Economics Provided in Cooperation with: Baltic International Centre for Economic Policy Studies (BICEPS), Riga Suggested Citation: Bodnár, Katalin et al. (2021) : The impact of credit shocks on the European labour market, Baltic Journal of Economics, ISSN 2334-4385, Taylor & Francis, London, Vol. 21, Iss. 1, pp. 1-25, https://doi.org/10.1080/1406099X.2020.1871213 This Version is available at: https://hdl.handle.net/10419/267585 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=rbec20 Baltic Journal of Economics ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rbec20 The impact of credit shocks on the European labour market Katalin Bodnár, Ludmila Fadejeva, Marco Hoeberichts, Mario Izquierdo Peinado, Christophe Jadeau & Eliana Viviano To cite this article: Katalin Bodnár, Ludmila Fadejeva, Marco Hoeberichts, Mario Izquierdo Peinado, Christophe Jadeau & Eliana Viviano (2021) The impact of credit shocks on the European labour market, Baltic Journal of Economics, 21:1, 1-25, DOI: 10.1080/1406099X.2020.1871213 To link to this article: https://doi.org/10.1080/1406099X.2020.1871213 © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group Published online: 17 Jan 2021. Submit your article to this journal Article views: 1474 View related articles View Crossmark data The impact of credit shocks on the European labour market* Katalin Bodnár a , Ludmila Fadejeva b , Marco Hoeberichts c , Mario Izquierdo Peinado d , Christophe Jadeau e and Eliana Viviano f a European Central Bank, Frankfurt, Germany; b Monetary Policy Department, Latvijas Banka, Riga, Latvia; c Economics and Research Division, De Nederlandsche Bank, Amsterdam, Netherlands; d Banco de España, Madrid, Spain; e Engineering and Statistical Methodology Department, Banque de France, Paris, France; f Economics and Research Department, Bank of Italy, Rome, Italy ABSTRACT The sovereign debt crisis led to financial difficulties for European firms andadeclineintheuseoflabourinput.Weusequalitativefirm-level data for 24 European countries, collected within the third wave of the Wage Dynamics Network (WDN3) of the ESCB, to propose a crosscountry analysis of the relationship between a credit shock and labour markets. We firstderiveasetofindicesmeasuringdifficulties in accessing the credit market for the period 2010–2013. Second, we provide a description of the relationship between credit difficulties and changes in labour input, both along the extensive and the intensive margins as well as on wages. We find strong and significant correlation between credit difficulties and adjustments along both the extensive and the intensive margin. In the presence of credit market difficulties, firms also cut wages by reducing the variable part of wages. This evidence suggests that credit shocks can affect not only the real economy, but also nominal variables. ARTICLE HISTORY Received 15 June 2020 Accepted 18 December 2020 KEYWORDS Credit difficulties; labour input adjustment; intensive margin; wage adjustment; survey data JEL D53; E24; E44; G31; G32 1. Introduction In many European countries, the early 2010s have been characterized by significant difficulties in accessing credit by firms, as well as households and governments. The global financial crisis, having originated in 2007 in the US subprime market, and the subsequent sovereign debt crisis, which hit Europe in the summer of 2011, forced European banks to considerably tighten their credit conditions for firms in many economies and for several years. The global financial crisis and sovereign debt crisis renewed the interest for the effect of credit shocks on the real economy. Before the global financial crisis, the relationship between credit constraints and employment was investigated in the literature, analysing the link between financial development and growth (e.g. Beck & Demirguc-Kunt, 2006; Klapper et al., 2006). Following the global financial crisis, many papers focussed on the effect of credit shocks on the real economy (Acharya et al., 2018; Berg, 2018; Bottero © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. CONTACT Ludmila Fadejeva [email protected] Monetary Policy Department, Latvijas Banka, Riga, Latvia; 2a Kr.Valdemara street, Riga LV1050, Latvia *The views expressed in this paper are those of th e authors and do not involve the responsibility of their institutions or the Eurosystem. BALTIC JOURNAL OF ECONOMICS 2021, VOL. 21, NO. 1, 1–25 https://doi.org/10.1080/1406099X.2020.1871213 et al., 2020; Cingano et al., 2016; Degryse et al., 2016), and the labour market in particular (Bentolila et al., 2018, June; Berton et al., 2018; Buera et al., 2015; Chodorow-Reich, 2014; Duygan-Bump et al., 2015; Hochfellner et al., 2016; Pagano & Pica, 2012; Popov & Rocholl, 2018). The existing literature builds primarily on linked firm-bank data and examines the impact of exogenous credit supply shocks, making use of the sticky lender–borrower relationship. Different types of financial shocks have been examined: Popov and Rocholl (2018) focus on the effect of the funding shock of German savings banks during the US mortgage crisis, Chodorow-Reich (2014) take advantage of the different exposures of the lenders on the syndicated market to mortgage-backed securities in the US. Acharya et al. (2018) look at bank-firm relationships in Europe to analyse several outcomes including employment. Bentolila et al. (2018, June) use the differences in Spanish banks’health at the start of the Great Recession. Other papers derive local-level measures of credit supply (e.g. Greenstone et al., 2020). These papers typically focus on a single country, and do not analyse the heterogeneity in firms’adjustments across countries. This paper looks at the link between credit shocks and adjustment in labour from a European perspective, which is fundamental for the European policy makers. To do so, we use a unique, fully harmonized survey conducted in 25 European countries by the Wage Dynamics Network (WDN), a research network of the European System of Central Banks. 1 The survey, which was the third one conducted by the network and thus is labelled as WDN3 in this paper, focusses on the period between 2010 and 2013 and asks firms to report both their difficulties in accessing credit and the ways of adjusting labour costs, be it through employment or through wages. We construct an index of credit difficulties which is cleaned from the impact of demand and other shocks. This index is fully comparable across firms in different countries, and we use it to analyse the intensity of credit restrictions in different EU countries. Then we relate our index of credit difficulties to firms’labour cost adjustments. First, we find that credit difficulties were extremely heterogeneous both within and across countries. According to our estimates, in countries with low average values of the credit difficulty index, the withincountry variability was also quite low. On the contrary, in the most severely hit countries (mainly Southern European and some Eastern European countries) the within-country variability was remarkably high. If we compare countries, we find that the interquartile range (the difference between the 25th and the 75th percentiles) of the distribution of our index in Austria, the country registering the lowest level of credit difficulties, was three times lower than that observed in the most severely hit countries. Second, we find that European firms hit by a credit shock report more frequently a reduction in both employment and wages than firms without financing difficulties. We estimate that a 1 point increase in our index of credit difficulties is associated with an increase in the probability to adjust employment by close to 2 pp (over a mean probability of 16%). As the survey collects detailed information on the strategies to adjust labour costs, we can distinguish also between adjustment along the extensive (i.e. change in headcount) and the intensive margin (i.e. reduction in hours per employee). Consistently with Berton et al. (2018), who focus on one Italian region, we find that credit supply shocks affected both the extensive and the intensive margin. More importantly, we find that the reduction of the intensity of the use of labour as a response to a credit shock was not confined to Italy, as found by Berton et al. (2018), where the subsidized reduction 2K. BODNÁR ET AL. of hours was widely used, but also happened in other European countries, mainly through non-subsidized reduction of hours (i.e. part-time work arrangements). Workers are affected heterogeneously by firms’credit difficulties. We find that the probability of an adjustment in case of an adverse credit shock was higher for temporary workers. This finding is consistent with Bentolila et al. (2018, June), who focus on Spain, and Caggese and Cuñat (2008), who examine the case of Italy. Firms also decreased their hiring, with a particularly significant effect on the employment opportunities of younger job-seekers. Labour market adjustment as a response to credit constraints is thus a potential explanation behind the considerable rise of youth unemployment in most European countries (see Hoynes et al., 2012 for the US and Verick, 2009 for European countries for a description of how youth unemployment developed following the financial crisis). In addition to the adjustment in employment, we find that firms also adjust wages when they face credit difficulties. The relationship between credit shocks and wage dynamics has been less investigated so far by the literature, probably because of data constraints and because it has not been clear whether European firms have margins to adjust wages. An exception are Hochfellner et al. (2016) and Moser et al. (2020), who use employer-employee matched data for a sample of German firms and, by the use of different strategies, examine the impact of credit shocks on earnings, and Adamopoulou et al. (2020) who focus on Italy. We find that in our sample an increase of 1 point of our credit difficulty index is associated with an increase in the probability to cut wages, by around 1pp. over an average of 14%. The impact of credit difficulties is stronger for the flexible part of workers’compensation, whereas the impact on base wage is small and not precisely estimated. This is probably related to the institutional rigidities which prevent cuts of base wages. We find that the effect of credit market conditions on the extensive margin of adjustment was rather similar along several firm characteristics: the interaction of the extensive margin of adjustment with firm’s size, sector and autonomy is insignificant. For firms which are neither parent nor subsidiary/affiliate we find a higher probability to adjust the extensive margin when the credit difficulty index increases. We also find that the interaction term between the credit difficulty index and the extensive margin is higher in case of domestic firms than foreign owned ones. For a given credit supply shock, the reaction of firms is show some degree of heterogeneity across European countries (controlling for differences in the economic structure, i.e. firm size and sector). The most significant differences can be detected between the Eastern European / Baltic group of countries. This suggests that the different impact of credit shocks on employment and wages can occur not only due to heterogeneity in the intensity of the shocks across countries but also due to country specific factors such as labour policy protection or wage bargaining. Our estimation strategy is also supported by some additional exercises based on matching banks in the credit registers of France and Italy with firms in the WDN3 survey. We are aware that estimates using WDN3 are based on qualitative self-reported information, which do not allow for the identification of the effect of an exogenous credit supply shock to firms’labour costs. Nevertheless our findings are robust as confirmed by the use of quantitative information from the credit register data. Based on these data, we construct a credit supply shock index and we show that it correlates BALTIC JOURNAL OF ECONOMICS 3 with our survey-based measure of credit difficulties. Within this rather different setting our main results are fully confirmed. This paper is organized as follows. Section 2 describes the main features of the WDN3 data. Section 3 explains the methodology for calculating the index of credit difficulties. In Section 4, we show how the index correlates with measures about employment and wage adjustments in our sample of firms. In Section 5, we focus on France and Italy and on a sample of WDN3 firms matched with credit register data. Last, Section 6 briefly concludes. 2. The WDN3 survey We use firm-level survey data collected by the Wage Dynamics Network (WDN). WDN is a research network of the European System of Central Banks, dedicated to the study of the features and sources of wage and labour cost dynamics and their implications for monetary policy in the euro area. The first and the second surveys on firms’price and wage setting practices have been carried out in 2007 and 2009. The third survey, the results of which are used in this paper, was conducted in 2014 by national central banks in 25 countries of the European Union. It covers the 2010–2013 period, and was answered by over 25,000 firms (see Izquierdo et al., 2017 for details). Following the global financial crisis, several European countries have been confronted with a severe sovereign debts crisis. The latter, together with more stringent regulation about capital requirements and the related tightening of credit standards, transmitted into the second phase of the double-dip recession from the last quarter of 2011 until the first quarter of 2013 in the European Union as a whole. Firms were hit by adverse demand and credit shocks, and both types of shocks affected their strategies to adjust their labour input during 2010–2013. Therefore, the third wave of the WDN survey (WDN3) was designed specifically to differentiate by types of shock (to product demand, demand volatility, credit availability, customers’ability to pay and supply availability) and their intensity, as well as to explore firms’adjustment strategies during this period. Special attention was given to firms’adjustments of labour input, wage dynamics and wage settings practices. For a more detailed description of the WDN3 survey, see Appendix 1. This paper uses four sets of questions from the survey (see Table A2 in the Appendix for the exact list of the questions). First, we use the questions on credit availability and credit conditions. Six questions aim at capturing the taxonomy in the severity of credit constraints. They consider both the worsening in the quantity or access to credit and the costs and conditions of credit supplied by the banks. Firms were also asked to qualify the intensity of the difficulties. Both the questions on access to credit and credit conditions were asked in relation to three types of requested credit (financing working capital, financing new investments, refinance debt). Second, a group of questions was asked on the changes in economic conditions faced by the firms during 2010–2013. Firms could choose between five symmetrical responses describing the change in level of demand, volatility of demand, customers’ability to pay and availability of supplies (the potential answers were: strong decrease, moderate decrease, unchanged, moderate increase, strong increase). We use these questions to 4K. BODNÁR ET AL. control for the effects of other shocks, deriving an uncorrelated measure of credit difficulties. Third, several questions were asked on the channels of labour market adjustments used by the firms during 2010–2013. Firms were asked if they needed to significantly reduce their labour input or to alter its composition. Firms that needed to adjust their labour input were asked about the exact way of doing so (e.g. layoffs, reduction of hours, freeze of new hires, etc.). We use these questions to make a distinction between the extensive and the intensive margins of labour adjustment. Adjustment along the extensive margin is defined as individual or collective layoffs, while the intensive margin is defined as a reduction of working hours per worker (be it unsubsidized or carried out in the framework of subsidized schemes). Finally, some questions allow us to measure the propensity of firms to adjust base and variable wages. (For a description of the answers on labour adjustment please see Table A3 in the Appendix.) We combine these four sets of data and control for firm-level characteristics to examine the connection between credit shocks and labour adjustment. 3. Measuring credit difficulties using WDN3 Using the questions about credit difficulties, we construct firm-specific indices of credit constraints, comparable for 24 EU countries included in the WDN3 survey. Data for Ireland are excluded as answers about the availability of credit are not collected for this country. We focus on firms in manufacturing, trade and business services sectors (we call the latter two sectors together private services). Our final sample consists of around 19,000 firms. 2 See Table A1 for a description of our sample. A look at the raw data confirms the presence of strong cross-country heterogeneity. Table 1 reports the share of firms stating that the lack of credit for a given purpose, or the cost of credit was a relevant or very relevant problem. Over 40% of firms in Greece, Bulgaria, Poland and Slovenia report that credit difficulties were relevant or very relevant for their activity, but the values are also high for Italy, Spain, Portugal and Cyprus. While in Greece, Slovenia, Italy, Spain, Portugal, and Cyprus the high values are likely to reflect the impact of the sovereign debt crisis on financial intermediation, in Poland the reason may be the high share of self-financing (Strzelecki & Wyszyński, 2016). In Malta and Austria, on the other hand, only a minor proportion of firms faced difficulties in getting credit. Within firms, the responses about the difficulties to obtain credit for different purposes are highly correlated. This explains why the average share of firms reporting problems to obtain credit is similar for different credit types within one country. We derive a unique comparable measure of credit difficulties across European countries. We take advantage of the high correlation between the six credit availability measures. We combine both conditions and quantity aspects of credit availability via principal component analysis (PCA). 3 Before applying PCA, we remove the part of the correlation which could be due to other shocks hitting firms and affecting also their ability to access credit. To do so, we first regress our basic measures of credit restrictions on variables measuring demand and demand volatility shocks, customers’ability to pay, the availability of supplies and firms’characteristics, such as sector and firm size. We use the residuals of these six regressions to carry out the PCA. The descriptive statistics of the obtained components are given in Table 2. The first principal component explains BALTIC JOURNAL OF ECONOMICS 5 70% of the total variance in credit difficulty measures and has positive loadings of roughly similar size for all the six questions, therefore representing the overall credit difficulty for a firm. Figure 1 reports the average firm scores of the ‘credit difficulty index’by country, as measured by the first principal component. Countries are ranked according to their average level of firms’credit difficulties. The values are normalized around the average level of credit indexes for all countries. Thus, values above zero reflect above-average levels of credit difficulty. The index has a value above the whole-sample average in Italy, Spain, Portugal, Poland, Slovenia, Bulgaria and Greece. The distribution of countries by credit difficulty is quite symmetric, with a roughly similar number of countries experiencing above-average and below-average level of credit problems. Table 1. Share of firms in manufacturing, trade and business services, who viewed that credit access problem in 2010–2013 (as described in the credit accessibility questions) was relevant or very relevant, %. Credit was NOT available to Credit was available to finance working capital finance new investment refinance debt finance working capital, finance new investment, refinance debt, but conditions were too onerous AT 5.4 3.6 1.5 4.3 2.0 0.9 BE 16.3 20.8 15.9 17.7 18.0 12.0 BG 52.2 51.5 44.3 53.4 52.6 49.6 CY 36.8 35.0 30.8 35.7 31.0 28.9 CZ 12.3 13.7 10.5 18.2 18.5 15.2 DE 10.0 9.2 8.9 7.7 6.6 5.9 EE 11.0 13.3 7.3 14.1 13.3 8.0 ES 32.5 32.8 29.5 38.4 38.2 34.3 FR 14.0 16.1 11.4 8.2 8.3 6.7 GR 56.3 53.1 46.5 54.2 41.9 46.2 HR 30.9 28.7 22.1 39.3 41.1 35.3 HU 9.1 10.5 9.5 26.5 26.4 24.4 IT 29.3 39.2 27.0 34.9 27.6 33.4 LT 24.1 19.1 12.3 27.9 21.4 14.6 LU 17.1 23.0 13.5 15.9 15.1 10.3 LV 33.0 22.8 17.3 28.8 24.3 18.4 MT 4.6 3.1 1.5 6.1 6.2 2.3 NL 23.4 26.2 16.8 18.4 19.4 13.5 PL 51.3 46.8 23.5 47.7 43.5 26.7 PT 31.4 31.3 25.3 42.8 40.5 33.7 RO 21.2 21.0 16.2 31.7 29.4 24.7 SI 46.4 46.6 36.7 47.3 47.4 40.9 SK 26.4 34.5 19.8 33.6 38.8 26.8 UK 28.7 26.6 21.6 24.4 24.3 24.6 Note: Frequency. Data weighted by employment weight. Table 2. Principal component analysis of the credit difficulty measures. Component Eigenvalue Difference Proportion Cumulative Loading 1 1 4.353 3.605 0.726 0.726 0.403 2 0.748 0.392 0.125 0.850 0.403 3 0.356 0.110 0.059 0.910 0.404 4 0.246 0.088 0.041 0.950 0.416 5 0.158 0.018 0.026 0.977 0.411 6 0.140 0.023 1.000 0.412 Note: PCA on answers about credit difficulties, after removing variables measuring demand and volatility shocks, difficulties in customers’ability to pay, availability of supplies, sector and size dummies. 6K. BODNÁR ET AL. The lower graph of Figure 1 shows the distribution of the obtained credit difficulty indices by country, with the lower and upper borders representing the 25th and the 75th percentiles, respectively. The line in the box shows the median. In all countries, the distribution of the credit difficulty index has a long positive tail, suggesting that even in countries where a majority of the firms had no credit difficulties, quite a large minority faced financing problems. Austria and Malta are extreme cases, where over 75% of firms had the same low level of credit difficulty. 4 In Poland, Slovenia, Bulgaria and Greece, the distribution of the credit difficulty index is more even. In these countries the occurrence of both the very low and the very high values of the credit difficulty indices was rare and the majority of firms had similar, relatively severe credit access problems. Overall, the figure shows the presence of high heterogeneity of credit difficulties across EU countries: the interquartile range (the difference between the 25th and the 75th percentile of the distribution) of the index of credit difficulties in Austria is around three times lower than that observed in Italy or in Greece. Figure 2 shows the distribution of the derived credit difficulty index for all countries in the sample, weighing observations to reflect total employment in the countries. The large mass in the negative interval reflects the high weight of France, Austria and Germany in the total sample of firms and rather good credit availability in these countries. The right tail is much longer and mostly positive, reflecting the overall severity of credit conditions for many firms. To cross-check whether our index indeed captures the credit difficulties that we intend to measure, we compare it with external data sources. For this cross-check, we first look at the Survey on the access to finance of enterprises (SAFE), conducted by the ECB and the Figure 1. Country averages of credit difficulty index and box-plot analysis of firm level variation. Note: Sample is restricted to manufacturing, trade and business service firms. BALTIC JOURNAL OF ECONOMICS 7 difficulties in the different areas. The results of this exercise are reported in Table 8. Interestingly, we do not find much evidence that the elasticity of employment to credit shocks was different across these areas. In particular, no significant differences are found between group (i) and (iii) in any method of labour cost adjustment, although the impact of credit difficulties on the employment adjustment in the intensive margin and flexible wage adjustment seems to be lower in the Eastern European and Baltic countries. For completeness, we also provide estimates in which the credit difficulty index is interacted with country dummies (see Table 9). We take France as baseline. The coefficients within Southern European countries are quite homogenous, the only difference being smaller effect on extensive margin in Italy. In general, some cross-country heterogeneity emerges, with no clear pattern. Some significant differences can be detected only within the Eastern European/Baltic group. Firms in Slovenia, Slovakia Poland, Lithuania and Romania tend to rely less on labour or flexible wage margins of adjustment in response to worsening credit supply conditions. The use of labour and wage adjustment margins in two Baltic countries, i.e. Latvia and Estonia, as well as Hungary, Croatia, Bulgaria does not differ significantly from the baseline. This result suggests that the heterogeneous reaction of the EU labour markets in response to the sovereign debt crisis is explained not only by the differences in the intensity of the crisis across countries, but also by country-specific factors. This conclusion is in line with the results by Mathä et al. (2019) showing that strict employment protection and high centralization or coordination of wage bargaining make it less likely that firms reduce wages when facing negative shocks (irrespective of the type of negative shock). 5. Robustness checks: evidence from France and Italy Our index of credit difficulties has been calculated on qualitative data from the survey. In this section, based on French and Italian data on loans to firms we construct a quantitative index of credit supply and we test whether our results are still confirmed. We combine the WDN3 sample with data from the credit register in the two countries, administered by the Banque de France and the Bank of Italy, respectively. The data includes all credit commitments by all banks operating in each country exceeding 25,000 euros in France and 30,000 6 euros in Italy. The thresholds, which are quite similar, therefore exclude only very small companies with low credit facilities granted. Both databases include firm identifiers (tax codes) that make it possible to link firms with WDN3 data and identify their lending banks. The data have a monthly Table 6. Wage adjustments and credit availability, base wage and variable wage components. Probit marginal effects. (1) (2) (3) Wages (total) Base wages Flexible wages Index of credit difficulties 0.009** 0.002 0.009*** [0.016] [0.188] [0.004] Observations 18,282 18,282 18,282 Mean probability 0.142 0.050 0.123 Note: Robust p-values in brackets ***p< 0.01, **p< 0.05, *p< 0.1. The models include country, sector and size dummies. 14 K. BODNÁR ET AL. Table 7. Credit availability and labour market adjustments by type of firm. Probit marginal effects. (1) Adjust labour input Index of credit difficulties 0.005 0.028*** 0.024*** [0.565] [0.000] [0.000] Interaction dummy with credit difficulty index Autonomy (subsidiary) Autonomy (other) Ownership (foreign) Structure (multi-establishment) 0.013 0.038*** −0.026** −0.009 [0.322] [0.002] [0.049] [0.413] Observations 15,646 15,646 15,734 Mean probability 0.339 0.339 0.338 (2) Adjust the extensive margin Index of credit difficulties 0.013* 0.026*** 0.018*** [0.064] [0.000] [0.001] Interaction dummy with credit difficulty index Autonomy (subsidiary) Autonomy (other) Ownership (foreign) Structure (multi-establishment) −0.006 0.025** −0.025** 0.001 [0.571] [0.020] [0.022] [0.907] Observations 15,728 15,728 15,817 Mean probability 0.181 0.181 0.181 (3) Adjust the intensive margin Index of credit difficulties 0.015*** 0.015*** 0.014*** [0.009] [0.000] [0.001] Interaction dummy with credit difficulty index Autonomy (subsidiary) Autonomy (other) Ownership (foreign) Structure (multi-establishment) −0.008 −0.002 −0.008 −0.008 [0.371] [0.755] [0.325] [0.258] Observations 15,728 15,728 15,817 Mean probability 0.113 0.113 0.112 Note: Robust p-values in brackets ***p< 0.01, **p< 0.05, *p< 0.1. The models include area, country, sector and size dummies. The baseline for autonomy –parent firm, baseline for ownership – domestic firm, baseline for structure –single establishment. BALTIC JOURNAL OF ECONOMICS 15 frequency. For each company-bank pair, the total credit granted at the end of the year is recovered. 7 Credit register data are used to construct a quantitative index of credit supply, which can be assigned to the firms in the WDN3 sample. We consider the universe of banks in Italy and France from 2007 to 2013, i.e. before the burst of the global financial crisis and after the sovereign debt crisis. Aggregating loan data by bank, we calculate the three-year percentage change in total loans for each bank. This way we remove bank fixed effects. We then carry out the simple regression [2] to remove a time trend t, aimed at capturing aggregate common factors (e.g. credit demand). D3Lbt = a + b t+ut(2) Last, we take the residuals of [2], and in particular residuals in year 2013, ˆ ub,t=2013. The use of residuals instead of D3Lbt allows us to get a quite conservative measure of credit supply, since, for each bank, we consider how much bank’sbcredit grows compared to the aggregate trend. Finally, we assign the residuals for the period 2010–2013 to each WDN3 firm. Firms have multiple bank relationships, thus, we weight each residual with the share of loans Lfb,t=2009 of firm f(in the WDN3 sample) in the total amount of loans of the firm with any bank bin the register in 2009, i.e. the year preceding the survey reference period. This strategy allows us to limit the impact of possible selection bias in the firm-bank relationship in response to the global financial and/or the sovereign debt crisis. In particular we calculate the index of credit supply to firm f,CSf, as: CSf=Lfb,t=2009 Lf,t=2009 ∗ˆ ub,t=2013 (3) This procedure leads to an imperfect match between the twodatasets and weget around 750 observations per country. 8 We normalize the two indices and pool the two datasets. Figure 5 compares the index of credit difficulties drawn from the WDN3 survey with this measure of credit supply change. As expected, the two indices are negatively correlated. Our estimation results are reported in Table 10 (the regression models also include sector and size dummies and a country dummy). The first column refers to the correlation of CSfand our credit difficulty index, entirely based on WDN3 data: firms matched with banks with lower credit growth are more likely to report difficulties in accessing to Table 8. Credit availability and labour market adjustments by geographical area. Probit marginal effects. (1) (2) (3) (4) Extensive margin Intensive margin Base wage Flexible wage Credit difficulties index 0.019*** 0.013*** 0.001 0.009** [0.001] [0.006] [0.700] [0.036] Southern Europe * Credit diff. Index 0.003 −0.003 0.001 0.010 [0.751] [0.715] [0.835] [0.211] Eastern/Baltic * Credit diff. Index −0.014* −0.007 0.002 −0.014** [0.061] [0.196] [0.423] [0.017] Observations 18,282 18,282 18,282 18,282 Mean probability 0.153 0.105 0.050 0.123 Note: Robust p-values in brackets ***p< 0.01, **p< 0.05, *p< 0.1. Southern Europe includes: Spain, Italy; Greece, Portugal, Cyprus. Eastern Europe/Baltic countries includes: Czech Republic, Estonia, Croatia, Hungary, Romania, Bulgaria, Latvia, Lithuania, Poland, Slovenia, and Slovakia. The models include area, country, sector and size dummies. 16 K. BODNÁR ET AL. credit and the correlation between the two indices is statistically significant.. This is an indirect validation of our approach based on qualitative self-reported information. Columns 2–5 show the estimated probabilities to reduce the extensive margin, the intensive margin, the base wage and the variable wage, respectively. The results are substantially confirmed. The correlation between credit supply and the probability to reduce Table 9. Credit availability and labour market adjustments by country. Probit marginal effects. (1) Extensive margin (2) Intensive margin (3) Base wage (4) Flexible wage Credit difficulties index 0.035*** 0.024*** 0.003 0.010 [0.001] [0.008] [0.649] [0.166] Austria * Credit diff. Index 0.023 −0.072** −0.019 −0.190*** [0.469] [0.049] [0.211] [0.000] Belgium * Credit diff. Index 0.012 0.002 0.011 −0.001 [0.453] [0.888] [0.217] [0.951] Germany * Credit diff. Index −0.007 −0.016 0.003 0.014 [0.585] [0.137] [0.681] [0.139] Luxemburg * Credit diff. Index −0.042* 0.002 0.000 0.005 [0.050] [0.865] [0.983] [0.807] Malta * Credit diff. Index 0.030 0.007 −0.055*** −0.297*** [0.432] [0.789] [0.000] [0.001] The Netherlands * Credit diff. Index −0.018 −0.019 −0.006 0.015 [0.343] [0.206] [0.559] [0.278] The United Kingdom * Credit diff. Index −0.042** −0.015 −0.013 −0.032*** [0.016] [0.321] [0.108] [0.009] Cyprus * Credit diff. Index −0.034* −0.025* −0.007 −0.017 [0.090] [0.072] [0.484] [0.264] Spain * Credit diff. Index 0.021 −0.015 0.003 0.018 [0.169] [0.268] [0.745] [0.160] Greece * Credit diff. Index −0.018 −0.022 0.002 0.004 [0.261] [0.141] [0.841] [0.746] Italy * Credit diff. Index −0.039** −0.014 −0.009 0.004 [0.027] [0.285] [0.345] [0.780] Portugal * Credit diff. Index −0.020 −0.004 −0.001 −0.005 [0.138] [0.747] [0.919] [0.597] Bulgaria * Credit diff. Index −0.002 −0.006 0.005 −0.000 [0.920] [0.645] [0.519] [0.980] Czechia * Credit diff. Index −0.025* −0.016 0.004 −0.014 [0.052] [0.144] [0.616] [0.149] Estonia * Credit diff. Index −0.015 −0.018 −0.004 −0.017 [0.414] [0.223] [0.707] [0.330] Croatia * Credit diff. Index −0.027 −0.012 0.003 0.000 [0.110] [0.430] [0.719] [0.998] Hungary * Credit diff. Index −0.004 0.001 0.006 −0.007 [0.745] [0.935] [0.456] [0.451] Lithuania * Credit diff. Index −0.011 −0.024** −0.016 −0.035** [0.598] [0.039] [0.117] [0.027] Latvia * Credit diff. Index −0.021 −0.021 0.000 0.007 [0.214] [0.141] [0.966] [0.614] Poland * Credit diff. Index −0.035** −0.017 −0.004 −0.029** [0.011] [0.100] [0.633] [0.021] Romania * Credit diff. Index −0.030*** −0.016 −0.001 −0.000 [0.010] [0.107] [0.913] [0.996] Slovenia * Credit diff. Index −0.030** −0.028** 0.003 −0.004 [0.017] [0.021] [0.650] [0.626] Slovakia * Credit diff. Index −0.045* −0.056*** −0.003 −0.034** [0.054] [0.001] [0.712] [0.029] Observations 18,282 18,282 18,282 18,282 Mean probability 0.154 0.105 0.050 0.123 Note: Robust p-values in brackets *** p< 0.01, ** p< 0.05, * p< 0.1. Base country –France. The models include area, country, sector and size dummies. BALTIC JOURNAL OF ECONOMICS 17 labour is negative. The same holds for flexible wages, but not for base wages, as found in the examination of all countries. 6. Conclusions In this paper, we provide empirical evidence about a strong correlation between credit shocks and labour market adjustments in Europe during the sovereign debt crisis. We rely on WDN3 survey data, which has the advantage to offer a unique European perspective, providing comparable, harmonized results for 24 countries. We are aware of the limits of our approach. First, the data allow us to calculate only the probability of an adjustment and not how much of the observed employment drop Figure 5. Correlations between the change in credit supply and the index of credit difficulties. Note: Credit supply (measured on credit registers and normalized between zero and one in both countries) and index of credit difficulty. Table 10. Italy and France. Alternative index of credit supply and labour market adjustments. Pooled data. (1) (2) (3) (4) (5) Index of credit difficulties Extensive margin Intensive margin Base wage Flexible wage Alternative index of credit supply −1.212** −0.009*** −0.008*** −0.001 −0.008*** [0.000] [0.001] [0.004] [0.277] [0.003] Observations 1558 1558 1558 1603 1603 Note: Index of credit supply, based on credit registers. Robust p-values in brackets ***p< 0.01, **p< 0.05, *p< 0.1. The models include country, sector and size dummies. 18 K. BODNÁR ET AL. can be imputed to credit difficulties. Second, since we use survey data on self-reported credit difficulties and other shocks, our index of credit difficulty does not allow for a proper identification of the credit supply shock hitting the various countries, net of any demand effect. Thus, our main estimates are simple correlations between credit difficulties and firms’labour cost adjustment strategies. Even with this limitation, our results confirm that credit shocks are important determinants of labour market fluctuations in Europe. Second, credit market difficulties are associated not only with a decrease in employment, but also with a decline in the intensity of the use of labour in terms of hours worked. One consequence of this finding is that after large-scale shocks like the sovereign debt crisis, special attention may be needed on measures of labour market slack, other than the unemployment rate (which is a simple headcount ratio of how many people are without a job over active population). Finally, our results suggest also that, in response to credit difficulties, base wages are quite rigid and do not adjust on average. Nevertheless, European firms reduce nominal wages by cutting the variable part of employee compensation (bonuses, performancerelated premia, etc.). Thus, credit difficulties may have consequences not only on real variables but also on nominal ones (e.g. Adamopoulou et al., 2020; Moser et al., 2020; for an analysis on the impact of credit difficulties on prices, see also Duca et al., 2017). Notes 1. Previous WDN surveys do not allow to make a similar analysis on the impact of credit constrains on labour cost adjustments. WDN1 survey did not include any question regarding difficulties in access to finance while WDN2, which was an update of WDN1 with small sample sizes and conducted only in 10 European countries, included merely one question regarding the extent of difficulties in access to finance for firms. Using this dataset, Fabiani et al. (2015), although they focus on demand shocks, find that negative finance shocks increase the likelihood to adjust margins and costs at the firm level and once the impact of demand shocks is taken into account, financially constrained firms are more likely to adjust non-labour costs. 2. Firms’non-response to the credit availability questions is not homogenous across countries. In the UK, almost 30% of firms in manufacturing, trade and business services sectors haven’t provided answers to this block of questions. In Greece this share is 12%, followed by Hungary (9%), Latvia (9%) and Italy (8%). In the remaining of the countries the non-response rate was smaller. 3. As robustness check we derive credit difficulty index using factor analysis. The obtained results lead to the same findings. The difference in the size of the marginal effects using both measures (with standardized variance) is negligible. 4. The countries with low credit difficulties on average are characterized by a good cyclical position at the start of the sovereign debt crisis, as shown for example by their low unemployment rates. At the same time, in Austria and Malta, the low share of state-owned banks and in the latter, the high share of foreign owned banks have also likely played a role. In Austria, only a low percentage of the firms applied for a credit in the period examined in the WDN3 survey. See Stiglbauer (2017) and Micallef and Caruana (2017). 5. The dummy for the adjustment on the extensive margin is equal to 1 if the firm answered that individual and/or collective layoffs were used moderately or strongly, and 0 otherwise. The dummy for the intensive margin is equal to 1 if the firm answered that the decrease of BALTIC JOURNAL OF ECONOMICS 19 hours worked per worker, either subsidized or non-subsidized, was used moderately or strongly, and 0 otherwise. 6. Prior to 2009, the Italian credit register only included loans in excess of €75,000. Companies not present in the credit register before 2009 are therefore excluded from the sample. However, this selection has an extremely limited impact since the Italian WDN3 sample mostly includes companies with at least five employees (see Izquierdo et al., 2017 for details), which meet the requirement for inclusion in the register. 7. When dealing with bank data one of the main problems concerns bank mergers and how to reattribute loans to firme made by the banks involved in the M&A. The data used in this paper presents the same problem and in case of merge the information related to the bank-firm match is dropped (and the firm is excluded from the sample). However, in the period under scrutiny, during which two deep financial crises occurred, merges in both countries were rather limited. Even if this potentially generate some selection bias, the type of selection induced by bank M&A is not clear and likely not correlated to labour market outcomes. 8. Most firms drop out because of errors in tax codes. Disclosure statement No potential conflict of interest was reported by the author(s). Notes on contributors Katalin Bodnár is an economist in the Supply Side, Surveillance and Labour Markets Division at the European Central Bank. Her interests include labour economics and potential output. Ludmila Fadejeva is a research economist at Latvijas Banka. Her research interests include monetary and labour economics, particularly the study of monetary policy transmission and income/wealth inequality. Marco Hoeberichts is an economist in the Economics and Research Division at De Nederlandsche Bank. His research interests I include labour economics and inflation dynamics. Mario Izquierdo Peinado is Head of the Supply and Labor Market Analysis Unit at the Directorate General of Economics and Statistics of the Bank of Spain. His research interests include various labor market issues, such as wage dynamics, wage inequality and the impact of labor market institutions. Christophe Jadeau is an economist at Banque de France, Engineering and Statistical Methodology Department. Eliana Viviano is Head of the Labour Market Division at the Economics and Research Department of the Bank of Italy. She has published both micro and macro papers on various labor market issues, such as labor market reforms, determinants of wage dynamics, methodological aspects related to the estimation of unemployment and employment rates. ORCID Ludmila Fadejeva http://orcid.org/0000-0002-6822-1577 References Acharya, V. V., Eisert, T., Eufinger, C., & Hirsch, C. (2018). Real effects of the sovereign debt crisis in Europe: Evidence from syndicated loans. The Review of Financial Studies,31(8), 2855–2896. https://doi.org/10.1093/rfs/hhy045 20 K. BODNÁR ET AL. Adamopoulou, E., De Philippis, M., Sette, E., & Viviano, E. (2020). The long run earnings effects of a Credit Market Disruption, IZA DP No. 13185. Beck, T., & Demirguc-Kunt, A. (2006). Small and medium-size enterprises: Access to finance as a growth constraint. Journal of Banking & Finance,30(11), 2931–2943. https://doi.org/10.1016/j. jbankfin.2006.05.009 Bentolila, S., Jansen, M., & Jimenez, G. (2018, June). When credit dries up: Job losses in the Great Recession. Journal of the European Economic Association,16(3), 650–695. https://doi.org/10. 1093/jeea/jvx033 Berg, T. (2018). Got rejected? Real effects of not getting a loan. Review of Financial Studies,31(12), 4912–4957. https://doi.org/10.1093/rfs/hhy038 Berton, F., Mocetti, S., Presbitero, A., & Richiardi, M. (2018). Banks, firms and jobs. Review of Financial Studies,31(6), 2113–2156. https://doi.org/10.1093/rfs/hhy003 Bottero, M., Lenzu, S., & Mezzanotti, F. (2020). Sovereign debt exposure and the bank lending channel: Impact on credit supply and the real economy. Journal of International Economics, 103328 Brown, J. R., Fazzari, S. M., & Petersen, B. C. (2009). Financing innovation and growth: Cash flow, external equity, and the 1990s R&D boom. Journal of Finance,64(1), 151–185. https://doi.org/ 10.1111/j.1540-6261.2008.01431.x Buera, F., Fattal-Jaef, R., & Shin, Y. (2015). Anatomy of a credit crunch: From capital to labor markets. Review of Economic Dynamics,18(1), 101–117. https://doi.org/10.1016/j.red.2014. 11.001 Caggese, A., & Cuñat, V. (2008, Nov.). Financing constraints and fixed-term employment contracts. The Economic Journal,118(533), 2013–2046. https://doi.org/10.1111/j.1468-0297. 2008.02200.x Chodorow-Reich, G. (2014). The employment effects of credit market disruptions: Firm-level evidence from the 2008–9financial crisis. Quarterly Journal of Economics,129(1), 1–59. https://doi. org/10.1093/qje/qjt031 Cingano, F., Manaresi, F., & Sette, E. (2016). Does credit crunch investments down? New evidence on the real effects of the bank-lending channel. Review of Financial Studies,29(10), 2737–2773. https://doi.org/10.1093/rfs/hhw040 Degryse, H., De Jonghe, O., Jakovljević, S., Mulier, K., & Schepens, G. (2016). ‘The impact of bank shocks on firm-level outcomes and bank risk-taking’, Paris December 2016 Finance Meeting EUROFIDAI - AFFI. Available at SSRN: https://ssrn.com/abstract=2788512 Duca, I., Montero, J. M., Riggi, M., & Zizza, R. (2017). I will survive. Pricing strategies of financially distressed firms. Temi di discussione (Economic working papers) 1106, Bank of Italy, Economic Research and International Relations Area. Duygan-Bump, B., Levkov, A., & Montoriol-Garriga, J. (2015). Financing constraints and unemployment: Evidence from the Great Recession. Journal of Monetary Economics,75,89–105. https:// doi.org/10.1016/j.jmoneco.2014.12.011 Fabiani, S., Lamo, A., Messina, J., & Rõõm, T. (2015). European firm adjustment during times of economic crisis, ECB Working Paper no 1778. Greenstone, M., Mas, A., & Nguyen, H. L. (2020). Do credit market shocks affect the real economy? Quasi experimental evidence from the great recession and ‘normal’economic times. American Economic Journal: Economic Policy 12,200(225). doi:10.1257/pol.20160005 Hochfellner, D., Montes, J., Schmalz, M., & Sosyura, D. (2016). Winners and losers of financial crises: Evidence from individuals and firms, mimeo, University of Michigan. Hoynes, H., Miller, D., & Schaller, J. (2012). Who suffers during recessions? Journal of Economic Perspectives,26(3), 27–48. https://doi.org/10.1257/jep.26.3.27 Izquierdo, M., J. F. Jimeno, T. Kosma, A. Lamo, S. Millard, T. Rõõm, & E. Viviano (2017)Labour market adjustment in Europe during the crisis: microeconomic evidence from the Wage Dynamics Network survey, ECB Occasional Paper n 192. Klapper, L., Laeven, L., & Rajan, R. (2006). Entry regulation as a barrier to entrepreneurship. Journal of Financial Economics,82(3), 591–629. https://doi.org/10.1016/j.jfineco.2005.09.006 BALTIC JOURNAL OF ECONOMICS 21 Mathä, T. Y., Millard, S., Rõõm, T., Wintr, L., & Wyszyński, R. (2019). Shocks and labour cost adjustment: evidence from a survey of European firms, ECB Working Paper Series no 2269. Micallef, B., & Caruana, K. (2017). Results of the 2014 Wage Dynamics Network for Malta, ECB WDN3 Country Report. Moser, C., Saidi, F., Wirth, B., & Wolter, S. (2020). Credit supply, firms, and earnings inequality, MPRA Paper 100371. Pagano, M., & Pica, G. (2012). Finance and employment. Economic Policy,27(69), 5–55. https://doi. org/10.1111/j.1468-0327.2011.00276.x Popov, A., & Rocholl, J. (2018). Do credit shocks affect labor demand? Evidence for employment and wages during the financial crisis. J. Financial Intermediation,36,16–27. https://doi.org/10.1016/j. jfi.2016.10.002 Stiglbauer, A. (2017). The third wage dynamics network firm survey: Country Report on Austria, ECB WDN3 Country Report. Strzelecki, P., & Wyszyński, R. (2016). Poland’s labour market adjustment in times of economic slowdown –WDN3 survey results, NBP Working Papers 233, Narodowy Bank Polski, Economic Research Department. Verick, S. (2009). Who is hit hardest during a financial crisis? The vulnerability of young men and women to unemployment in an economic downturn, IZA DP No. 4359, August 2009. Appendix 1. The WDN3 survey In this paper, we use so-called WDN3 survey, conducted in 2014 by national central banks in 24 countries of the European Union. This survey constitutes the main data source we use to deal with these issues. This survey is the third wave of enquiries led by the Wage Dynamics Network (WDN) of the European System of Central Banks, a research network dedicated to the study of the features and sources of wage and labour cost dynamics and their implications for monetary policy in the euro area. The first survey on firms’price and wage setting practices has been carried out by 17 national central banks in 2007–2008. Additional questions –mainly to respondents of the first wave –have then been issued in a short second wave in 2009, in order to assess the firms’ reaction to the global financial crisis of 2007–2008. Since late 2009, the European countries have been confronted to the sovereign debts crisis, and labour market reforms have occurred: the third wave was designed to measure the nature of shocks and the firms’reaction during the period 2010–2013, and especially the adjustments they made in their price and wage settings practices. The harmonized questionnaire contains three main parts: the nature of the shocks (changes in demand, in accessibility of funding, costs and mainly elements of the labour costs), the adjustments on employment and wages, and the main obstacles to hiring. Each participating national central bank was responsible for the translation of the questionnaire and for the conduct of the survey in the country. Each central bank chose both the sample computation and the data collection method for its national data, leading to a large variety of sample computation and data characteristics. More than 24,000 firms were surveyed during the year 2014: if the perimeter of sectors can differ from a country to another, the manufacturing, trade, business services and, to a lesser extent, construction are well represented across the participating countries. To improve firm comparability between countries we restrict our analysis to the three main sectors – manufacturing, trade and business services (see Table A1 for detailed information on sample). In most countries, firms with less than 5 employees were excluded from the survey: they only represent 2% of the data. 29% of firms have 5–19 employees, 24% 20–49, 25% 50–199 and 20% more than 200 employees. For all countries we only include firms with at least 5 employees. 22 K. BODNÁR ET AL. Table A1. Survey sample by country, sector and size (firms that provided answers about credit availability). Country Number of firms that provided answers about credit availability (all sectors) Share of firms in manufacturing, trade or business services (%) Firms in manufacturing, trade or business services Average item non-response (% of total number of firms) Distribution by sector (%) Distribution by employee number (%) for credit availability questions for labour input adjustment questionsManufacturing Trade Business services <5 5–19 20–49 50–199 200 and > AT 744 83.9 34.8 25.8 39.4 0.6 18.1 22 31.6 27.7 1.9 3.2 BE 958 77.5 54.3 14.8 30.9 –22.8 23.9 42.6 10.8 2.1 0.6 BG 507 75.9 14 59.5 26.5 –72.5 18.2 7 2.3 1.4 0 CY 167 85 22.5 32.4 45.1 26.1 41.5 15.5 9.9 7 5.4 2.6 CZ 944 91.5 54.5 16.1 29.4 –15.4 18.9 25.8 39.9 4.7 1.8 DE 2297 80.8 34 28.1 37.9 9.2 24 28 27.2 11.7 3.8 2.9 EE 500 76.6 35 24 41 –36 34.7 23.2 6 0 0 ES 1975 99.1 25.9 30.7 43.5 –73.1 18 6.6 2.3 0 0 FR 1120 85.4 51.4 25.2 23.4 –18.4 22.2 27.7 31.7 2.6 0.8 GR 348 100 39.4 35.3 25.3 –11.2 36.2 34.8 17.8 11.5 2.0 HR 301 90.4 42.6 21 36.4 –30.1 25.7 33.1 11 0 0 HU 1782 90.1 43.7 23.6 32.7 –10.5 29.5 40.1 19.9 9.2 0 IT 919 97.6 51.7 21.1 27.2 –6.7 51.4 29 12.5 8.3 0.5 LT 515 77.3 19.1 42.5 38.4 –57.5 19.3 18.3 4.8 0 0 LU 661 64.9 17.2 35.9 46.9 23.5 35.7 21.9 14.9 4 1.3 0 LV 463 85.3 20.8 36.7 42.5 –47.6 25.8 20.8 5.8 8.6 0 MT 178 73 24.6 20 55.4 –13.8 24.6 37.7 23.8 0 0 NL 727 58.2 22.9 34.8 42.3 –45.6 25.8 24.3 4.3 0 0 PL 1414 84.4 33.9 34 32.1 20.8 27.9 15.3 22.4 13.7 3.5 4.8 PT 1261 70.9 47.5 20.4 32.1 –13.4 23.6 36.4 26.6 3.7 0 RO 2030 89.4 60.4 16.1 23.5 –– 8.2 14.7 77.1 0.4 0.1 SI 1269 80.9 40.8 20.1 39.1 –48.3 20 20.5 11.2 0 0 SK 601 84.7 37.3 24.4 38.3 –25.9 27.3 32.6 14.1 2.1 0 UK 395 72.4 23.1 19.2 57.7 5.6 6.6 24.1 28.7 35 29.6 0.8 BALTIC JOURNAL OF ECONOMICS 23