The impact of the increase on wages, employment and average hours paid

Case study for

Author: Mojca LindiCµ [222198] [B.Sc. University of Ljubljana 2009]

Supervisor: Prof. Dr. Ir. Jan C. van Ours,

A thesis submitted in partial ful…llment of the requirements for the degree of Master of Science in Economics

Faculty of Economics and Business Administration

Tilburg University

Number of words (including references and appendix): 19.379

Date: July 22, 2010 Contents

1 Introduction 1

2 Theoretical background 2 2.1 A competitive labour market ...... 2 2.2 A non-competitive labour market ...... 3

3 Literature review 4 3.1 Research studies before the year 1990 ...... 4 3.2 Research studies after the year 1990 ...... 6 3.2.1 Negative minimum wage e¤ects ...... 7 3.2.2 Positive or undetermined minimum wage e¤ects and their critics . . . . 7 3.2.3 E¤ects of the minimum wage on teenagers ...... 9

4 The data 16

5 Minimum wage in Slovenia 16 5.1 Stylised facts about Slovenia ...... 16 5.2 De…nition of Slovenian minimum wage ...... 17 5.3 Comparison with other countries ...... 19 5.4 Empirical evidence ...... 19

6 The model 25 6.1 Research question ...... 25 6.2 Method ...... 26

7 Parameter estimates 29 7.1 Basic regressions ...... 29 7.2 Sensitivity analyses ...... 30 7.2.1 Impact of the extraordinary minimum wage increase on average gross wages ...... 31 7.2.2 Impact of the extraordinary minimum wage increase on employment . . 32 7.2.3 Impact of the extraordinary minimum wage increase on the average number of hours paid ...... 34 7.3 Concerns regarding the di¤erence-in-di¤erences method ...... 34

8 Discussion and recommendations 35

9 Conclusions 38

i 10 References 40

A Appendix 45 A.1 Descriptive statistics ...... 45 A.2 Description of “SKD united” ...... 48 A.3 Comparison of time trends in treatment and comparison groups ...... 49 A.4 The results of treatment groups 1 and 2 ...... 54 A.5 The results of sensitivity analyses ...... 56

ii Acknowledgements

This thesis would not have been possible to write without the help of many people, to only some of whom it is feasible to give particular mention here. I would like to thank my supervisor, Prof. Dr. Ir. Jan van Ours for guidance, useful comments and advice about my work, which did not only help me a great deal when writing the thesis but will also be very useful for my future work. Great thanks also to my family, friends and fellow students who were encouraging me during the study at the Tilburg University. Your friendship and spiritual support meant a great deal to me. Special gratitude goes to my parents, Matej for his personal support, Carlos for reviewing the draft, giving constructive criticism and encouragement, Ina for re-evaluating the thesis and Igor for the conversation that clari…ed my thinking. I would like to acknowledge the …nancial and academic support of the Institute of Macro- economic Analysis and Development of the Republic of Slovenia.

iii Executive Summary

Studying the minimum wage e¤ects has been a matter of interest for many researchers. At …rst, their main focus was aimed at …nding whether the low-wage market has more char- acteristics of a competitive or a non-competitive labour market. Later, researchers focused more on …nding what kind of e¤ects the minimum wage has. Although the majority of the researchers found adverse employment e¤ects, the debate still has not reached an agreement. Slovenian minimum wage law, declared in 1995, faced the most recent changes in February 2010, especially concerning the minimum wage level. Although employers can adapt to the changes in a few years, it is expected that the law will have a major in‡uence on the labour market. However, due to the lack of research on a Slovenian case study and the fact that the literature does not provide an unambiguous answer concerning the minimum wage e¤ect, it is hard to predict the results. In order to make the possible conclusions concerning what might happen after the new minimum wage law enacted in the beginning of the 2010, the extraordinary minimum wage increase that took place in March 2008 was taken into account to see how the Slovenian labour market reacts to the extraordinary minimum wage increase. The model was based on the di¤erence-in-di¤erences method, consisting of several treatment groups of the most a¤ected industries and taking into consideration three di¤erent e¤ects of the minimum wage; the e¤ect on the average wages, employment and the average hours paid. The observation period lasted from June 1999 till December 2009. On average, the results showed that the extraordinary minimum wage increase had a statistically signi…cant negative e¤ect on the employment and positive and statistically insigni…cant e¤ect on the average wages and average number of hours paid. However, due to the inconsistent results, the e¤ect of the extraordinary minimum wage increase on the Slovenian labour market remains a conundrum.

iv 1 Introduction

Already at the end of 1770s, Adam Smith emphasised the importance of an appropriate wage that would assure workers a decent living, “It is but equity...that they who feed, clothe and lodge the whole body of the people, should have such a share of the produce of their own labour as to be themselves tolerably well fed, clothed and lodged.” Nowadays, the opinions concerning the minimum wage are split; some economists and politicians are in favour of the law; former President Bill Clinton for example stated “I’ve studied the arguments and the evidence for and against a minimum wage increase. I believe that the weight of the evidence is that a modest increase does not cost jobs, and may even lure people into the job market. But the most important thing is, you can’t make a living on $4.25 an hour.” However, there are also many sceptics; among them some Nobel Prize recipients, “The high rate of among teenagers, and especially black teenagers, is both a scandal and a serious source of social unrest. Yet it is largely a result of minimum wage laws. We regard the minimum wage law as one of the most, if not the most, antiblack laws on the statute books.” (Milton Friedman) and “...no self-respecting economist would claim that increases in the minimum wage increase employment.”(James M. Buchanan). Minimum wage has therefore been a subject of interest for policy makers, economists and researchers for decades. In the beginning, the marginalists and institutionalists de- bated whether the low-wage labour market can be characterized as a competitive or a non- competitive labour market. Later, the focus of discussion moved to distinguishing the e¤ects of the minimum wage increase. The majority of the researchers found that the minimum wage on average has a negative impact on employment (in favour of this were especially Neumark and Wascher) while some researchers (especially Card, Katz and Krueger) concluded the min- imum wage increase has no or even a positive in‡uence on the employment. The in‡uence of the minimum wage on the working hours and on wages was on average proved to be positive. Slovenian minimum wage law, introduced in 1995, was recently modi…ed in February 2010. One of its major changes was the increase in the minimum wage level. Due to the lack of research on the Slovenian case study, it is hard to predict the e¤ects of the minimum wage increase in the following years but it is expected that the new law will mostly have a negative impact on the Slovenian labour market. In order to receive plausible indications about what kind of e¤ects the new law will have, the extraordinary minimum wage increase from March 2008 was taken into account. The applied model used the di¤erence-in-di¤erences method, where the di¤erent treatment groups included the industries that were a¤ected the most by the extraordinary minimum wage increase. The thesis will …rst present the economic theory concerning the e¤ects of the minimum wage increase on the competitive and non-competitive labour markets, followed by a literature

1 review over research studies that studied minimum wage e¤ects. In section 4, an overview of the data is given. The thesis will then proceed with an overview of Slovenian minimum wage policy and present the empirical evidence and comparison with other EU countries. Section 6 will introduce the model, present the research question and the method used. Results of the basic regressions and the sensitivity analyses will be introduced in section 7. Section 8 will discuss the results and provide some recommendations while section 9 will summarise and conclude. References and appendix are included in sections 10 and 11.

2 Theoretical background

2.1 A competitive labour market

In a competitive labour market, a minimum wage that is higher than the market equilib- rium wage is going to raise the equilibrium wage level and cause unemployment. There are two reasons for the latter; some workers who were up to that time working at a lower wage are now made redundant due to the higher average labour costs while workers who were not working at the market equilibrium wage are now encouraged to work at the minimum wage (Boeri and van Ours, 2008: 33). Hence, the minimum wage causes unemployment owing to the people who lost their jobs and those who are now willing to work because of the higher minimum level. This unemployment is usually persistent due to the fact that the participants in the labour market have no inducements to change their behaviour. Firms do not want to hire new workers while the unemployed do not want to work at the wage, lower than the minimum wage. The unemployment rate will be higher, the higher the minimum wage level is and the more elastic labour demand and labour supply are (Borjas, 2008: 137-138). Figure 1 presents the situation in the competitive labour market, where the new minimum wage w is higher than the competitive wage w. Higher minimum wage compels employers to dismiss a share of workers (from E to E) but on the other hand it also induces some workers to enter the labour market (ES E ). The result of both events is unemployment (from E  to ES). The theory predicts that due to the minimum wage, the average costs will increase. There- fore, …rms will now have fewer funds available to provide nonwage bene…ts, such as fringe ben- e…ts or trainings. Moreover, if unskilled workers that receive minimum wage get less training, it might lower their overall life earnings and push them in so called “dead-end jobs”(Filer et al., 1996: 173). On the other hand, minimum wage can also have positive e¤ects on welfare, if it induces employees to invest in education in order to gain higher productivity and to evade being displaced (Boeri and van Ours, 2008: 38). If skilled workers (along with the capital) are substitutes for less skilled workers, the latter would appear relatively more expensive compared to the skilled after the minimum

2 Figure 1: The impact of minimum wage on employment in the competitive labour market [Source: Borjas, 2008: 138]

wage increase. Hence, companies will most probably replace the unskilled workers with more skilled ones, along with increasing the earnings and employment of the more skilled workers. The theory therefore predicts that only the less skilled and inexperienced workers will be a¤ected by the minimum wage increase (Filer et al., 1996: 171). Although the minimum wage is mainly aimed at increasing the incomes of the least skilled workers –who usually come from lower-income households –its welfare e¤ect might be low, especially for this group of workers. Namely, because of the minimum wage, least skilled workers are now more exposed to being discharged. Yet, the workers who preserve their positions receive higher wages and increase their incomes (Borjas, 2008: 138).

2.2 A non-competitive labour market

The introduction of a minimum wage in a non-competitive labour market – such as a labour market with one employer –has more inde…nite consequences. Similar to a monopolist, who matches prices to the marginal revenues, the monopsonist equalises the marginal hiring costs and the marginal revenues of labour. Furthermore, while monopolists’prices are higher than in perfect competition, the monopsonist sets wages that are lower than in the competitive labour market. If the minimum wage in a monopsony is set between the monopsonistic level and the level of a competitive labour market, this leads to a higher level of wages and employment. If the monopsonistic level of the minimum wage is set higher than the wage in the competitive labour market, this leads to decrease in the employment (Boeri and van Ours, 2008: 34-36). If the minimum wage is set at the level of the competitive wage, the market power of the monopsonist would be eliminated and the exploitation of workers diminished

3 (Borjas, 2008: 203). Nevertheless, presented e¤ects of the minimum wage might change on the long run in a monopsony market. Namely, the pro…ts then decline due to the higher average costs of labour and some employers are forced to leave the market. If many …rms ought to shut down their businesses, the employment will decrease due to the minimum wage on the long run (Ehrenberg and Smith, 2003: 79). The impact of the minimum wage on the distribution of wages and employment in the imperfect labour markets could not be estimated with a clear-cut answer also by reason of several factors. As for the impact on wage distribution, there are di¤erent e¤ects and reasons liable for the …nal outcome. For example, the spike in the wage distribution at the minimum wage level depends on the presence of di¤erent segments of labour market (such as skill-di¤erentiated), increase in the e¤ort or the level of the minimum wage. As for the employment e¤ect, the strength of the minimum wage e¤ect among others depends on the sensitivity of wages and elasticity of the supply of labour. The minimum wage will in‡uence the employment more negatively if the dispersion of wages is bigger among industries. Therefore, some economists suggest it would be more useful if there were di¤erent minimum wages for di¤erent industries (Manning, 2003: 325-347). Apart from rare one-company central planned towns in Russia, monopsony is hardly ever observed in practice. However, there are labour markets with characteristics of the monop- sonistic market; for example mobility costs, job creation costs and hiring costs aggravate employees to change jobs or/and the employers to create new vacancies or hire a new person. The increase in the employment due to the minimum wage is present in these “non-perfect” monopsonies (Boeri and van Ours, 2008: 36-37). To sum up, economic theory predicts that the employment will decrease due to the min- imum wage in the competitive markets while the outcome is not so clear-cut when markets have characteristics of a monopsony.

3 Literature review

3.1 Research studies before the year 1990

Minimum wage and its e¤ects have been the source of heated discussions for many years. The …rst debates started after the set up of the minimum wage in the US in the 1930s. Some economists (i.e. the marginalists) argued that the low-wage labour market can be distin- guished as a competitive market while the other pole (i.e. the institutionalists) claimed this does not hold (Neumark and Wascher, 2007: 1). In his well-cited paper, Stigler (1946) mostly agreed with the model of competitive wage determination and asserted that the minimum

4 wage most probably reduces the aggregate output, raises the earnings of the workers whose wages were just below the new minimum wage but reduces the earnings of those who earned well beneath the new minimum. In addition, Stigler claimed that the minimum wage would have two e¤ects; one is that the services of some of the workers will now become too expensive so they will become unemployed and the other is that the productivity of low e¢ ciency work- ers will increase; either by working harder or by the introduction of special entrepreneurial techniques. Stigler also assumed that the higher the minimum wage, the more people will be discharged from their working positions (Stigler, 1946: 358-362). Burkhauser and Finegan (1993) con…rmed Stigler’spredictions with an empirical research and concluded that the min- imum wage does not help the working poor and might even provide less income to low-income households. Lester (1947) commented on Stigler’s paper, holding the opposing views. Lester (1947) claimed that manufacturing …rms did not replace their regular employees with more e¢ cient ones because of the minimum wage. In addition, he supposed that only the minor part of the industry is a¤ected and that it is less likely that the …rms would have to decrease their output. Lester (1947) argued that instead of converting the workforce, the management’s e¢ ciency will change in order to achieve higher pro…tability (Lester, 1947: 142-146). Thereupon, many economists were stimulated to …nd empirical substantiation of the minimum wage e¤ects (Neumark and Wascher, 2007: 1). In the aftermath, the core of the economic debate concerning minimum wage moved from the discussion about whether the labour market for low-skilled workers is competitive or not to the empirical e¤ects of the minimum wage. One of the main apples of discord was whether the minimum wage increase has an adverse e¤ect on the employment or not. The bulk of the research implied that the increase in the minimum wage has a negative impact on the low-skilled workers (Neumark and Wascher, 2007: 1). Brown, Gilroy and Kohen (1982) made a review of several research studies about the time- series and cross-section studies of teenagers and youth which all concluded that the increase in the minimum wage causes employment reductions. However, the estimates concerning e¤ects on unemployment vary among surveys and it would be hard to determine the signi…cant e¤ect of the minimum wage increase. Moreover, authors were cautious about the results since the majority of the studies did not include sensitivity analyses and the results seemed to change with di¤erent extensions of the observation period (Brown, Gilroy and Kohen, 1982: 32-44). Furthermore, concerning the minimum wage and adult employment, studies also found negative employment e¤ects but were somehow smaller than the ones for teenagers, whereas employment e¤ects di¤ered among low-wage industries (Brown, Gilroy and Kohen, 1982: 47-69). Studies by Meyer and Wise (1982) and Welch and Cunningham (1987) both concluded that

5 the minimum wage had negative e¤ects on the employment of youth in the period before the 1980s. Contrary to the common belief that the negative employment e¤ects of the minimum wage are usually counterbalanced with increased earnings, Meyer and Wise (1982) did not …nd an earnings e¤ect. Mincer (1976) also con…rmed the negative employment e¤ects due to the minimum wage for the majority of the observed groups. He concluded that non-white males, followed by non-white teenagers, white males and white teenagers faced the highest increase in the unemployment rate.

3.2 Research studies after the year 1990

Researchers started to re-examine the employment e¤ects of the minimum wage in the 1990s. In the new wave of research, economists improved the already existing research studies with more recent data series, included new methods, started to use the variation in the minimum wage levels among the United States’states and took into account the economic conditions in order to get more robust results (Neumark and Wascher, 2007: 2-3). After more than a decade, Brown (1999) made another précis of the literature, related to the minimum wage e¤ects. He concluded that the new studies indicate smaller negative e¤ect on the teenage employment than the previous ones. Moreover, studies that took time and state variables into account, found smaller minimum-wage e¤ects than those which controlled for the two e¤ects separately. The reasons for a weaker in‡uence of the minimum wage might be the incomplete minimum wage coverage, substitution of less-skilled teenagers for more-skilled ones, not taking the possible fall in weakly hours into account, the possibility that the labour market has more in common with the monopsony than with the competitive market (however the majority of the research studies assumed the latter) and …nally, that the demand for low- skilled workers might not be elastic in the short period (though, evidence for this statement is lacking). As the evidence of the employment e¤ects due to the minimum wage remains more or less contentious, the minimum wage evidently has an in‡uence on the wage distribution and incomes, particularly for teenagers. Brown (1999) concludes that the e¤ect of the minimum wage is less obvious when observing the wage distribution over a several-years period. Neumark and Wascher (2007) also made an exhaustive review of the minimum wage studies. Among more than one hundred studies, almost two thirds of them proved consistent but not always signi…cant disemployment e¤ects of the minimum wage while less than ten found positive employment e¤ects. Negative employment e¤ect appears to be particularly strong for the least-skilled workers. Furthermore, Neumark and Wascher (2007) agree with the marginalists that the labour market for low-wage workers can be compared to the competitive model. Authors conclude that in order to estimate the e¤ectiveness of the minimum wage policy, one needs to take into account not only the employment e¤ect of the policy but also

6 its e¤ects on the distribution of wages, working hours, incomes, human capital accumulation and other possible variables.

3.2.1 Negative minimum wage e¤ects

There were many studies that observed employment e¤ects of the minimum wage. Abowd et al. (2009) used panel data and di¤erence-in-di¤erences estimate when examining the em- ployment e¤ects in the United States and France. Researchers found no employment e¤ects in the United States whereas France was facing a strong and negative employment e¤ects. Con- cerning research studies made on the basis of the US data, Deere, Murphy and Welch (1995) and Kim and Taylor (1995) concluded that when the costs of low-wage workers increase, the employment of these workers decreases. Singell and Terborg (2007) found negative employ- ment e¤ects for the industries where the minimum wage is relatively binding, but not for those that have less strict legislation. David Neumark conducted many research studies concerning the minimum wage. In the one with Neumark, Schweitzer and Wascher (2004), they concluded that the low-wage work- ers are more strongly a¤ected by the minimum wage increase. Workers, whose income lies between the previous and the new minimum wage level, face the wage increase. However, they are also confronted with the diminishing employment and declining number of working hours. Altogether, the net e¤ects of the minimum wage increase for the low-wage workers are nega- tive. Neumark and Nizalova (2006) took into account the long-term e¤ects of the minimum wage recipients and found that workers in their late twenties earned less the longer they have received the minimum wage when they were younger and that the minimum wage can have negative or positive long-term e¤ects; the …rst due to decreased labour market experience and reduced education and training, whilst the latter from the potential skill acquirement. When doing a research about the minimum wage in Brazil, Neumark, Cunningham and Siga (2006) could not con…rm that the minimum wage increased earnings of the low-income fami- lies. Conversely, several facts indicated negative consequences of the minimum wage for these households. A similar study was made by Neumark, Schweitzer and Wascher (2005) for the US data, where the results also indicated the negative net e¤ect of the increase in the mini- mum wage for the low-income families, although the earnings of some of these families near or below the poverty line increased.

3.2.2 Positive or undetermined minimum wage e¤ects and their critics

Surveys that did not …nd negative employment e¤ects but more condensed distribution of earnings due to the minimum wage, were made by Dickens, Machin and Manning (1999) on British data and Lemos (2007) for Brazil, while the research by Stewart (2004) only con…rmed

7 the …rst part; i.e. no adverse employment e¤ects in the UK case. Metcalf (2004) wrote an overview of the literature for the UK and concluded no adverse e¤ect on employment but increased intensity of training due to the minimum wage. Levin-Waldman (2000) took the alternative approach and examined the consequences of the minimum wage increase on the employers. His study concluded that the treated minimum wage increase did not have harmful impacts on employers. However, the survey showed that there would be one if the minimum wage increase was bigger. Research studies made by Card, Katz and Krueger received a great deal of attention and led to an intense debate among researchers concerning the e¤ects of the minimum wage. The research by Katz and Krueger (1992) studied the impact of the minimum wage change on the fast-food restaurants in Texas. Their results indicate that employers increased the wages of fast-food workers for more than it was determined by the new law. Moreover, the new law condensed the distribution of the starting wages in the industry. Employment increased in those companies that were supposedly more a¤ected by the law while the changes in prices were proven not to be the consequence of the law. Katz and Krueger (1992) supplement their results with the comment that larger increases in the minimum wage might cause the decrease in employment and increase in prices. Card and Krueger (1994) also made a research on the fast-food industry. When comparing the less a¤ected New Jersey fast-food restaurants with more a¤ected ones or when comparing New Jersey fast-food restaurants with Pennsylvanian, the results indicated that the minimum wage increase had a positive e¤ect on the employment. Moreover, Card and Krueger (1994) found that prices of meals in fast-food restaurants increased due to the new law, but the increase was not higher in more a¤ected restaurants. Similar conclusions than in the previous two studies were made by Card (1992) on the case of the minimum wage increase in California. The results indicated that the raise of the minimum wage increased the earnings of low-wage workers and did not reduce the employment of teenage workers. Many researchers were stimulated to replicate the presented three studies. Neumark (1993), Neumark and Wascher (2000) and Clark, Kaas and Madden (2006) all concluded, that there in fact were negative employment e¤ects due to the increased minimum wage in the three presented studies of Card, Katz and Krueger. Neumark and Wascher exchanged several comments with Card and Krueger in the continuing period, where the …rst two re- searchers usually concluded that the minimum wage has a negative e¤ect on the employment while Card and Krueger advocated the opposite (Neumark and Wascher, 2007: 29-34).

8 3.2.3 E¤ects of the minimum wage on teenagers

Concerning e¤ects of the minimum wage on the teenage employment, Baker, Benjamin and Stanger (1999) found a decrease in the teenage employment in Canada. The same con- clusions were made by Burkhauser, Couch and Wittenberg (2000) in the US example and by Pereira (2003) who found that the new minimum wage law in Portugal decreased the employ- ment of teenagers who were a¤ected by the law, but increased the employment of younger adults aged 20 to 25. Bazen and Martin (1991) could not signi…cantly con…rm the negative e¤ects on the youth employment in France but proved the positive in‡uence on real youth earnings due to the minimum wage. Short review of the presented studies, written after the 1990 is shown in the table 1.

Table 1: Review of the minimum wage studies

Author(s) Country Observed group Type of data and E¤ects of the methods used minimum wage Negative minimum wage e¤ects (in a chronological order) Abowd et al. The United Focus on Panel data The US: no (2009) States workers, whose (individual wage employment and France. wages are close and employment e¤ects. to the minimum data). The France: strong wage. United States: negative 1981-2000, employment France: e¤ects. 1990-1998. Di¤erence-in- -di¤erences method. Singell and The United Eating and Monthly Negative Terborg (2007) States drinking employment data employment (referendum for industry; hotel (1997-2001); e¤ects in the minimum and lodging help-wanted ads industries where wage increase; workers. (1994-2001). the minimum wage 1996 in Oregon Employment is more binding and 1998 in growth model but not where Washington). (with time-varying minimum wage is and seasonally less binding. observed factors).

9 Table 1: Review of the minimum wage studies

Author(s) Country Observed group Type of data and E¤ects of the methods used minimum wage Neumark and The United Individuals, Observation period: Long-term e¤ects: Nizalova (2006) States (states aged 16-29. 1979-2001 (panel individuals in late and federal). data). The minimum twenties earn wage history for less the longer every individual. they received The model: e¤ects higher minimum of the minimum wage (the wage on wages, consequences employment, hours are stronger for and earnings. African-Americans). Neumark, Brazil (increase All workers. Monthly panel data No positive Cunningham in the minimum for six states e¤ects on and Siga (2006) wage in 2003 (1996-2001). low-income by the households. President). Neumark, The United All families. Di¤erence-in- The minimum Schweitzer and States di¤erences estimates wage raises the Wascher (2005) (comparison of the e¤ects of the incomes of some between states minimum wage poor families, that faced and increase. but also did not face the Observation period: increases the minimum wage 1986-1995. share of families increase). below or near the poverty line. Neumark, The United Workers from Panel data on Employment and Schweitzer and States (federal di¤erent individuals in the number of Wascher and states). categories of matched monthly working hours (2004) the wage group (1979-1997). of the low-wage distribution. workers decreased.

10 Table 1: Review of the minimum wage studies

Author(s) Country Observed group Type of data and E¤ects of the methods used minimum wage Deere, Murphy The United Teens and adult Panel data, With the and Welch States (federal high school controlling for increase in costs (1995) minimum wage dropouts. demographic and of low-wage increases in geographic trends workers, fewer of 1990 and 1991). (1985 to 1993). these workers The changes in are employed. employment of high- and low-wage states. Kim and Taylor The United Real trade Panel data (1988 to Lower growth (1995) States industry. 1989). The change of employment (minimum wage between the US and in low-wage increase in California’s retail trade California in employment in real sector. 1988). trade industry. Burkhauser and The United 17-64 year old Samples of the data Con…rmed …ndings Finegan (1993) States workers. from the years 1940, of Stigler (1946)- 1950, 1960, 1970, the minimum wage 1980 and 1988. The has negative research controlled consequences for for the income of the working poor other members of and the low- households. income households. Positive or undetermined minimum wage e¤ects (in a chronological order) Lemos (2007) Brazil Private and Panel data More compressed public sector, (1982-2000). The wage distribution workers of model controlled for in both sectors all ages. in‡ation, and no negative unemployment rate impacts on and region and time employment …xed e¤ects. (di¤erent groups of workers observed).

11 Table 1: Review of the minimum wage studies

Author(s) Country Observed group Type of data and E¤ects of the methods used minimum wage Metcalf (2004) The UK Workers that Overview of the Between 1998 were a¤ected di¤erent UK studies. and 2000, low- by the minimum wage workers wage increase. experienced above average pay rises and no overall employment e¤ect. Probability and intensity of training increased. Stewart (2004) The UK Workers whose Longitudinal data No signi…cant (studying the wages were (three di¤erent data negative e¤ects on impact of the a¤ected by the sets). The research the employment minimum wage new law. used the di¤erence- when observing introduction in in-di¤erences di¤erent the UK in estimator. demographic April 1999). groups or data sets. Levin- The United Employers in Sample of 560 small Minimum wage Waldman States (the small businesses (phone increase did not (2000) 1997 minimum businesses. survey). have severe wage increase). Respondents were consequences. asked if the latest However, minimum wage employers increase in‡uenced expected them them. if the increase was bigger.

12 Table 1: Review of the minimum wage studies

Author(s) Country Observed group Type of data and E¤ects of the methods used minimum wage Dickens, The UK Workers in the Panel data Compression Machin and covered (1975-1992). The e¤ect in the Manning (1999) industries. employment earnings equation included distribution and council-speci…c non-negative …xed e¤ect, in‡uence on macroeconomic employment due e¤ects, real sales to the minimum and trends for wage. speci…c sectors. Card and The United Fast-food Survey of 410 Positive impact Krueger (1994) States (1992 restaurants in fast-food on the increase in the New Jersey restaurants, employment. New Jersey and comparison of the minimum Pennsylvania. employment e¤ects wage). within New Jersey and between New Jersey and Pennsylvania fast-food restaurants. Card (1992) The United Teen Comparison of Earnings of States (1988 employment changes for workers low-wage minimum wage and retail trade in California with workers increase in employment. ones that did not increased and California). face the minimum there was no wage increase loss in teenage (1987-1989 and employment or 1984-1990). in positions in retail trade.

13 Table 1: Review of the minimum wage studies

Author(s) Country Observed group Type of data and E¤ects of the methods used minimum wage Katz and The United Fast-food Survey of Positive e¤ect on Krueger (1992) States (impact restaurants in restaurants in 1990 the employment, of federal Texas. and 1991 (critics more compressed increase in claim large amount distribution of 1991). of measurement starting wages errors; e.g. and no in‡uence Neumark and on changes in Wascher, 2007: 142). the meal prices. Critics of studies that concluded positive minimum wage e¤ect (in a chronological order) Clark, Kaas The United Fast-food Model with a Reduction in and Madden States (1992 restaurants in two-stage game wages and (2006) New Jersey New Jersey (stage I: employment. minimum wage and determination of increase). Pennsylvania. wages and employment; stage II: determination of prices and output). Neumark and The United Fast-food Restaurants from Decrease in the Wascher States (1992 restaurants in the same geographic New Jersey (2000) New Jersey New Jersey areas and chains employment in minimum wage and as in Card and fast-food increase). Pennsylvania. Krueger (1994). restaurants, Di¤erence-in- compared to the di¤erences model. Pennsylvanian. Neumark The United Teenagers and Observation period: Negative (1993) States (federal young adults. 1973-1989, a employment and states). re-examination of e¤ects on existing evidence teenagers and and comments on young adults. work by Card, Katz and Krueger.

14 Table 1: Review of the minimum wage studies

Author(s) Country Observed group Type of data and E¤ects of the methods used minimum wage E¤ects of the minimum wage on teenagers (in a chronological order) Pereira (2003) Portugal Individuals, Observation period: A negative (elimination of aged 18-19 and 1986-1989. Model impact on the teenage 20-25. collated the teenage employment of subminimum employment growth teenagers, but the wage in 1987). with the employment increase in the growth of older employment of workers. 20-25 year olds. Burkhauser, The US Teenagers. Monthly panel data Adverse (but Couch and (federal (1979-1997). The modest) e¤ects Wittenberg minimum wage model controlled for on the teenage (2000) increases). the di¤erences in employment. minimum wages between states and the federal state, changes in federal policy and year speci…c e¤ects. Baker, Canada (across Teenagers Panel data (1975 Decrease in the Benjamin and provinces). (15-19). to 1993). Measure teenage Stanger (1999) for the minimum employment. wage: the ratio of the adult minimum wage to the average manufacturing wage. Bazen and France Youth labour Observation period: Positive long- Martin (1991) market. 1963-1986, annual run impact time-series data. on earnings of youth but no signi…cant negative e¤ects on the employment.

15 4 The data

The data used in this study was from the Statistical O¢ ce of the Republic of Slove- nia (SURS) and from the Agency of the Republic of Slovenia for Public Legal Records and Related Services (AJPES). SURS is the core provider of statistical data in Slovenia, whereas AJPES carries out di¤erent statistical tasks and research studies, mainly related to companies and sole proprietors. Only the variable Harmonised Index of Consumer Prices (HICP) was acquired from the EUROSTAT database. All included variables are of monthly-type, except for the number of population (half-year data) and GDP by industries (quarterly data) which were transformed into monthly data with the assumption that the population and GDP by industries does not change signi…cantly during the half-year or quarterly period respectively. Therefore, the data is of panel type; controlling for the monthly time variable and cross- sectional type of the industry. The observation period is from June 1999 until December 2009. The descriptive statistics of the main variables is included in appendix 1 (table 6). In January 2008 the new classi…cation of activities of business entities NACE Rev. 2 replaced NACE Rev 1.1 in all EU Member States. The national version of the standard classi…cation (SKD 2008) also came into force on the cited date in Slovenia. SKD 2008 includes the entire European classi…cation of activities but also adds some national subclasses. Therefore, it was necessary to combine both classi…cations in order to have consistent results. The combination of both classi…cations (“SKD united”) was done as a rough comparison, suggested by SURS (SKD). A more accurate arrangement was not possible due to the lack of subcategories among some of the variables. The table for converting into “SKD united”is included in appendix 2 (table 7).

5 Minimum wage in Slovenia

5.1 Stylised facts about Slovenia

Slovenia was a part of Yugoslavia from the beginning of 20th century and gained its independence in 1991. Because of the historical bonds with Western Europe, strong economy and stable democracy, Slovenia became a member of the EU and NATO in May 2004 and signed the accession treaty to the OECD in June 2010. Slovenia was also the …rst of the “new” Member States that received the Euro in 2007. Until recently, Slovenian …ne infrastructure, a well-educated workforce and the strategic location between the Balkans and Western Europe were some of the main factors for higher foreign involvement, high GDP per capita growth and decreasing unemployment. Despite the outstanding economic performance, foreign direct investment seemed to be relatively low when compared to other countries in the region, taxes were considered high and it was claimed that the labour market was in‡exible (CIA).

16 Signs of turmoil started to appear after receiving the Euro, with energy and food price shocks, the highest in‡ation within the Euro area and unemployment falling below the es- timated natural rate. In the era of the global …nancial crisis, Slovenia was distressed by decreased foreign demand (especially from Germany). The most a¤ected industries were those making cyclically-sensitive goods (e.g. the automobile sector) and the banking sector (OECD). Before the crisis, Slovenian labour market was faced with increasing employment and de- creasing unemployment. However, the situation changed after the crisis. Nowadays, there are worries concerning low labour force participation rate of the elderly and younger population and high hiring and …ring costs. OECD suggested Slovenia to increase the retirement age, restrict the early retirement programs and decrease the length of tertiary studies in order to improve the labour market (OECD). In the beginning of 2010, Slovenian population was 2 046 976 inhabitants, where 935 795 of them participated in the labour force (836 902 in paid employment and 98 893 registered as unemployed). Registered unemployment rate in March 2010 was 10.6%, compared to September 2008 when it was 6.3%. The GDP has decreased from e37 135 million in 2008 to e34 894 million in 2009 (SURS). The recovery of the Slovenian economy is slower than in other countries of the Euro zone. While the Euro zone was on average showing signs of a positive economic growth in the beginning of 2010, Slovenia was still facing a real downturn of the GDP growth (GDP growth was 0.5% lower in the …rst quarter of 2010 compared to the previous quarter). Forecasts show the healing of Slovenian economy will be slow, with the economic growth estimate being 0.6% in the 2010 (UMAR).

5.2 De…nition of Slovenian minimum wage

Slovenian law declares the minimum wage as the lowest amount of payment, received by full-time employees. Many international organizations and contracts declare that every employee has the right to receive a decent payment for his or her work. Although the level of minimum wage is usually set above the poverty line, the de…nition of the minimum wage di¤ers from country to country especially concerning whether the minimum wage should assure a decent standard of life only to the recipient of the minimum wage or to his or her family as well (Brezigar Masten et al., 2010: 2). Moreover, de…nitions also vary depending on the type of the minimum wage in one country. Namely, a minimum wage can be government-legislated or the outcome of collective bargaining (Boeri and van Ours, 2008: 29-30). The Slovenian minimum wage is adjusted annually by the government, based on the arrangement with social partners; i.e. so called “tripartite agreement”. In addition, the minimum wage a¤ects all employees with no exemptions (there is no di¤erentiation between regions, sectors, age or quali…cation) and the coverage is approximately 2.7% of all employees (Funk and Lesch, 2005). In the case

17 of non compliance with the legislation, the highest punishment for a …rm that breaks the law is e20.000 (Uradni list RS, 2010). The Slovenian minimum wage was enacted in 1995. From 1997 until 2003, the minimum wage was increased once a year, taking into account the consumer price indices and the real GDP growth. As a consequence of calculating the minimum wage level in this way, the minimum wage growth was higher than the growth of wages in the private sector; the ratio between the minimum wage and average gross wage in the private sector was for example 42.5% in 1995 and it increased to 46.2% in 2005. The aim of this calculation was to make sure that employers would not try to attain higher pro…ts by decreasing the labour costs but by increased added value. On the other hand, this way of calculating the minimum wage level presented a huge burden for employers, which resulted in a higher share of employees that received lower wages. From 2006, the minimum wage increased once a year every August only due to the expected in‡ation. As a consequence, there was a decrease in the ratio of the minimum wage and the average gross wage in years 2006 and 2007 and the decrease in the real minimum wage in 2007. In 2008, labour unions forced the government to make an extraordinary increase of the minimum wage in March 2008. Although the ratio between the minimum wage and the average gross wage increased again, there was a pretension to introduce a new law, which was accepted in February 2010. The main di¤erence is that the minimum wage will now be adjusted once a year, after consulting with social partners and taking into account consumer price indices, wage movement, growth rate and the movement of employment. The new law also signi…cantly enhanced the level of the minimum wage but allowed employers to adapt to it gradually until January 2012. Projections show the new law will have a great and mostly negative in‡uence on Slovenian labour market (Brezigar Masten et al., 2010: 3-5). When carrying out an economic survey of Slovenia, OECD concluded the ratio of the minimum wage and the average gross wage should be reduced further (OECD, 2009), while it is assumed the ratio will increase due to the new law (Brezigar Masten et al., 2010: 4-5). The levels of monthly minimum wages in the last few years were as follows: from the 1st August 2007 until the 29th February 2008, the minimum wage was e538.53 gross. The extraordinary increase on the 1st March set the minimum wage to e566.53 gross, from the 1st August 2008 until the 1st July 2009, the minimum wage increased to e589.19 gross and from the 1st August 2009, the minimum wage was e597.43 gross. The new law increased the minimum wage to e734.15 gross on the 1st March 2010 (DURS). However, if the employer is allowed to increase the minimum wage gradually, the current minimum wage value is e654.69 gross (Uradni list RS, 2010).

18 5.3 Comparison with other countries

The majority of the EU countries have some form of the statutory minimum wage, varying in types, coverage and the groups of exempted employees. Slovenia has one of the highest levels of the minimum wage among member states, which joined the EU in 2004 and 2007, but when taking into account all Member States; its minimum wage is somewhat average. Slovenia is in a group of EU countries that treat the minimum wage as a mean of social protection whereas employers see the minimum wage as an obstruction that increases labour costs and hinders hiring. The majority of the EU Member States (Slovenia being one of them) reported that the minimum wage had a positive in‡uence on female incomes and the decline in the gender wage gap (Funk and Lesch, 2006). When comparing the minimum wages as a proportion of average monthly earnings as reported by EUROSTAT for 2008, the ratio among Member States ‡uctuated between 50.2% in Malta and 30.5% in Romania. Slovenia is in the upper half in the EU by this indicator (…gure 2).

Figure 2: Minimum wage as a proportion of average monthly earnings across the EU Member States [Source: EUROSTAT]

MT=Malta, LU=Luxembourg, NL=, SI=Slovenia, BG=Bulgaria, LT=Lithuania, PT=Portugal, HU=Hungary, GB=United Kingdom, LV=Latvia, ES=Spain, PL=Poland, CZ=Czech Republic, EE=Estonia, RO=Romania

5.4 Empirical evidence

The relation between the minimum wage and the average gross wage in Slovenia ‡uctuated through the observation period (…gure 3) but did not deviate on average for more than a

19 magnitude of 0.04 from the initial value in December 1999 (0.42), which indicates that it stayed more or less the same through the observed period. The ratio between the minimum wage and the average gross wage declined between the time that the nominal ‡oor was set and the next time that the government raised it (usually every August). Therefore, the biggest decline was usually every July (similar pattern is visible in the US data, Borjas, 2008: 137). This trend implies that the economic impact of minimum wage diminishes from the last rise (Borjas, 2008: 137). The vertical line in …gure 3 presents the extraordinary increase of the minimum wage that took place in March 2008. In order to smooth out the peaks of the average wages caused by the 13th yearly wage and Christmas bonuses, the wages in …gure 3 are calculated as moving averages.

Figure 3: The movement of the minimum wage in Slovenia [Source: SURS, DURS, own cal- culations]

Concerning the wage distribution in September 2007 and in September 2008, the highest real average wage was in the industry D (Electricity, gas and water supply) and the lowest was in the industry G (Hotels and restaurants) (table 2).

20 Table 2: Lowest and highest real gross earnings (June 1999=100), September 2007 and 2008 [Source: SURS, own calculations] September 2007 September 2008 Real average monthly Real average monthly gross earnings gross earnings (June 1999 =100) (June 1999 =100) Minimum wage Minimum wage (June 1999 =100) : 347 (June 1999 =100) : 360 highest average lowest highest average lowest (full (full time) time) Industry ("SKD united") EUR EUR A Agriculture, hunting, 1272 690 472 1378 738 508 forestry and …shing B Mining and quarrying 1861 988 492 2227 1129 528 C Manufacturing 1495 695 485 1560 729 515 D Electricity, gas and 2385 939 627 2446 961 648 water supply E Construction 1065 674 460 1105 711 485 F Wholesale, retail; 1033 737 513 1070 769 535 certain repair G Hotels and 860 590 414 882 619 438 restaurants H Transport, storage 1160 910 558 1195 951 576 and communication I Financial 1974 1189 638 1993 1255 657 intermediation J Real estate, renting 1187 812 625 1213 849 652 and business activities K Public administration 1852 990 530 1967 1062 536 and defence; compulsory social security L Education 1558 1002 489 1574 1019 494 M Health and social work 1659 894 553 1797 1029 601 N Other social and 1162 887 578 1187 917 609 personal services

21 Table 3 presents people in paid employment by amount of gross wages in September 2007 and September 2008. Since the lowest wage class from the year 2007 did not exist anymore after the extraordinary minimum wage increase in March 2008, one could predict what might happen in 2008 by summing the lowest two wage classes from the September 2007. After summing the percentages in classes up to 550, and from 551 to 615 (columns 1 and 2, while the sum is in the column 3), we could expect that, due to the extraordinary increase in 2008, the number of minimum wage recipients would increase the most in the industries E (Construction), G (Hotels and restaurants) and J (Real estate, renting and business activities). The results in the third column would also be in accordance to the presented literature, since the research studies usually found the condensation of wage distribution around the minimum wage after the increase of the minimum wage. However, when comparing the third column with the year 2008 (column 6), the percentages of employees that received the actual wages from e551 to e615 did not increase as much as one would expect from the 2007 data. Table 3 indicates that the wage distribution adapted to the extraordinary wage increase very quickly; probable reasons could be the anticipation e¤ect or that the new minimum wage increase was not su¢ cient. The highest relative share of the minimum wage recipients in September 2007 and in September 2008 was in the industry J (Real estate, renting and business activities), whereas the lowest was in the industries B (Mining and quarrying) and L (Education) in September 2007 and A (Agriculture, hunting, forestry and …shing), B (Mining and quarrying) and L (Education) in September 2008 (table 4).

22 Table 3: Persons in paid employment by amount of gross wages, by industries, Slovenia, Sep- tember 2007 and September 2008 [Source: SURS, own calculations] 2007 2008 (1) (2) (3) (4) (5) (6) (7) (8) 551 4101 Total 551 4101 Total to to to and to and 550 615 615 more 615 more Industry ("SKD united") % % Total 5.1 5.3 10.4 89.6 100 5.3 94.7 100 A Agriculture, hunting, 5.1 6.5 11.6 88.4 100 4.6 95.4 100 forestry and …shing B Mining and quarrying 0.6 0.7 1.3 98.7 100 0.5 99.5 100 C Manufacturing 6 7 13 87 100 5.7 94.3 100 D Electricity, gas and 0.5 0.8 1.3 98.7 100 0.6 99.4 100 water supply E Construction 13.3 8.1 21.4 78.6 100 15 85 100 F Wholesale, retail; 4.6 6.4 11 89 100 4.5 95.5 100 certain repair G Hotels and 8.3 10.4 18.7 81.3 100 10.4 89.6 100 restaurants H Transport, storage 3.2 2.8 6 94 100 4.4 95.6 100 and communication I Financial 2 1 3 97 100 1.4 98.6 100 intermediation J Real estate, renting 9.4 7.3 16.7 83.3 100 10.3 89.7 100 and business activities K Public administration 0.6 1.1 1.7 98.3 100 0.5 99.5 100 and defence; compulsory social security L Education 0.9 1.3 2.2 97.8 100 0.6 99.4 100 M Health and social work 3.2 3.9 7.1 92.9 100 1.3 98.7 100 N Other social and 3.7 3.5 7.2 92.8 100 3.6 96.4 100 personal services

23 Table 4: Number of persons in paid employment with minimum wage by industries, September 2007 and 2008 (MW = minimum wage, WP = working population) [Source: SURS, own calculations] September 2007 September 2008 Persons in paid employment Persons in paid employment With MW Employees Working Share of MW With MW Employees Working Share of MW population recipients population recipients (MW/WP) (MW/WP) Industry ("SKD united") Number % Number % Total 22869 771066 858870 0.027 20782 795591 884790 0.023 A Agriculture, hunting, 158 6351 40979 0.004 67 6552 40654 0.002 forestry and …shing B Mining and quarrying 8 3673 3722 0.002 7 3496 3540 0.002 C Manufacturing 8286 222113 229736 0.036 6655 220860 228458 0.029 D Electricity, gas and 38 11363 11390 0.003 42 11461 11497 0.004

24 water supply E Construction 3355 70562 80884 0.041 3470 79512 90562 0.038 F Wholesale, retail; 2400 104188 112158 0.021 2080 107740 115675 0.018 certain repair G Hotels and restaurants 1005 28812 32854 0.031 1009 29539 33537 0.03 H Transport, storage 818 50732 56252 0.015 988 53404 58862 0.017 and communication I Financial intermediation 160 21969 22378 0.007 133 22831 23307 0.006 J Real estate, renting 5071 68871 76304 0.066 5185 74449 82625 0.063 and business activities K Public administration and 223 50205 50206 0.004 301 51021 51024 0.006 defence; compulsory social security L Education 136 58467 58712 0.002 144 59625 59925 0.002 M Health and social work 714 49713 51433 0.014 282 50612 52407 0.005 N Other social and 497 24047 31862 0.016 419 24489 32717 0.013 personal services Furthermore, …gure 4 shows that there were more employees with lower wages in September 2007 compared to 2008 while the opposite stands for employees with higher wages. This could be due to the extraordinary minimum wage increase in March 2008, or more likely due to the overall real wage increase since table 3 does not report remarkable change in wage distribution in September 2008 compared to September 2007. The legend presents classes of real wages, where the September 2008 wages are transformed into September 2007 wages. The upper class of the legend in …gure 4 stands for September 2007 whereas the lower one is for September 2008.

Figure 4: Persons in paid employment by amount of gross wages (cumulative distribution) [Source: SURS, own calculations]

6 The model

6.1 Research question

After demonstrations by workers in 2007, 2008 and 2009, where one of the demands by protestors was the increase in the minimum wage, the government legalized new minimum wage law in the beginning of 2010 and gave the companies time to adjust to it until January 2012. Owing to the short observation period and the fact that …rms can set the new minimum wage gradually, it is not yet possible to answer whether the increase in the minimum wage and the change of the law was sensible in the period of …nancial crises. However, since the extraordinary increase in the minimum wage happened also in March 2008, one could see how Slovenian labour market adjusted to that increase and then make inferences about what

25 might happen after February 2010 increase. The main focus of this study will be aimed at answering the question regarding the outcome of the extraordinary minimum wage increase in March 2008 on the employment, wages and the average number of hours paid in Slovenia. As some researchers proposed, the employers can also adjust the number of working hours instead of lowering the employment (Neumark and Wascher, 2007: 34-36), therefore the variable average number of hours paid was also included in the model. Since the literature does not provide a clear cut answer concerning the impacts of the minimum wage increase, and due to the lack of research on the Slovenian case, it is hard to predict the conclusions. Concerning the e¤ect of the minimum wage on average wages, some researchers found the increase in wages for those employees who retained their positions (e.g. Bazen and Martin (1991), Card (1992) and Neumark, Schweitzer and Wascher (2005)), some found more compressed wage distribution (e.g. Dickens, Machin and Manning (1999), Katz and Krueger (1992) and Lemos (2007)), whereas some found the minimum wage had a negative in‡uences on wages (e.g. Clark, Kaas and Madden (2006) and Neumark and Nizalova (2006)). Moreover, if referring to the economic theory presented in the section 2, one could assume the extraordinary minimum wage increase would have a negative impact on employment since the study will take into account Slovenian economy as a whole and not only one part of the sector as for example in Card and Krueger (1994) or Katz and Krueger (1992). Therefore, it would be hard to imagine the economy as a whole having monopsonistic characteristics. In the end, should the results of the majority research studies hold, one could expect that the extraordinary increase of the minimum wage would decrease the employment and increase the number of average hours paid and the average gross wages (Neumark and Wascher, 2007).

6.2 Method

When an exogenous event, which is usually a consequence of the government’smodi…ca- tion of a policy, changes the environment of individuals, families, …rms or cities, the ensuing conditions have characteristics of the so-called natural experiment. Such experiment has two groups; a control group that is not a¤ected by the policy change and a treatment group which is. Usually the method used with the natural experiment is the di¤erence-in-di¤erences which compares the di¤erence in results of the a¤ected group with the una¤ected one before and after the interference (Wooldridge, 2009: 451-454). When the (extraordinary) increase in min- imum wage occurs, the shock o¤ers a natural experiment for observing the consequences of the minimum wage increase (Pereira, 2003: 230). The advantages of the di¤erence-in-di¤erences method are its simplicity (Bertrand, Du‡o and Mullainathan, 2002: 2), removing the biases after the intervention which could be due to the undeviating diversities in the treatment and

26 comparison group and removing the over-time di¤erences in the treatment group that could be attributable to trends (Wooldridge, 2007: 2-4). Many researchers have evaluated the impacts of minimum wage increase with the di¤erence- in-di¤erences method. Some of these were Card and Krueger (1994), Neumark and Wascher (2000), Pereira (2003), Neumark, Schweitzer and Wascher (2005) and Abowd (2009). The three di¤erent models used in this study also studied the e¤ects of the minimum wage increase with the di¤erence-in-di¤erences method; the …rst measured the in‡uence on the average gross wages, the second on the employment and the third on the average number of hours paid. The economic theory suggests that the industry will be more a¤ected due to the minimum wage increase if the raise in the average costs is higher. Therefore, with the extraordinary minimum wage increase, the average costs would increase relatively more in those industries that have a higher share of the minimum wage recipients. Since Haltiwanger and Vodopivec (2003) concluded that …rms with less compressed wage-range have lower employment insta- bility, one could expect that the more a¤ected industries due to the extraordinary minimum wage increase would also be those that had more compressed wage-range in the period before the extraordinary increase. Given that the data of wage distribution is done only once a year, the treatment groups were chosen only by the share of the minimum wage recipients. Every model was observed via three di¤erent treatment groups. The treatment group 1 included the industries that had more than 3% of the minimum wage recipients in March 2008 (that is in the month of the extraordinary minimum wage increase), treatment group 2 presented the industries that had more than 3% of the minimum wage recipients in February 2008 and the treatment group 3 presented the industries with more than 6% of the minimum wage recipients in February 2008. Industries in the treatment group 1 were: C (Manufacturing), E (Construction), G (Hotels and restaurants) and J (Real estate, renting and business activ- ities), the treatment group 2 consisted of industries E, G, J whereas in the treatment group 3, there was the industry J. As the industry J has the highest share of minimum wage recipi- ents, one could expect the results to be the most signi…cant for the treatment group 3. The presumption is that the …nancial crisis a¤ected the treatment and comparison group in the same way.

a) Model 1: the impact of an extraordinary minimum wage increase on the average gross wages

ln awgrossit = 0 + 1after + 2treatX + 3aftertreatX + (1)

+ 4treatXt + 5compXt + other variables

The subscription i represents the industries (from A to N) and t represents the months

27 (from June 1999 until December 2009). The dependent variable lnawgrossit presents the logarithm of the average gross wage, after is the dummy variable that is 1 in the period after the extraordinary minimum wage increase (thus starting in March 2008) and therefore presents a treatment period, treatX presents three di¤erent treatment groups, where X corresponds to 1 when dealing with the treatment group 1 and 2 and 3 when dealing with treatment groups 2 and 3 correspondingly. Variable aftertreatX presents the multiplication of the dummy variable after and treatX and is so-called “average treatment e¤ect”with 3 being the coe¢ cient of interest. Variable aftertreatX measures the e¤ect of the minimum wage increase on the average gross wages. Since the graphic analyses (added to the appendix 3) showed there were di¤erent time trends between treatment and comparison groups, two di¤erent time trends to control for group speci…c time trends were included into the model. The variable treatXt is the multi- plication of the dummy variable treatX and time trend, speci…c for the equivalent treatment group, while the variable compXt presents the time trend, speci…c for the comparison group of the matching treatment group. Other factors include the logarithm of GDP per employee in di¤erent industries (lngdpempit), the logarithm of harmonized index of consumer prices

(lnhicpt) and the dummy variables for industries (da, db, dc, etc.).

b) Model 2: the impact of an extraordinary minimum wage increase on the number of employees

ln empit = 0 + 1after + 2treatX + 3aftertreatX + (2)

+ 4treatXt + 5compXt + other variables

Where the variable lnempit is the logarithm of the number of employees by industries and the other variables are the same as in the model 1.

c) Model 3: the impact of an extraordinary minimum wage increase on average number of hours paid

ln hourspaidit = 0 + 1after + 2treatX + 3aftertreatX + (3)

+ 4treatXt + 5compXt + other variables

The variable lnhourspaidit presents the logarithm of the average number of hours paid by industries. The other variables are the same as in the model 1.

28 7 Parameter estimates

7.1 Basic regressions

The results vary among the three treatment groups. Results of the models with the treatment group 3 are in accordance with the expected and more statistically signi…cant in the most important coe¢ cients, comparing to the results of the treatment groups 1 and 2. Outcomes therefore con…rm the treatment group 3 (industry J; Real estate, renting and business activities) was the most a¤ected by the extraordinary minimum wage increase which was also expected since it had the highest share of the minimum wage workers. The results of the models with the treatment groups 1 and 2 are included in the appendix 4. All regressions have heteroskedastic robust standard errors. The average treatment e¤ect (aftertreat3) is statistically signi…cant only in the model 2 and shows there was an approximate 5.4% reduction in the number of employees due to the extraordinary minimum wage increase. Model 1 and model 3 indicate that the average gross wages and the average number of hours paid increased. However, given that the results are not statistically signi…cant, one could not conclude how strongly the extraordinary increase of the minimum wage in‡uenced the average gross wages and the average number of hours paid. The treatment and comparison group had on average similar e¤ects on the dependent variable.

29 Table 5: The results of the models with the treatment group 3 Model 1 Model 2 Model 3 (y=lnawgross) (y=lnemp) (y=lnhrspaid) after 0.016 -0.019 0.007 (2.6)*** (-2.0)** (2.4)** treat3 -0.153 1.056 -0.014 (-9.2)*** (44.4)*** (-1.4) aftertreat3 0.001 -0.054 0.002 (0.1) (-2.9)*** (0.2) lngdpemp 0.078 -0.536 0.030 (4.9)*** (-19.2)*** (3.9)*** lnhicp 0.870 0.166 0.059 (18.0)*** (2.0)** (2.0)** treat3t 0.001 0.007 -0.001 (5.3)*** (21.7)*** (-4.8)*** comp3t 0.002 0.003 -0.001 (7.7)*** (11.0)*** (-6.3)*** cons 1.876 13.77 4.596 (8.2)*** (33.4)*** (33.5)*** Number of 1778 1778 1680 observations R-squared 0.9548 0.9923 0.3108

Notes: a)* statistically signi…cant at 10%, ** statistically signi…cant at 5%, *** statistically signi…cant at 1% b)t-test values are shown in parentheses c)The dependent variables are: the logarithm of the average gross wages (lnawgross; model 1), the logarithm of the number of employees (lnemp; model 2) and the logarithm of the average number of hours paid (lnhourspaid; model 3). d)The independent variables are: dummy variable for the treatment period, being 1 in the period after March 2008 (after), dummy variable for the treatment group 3 (industry J; Real estate, renting and business activities) (treat3), the average treatment e¤ect (aftertreat3), the logarithm of the GDP per employee by industries (lngdpemp), the logarithm of the harmonized index of consumer prices (lnhicp) and group speci…c time trends for the treatment and comparison groups (treat3t and comp3t correspondingly). Other regressors are dummy variables for the speci…c industry.

7.2 Sensitivity analyses

The sensitivity analyses included di¤erent dependent variables and diverse combinations of explanatory variables with the treatment group 3 (industry J; Real estate, renting and business activities). The reason for using only the treatment group 3 is that it had the most statistically signi…cant results in the basic regressions comparing to the other two treatment groups and is therefore the most representative one of the three. The analyses were composed of the basic regressions and regressions with the di¤erent treatment period that was testing for the anticipation e¤ect (the variable after3). The treat- ment period of the variable after3 starts from November 2007 since no references were found

30 in the media or in o¢ cial documents to the extraordinary minimum wage increase before that date. Yet, the regressions did not con…rm any anticipation e¤ects. Furthermore, regres- sions with shorter observation period (from January 2005 until December 2009), calculated with both treatment periods after and after3 were included as well, in order to have more time-balanced treatment and pre-treatment period. However, as the number of observations decreased considerably (on average to 840 observations), the results were unreliable for draw- ing conclusions. Moreover, a dummy variable for March 2008 was incorporated so as to control for possible di¤erent trends in this month. But since the coe¢ cients with the included dummy variable for March 2008 did not vary from the basic ones, they are not included in the results. Finally, due to the results of Haltiwanger and Vodopivec (2003), a new treatment group (treat4) was formulated which included the industry G (Hotels and restaurants) that had the most compressed wages in September 2007 (table 2). However, the results of the latter treatment group should be used with caution since the research is done only once a year every September. All regressions have heteroskedastic robust standard errors.

7.2.1 Impact of the extraordinary minimum wage increase on average gross wages

Several di¤erent dependent variables were treated when looking for the wage e¤ect. The …rst was the same as in the basic regressions (variable lnawgross; the logarithm of the average gross wages). The average treatment e¤ect changed signs depending on which independent variables one included into the regressions. Additionally, all coe¢ cients of the average treat- ment e¤ect were statistically insigni…cant, so it is not clear what the exact e¤ect of the extraordinary minimum wage increase on the average gross wages was. The treatment and comparison group on average had similar e¤ects on the dependent variable. It is sometimes argued that because of the increase in the minimum wage, the negative employment e¤ects (which will be presented later) are o¤set by increased earnings. However, similar to Meyer and Wise (1982), no statistically signi…cant increase was found in the earnings due to the ex- traordinary increase in the minimum wage. Consequent regressions with the addition of the variable after3, with the shorter observation period or with the treatment group 4 (industry G; Hotels and restaurants) were similar to the one presented in the appendix 5, table 10. The second compilation checked whether the results would be more statistically signi…- cant with the inclusion of a new dependent variable, the logarithm of the ratio between the minimum wage level and the average gross wage (variable lnmwawgross). Comparing di¤er- ent regressions, the average treatment e¤ect was on average negative but again statistically insigni…cant. As a result, one cannot conclude the e¤ect of the extraordinary minimum wage increase on the average gross wages from these regressions (appendix 5, table 11). In view of the fact that average wages had spikes every year in November and December

31 due to the 13th wage or Christmas bonuses, the moving averages were calculated in order to smooth the data. Since Slovenia had a very high in‡ation in 2007, 2008 and 2009 (above the average in the Euro area) (EUROSTAT), the nominal average wages were transformed into real average wages with the reference date being June 1999 (variable lnrealawmoving). The average treatment a¤ect was positive in all cases, but on average statistically insigni…cant (appendix 5, table 12). Since the variable lnrealawmoving is calculated as the moving average, is in real terms and gives the most consistent results, one could conclude that the extraordinary increase in the minimum wage increased the average gross wages. Though, the conclusion should not be taken as a matter of course as the majority of the results are not statistically signi…cant.

7.2.2 Impact of the extraordinary minimum wage increase on employment

When observing the in‡uence of the extraordinary minimum wage increase on the em- ployment, several dependent variables were included in the sensitivity analyses. First, when using the logarithm of the number of employees (lnemp) as the dependent variable, the av- erage treatment e¤ect was always negative and statistically signi…cant, with two exceptions; the regressions with a shorter observation period and with an inclusion of the treatment group treat4. There, the average treatment e¤ect was on average positive and statistically insigni…cant. The results of the basic regressions are shown in the appendix 5 (table 13). The treatment group on average had bigger e¤ect than comparison group in all cases but the e¤ect was not radically higher. Concerning the …rst column in the table 13, the extraordinary minimum wage increase decreased the employment on average by 5:4%. The results of the regressions with the treatment variable treat4 are presented in appendix 5, table 14. The average treatment e¤ect is only statistically signi…cant in two cases but always positive. The average treatment e¤ect in the …rst column of the table 14 indicates that the extraordinary minimum wage increase on average increased the employment by 6:0%. The treatment and comparison group on average had comparable e¤ects on the dependent variable. The second dependent variable was the logarithm of the ratio between the number of minimum wage recipients and the number of employees (lnmwcover), where the average treatment e¤ect was on average negative and statistically insigni…cant. Hence, one could not conclude what was the e¤ect of the extraordinary minimum wage increase on the share of the minimum wage recipients. However, the dependent variable lnmwcover might not be the most appropriate since one cannot distinguish which of the two variables had prevailing e¤ect on the ratio; it might be the decrease in the number of minimum wage recipients or the increase in the number of employees. Table 15 in the appendix 5 presents results of basic regressions with the treatment group 3 whereas the table 16 presents results of the basic regressions with the treatment group 4. The average treatment e¤ect was positive and statistically signi…cant

32 in the latter case. The coe¢ cient of the average treatment e¤ect in the …rst column of the table 16 indicates that the extraordinary minimum wage increase increased the share of the minimum wage recipients on average by 21:1%. The third variable was the logarithm of the whole working population (lnworkpop), which, besides employees, also includes self employed. The average treatment e¤ect is negative and statistically signi…cant through all variations, except when doing regressions with the time variable after3 and the shorter observation period and the treatment variable treat4. Table 17 in the appendix 5 summarises the results of the most basic regression. The average treatment e¤ect in the column one shows that, on average, the extraordinary increase in the minimum wage decreased the working population for 3:9%. Treatment group had a bit stronger e¤ect than comparison group. When comparing the results of dependent variables lnworkpop (the logarithm of the whole working population) and lnemp (the logarithm of the number of employees) (tables 13 and 17 in the appendix 5) one sees that the negative e¤ect of the extraordinary minimum wage increase diminishes when having a lnworkpop (the logarithm of the whole working population) as a dependent variable. The …nal dependent variable in this group was the logarithm of the ratio between the number of minimum wage recipients and working population (lnshareworkpop). The average treatment e¤ect in these regressions was on average negative and statistically signi…cant. Once more the treatment group had a bit stronger e¤ect than the comparison group. First column in the table 18 (appendix 5) presents that the ratio between minimum wage recipients and the working population on average decreased by 3.9% due to the extraordinary minimum wage increase. Nonetheless, the same concerns relate to the dependent variable lnshareworkpop as to the dependent variable lnmwcover (the ratio between the number of minimum wage recipients and the number of employees). The impact of the extraordinary minimum wage increase on the employment of the young population was also tested. However, it was later excluded from the results due to the in- compatibility of the data. Namely, the data about employment, average wages and GDP has always been obtained from SURS or AJPES, but neither of the institutions reports the data of the employment for the young population. On the other hand, EUROSTAT does, but the data is not compatible when comparing it with the data for the overall employment from the SURS. Moreover, EUROSTAT reports only quarterly data. Therefore, the data was found to be irreconcilable for achieving consistent results. Concisely, taking into account the constraints of variables lnmwcover (the ratio between the number of minimum wage recipients and the number of employees) and lnshareworkpop (the ratio between the number of minimum wage recipients and working population), one should rely more on the results of regressions with dependent variables lnemp (the logarithm of the number of employees) and lnworkpop (the logarithm of the whole working population).

33 The latter both indicate the extraordinary minimum wage increase on average had a negative impact on the employment. However, the exceptions were regressions with the treatment variable treat4.

7.2.3 Impact of the extraordinary minimum wage increase on the average num- ber of hours paid

The in‡uence of the extraordinary minimum wage increase on the logarithm of the av- erage number of hours paid (variable lnhrspaid) was, on average, positive and statistically insigni…cant across di¤erent regressions. Treatment and comparison groups on average had similar e¤ects on the dependent variable. Since all results of the average treatment e¤ect are statistically insigni…cant, it is not possible to conclude what the e¤ect of the extraordinary minimum wage increase on the average number of hours paid was. Table 19 in the appendix 5 presents the results for the dependent variable lnhrspaid for the basic regressions. In short, regressions with the dependent variable lnhrspaid indicated that the extraordi- nary minimum wage increase had, on average, a positive impact on the average number of hours paid (negative concerning shorter period and the treatment variable treat4). However, given that the results are not statistically signi…cant and inconsistent, one cannot conclude what the real e¤ect was.

7.3 Concerns regarding the di¤erence-in-di¤erences method

The drawbacks of the di¤erence-in-di¤erences method and the possible solutions for re- solving them were introduced only recently (Wooldridge, 2007). Donald and Lang (2007) and Bertrand, Du‡o and Mullainathan (2002) for example presented some methods that improve the di¤erence-in-di¤erences method. Bertrand, Du‡o and Mullainathan (2002) made a re- view of the recent papers that used di¤erence-in-di¤erences in their research and concluded the vast majority of them did not deal with the problem of serial correlation adequately. To test how serious the problem of the serial correlation is, Bertrand, Du‡o and Mullainathan (2002) created placebo laws (i.e. the imaginary laws, founded out by Bertrand, Du‡o and Mullainathan that did not exist in reality) and tested them with the di¤erence-in-di¤erences method. The method found a signi…cant e¤ect in the 45% of the placebo laws. Although the presented solutions are beyond the scope of this study, one should bear the drawbacks of the method in mind. There might also be concerns regarding high R-squared values, especially when observing the employment e¤ect, where the R-squared was often higher than 0.99. One of the …rst researchers that indicated the concerns regarding high R-squared values were Granger and Newbold (1974). Granger and Newbold (1974) also brought forward the phrase “spurious re-

34 gression”which relates to a situation where two variables are related due to their correlation with a third variable (Wooldridge, 2009: 636). This might indicate the problem of the inte- gration of order zero (I(0)) or the integration of order one (I(1)) (Wooldridge, 2009: 363-637). Since it is hard to determine whether a time series is integrated of order zero or of order one (Wooldridge, 2009: 394), it was checked whether the result changes with …rst-di¤erencing the dependent variable lnemp (the logarithm of the number of employees); one of the methods, proposed by Wooldridge (Wooldridge, 2009: 393-395). The R-squared decreased, while the other coe¢ cients did not change signi…cantly; the average treatment e¤ect still indicated the negative employment e¤ect due to the extraordinary minimum wage increase. Other methods proposed are beyond the scope of this study.

8 Discussion and recommendations

The results from the previous section indicate the extraordinary minimum wage increase from March 2008 on average had a negative impact on the employment and a positive impact on the average gross wages and the average number of hours paid. However, the conclusions are not unambiguous since the coe¢ cients of the average treatment e¤ect vary substantially when observing the employment e¤ect, whereas they are usually statistically insigni…cant when observing the e¤ect on the average gross wages and the e¤ect on the average hours paid. There are several concerns regarding the results presented in the previous section. First, the results with the treatment group 4 (variable treat4: industry G; Hotels and restaurants) seem to contradict majority of the results when observing the impacts of extraordinary mini- mum wage increase on the employment and on the average hours paid. Moreover, the average treatment e¤ect of the employment e¤ect is sometimes too high to be realistic (for example 0:211 when the dependent variable was lnmwcover). One reason for this phenomenon could be the characteristics of the industry G (Hotels and restaurants). Namely, Card and Krueger (1994) also found a positive impact on the employment because of the minimum wage increase in the fast-food restaurants. Critics claimed the positive employment e¤ect appeared also due to the characteristics of the industry which supposedly has monopsonistic qualities. However, Card and Krueger (1994) observed only one spectre of the industry while this study looked at it as a whole, so this argument might not be reliable. The other explanation could be that, contrary to the other industries, the crisis in‡uenced this industry in a positive way. The reason could be that Slovenian tourists rather spent their holidays in Slovenian tourist resorts than abroad because of the crisis. Therefore, since the industry G (Hotels and restaurants) greatly depends on the trends in tourism, higher domestic demand could have had a positive impact on this industry. When observing the statistical data, the percentage of Slovenian

35 tourists in 2008 and 2009 on average increased comparing to the same months of the previous year, while the number of foreign tourists on average decreased when comparing months in 2009 with 2008. When evaluating the overall number of tourists in Slovenia, the number increased by 3:17% (the number of domestic tourists increased by 7% while the number of foreign tourists increased by 1:14%) in 2008 comparing to 2007. Conversely, the overall num- ber of tourists decreased by 1:6% in 2009 compared to 2008 (although the increase in the number of domestic tourists was 5:92%, the decrease of 5:82% of foreign tourists prevailed). Therefore, the reason for increased employment in the treatment period in the industry G (Hotels and restaurants) could appear due to the increased number of domestic tourists and not due to the extraordinary minimum wage increase. The results thus indicate the treatment variable treat4 might not be reliable since the crisis might in‡uence the industry G (Hotels and restaurants) in a di¤erent way than the other industries. Although the results for the dependent variable average number of hours paid (lnhrspaid) are statistically insigni…cant, the outcomes are congruent with some other research studies (Neumark and Wascher, 2007: 34-36). When the extraordinary minimum wage increase had a negative in‡uence on employment, it on average had a positive in‡uence on average number of hours paid (and vice versa). This indicates that when employers had to decrease the employment due to the higher labour costs, they compensated the decrease with the increased number of hours paid. However, since these coe¢ cients among others also relate to the treatment group 4 (variable treat4: industry G; Hotels and restaurants) and some are not statistically signi…cant, the conclusions should not be taken for granted. Another remark can be made about the vast varieties among elasticities for average treat- ment e¤ect when observing the employment e¤ect. The already mentioned concerns regarding the treatment group 4 (variable treat4: industry G; Hotels and restaurants) and regressions with shorter observation period could explain the diversities in coe¢ cients for these regres- sions, while the inconsistency in coe¢ cients of basic regressions might indicate the problems with the data. Ultimately, disregarding the peculiar results of the regressions with the shorter period and the variable treat4, the results of the employment e¤ect for the basic regressions do not seem to be compatible with the results of the wage e¤ect. Speci…cally, since the average treatment e¤ect is mostly consistent in being negative and statistically signi…cant in the …rst case, one would expect it would also be statistically signi…cant in the latter. However, wage e¤ect is on average positive but always statistically insigni…cant when observing the basic regressions. One reason for this could be that the minimum wage receivers were getting a hike above in‡ation while other workers were getting a raise that did not compensate for in‡ation. Facts that are in favour of this statement are already mentioned high Slovenian in‡ation in 2007, 2008 and 2009 and the increase in the ratio between the minimum wage and the average

36 gross wage after the extraordinary minimum wage increase in March 2008. Therefore, the reason could be that the most a¤ected workers by the extraordinary minimum wage increase received a nominal wage increase, hence nobody’s salary was reduced, but the overall e¤ect on the real wages was zero since the wage of the minimum wage receivers increased in the real terms, whereas the rest of the workers faced a real wage decrease. The data con…rm the forecasts since the real average gross wages in the industry J (Real estate, renting and business activities) decreased in real terms several times after February 2008, especially when observing the moving average of the average gross wages, the data shows the increase was smaller every year and in the last months of 2009, there was a decrease in the moving average of real average gross wages in the industry J (Real estate, renting and business activities). Moreover, in favour of this explanation are also facts that the industry J (Real estate, renting and business activities) had a higher share of high wage earners than the average industry in September 2007 and in September 2008 (SURS) and that the real wages in this industry grew slower than the real minimum wages and real average gross wages in the average industry (…gure 5).

Figure 5: Comparison of the real minimum wage and moving averages of the real average gross wages and the real average gross wages in the industry J (Real estate, renting and business activities) [Source: SURS, own calculations]

The other explanation for the phenomenon that the results of the employment e¤ect for the basic regression are statistically signi…cant and negative while the results of the wage e¤ect are statistically insigni…cant could be that due to the crisis, the amount of the grey economy increased in the industry J (Real estate, renting and business activities) disproportionally more

37 than in other industries. It would be hard to prove this argument, since the data concerning the gray economy does not exist. However, several Slovenian newspapers reported increased gray economy in the last few years (for example, STA, 2007 and STA, 2010). Moreover, since the industry J (Real estate, renting and business activities) had the highest share of minimum wage recipients, economic theory suggests the employers in this industry faced the highest constraints concerning increased labour costs and therefore might o¢ cially …re some minimum wage employees but hire them illegally and pay them the wage that is lower than the minimum wage. The o¢ cial employment rate of the industry would therefore lower, while the average gross wages would stay more or less the same since employers could not a¤ord to give a signi…cant wage raise to the remaining legally employed workers as they still had to pay the illegally employed ones. Given that the extraordinary minimum wage level increase in March 2008 was not as high as the one in February 2010 and was only a one-time event whereas employers can adjust to the new minimum wage level gradually, it would be hard to infer from the March 2008 increase, what the consequences of the new law will be. However, the extraordinary minimum wage increase in March 2008 still o¤ers a proper prediction about the e¤ects of the minimum wage increase. Finally, the data in this study seems to be insu¢ cient. Maybe this is due to the change in the classi…cations in 2008, too short post-treatment observation period or the lack of the inclusion of the other relevant factors, such as human capital, for which no appropriate data was found.

9 Conclusions

The economic theory predicts that in the competitive labour market, the minimum wage, set above the equilibrium wage level, will lower the employment and increase the labour costs and equilibrium wage level. In a non-competitive labour market, the outcome is more indistinct. If the minimum wage in a monopsonistic market is set between the competitive wage level and the monopsonistic level, the employment and wages will increase, while if the minimum wage in a monopsonistic market is set higher than the wage of the competitive labour market, the employment decreases. When the minimum wage in a monopsonistic market is set to a wage level of the competitive labour market, the characteristics of a competitive labour market prevail over characteristics of a monopsony. In the …rst wave of research on the minimum wage economists debated whether the low- wage labour market should be distinguished as a competitive market (i.e. the marginalists) or a non-competitive one (i.e. the institutionalists). In the beginning of the 1990s, Card, Katz and Krueger stimulated the commencement of a still ongoing debate concerning the

38 in‡uences of the minimum wage. Some researchers found the minimum wage has a positive or insigni…cant impact on the employment while the majority of the results indicate the minimum wage has an adverse e¤ect on the employment. The wage e¤ect seems to be less controversial. Some authors have also mentioned the importance of the working hours e¤ect since, besides the employment, employers can also adjust the number of hours paid by their employees. Slovenian minimum wage law was passed on in 1995 and until 2010 there have been modest changes in the law. However, in February 2010, the new minimum wage law has been accepted and one of its main changes was a signi…cant raise in the minimum wage level. In order to lessen the shock, employers can set the new minimum wage level progressively in the following few years. The motivation was to check the possible consequences of the new minimum wage law with making inferences from the extraordinary minimum wage increase that happened in March 2008. In this study, the focus was on …nding the e¤ects that the extraordinary minimum wage increase had on wages, employment and average number of hours paid in Slovenia. Since the literature does not provide an unambiguous answer concerning the minimum wage e¤ect and the fact that there have not been many research studies done on a Slovenian case study, it was hard to predict the results. As in many of the other research studies that studied the minimum wage e¤ects, the model in this study was also based on the di¤erence-in-di¤erences method. Several treatment groups consisted of the industries that had the highest share of minimum wage recipients and should therefore be more a¤ected by the extraordinary minimum wage increase. The treatment group that generated the most consistent results comprised the industry J (Real estate, renting and business activities). The observation period was from June 1999 until December 2009. The results on average indicated the extraordinary minimum wage increase had a positive and statistically insigni…cant in‡uence on wages and average number of hours paid but a statistically signi…cant negative in‡uence on the employment. Even though there were some possible explanations provided in the discussion to the dilemmas that appeared during the research some puzzles still remain, since the results of all three models did not produce the expected sign nor were they consistent. Therefore, in order to obtain more signi…cant results, future researchers should have a longer post-treatment observation period, include other appropriate factors, for instance human capital, and take the drawbacks of the di¤erence-in-di¤erences method into consideration. As a result, it is hard to make signi…cant conclusions concerning the e¤ects of the new minimum wage law from the extraordinary minimum wage increase that took place in March 2008, because of the present voids. Nonetheless, this study sheds some light on the possible outcomes, what might be the shortcomings and giving some recommendations for future research.

39 10 References

1. Abowd, J. M. et al. (2009), Minimum Wages and Employment in France and the United States. National Bureau of Economic Research, NBER Working Paper.

2. AJPES, Agency of the Republic of Slovenia for Public Legal Records and Related Ser- vices. [Online] Available at:

http://www.ajpes.si/ [Accessed on February 15, 2010]

3. Baker, M., Benjamin, D. and Stanger, S. (1999), The Highs and Lows of the Minimum Wage E¤ect: A Time-Series Cross-Section Study of the Canadian Law. Journal of Labor Economics, 17, (2), 318-350.

4. Bazen, S. and Martin, J. P. (1991), The impact of the minimum wage on earnings and employment in France. OECD Economic Studies No. 16, 199-221.

5. Bertrand, M., Du‡o, E. and Mullainathan, S. (2002), How much should we trust di¤erences-in-di¤erences estimates? National Bureau of Economic Research, NBER Working Paper, No. 8841.

6. Boeri, T. and van Ours, J. (2008), The Economics of Imperfect Labor Markets. Princeton and Oxford: Princeton University Press.

7. Borjas, G. J. (2008), Labor economics. 4th ed. Boston: McGraw-Hill/Irwin.

8. Brezigar Masten, A. et al. (2010), Ocene posledic dviga minimalne plaµce v Sloveniji, Zbirka Delovni zvezki UMAR, 19, (3).

9. Brown, C. (1999), Minimum wages, employment, and the distribution of income, Hand- book of Labor Economics, 3, (2), 2101-2163.

10. Brown, C., Gilroy, C. and Kohen, A. (1982), The e¤ect of the minimum wage on em- ployment and unemployment: a survey, National Bureau of Economic Research, NBER Working Paper, No. 846.

11. Burkhauser, R. V., Couch, K. A. and Wittenberg, D. C. (2000), A Reassessment of the New Economics of the Minimum Wage Literature with Monthly Data from the Current Population Survey, Journal of Labor Economics, 18, (4), 653-680.

12. Burkhauser, R. V., Couch, K. A. and Wittenburg, D. C. (1996), "Who Gets What" from Minimum Wage Hikes: A Re-Estimation of Card and Krueger’sDistributional Analysis in Myth and Measurement: The New Economics of the Minimum Wage, Industrial and Labor Relations Review, 49, (3), 547-552.

40 13. Burkhauser, R. V. and Finegan, T. A. (1993), The Economics of Minimum Wage: Leg- islation Revisited, Cato Journal, 13, (1), 123-129.

14. Card, D. and Krueger, A. B. (1994), Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania, The American. Economic Review, 84, (4), 772-793.

15. Card, D. (1992), Do Minimum Wages Reduce Employment? A Case Study of California, 1987-89, Industrial and Labor Relations Review, 46, (1), 38-54.

16. CIA (The World Factbook –Slovenia). [Online] Available at:

https://www.cia.gov/library/publications/the-world-factbook/geos/si.html [Accessed on June 22, 2010]

17. Clark, K., Kaas, L. and Madden, P. (2006), Minimum wage increases can lead to wage reductions by imperfectly competitive …rms, Economics Letters, 91, (2), 287-292.

18. Deere, D., Murphy, K. M. and Welch, F. (1995), Reexamining Methods of Estimat- ing Minimum-wage E¤ects, Employment and the 1990-1991 Minimum-Wage Hike, The American Economic Review, 85, (2), 232-237.

19. Dickens, R., Machin, S. and Manning, A. (1999), The E¤ects of Minimum Wages on Employment: Theory and Evidence from Britain, Journal of Labor Economics, 17, (1).

20. Donald, S. G. and Lang, K. (2007), Inference with Di¤erence-in-Di¤erences and Other Panel Data, The Review of Economics and Statistics, 89, (2), 221-233.

21. DURS, Tax Administration of the Republic of Slovenia, Ministry of Finance. [Online] Available at:

http://www.durs.gov.si/si/aktualno/minimalna_in_zajamcena_placa/minimalna_

placa_po_mesecih/ [Accessed on May 2, 2010]

22. Ehrenberg, R. G. and Smith, R. S. (2003), Modern labor economics: theory and public policy. 8th ed. Boston [etc.]: Addison-Wesley.

23. Filer, R. K., Hamermesh, D. S. and Rees, A. E. (1996), The economics of work and pay. 6th ed. New York: HarperCollins College Publishers.

24. Funk, L. and Lesch, H. (2006), Minimum Wage Regulations in Selected European Coun- tries, Intereconomics, 41, (2), 78-92.

41 25. Funk, L. and Lesch, H. (2005), Minimum wages in Europe. [Online] Available at: http://www.eurofound.europa.eu/eiro/2005/07/study/tn0507101s.htm [Accessed on April 15, 2010]

26. Granger, C. W. J. and Newbold, P. (1974), Spurious regressions in econometrics, Journal of Econometrics, 2, (2), 111-120.

27. Haltiwanger, J. and Vodopivec, M. (2003), Worker ‡ows, job ‡ows and …rm wage poli- cies, An analysis of Slovenia, Economics of Transition, 11, (2), 253-290.

28. Katz, L. F. and Krueger, A. B. (1992), The e¤ect of the minimum wage on the fast food industry, National Bureau of Economic Research, NBER Working Paper, No. 3997.

29. Kim, T. and Taylor, L. J. (1995), The Employment E¤ect in Retail Trade of California’s 1988 Minimum Wage Increase, Journal of Business & Economic Statistics, 13, (2), 175- 182.

30. Lemos, S. (2007), Minimum wage e¤ects across the private and public sectors in Brazil, Journal of Development Studies, 43, (4), 700-720.

31. Lester, R. A. (1947), Marginalism, Minimum Wages, and Labor Markets, The American Economic Review, 37, (1), 135-148.

32. Levin-Waldman, O. M. (2000), The E¤ects of the Minimum Wage: A Business Response, Journal of Economic Issues, 34, (3), 723-730.

33. Manning, A. (2003), Monopsony in Motion: Imperfect competition in labor markets. Princeton and Oxford: Princeton University Press.

34. Metcalf, D. (2004), The impact of the national minimum wage on the pay distribution, employment and training, The Economic Journal, 114, 84-86.

35. Meyer, R. H. and Wise, D. A. (1982), The e¤ects of the minimum wage on the employ- ment and earnings of youth, National Bureau of Economic Research, NBER Working Paper, No. 849.

36. Mincer, J. (1976), Unemployment E¤ects of Minimum Wages, Journal of Political Econ- omy, 84, (4), 87-104.

37. Neumark, D. and Wascher, W. (2007), Minimum Wages and Employment, IZA Discus- sion Paper Series, IZA DP No. 2570.

38. Neumark, D. and Nizalova, O. (2006), Minimum Wage E¤ects in the Longer Run, The Journal of Human Resources, 42, (2), 435-452.

42 39. Neumark, D., Cunningham, W. and Siga, L. (2006), The e¤ects of the minimum wage in Brazil on the distribution of family incomes: 1996–2001, Journal of Development Economics, 80, 136-159.

40. Neumark, D., Schweitzer, M. and Wascher, W. (2005), The E¤ects of Minimum Wages on the Distribution of Family Incomes, A Nonparametric Analysis, The Journal of Human Resources, 40, (4), 867-894.

41. Neumark, D., Schweitzer, M. and Wascher, W. (2004), Minimum Wage E¤ects through- out the Wage Distribution, The Journal of Human Resources, 39, (2), 425-450.

42. Neumark, D. and Wascher, W. (2000), Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Comment, American Economic Review, 90, (5), 1362-1396.

43. Neumark, D. (1993), Employment E¤ects of Minimum and Subminimum Wages. Recent Evidence. Washington, D.C.: Employment Policies Institute.

44. OECD (2009), Economic Survey of Slovenia, 2009. [Online] Available at:

http://www.oecd.org/dataoecd/2/17/43039695.pdf [Accessed on April 15, 2010]

45. Pereira, S. C. (2003), The impact of minimum wages on youth employment in Portugal, European Economic Review, 47, 229-244.

46. Singell, L. D. and Terborg, J. R. (2007), Employment e¤ects of two northwest minimum wage initiatives, Economic Inquiry, 45, (1), 40-55.

47. Stewart, M. B. (2004), The impact of the introduction of the U.K. minimum wage on the employment probabilities of low-wage workers, Journal of the European Economic Association, 2, (1), 67-97.

48. Stigler, G. J. (1946), The economics of minimum wage legislation, The American Eco- nomic Review, 36, 358-365.

49. SKD [Anon.] (2008), Standardna klasi…kacija dejavnosti –SKD, Razlike med SKD 2008 in SKD 2002. [Online] Available at:

http://www.stat.si/doc/klasif/revizija/Razlike_SKD2008_SKD2002.pdf [Accessed on April 1, 2010]

50. STA (2010), Obrtniki: Plaµcilna nedisciplina in siva ekonomija neobvladljivi, Delo.si. [Online] Available at:

http://www.delo.si/clanek/105386 [Accessed on June 7, 2010]

43 51. STA (2007), Siva ekonomija predstavlja med 20 in 24 odstotkov BDP, Finance.si. [On- line] Available at:

http://www.…nance.si/189731/Siva-ekonomija-predstavlja-med-20-in-24-odstotkov-BDP [Accessed on June 7, 2010]

52. SURS, Statistical O¢ ce of the Republic of Slovenia. [Online] Available at:

http://www.stat.si/eng/index.asp [Accessed on February 15, 2010]

53. UMAR, Institute of Macroeconomic Analysis and Development of the Republic of Slove- nia. [Online] Available at:

http://www.umar.gov.si/informacije_za_javnost/obvestila_in_sporocila_za_javnost/

obvestilo/zapisi/ekonomsko_ogledalo_okrevanje_domacega_gospodarstva_

zaostaja_za_evropskim/31/?tx_ttnews%5Bsyear%5D=2010&tx_ttnews%5

Bsmonth%5D=6&cHash=0638ed5a51 [Accessed on June 22, 2010]

54. Uradni list RS [Anon.] (2010), Zakon o minimalni plaµci, Uradni list RS, 13 (1457). [Online] Available at:

http://www.uradni-list.si/1/objava.jsp?urlid=201013&stevilka=519 [Accessed on May 21, 2010]

55. Welch, F. and Cunningham, J. (1978), E¤ects of Minimum Wages on the Level and Age Composition of Youth, The Review of Economics and Statistics, 60, (1), 140-145.

56. Wooldridge, J. M. (2009), Introductory Econometrics: A Modern Approach. 4th ed. Mason, OH [etc.]: South-Western/Cengage Learning.

57. Wooldrirge, J. M. (2007), What’s New in Econometrics? (Lecture 10: Di¤erence-in- Di¤erences Estimation), NBER Summer Institute, 2007. [Online] Available at:

http://www.nber.org/WNE/Slides7-31-07/slides_10_di¢ ndi¤s.pdf [Accessed on June 23, 2010]

44 A Appendix

A.1 Descriptive statistics

Table 6: Descriptive statistics by industries [Source: SURS, EURPSTAT, AJPES, own cal- culations]

Variable Obs Mean Std. Dev. Min Max mwcoveragea 113 0.023 0.01 0.005 0.054 mwcoverageb 112 0.001 0.001 0 0.004 mwcoveragec 113 0.034 0.008 0.016 0.06 mwcoveraged 113 0.003 0.001 0 0.007 mwcoveragee 113 0.032 0.008 0.016 0.052 mwcoveragef 112 0.02 0.004 0 0.028 mwcoverageg 113 0.026 0.005 0.016 0.041 mwcoverageh 113 0.011 0.003 0.005 0.021 mwcoveragei 113 0.005 0.002 0.002 0.009 mwcoveragej 112 0.069 0.01 0 0.086 mwcoveragek 113 0.004 0.003 0.001 0.012 mwcoveragel 113 0.003 0.002 0.002 0.015 mwcoveragem 113 0.01 0.005 0.001 0.026 mwcoveragen 113 0.021 0.005 0.012 0.034 mwcoveragesum 113 0.025 0.004 0.013 0.036 gdpa 127 154 25 109 212 gdpb 127 29 4 21 37 gdpc 127 1396 280 860 1936 gdpd 127 165 49 76 243 gdpe 127 401 140 243 693 gdpf 127 675 190 360 1003 gdpg 127 128 37 72 226 gdph 127 428 128 244 652 gdpi 127 273 79 148 394 gdpj 127 962 292 496 1468 gdpk 127 336 87 189 482 gdpl 127 309 80 173 434 gdpm 127 284 69 155 411 gdpn 127 197 45 133 275

45 Table 6: Descriptive statistics by industries [Source: SURS, EURPSTAT, AJPES, own cal- culations]

Variable Obs Mean Std. Dev. Min Max gdpsum 121 5641 1425 3405 8083 awgrossa 127 943 180 660 1383 awgrossb 127 1361 334 794 2354 awgrossc 127 933 197 592 1363 awgrossd 127 1282 316 813 2261 awgrosse 127 911 176 604 1226 awgrossf 127 994 192 674 1389 awgrossg 127 835 134 573 1093 awgrossh 127 1255 252 757 1833 awgrossi 127 1654 408 1020 3022 awgrossj 127 1140 202 769 1572 awgrossk 127 1368 239 921 1818 awgrossl 127 1338 271 783 1739 awgrossm 127 1313 225 828 1806 awgrossn 127 1284 188 894 1636 awgrosssum 121 1081 203 708 1521 hrspaida 120 170 5 157 180 hrspaidb 120 170 4 162 183 hrspaidc 120 172 6 156 182 hrspaidd 120 353 14 320 376 hrspaide 120 175 6 160 188 hrspaidf 120 170 5 158 179 hrspaidg 120 168 5 155 178 hrspaidh 120 343 11 317 360 hrspaidi 120 169 7 154 184 hrspaidj 120 507 16 471 537 hrspaidk 120 170 6 154 179 hrspaidl 120 167 3 154 174 hrspaidm 120 176 4 166 183 hrspaidn 120 337 10 311 355

Notes : a)mwcoverageX presents the ratio between the number of minimum wage recipients and the number

46 of employees in corresponding industry b)gdpX presents the GDP in millions in the corresponding industry c)awgrossX presents the average gross wages in the corresponding industry d)hrspaidX presents the average number of hours paid in the corresponding industry e)sum stands for the total sum of all industries

47 A.2 Description of “SKD united”

Table 7: Description of "SKD united" [Source: SKD] SKD "united" SKD 2002 SKD 2008 Description A A+B A AGRICULTURE, FORESTRY AND FISHING B C B MINING AND QUARRYING C D C MANUFACTURING D E D+E ELECTRICITY, GAS AND WATER SUPPLY E F F CONSTRUCTION F G G WHOLESALE AND RETAIL, REPAIR MOTOR VEHICLES, PERSONAL AND HOUSEHOLD GOODS G H I HOTELS AND RESTAURANTS H I H+J TRANSPORT, STORAGE AND COMMUNICATION I J K FINANCIAL INTERMEDIATION J K L+M+N REAL ESTATE, RENTING AND BUSINESS ACTIVITIES K L O PUBLIC ADMINISTRATION AND DEFENCE, COMPULSORY SOCIAL SECURITY L M P EDUCATION M N Q HEALTH AND SOCIAL WORK N O R+S OTHER COMMUNITY, SOCIAL AND PERSONAL SERVICE ACTIVITIES

48 A.3 Comparison of time trends in treatment and comparison groups

For clari…cation, average wages and average number of hours paid are calculated as moving averages. Namely, average wages have every-year peaks in November and December due to the 13th yearly wage or Christmas bonuses that are paid in these months; whereas the number of average hours paid depends on the number of working days in a particular month and therefore has nadirs every February.

Figure 6: Average gross wages in treatment group 1 and comparison group 1 [Source: SURS, own calculations]

49 Figure 7: Average gross wages in treatment group 2 and comparison group 2 [Source: SURS, own calculations]

Figure 8: Average gross wages in treatment group 3 and comparison group 3 [Source: SURS, own calculations]

50 Figure 9: Employment in treatment group 1 and comparison group 1 [Source: SURS, own calculations]

Figure 10: Employment in treatment group 2 and comparison group 2 [Source: SURS, own calculations]

51 Figure 11: Employment in treatment group 3 and comparison group 3 [Source: SURS, own calculations]

Figure 12: Average number of hours paid in treatment group 1 and comparison group 1 [Source: SURS, own calculations]

52 Figure 13: Average number of hours paid in treatment group 2 and comparison group 2 [Source: SURS, own calculations]

Figure 14: Average number of hours paid in treatment group 3 and comparison group 3 [Source: SURS, own calculations]

53 A.4 The results of treatment groups 1 and 2

Table 8: The results of the treatment group 1 Model 1 Model 2 Model 3 (y = lnawgross) (y = lnemp) (y = lnhrspaid) after 0.020 -0.023 0.009 (2.6)*** (-2.3)** (2.7)*** treat1 -0.170 1.191 -0.009 (-11.1)*** (49.9)*** (-1.2) aftertreat1 -0.010 -0.004 -0.008 (-1.1) (-0.2) (-1.4) lngdpemp 0.082 -0.552 0.027 (5.0)*** (-19.8)*** (3.5)*** lnhicp 0.867 0.181 0.062 (17.9)*** (2.1)** (2.2)** treat1t 0.001 0.005 -0.001 (7.4)*** (13.4)*** (-6.2)*** comp1t 0.001 0.003 -0.001 (7.4)*** (10.7)*** (-6.0)*** cons 1.862 13.851 4.606 (8.1)*** (33.2)*** (33.6)*** Number of 1778 1778 1680 observations R-squared 0.9547 0.992 0.3134

Notes: a)* statistically signi…cant at 10%, ** statistically signi…cant at 5%, *** statistically signi…cant at 1% b)t-test values are shown in parentheses c)The independent variables are: dummy variable for the treatment group 1 (C (Manufacturing), E (Con- struction), G (Hotels and restaurants) and J (Real estate, renting and business activities)) (treat1), the average treatment e¤ect (aftertreat1) and the group speci…c time trends for the treatment and comparison groups correspondingly (treat1t and comp1t). The description of the other variables is under the table 5.

54 Table 9: The results of the treatment group 2 Model 1 Model 2 Model 3 (y = lnawgross) (y = lnemp) (y = lnhrspaid) after 0.016 -0.029 0.007 (2.2)** (-3.0)*** (2.2)** treat2 -0.140 1.094 -0.006 (-8.8)*** (45.5)*** (-0.7) aftertreat2 0.000 0.039 -0.002 (0.0) (2.0)** (-0.3) lngdpemp 0.067 -0.497 0.0260 (4.1)*** (-17.2)*** (3.4)*** lnhicp 0.880 0.132 0.063 (18.2)*** (1.6) (2.2)** treat2t 0.001 0.005 -0.001 (6.2)*** (15.5)*** (-6.4)*** comp2t 0.002 0.003 -0.001 (8.0)*** (10.1)*** (-6.0)*** cons 1.922 13.618 4.608 (8.4)*** (33.2)*** (33.6)*** Number of 1778 1778 1680 observations R-squared 0.9551 0.9928 0.3124

Notes: a)* statistically signi…cant at 10%, ** statistically signi…cant at 5%, *** statistically signi…cant at 1% b)t-test values are shown in parentheses c)The independent variables are: dummy variable for the treatment group 2 (E (Construction), G (Hotels and restaurants) and J (Real estate, renting and business activities)) (treat2), the average treatment e¤ect (aftertreat2) and the group speci…c time trends for the treatment and comparison groups correspondingly (treat2t and comp2t). The description of the other variables is under the table 5.

55 A.5 The results of sensitivity analyses

Table 10: Basic regressions with heteroskedastic robust standard errors (dependent variable: lnawgross) y=lnawgross after 0.016 0.011 -0.014 -0.03 (2.6)*** (1.7)* (-2.3)** (-4.9)*** treat3 -0.153 -0.205 (-9.2)*** (-11.2)*** aftertreat3 0.001 0.009 -0.006 0.009 (0.1) (0.6) (-0.4) (0.6) lngdpemp 0.078 0.141 (4.9)*** (8.7)*** lnhicp 0.87 0.939 (18.0)*** (20.2)*** treat3t 0.001 0.001 0.004 0.005 (5.3)*** (5.1)*** (25.8)*** (27.1)*** comp3t 0.001 0.002 0.005 0.005 (7.7)*** (9.0)*** (38.1)*** (80.6)*** cons 1.876 2.218 5.572 6.801 (8.2)*** (9.7)*** (39.2)*** (874.2)*** Number of 1778 1778 1778 1778 observations R-squared 0.9548 0.954 0.9471 0.9444

Notes: a)* statistically signi…cant at 10%, ** statistically signi…cant at 5%, *** statistically signi…cant at 1% b)t-test values are shown in parentheses c)The dependent variable is the logarithm of the average gross wages (lnawgross). The description of the independent variables is under the table 5.

56 Table 11: Basic regressions with heteroskedastic robust standard errors (dependent variable: lnmwawgross) y=lnmwawgross after -0.031 -0.026 -0.048 -0.043 (-4.9)*** (-4.0)*** (-7.8)*** (-7.1)*** treat3 0.149 0.120 (8.7)*** (7.1)*** aftertreat3 -0.001 -0.009 -0.005 -0.009 (-0.1) (-0.6) (-0.3) (-0.6) lngdpemp -0.074 -0.04 (-4.6)*** (-2.5)*** lnhicp 0.476 0.410 (9.6)*** (8.6)*** treat3t -0.000 -0.000 0.001 0.001 (-2.0)** (-1.9)* (7.4)*** (7.0)*** comp3t -0.001 -0.001 0.001 0.001 (-4.1)*** (-5.3)*** (7.1)*** (8.6)*** cons -2.737 -3.061 -0.713 -1.059 (-11.7)*** (-13.1)*** (-5.3)*** (-144.6)*** Number of 1778 1778 1778 1778 observations R-squared 0.9092 0.9079 0.9048 0.9044

Notes: a)* statistically signi…cant at 10%, ** statistically signi…cant at 5%, *** statistically signi…cant at 1% b)t-test values are shown in parentheses c)The dependent variable is the logarithm of the ratio between the height of the minimum wage and the average gross wage (lnmwawgross). The description of the independent variables is under the table 5.

57 Table 12: Basic regressions with heteroskedastic robust standard errors (dependent variable: lnrealawmoving) y=lnrealawmoving after 0.020 0.012 (5.8)*** (3.5)*** treat3 -0.086 (-14.6) aftertreat3 0.001 0.009 (0.2) (1.6) lngdpemp 0.072 (7.5)*** treat3t 0.001 0.001 (13.5)*** (20.2)*** comp3t 0.001 0.001 (14.2)*** (38.4)*** cons 6.095 6.725 (72.9)*** (1222.4)*** Number of 1694 1694 observations R-squared 0.9734 0.9721

Notes: a)* statistically signi…cant at 10%, ** statistically signi…cant at 5%, *** statistically signi…cant at 1% b)t-test values are shown in parentheses c)The dependent variable is the logarithm of the moving average of the real average gross wages (lnrealawmoving). The description of the independent variables is under the table 5.

58 Table 13: Basic regressions with heteroskedastic robust standard errors (dependent variable: lnemp) y=lnemp after -0.019 0.021 -0.025 0.034 -0.018 0.027 (-2.0)** (1.9)* (-2.5)*** (3.0)*** (-1.8)* (2.4)** treat3 1.056 1.046 1.042 (44.4)*** (45.8)*** (42.3)*** aftertreat3 -0.054 -0.111 -0.055 -0.111 -0.054 -0.109 (-2.9)*** (-4.6)*** (-2.9)*** (-4.8)*** (-2.9)*** (-4.6)*** lngdpemp -0.536 -0.524 -0.529 (-19.2)*** (-19.3)*** (-18.9)*** lnhicp 0.166 -0.31 0.245 (2.0)** (-3.4)*** (2.9)*** lnawgross -0.091 -0.226 (-2.4)** (-5.1)*** treat3t 0.007 0.007 0.008 0.006 0.007 0.007 (21.7)*** (19.7)*** (70.7)*** (76.7)*** (21.6)*** (31.6)*** comp3t 0.003 0.002 0.004 0.000 0.004 0.002 (11.0)*** (5.0)*** (21.6)*** (4.8)*** (11.2)*** (7.5)*** cons 13.77 11.433 14.477 9.919 13.94 11.454 (33.4)*** (25.8)*** (61.0)*** (1161.1)*** (32.8)*** (38.0)*** Number of 1778 1778 1778 1778 1778 1778 observations R-squared 0.9923 0.9896 0.9923 0.9895 0.9923 0.9897

Notes: a)* statistically signi…cant at 10%, ** statistically signi…cant at 5%, *** statistically signi…cant at 1% b)t-test values are shown in parentheses c)The dependent variable is the logarithm of the number of employees (lnemp). The dependent variable lnawgross is the logarithm of the average gross wages. The description of other independent variables is under the table 5.

59 Table 14: Basic regressions with heteroskedastic robust standard errors and treatment vari- able treat4 (dependent variable: lnemp) y=lnemp after -0.033 0.012 -0.041 0.025 (-3.3)*** (1.0) (-4.0)*** (2.0)** treat4 -0.190 0.157 -0.181 0.157 (-7.5)*** (16.9)*** (-7.3)*** (16.2)*** aftertreat4 0.060 0.012 0.059 0.012 (2.0)** (0.9) (2.0)** (0.8) lngdpemp -0.618 -0.601 (-22.1)*** (-22.3)*** lnhicp 0.239 -0.310 (2.7)*** (-3.1)*** treat4t 0.004 0.003 0.005 0.002 (8.4)*** (7.6)*** (14.8)*** (24.5)*** comp4t 0.004 0.002 0.005 0.001 (12.1)*** (5.4)*** (25.0)*** (7.1)*** cons 14.113 11.414 15.139 9.9 (32.8)*** (23.4)*** (64.2)*** (1126.8)*** Number of 1778 1778 1778 1778 observations R-squared 0.9915 0.9876 0.9915 0.9875

Notes: a)* statistically signi…cant at 10%, ** statistically signi…cant at 5%, *** statistically signi…cant at 1% b)t-test values are shown in parentheses c)The dependent variable is the logarithm of the number of employees (lnemp). d)The independent variables: the dummy variable for the treatment group 4 (industry G, Hotels and restaurants) (treat4), the average treatment e¤ect (aftertreat4) and group speci…c time trends for the treatment and comparison groups correspondingly (treat4t and comp4t). The description of the other independent variables is under the table 5.

60 Table 15: Basic regressions with heteroskedastic robust standard errors (dependent variable: lnmwcover) y=lnmwcover after 0.029 0.038 0.000 0.005 0.032 0.004 (0.7) (0.9) (0.0) (0.1) (0.8) (0.1) treat3 1.352 1.234 1.279 1.234 1.308 1.229 (13.1)*** (34.7)*** (12.1)*** (34.4)*** (12.2)*** (29.8)*** aftertreat3 -0.03 -0.04 -0.037 -0.041 -0.03 -0.041 (-0.6) (-0.9) (-0.8) (-0.9) (-0.6) (-0.9) lngdpemp -0.139 -0.054 -0.113 (-1.1) (-0.4) (-0.9) lnhicp 1.121 0.996 1.338 (2.7)*** (2.4)** (3.3)*** lnawgross -0.245 -0.046 (-1.2) (-0.2) treat3t -0.002 -0.002 0.002 0.002 -0.002 0.002 (-1.3) (-1.2)