Is there publication selection bias in minimum wage research during the five-year period from 2010 to 2014?

Giotis, Georgios; Chletsos, Michael Working Paper Is there publication selection bias in minimum wage research during the five-year period from 2010 to 2014? Economics Discussion Papers, No. 2015-58 Provided in Cooperation with: Kiel Institute for the World Economy – Leibniz Center for Research on Global Economic Challenges Suggested Citation: Giotis, Georgios; Chletsos, Michael (2015) : Is there publication selection bias in minimum wage research during the five-year period from 2010 to 2014?, Economics Discussion Papers, No. 2015-58, Kiel Institute for the World Economy (IfW), Kiel This Version is available at: https://hdl.handle.net/10419/115362 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/3.0/ Received May 31, 2015 Accepted as Economics Discussion Paper August 27, 2015 Published August 31, 2015 © Author(s) 2015. Licensed under the Creative Commons License - Attribution 3.0 Discussion Paper No. 2015-58 | August 31, 2015 | http://www.economics-ejournal.org/economics/discussionpapers/2015-58 Is There Publication Selection Bias in Minimum Wage Research during the Five-year Period from 2010 to 2014? Georgios Giotis and Michael Chletsos Abstract The impact of minimum wages on employment has always been a field of conflicts among economists and this divergence of views has usually taken the form of competing studies. Doucouliagos and Stanley (Publication selection bias in minimum-wage research? A metaregression analysis, 2009) conducted a meta-analysis of 64 US studies which showed that literature is contaminated by publication selection bias, and once it is corrected, little or no evidence of a negative association between minimum wages and employment remains. This result contradicts the neoclassical theory and gives a Keynesian perspective which suggests that changes in minimum wages are not related with positive or negative employment effects. In their analysis, the authors use a meta-sample of 45 empirical studies published in academic journals in the 2010–2014 fiveyear period, to investigate whether minimum wage research has been affected by Doucouliagos and Stanley’s study. Their results indicate that there is evidence of publication selection in the elasticities’ meta-sample, but once it is corrected only a small negative effect remains and, in the coefficients’ meta-sample, publication selection bias is not found and the genuine effect is again negative but small. In addition, the authors find that study characteristics related to the data, the model specifications, the minimum wage and employment measure used, and the industry concerned, diversify the sign of the minimum wage effect. (Published in Special Issue Meta-Analysis in Theory and Practice) JEL J38 J21 C12 Keywords Minimum wage; employment; meta-analysis Authors Georgios Giotis, University of Ioannina, Department of Economics, P.O. Box 1186, 45110 Ioannina, Greece, [email protected] Michael Chletsos, University of Ioannina, Department of Economics and Laboratory of Applied Economics and Social Policy (LAESP), P.O. Box 1186, 45110 Ioannina, Greece Citation Georgios Giotis and Michael Chletsos (2015). Is There Publication Selection Bias in Minimum Wage Research during the Five-year Period from 2010 to 2014? Economics Discussion Papers, No 2015-58, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2015-58 2 1. Introduction The impact of minimum wages on employment has been a field of conflicts within the economic society and especially in labor economics. One side supports that minimum wages have a negative effect on employment, another smaller side argues that there can be a positive impact, while there is also a side which argues that minimum wages do not affect (or only slightly affect) employment. More specifically, until the early 90’s the neoclassical approach was the prevailing theory and strong consensus existed among economists that an increase in the minimum wage would cause a decrease in employment. However, the studies by Card (1992) and Katz and Krueger (1992) came to create a schism as they did not find evidence of adverse employment effects of minimum wages. Since then, a divergence of views exists in the literature, which is expressed by opposing views. In this frame of competing theoretical and empirical studies, Doucouliagos and Stanley (2009) conducted a meta-analysis of 64 US studies which provided 1,474 estimates of the employment elasticity and concluded that literature is contaminated by publication selection bias, and once this publication selection is corrected, little or no evidence of a negative association between minimum wages and employment remains. This result contradicts the neoclassical theory and gave an opening to oligopolistic or monopsonistic theories, or efficiency wages and job search models, which opposed to the dominant neoclassical theory. In addition, we could say that their study gave a Keynesian perspective in minimum wage research, in the view that changes in minimum wages are not related with positive or negative employment effects at all. Undoubtedly, Doucouliagos and Stanley meta-analysis in 2009 had an important impact on the economic research with the use of meta-analysis techniques, which are very useful statistical tools for reviewing empirical results, and boosted the meta-analysis studies in economics. The purpose of our paper is to see whether minimum wage research has been affected by their study and investigate the presence of publication selection bias in minimum wage literature since then. In addition, our objective is to find the genuine effect of minimum wage on employment, if any exists. In our analysis, we use a meta-sample of 45 empirical studies published in academic journals in the 2010-2014 five-year period. When 1,068 elasticities and 484 coefficients are combined to constitute the meta-sample from these studies, evidence 3 of publication selection bias is found in the elasticities’ meta-sample, but when it is corrected only a slight employment effect remains. On the other hand, no evidence of publication bias is found in the smaller coefficients’ meta-sample and the true effect of minimum wage on employment is negative but small. Furthermore, study characteristics related to the data, the model specifications, the minimum wage and employment measure used, and the industry concerned, diversify the sign of the impact. In what follows, we present the previous literature on this issue using metaanalysis methods, the coding and indexing procedure, and some characteristics of our meta-sample and then we enter into the FAT-PET tests and the multiple metaregression analysis. Finally, we present the concluding remarks of our analysis. 2. Review of meta-analysis literature on the employment effect of minimum wage Until March 31, 2015, in our research, we have found six studies that use meta-analysis methods to investigate the employment effect of minimum wages. First, Card and Krueger (1995) analyzed 15 earlier US time-series studies on minimum wages and found publication bias in favor of studies that provided a statistically significant negative employment effect. The authors suggested that the most recent studies, which had more data and lower standard errors, did not show the expected increase in t-statistic and almost all the studies had a t-statistic of about two, just above the level of statistical significance at 5%. That study created a schism among economists and created the base for the New Minimum Wage Theory. The second meta-analysis was conducted by Doucouliagos and Stanley (2009) using 64 US studies which offered 1,474 estimates of the employment elasticity and concluded that Card and Krueger’s initial claim of publication bias was right and once this publication selection was corrected, an adverse employment effect was not supported by this large and rich research record on the employment effects of minimum-wage regulation. One year later, Boockmann (2010) conducted a metaanalysis of 55 empirical studies estimating the employment effects of minimum wages in 15 industrial countries since 1995. Almost 2/3 of the estimations of the meta-sample provided negative sign, implying that studies were still affected by the traditional neoclassical theory. 4 Nataraj et al. (2014) dealt with the employment effect of minimum wage in low-income countries with meta-analysis methods. Their meta-sample utilized fifteen studies from individual countries and two cross-country studies, and the metaregression analysis showed an ambiguous effect of minimum wages on total employment, specifically a positive impact on informal employment and negative on formal employment. Another study published at the same year was conducted by Leonard et al. (2014). The authors used meta-analysis techniques to investigate the effect of increases in the UK minimum wage on employment. The meta-sample consisted of 16 studies which provided 710 partial correlations and 236 elasticities and according to the results no adverse effect of minimum wage could be found by the increases of the UK minimum wages apart from the residential home-care sector. In comparison to Doucouliagos and Stanley (2009), this study did not find evidence or publication bias as the larger US study did. Closing the review of previous meta-analysis on the employment impact of minimum wages, we should mention Belman and Wolfson (2014) who use data from 23 international studies since 2000. Their meta-sample generated 439 estimates and the majority of the studies concerned the USA. Generally, we could say that the authors found negative and statistically significant effects of minimum wage which were very small, though. Table 1, reports the basic characteristics of the previous meta-analysis studies discussed above. It seems quite interesting that in 2014 alone, three such studies have been published and, generally, we can say that the literature on the effect of minimum wages on employment is not only large but it is also growing. Within the framework of the large minimum wage research, we tried to investigate this issue using the highest possible number of related empirical studies published in academic journals during the last five years, in order to attempt to provide a general and objective picture of the literature after searching the academic journals thoroughly. Table 1. Previous meta-analysis on the employment effect of minimum wages. Author(s) Year Country(ies) examined Studies in the meta-sample Studies’ time-period 1 Card and Krueger 1995 USA 15 1970-1992 2 Doucouliagos and Stanley 2009 USA 64 1972-2007 3 Boockman 2010 Industrial countries 55 1995-2009 4 Nataraj et al. 2014 Low-income countries 17 1991-2011 5 Leonard et al. 2014 UK 16 1994-2012 6 Belman and Wolfson 2014 Different countries 23 2000-2013 5 3. The meta-sample and the indexing and coding of the data In general, meta-analysis is a very useful tool to examine all the available research to present an objective picture of the literature. It is more than a review as it employs statistical techniques to summarize the empirical evidence and explore the sources of heterogeneity among studies. Moreover it can be used to provide the genuine effect when the publication bias is corrected. The initial but very important steps are the process of the identification of the studies and the coding of the observations which will constitute the meta-sample. Stanley et al. (2013) present the guidelines that all meta-analyses in economics should follow. In our analysis we try to fully comply with these protocols expressed by the meta-analysis of economics research-network (MAER-Net). First of all, we began our research using the Google Scholar search machine, and afterwards the economic databases Econlit, Sciencedirect, RePEc and JSTOR. Our objective was to find those empirical studies published in academic journals which reported at least one estimate on the employment effect of minimum wage. The main the keywords used in the search were “minimum wage” and “employment” and we also used several other flections as a keyword. Before entering into the details of the identification, it has to be pointed out that we restricted the research only to those studies published during the five year period from 2010 to 2014 to investigate whether the study by Doucouliagos and Stanley in 2009 has affected the minimum wage literature on the employment impact onwards. Our search for studies was terminated January 31, 2015. We chose to restrict our meta-sample only to studies published in an academic journal and not to broaden our search to unpublished papers or to other publication outlets without refereeing process. Moreover, apart from those papers not published in an academic journal, we filtered out the studies where at least one of the following characteristics was present: Studies using unemployment rate or labor force participation rate or selfemployment as an employment measure. Studies which focus on the minimum wage effect on average wage, income inequality, income distribution, reservation wage, poverty, welfare, prices, profits, firm performance, job training or economic development. Theoretical studies which do not report regression estimates. 6 Studies which use statistical and mathematical models to examine the minimum wage effect, mostly with the use of correlations, descriptive statistics and diagrams. Studies written in any other language than English. Studies which do not mention a direct minimum wage effect but focus, in general, on the effect of distribution of wages on employment measures. After extensive reading and filtering, 58 empirical studies published in an academic journal remained. We then had to exclude 13 of them due to reasons described in detail in table A.2 of the appendix. The remaining 45 provided 1,068 elasticities and 484 coefficients which constitute our meta-sample. Most of the 13 excluded studies use a binary dependent variable, reporting employment probabilities (employing mostly probit and logit models). In our analysis we follow Doucouliagos and Stanley (2009) and we focus on employment elasticities (and coefficients) drawn from studies using a continuous measure of employment. Therefore, we had to exclude those studies from the meta-sample, in order to keep it homogeneous. In the appendix, table A.1 presents the 45 studies included in our meta-sample and table A.2 reports the 13 studies excluded with the reason of exclusion. Moving on to the type of measurement for the minimum wage effect we use both elasticities and regression coefficients in separate meta-samples. The use of elasticities as the metric to measure the employment effect is followed in most studies and is considered more appropriate, because they are assumed to be relatively stable parameters. 3 On the other hand, the regression coefficients show how many units the dependent variable changes, when the independent changes by one unit. In total, our meta-sample consists of 1,068 elasticities and 484 coefficients providing a sum of 1,552 estimates. Table 2 presents the summary statistics of the studies in the meta-sample, by year of publication. More specifically, it reports the name(s) of the author(s), the year of publication of the study, the country that the estimates concern, the number of estimates each study provided and the type and average estimate of each study. 3 See Doucouliagos and Stanley (2009), p. 412. 7 Table 2. Summary statistics of the studies in the meta-sample. No Author(s) Year Country Estimates Type of estimate Average estimate 1 Cadena 2014 USA 5 Elasticities 0.030 2 Even and Macpherson 2014 USA 71 Elasticities -0.082 3 Hoffman 2014 USA 9 Elasticities -0.030 4 Neumark, Salas and Wascher 2014 USA 54 Elasticities -0.199 5 Sabia 2014 USA 112 Elasticities -0.094 6 Addison, Blackburn and Cotti 2013 USA 34 Elasticities -0.030 7 Coomer and Wessels 2013 USA 21 Elasticities -0.816 8 Giuliano 2013 USA 8 Elasticities -0.370 9 Kalenkoski and Lacombe 2013 USA 3 Elasticities -0.216 10 Kambayashi, Kawaguchi and Yamada 2013 Japan 1 Elasticity -0.115 11 Laporsek 2013 EU Members 10 Elasticities -0.702 12 Magruder 2013 Indonesia 44 Elasticities 0.075 13 Rani, U., Belser, P. and Ranjbar 2013 6 countries 6 Elasticities 0.490 14 Addison, Blackburn and Cotti 2012 USA 13 Elasticities 0.114 15 Addison and Ozturk 2012 Cross-country 13 Elasticities -0.269 16 Bassanini 2012 OECD countries 4 Elasticities 2.314 17 Dinkelman and Ranchod 2012 South Africa 16 Elasticities -1.570 18 Dolton and Bondibene 2012 Cross-country 20 Elasticities -0.165 19 Majchrowska and Zolkiewski 2012 Poland 30 Elasticities -0.105 20 Papps 2012 Turkey 3 Elasticities 0.001 21 Sabia, Burkhauser and Hansen 2012 USA 35 Elasticities -0.423 22 Allegretto, Dube and Reich 2011 USA 132 Elasticities -0.057 23 Cuesta, Heras and Carcedo 2011 Spain 11 Elasticities 0.123 24 Draca, Machin and Van Reenen 2011 UK 2 Elasticities 0.046 25 Ni, Wang and Yao 2011 China 72 Elasticities 0.199 26 Sen, Rybczynski and Van de Waal 2011 Canada 23 Elasticities -0.237 27 Lee and Suardi 2011 Australia 6 Elasticities -0.202 28 Neumark and Wascher 2011 USA 18 Elasticities -0.033 29 Wang and Gunderson 2011 China 27 Elasticities -0.040 30 Belman and Wolfson 2010 USA 86 Elasticities -0.007 31 Dube, Lester and Reich 2010 USA 133 Elasticities -0.025 32 Myatt and McDonald 2010 Canada 44 Elasticities -0.176 33 Volorokosova 2010 Slovak Republic 2 Elasticities 0.111 34 Bhorat, Kanbur and Stanwix 2014 South Africa 3 Coefficients 7.277 35 Bhorat, Kanbur and Mayet 2013 South Africa 15 Coefficients -0.919 36 Boockman, Krumm, Neumann and Rattenhuber 2013 Germany 6 Coefficients 0.407 37 Frings 2013 Germany 12 Coefficients 0.004 38 Higuchi 2013 Japan 8 Coefficients -0.040 39 Nguyen 2013 Vietnam 8 Coefficients -3.779 40 Addison and Ozturk 2012 14 countries 14 Coefficients -0.211 41 Dolton and Bondibene 2012 Cross-country 10 Coefficients -0.209 42 Dolton, Bondibene and Wadsworth 2012 UK 258 Coefficients 0.012 43 Papps 2012 Turkey 4 Coefficients -13.329 44 Wang 2012 China 3 Coefficients 0.269 45 Wang and Gunderson 2012 China 21 Coefficients 0.123 46 Comola and De Mello 2011 Indonesia 8 Coefficients -0.002 47 Dolton, Bondibene and Wadsworth 2010 UK 60 Coefficients 0.006 48 Persky and Baiman 2010 USA 54 Coefficients 0.336 Note: The total number of the studies in the meta-sample is not 48 but 45 as three studies by Addison and Ozturk (2012), Dolton and Bondibene (2012) and Papps (2012) report both elasticities and coefficients. 8 4. Publication bias and FAT-PET tests In meta-analysis the simplest way to see if there is publication bias is the funnel graph, which is nothing more than a scatter diagram of all empirical estimates and these estimates’ inverse of the standard error. In figures 1 and 2 we present the funnel graphs of the estimated minimum wage elasticities and coefficients, respectively. Although it may seem that the elasticities are distributed symmetrically around zero, the majority of the estimated elasticities of the meta-sample are negative as presented in table 3. This means that the majority of the values are gathered in the left portion of the graph which reveals selection for negative employment effects of minimum wages in the published studies of our meta-sample. As we can see in table 3, almost 2/3 of the elasticities in our meta-sample have a negative sign, which could imply publication bias in favor of studies with negative minimum wage effects. With respect to the coefficients of our meta-sample, it seems that most of them gather around the zero value, but we certainly cannot jump into secure conclusions as the estimates are widely distributed. However, according to table 3, the coefficients are relatively equally divided into positive and negative values. Table 3. Characteristics of the estimates. Elasticities Number Percent Negative (Significantly negative at 10% level) 710 (261) 66.48% Zero 3 0.28% Positive (Significantly positive at 10% level) 355 (105) 33.24% Total 1,068 Coefficients Number Percent Negative (Significantly negative at 10% level) 230 (87) 47.52% Zero 1 0.21% Positive (Significantly positive at 10% level) 253 (111) 52.27% Total 484 15 Table 6. Multiple Meta-Regression-Analysis using Elasticities (Dependent variable: t-stat). Using General-to-Specific Methodology Column 1 OLS Column 2 Robust Column 3 REML Column 4 WLS 1/SE -0.030*** (0.009) -0.028** (0.014) -0.029*** (0.009) -0.026 (0.018) DID/SE 0.036*** (0.010) 0.033*** (0.012) 0.035*** (0.010) 0.043*** (0.007) USE/SE 0.135*** (0.012) 0.128*** (0.016) 0.133*** (0.012) 0.152*** (0.009) Europe/SE 0.034*** (0.010) 0.034** (0.015) 0.033*** (0.010) 0.037*** (0.005) MWlag/SE -0.048*** (0.012) -0.035* (0.020) -0.049*** (0.012) -0.071*** (0.009) Double/SE -0.103*** (0.010) -0.097*** (0.011) -0.101*** (0.010) -0.116*** (0.009) PanelCross/SE -0.042*** (0.015) Teens/SE -0.139*** (0.010) -0.136*** (0.013) -0.138*** (0.010) -0.114*** (0.008) Youth/SE -0.144*** (0.019) -0.142*** (0.024) -0.143*** (0.019) -0.137*** (0.015) Hours/SE 0.094*** (0.015) 0.091*** (0.012) 0.092*** (0.093) 0.091*** (0.013) Educ/SE -0.017*** (0.006) Kaitz/SE 0.026** (0.011) 0.025** (0.011) 0.048*** (0.006) Dummy/SE -0.110*** (0.035) -0.103*** (0.020) -0.107*** (0.035) -0.111*** (0.025) Retail/SE -0.161*** (0.040) -0.157** (0.072) -0.160*** (0.040) -0.207*** (0.032) OtherIndustrySE 0.771*** (0.216) 0.829*** (0.162) 0.915*** (0.235) 0.832** (0.332) Constant -0.248*** (0.094) -0.224** (0.093) -0.252*** (0.096) -0.703*** (0.117) Observations 1,068 1,068 1,068 1,068 R-squared 0.286 0.282 0.281 0.460 Notes: *, **, *** denote statistical significance at 10%, 5% and 1% level of significance respectively. Standard errors are reported in parentheses. Table 7. Multiple Meta-Regression-Analysis using Coefficients (Dependent variable: t-stat). Using General-to-Specific Methodology Column 1 OLS Column 2 Robust Column 3 REML Column 4 WLS 1/SE -0.083*** (0.028) -0.025*** (0.009) -0.083*** (0.029) -0.017*** (0.003) Endogen/SE -0.192*** (0.026) -0.233*** (0.015) -0.192*** (0.027) -0.256*** (0.024) USE/SE 0.035** (0.015) Europe/SE 0.061** (0.028) 0.061** (0.029) Youth/SE -0.220** (0.111) -0.281*** (0.042) -0.222* (0.114) FE/SE 0.022*** (0.003) 0.026** (0.010) 0.022*** (0.003) 0.020*** (0.003) Educ/SE 0.082*** (0.028) 0.027* (0.015) 0.083*** (0.029) 0.018*** (0.004) Kaitz/SE 0.010** (0.005) 0.010** (0.005) 0.002*** (0.004) Dummy/SE 0.345*** (0.130) 0.293*** (0.043) 0.346** (0.133) Retail/SE -0.806** (0.332) FoodDrink/SE -0.412** (0.189) -0.470*** (0.046) -0.418** (0.194) OtherIndustrySE -0.338** (0.130) -0.285*** (0.043) -0.339** (0.134) Constant 0.179 (0.294) 0.305 (0.188) 0.226 (0.320) -0.975 (0.731) Observations 484 484 484 484 R-squared 0.477 0.468 0.461 0.522 Notes: *, **, *** denote statistical significance at 10%, 5% and 1% level of significance respectively. Standard errors are reported in parentheses. 16 What is important in the multivariate meta-regression analysis is to examine the presence of publication bias and the genuine effect of minimum wage on employment, after any other potential explanatory factors are taken into account. Initially, about the existence of publication selection, we notice that the constant in the elasticities’ meta-sample is negative and statistically significant, indicating publication selection in favor of studies with negative minimum wage effects. However, in the smaller coefficients’ meta-sample the intercepts are not statistically significant suggesting no evidence of publication selection bias. Another significant dimension is the genuine effect in the multiple metaregressions, in order to see if the very small effect of minimum wages on employment found in the previous section remains. Firstly, in the elasticities’ meta-sample, the coefficient of 1/SE that represents the magnitude of the impact takes value from - 0.026 to -0.030 with three of them being statistically significant. Secondly, in the coefficients’ meta-sample, the sign of the impact is negative and statistically significant in all columns and the magnitude is from -0.017 to -0.083, which indicates small negative minimum wage effect. Except for the investigation of the existence of publication bias and the genuine effect, as we have explained earlier, the multivariate meta-regression analysis presented in tables 6 and 7 can be used to find the sources of heterogeneity of the results among studies. In both tables, we employ four estimation methods in order to provide robustness to the results, which generally do not seem to produce different results and variability in the sign of the estimates. Column 1 shows the estimations using the ordinary-least-squares estimation method, and column 2 reports the robust version of the OLS estimation. Column 3 presents the results when we use random effects model (REML), which is the benchmark method for estimating the betweenstudy variance, and column 4 reports the results using the weighted-least-squares estimation method, which is considered better than conventional random-effects metaanalysis methods when there is publication (or small-sample) bias. The estimation results in table 6 indicate that effects regarding DID specification, US studies, European studies and studies which refer to hours worked as dependent variable or use the Kaitz index as the minimum wage measurement, tend to report a positive employment impact of minimum wages. On the other hand, elasticities related to a lagged minimum wage measure, or drawn from double log specifications, or refer to younger age groups (teenagers and youth) or use a dummy 17 variable as minimum wage, report a negative employment effect of minimum wages. The results also depend on the type of industry or sector concerned, as elasticities from retail industry revealed negative minimum wage effects, food-beverage or drinking sector does not seem to affect the sign of the impact, while, if the elasticities relate to other industry or group of industries, they tend to report positive minimum wage effects. Furthermore, endogeneity, time effects, fixed effects, the unemployment rate and the use of the level of minimum wage as a measurement, do not appear to explain any heterogeneity of the minimum wage elasticities. Table 7 reports the estimation results from the meta-sample consisted of coefficients. Using this meta-sample, the sources of heterogeneity seem to diversify. Specifications using fixed effects, controlling for education and using the Kaitz index or a dummy variable, are found to produce positive regression coefficients. On the other hand, estimation methods dealing with endogeneity relating to the youth and coming from food-beverage and drinking industry, tend to report negative minimum wage effects. Moreover, the R-squares are over 46% in all four estimations methods, which are considered satisfactory in multiple meta-regressions. Comparing these two tables, we can notice some differences with respect to the sign and the magnitude of the results. For instance, in the elasticities’ metasample, the effect of minimum wage on employment is positive and particularly large when study is related to a specific industry apart from retail and food-beverage- drinking sector, while in the coefficients’ meta-sample the sign is negative, but still considerably large. Furthermore, studies which use a dummy variable as minimum wage measure display a negative and relatively stable magnitude around -0.110 in table 6, but the sign is positive and much higher in table 7. Therefore, the two metasamples generate some differences in the factors which account for the heterogeneity of the minimum effect on employment and thus results on the moderators should be treated with caution. In summary, we would say that, during the five-year period from 2010 to 2014, the minimum wage literature seems to be characterized by publication bias, in the elasticities’ meta-sample. However, once this publication bias is corrected, the genuine effect of minimum wage is negative but small. Nevertheless, in the coefficients’ meta-sample, evidence of publication bias is not supported and the magnitude of the impact is once again negative and small. Besides that, the sign of the impact in both meta-samples is affected by the study characteristics of the meta- 18 sample related to the data, the model specifications, the minimum wage and employment measures used, and the industry concerned. 6. Conclusions The minimum wage literature is not only large but also growing and the empirical studies on the employment effect of minimum wages still produce controversial results. During the last five years we have found dozens of published and unpublished studies on the employment effect of minimum wages which approach the impact from different point of views and use various methods and models. The objective of this paper is to investigate the relationship between minimum wages and employment using a meta-sample of 45 empirical studies published in academic journals within the 2010-2014 period. Our analysis points to the existence of publication bias in the elasticities’ meta-sample but once it is corrected, the impact is so small that is of no significant use. On the other hand, in the coefficients’ meta-sample, no evidence of publication bias is found in the multiple meta-regression analysis and again the minimum wage effect is negative but small. In addition, we identify as potential sources of heterogeneity of the results the study characteristics related to the data, the model specifications, the group of population, the minimum wage and employment measurements and the industry concerned. Our meta-analysis, which consists of 1,068 elasticities and 484 regression coefficients, indicates that the minimum wage has only a small effect on employment. In this frame, space has begun to be given to other theories in the minimum wage research such as the Keynesian perspective, according to which, changes in minimum wages are not related with either positive or negative employment effects. Therefore, perhaps the appropriate ground has been created to investigate the impact through the role of specific aspects of the minimum wage and employment characteristics. Further research on the role of education, the skills of the employees, the age, the specific industry or sector and several other factors need to be conducted in order to find a mechanism which will describe the procedure of any direct or indirect effect of minimum wages on the employment measures. Undoubtedly, the minimum wage is a very useful tool in the hands of policy makers and it can improve the economic conditions of the more vulnerable groups and the utility of the low-income families. The issue of the employment effect of minimum wages provides many avenues of research and a number of preferable 19 alternatives to a researcher. Naturally, like most research, the more the topics are reviewed the more questions are raised and it is essential a minimum wage setting model to be found which will improve the economic effectiveness and the social welfare. Consequently, the need of the minimum wage research not to compromise to the existing contradictory empirical results and to progress, is a matter of great importance in the constantly changing and demanding economic environment that we live in. Acknowledgements The authors are grateful to Dimitris Zikos for his helpful suggestions and generous comments. We have also benefited from valuable comments and discussions by Nikos Benos, Pantelis Kammas and Antonis Adam. Any remaining errors are ours. References Adam, A., Kammas, P. and Lagou, A. (2013). The effect of globalization on capital taxation: What have we learned after 20 years of empirical studies? Journal of Macroeconomics 35(3): 199-209. Belman, D. L. and Wolfson, P. (2014). What Does the Minimum Wage Do? Upjohn Institute Press. Benos, N. and Zotou, S. (2014). Education and economic growth: A meta-regression analysis. World Development 64: 669-689. Boockman, B. (2010). The combined employment effects of minimum wages and labor market regulation: A meta-analysis. IZA Discussion Paper, No. 4983. Card, D. (1992). Do minimum wages reduce employment? A case study of California, 1987-1989. Industrial and Labor Relations Review 46(1): 38-54. Card, D. and Krueger, A. B. (1995), Time-series minimum wage studies: A metaanalysis. American Economic Review 85(2): 238-243. Charemza, W. and Deadman, D. (1997). New directions in econometric practice. 2nd Edition. Cheltenham: Edward Elgar. 20 Doucouliagos, C. H. and Stanley, T. D. (2009). Publication selection bias in minimum-wage research? A meta-regression analysis. British Journal of Industrial Relations 47(2): 406-428. Katz, L. and Krueger, A., (1992). The effect of the minimum wage in the fast food industry. Industrial and Labor Relations Review 46(1): 6-21. Leonard, M., Stanley, T. D. and Doucouliagos, H. (2014). Does the UK minimum wage reduce employment? A meta-regression analysis. British Journal of Industrial Relations 52(3): 499-520. Nataraj, S., Perez-Arce, F. Srinivasan, S. V. and Kumar, K. B. (2014). The impact of labor market regulation on employment in low-income countries: A metaanalysis. Journal of Economic Surveys 28(3): 551-572. Stanley, T. D., Doucouliagos, C. and Jarrell, S. (2008). Meta-regression analysis as the socioeconomics of economics research. The Journal of Socio-Economics 37: 276-292. Stanley, T. D., Giles, M., Heckemeyer, J. H., Johnston, R. J., Laroche, P., Nelson, J. P., Paldam, M., Poot, J., Pugh, G., Rosenberger, R. S. and Rost, K. (2013). Meta-analysis of economics research reporting guidelines. Journal of Economic Surveys 27(2): 390-394. Stanley, T. D. and Doucouliagos, C. (2012). Meta-regression analysis in economics and business. Routledge Advances in Research Methods. Stanley, T. D. and Doucouliagos, C. (2015). Neither fixed nor random: weighted least squares meta-analysis. Statistics in Medicine 34: 2116-2127. 21 Appendix Table A.1: Studies included in the meta-sample, by year of publication. Study Country 1 Belman, D. L. and Wolfson, P. (2010). The effect of legislated minimum wage increases on employment and hours: A dynamic analysis. LABOUR 24(1): 1-25. USA Note: We excluded 4 employment elasticities and 2 volume (hour) elasticities from table 4, since we couldn't calculate (with the use of TINV function in excel) the t-statistics from their six p-values reported in the study which had value = 0. 2 Dolton, P., Bondibene, C. R. and Wadsworth, J. (2010). The UK national minimum wage in retrospect. Fiscal Studies 31(4): 509-534. UK 3 Dube, A., Lester, T. W. and Reich M. (2010). Minimum wage effects across state borders: Estimates using contiguous counties. The Review of Economics and Statistics 92(4): 945-964. USA Note: In the two minimum wage elasticities of table B.1, the sizes of the samples are not reported and we are unable to calculate them. 4 Myatt, T. and McDonald J. T. (2010). The robustness of provincial panel-data studies of minimum wages in Canada. Canadian Journal of Regional Science 33(3): 77-88. Canada Note: In table 2, the minimum wage elasticities do not report standard errors or t-stats and we are unable to calculate them. Therefore, these specific estimates were excluded from the meta-sample. 5 Persky, J. and Baiman, R. (2010). Do state minimum wage laws reduce employment? Mixed messages from fast food outlets in Illinois and Indiana. Journal of Regional Analysis and Policy 40(2): 132-142. USA 6 Vokorokosová, R. (2010). Do minimum wage changes influence employment? Economic Analysis 43(1-2): 83-90. Slovak Republic 7 Allegretto, S., Dube, A., and Reich, M. (2011). Do minimum wages really reduce teen employment? Accounting for heterogeneity and selectivity in state panel data. Industrial Relations 50(2): 205-240. USA Note: In some estimated elasticities (eight in table 5 and twenty in table 8), we are unable to calculate the sizes of the samples. 8 Comola, M. and De Mello, L. (2011). How does decentralized minimum wage setting affect employment and informality? The case of Indonesia. Review of Income and Wealth 57: 79-99. Indonesia Note: We had to exclude 7 estimates (coefficients) from the meta-sample, since their relative t-statistics were 0.000. Therefore, we could not calculate the values of their standard errors which are necessary for publication selection bias correction. 9 Cuesta, M. B., Heras, R. L. and Carcedo, J. M. (2011). Minimum wage and youth employment rates 2000-2008. Revista de Economía Aplicada 19(56): 35-57. Spain 10 Draca, M., Machin, S. and Van Reenen, J. (2011). Minimum wages and firm profitability. American Economic Journal: Applied Economics 3: 129-151. UK 11 Lee, W-S. and Suardi, S. (2011). Minimum wages and employment: Reconsidering the use of a time-series approach as an evaluation tool. British Journal of Industrial Relations 49(2): 376-401. Australia 12 Neumark, D. and Wascher, W. (2011). Does a higher minimum wage enhance the effectiveness of the earned income tax credit? Industrial and Labor Relations Review 64(4): 712-746. USA Note: We did not include the estimates of the interactions of minimum wage with EITC since these estimates show if a higher minimum wage enhances the employment effect of the earned income tax credit. We also excluded estimates of the interaction of minimum wage with the dummy KIDS, as these estimates show if minimum wage benefits more the 22 employment of families with children in comparison to those being childless. These estimates do not imply a direct employment impact of minimum wage and had to be excluded from the meta-sample. 13 Ni, J., Wang, G. And Yao, X. (2011). The impact of minimum wages on employment: Evidence from China. The Chinese Economy 44(1): 18-38. China 14 Sen, A., Rybczynski, K. and Van De Waal, C. (2011). Teen employment, poverty, and the minimum wage: Evidence from Canada. Labour Economics 18: 36-47. Canada 15 Wang, J. and Gunderson, M. (2011). Minimum wage impacts in China: Estimates from a prespecified research design, 2000-2007. Contemporary Economic Policy 29(3): 392-406. China 16 Addison, J. T., Blackburn, M. L. and Cotti, C. D. (2012). The effect of minimum wages on labour market outcomes: County-level estimates from the restaurant-and- bar sector. British Journal of Industrial Relations 50(3): 412-435. USA 17 Addison, J. T. and Ozturk O. D. (2012). Minimum wages, labor market institutions, and female employment: A cross-country Analysis. Industrial and Labor Relations Review 65(4): 779-809. Several OECD countries Note: Study is a cross-country analysis and 13 elasticities are based on cross-national data. However, there are 14 coefficients which concern a single country and are included in our meta-sample. Moreover we have to mention that we did not include the estimates with respect to the labor force participation rate as we do not consider it as an employment measure. 18 Bassanini, A. (2012). Aggregate earnings and macroeconomic shocks: The role of labour market policies and institutions. Review of Economics and Institutions 3(3): 1- 44. Several OECD countries 19 Dinkelman, T. and Ranchhod, V. (2012). Evidence on the impact of minimum wage laws in an informal sector: Domestic workers in South Africa. Journal of Development Economics 99: 7-45. South Africa Note: We did not include in the meta-sample estimates from table 4 as they indicate employment probabilities (probability of working as a domestic worker). 20 Dolton, P. and Bondibene, C. R. (2012). The international experience of minimum wages in an economic downturn. Economic Policy 27(69): 99-142. Several OECD countries Note: Minimum wage elasticities from table 4 are not included in our meta-sample, since they do not report standard errors or t-stats which are both needed for publication selection bias correction. 21 Dolton, P., Bondibene, C. R. and Wadsworth, J. (2012). Employment, inequality and the UK national minimum wage over the medium-term. Oxford Bulletin of Economics and Statistics 74(1): 78-106. UK 22 Majchrowska, A. and Zolkiewski Z. (2012). The impact of minimum wage on employment in Poland. Investigaciones Regionales 24: 211-239. Poland 23 Papps, K. L. (2012). The effects of social security taxes and minimum wages on employment: Evidence from Turkey. Industrial and Labor Relations Review 65(3): 686-707. Turkey 24 Sabia J. J., Burkhauser, R. V. and Hansen B. (2012). Are the effects of minimum wage increases always small? New evidence from a case study of New York state. Industrial and Labor Relations Review 65(2): 351-376. USA Note: We are unable to calculate the sizes of the samples in table 5. 25 Wang, X. (2012). When workers do not know - The behavioral effects of minimum wage laws revisited. Journal of Economic Psychology 33: 951-962. China 26 Wang, J. and Gunderson, M. (2012). Minimum wage effects on employment and China 23 wages: dif-in-dif estimates from eastern China. International Journal of Manpower 33(8): 860-876. 27 Addison, J. T., Blackburn, M. L. and Cotti, C. D. (2012). Minimum wage increases in a recessionary environment. Labour Economics 23: 30-39. USA 28 Bhorat, H., Kanbur, R. and Mayet, N. (2013). The impact of sectoral minimum wage laws on employment, wages, and hours of work in South Africa. IZA Journal of Labor & Development 2(1): 1-27. South Africa Note: We did not include in our meta-sample the minimum wage impact on employment presented in table 5 of their paper, because the dependent variable is a dummy variable equal to 1 if the individual is employed in the respective sector and equal to 0 otherwise, suggesting probability. 29 Boockmann, B., Krumm, R., Neumann, M. and Rattenhuber, P. (2013). Turning the switch: An evaluation of the minimum wage in the German electrical trade using repeated natural experiments. German Economic Review 14(3): 316-348. Germany Note: We did not include the coefficients of tables 4 and 7 in our meta-sample, as they refer to the impact on the probability of remaining in employment. We also excluded the estimates of table 5 concerning the effect on hiring and separations at the company (columns 1 and 2, respectively) which are not considered as employment measures in our analysis. 30 Coomer, N. M. and Wessels, W. J. (2013). The effect of the minimum wage on covered teenage employment. Journal of Labor Research 34: 253-280. USA Note: Estimates from table 2 were not included in the meta-sample, since they come out from logit estimation, implying probability. 31 Frings, H. (2013). The employment effect of industry-specific, collectively bargained minimum wages. German Economic Review 14(3): 258-281. Germany 32 Giuliano, L. (2013). Minimum wage effects on employment, substitution, and the teenage labor supply: Evidence from personnel data. Journal of Labor Economics 31(1): 155-194. USA Note: Apart from tables 4 and 6, the study reports additional estimates on relative employment of teenagers, on teenage employment flows, and on employment and hiring shares in tables 5, 7 and 9, respectively, which do not represent direct impact of minimum wages on employment measures. Therefore, these specific additional estimates were not included in the meta-sample. 33 Higuchi, Y. (2013). The dynamics of poverty and the promotion of transition from non-regular to regular employment in Japan: Economic effects of minimum wage revision and job training support. Japanese Economic Review 64(2): 147-200. Japan Note: We did not include in our meta-sample estimates with respect to the minimum wage impact on employment, since the authors use a logit model and a random-effect logit model which suggests probability. We included only the estimates concerning the minimum wage impact on hours worked. 34 Kalenkoski, C. M. and Lacombe, D. J. (2011). Minimum wages and teen employment: A spatial panel approach. Papers in Regional Science 92(2): 407-418. USA 35 Kambayashi, R., Kawaguchi, D. and Yamada, K. (2013). Minimum wage in a deflationary economy: The Japanese experience, 1994-2003. Labour Economics 24: 264-276. Japan Note: We did not include the minimum wage elasticities of new hires (it does not represent a direct employment measure) and employment (study reports the probability of being employed for a woman which is not a continuous measure of employment that we use in our meta-analysis). 36 Laporšek, S. (2013). Minimum wage effects on youth employment in the European Union. Applied Economics Letters 20(14): 1288-1292. EU Members 37 Magruder, J. R. (2013). Can minimum wages cause a big push? Evidence from Indonesia 24 Indonesia. Journal of Development Economics 100: 48-62. 38 Nguyen, C. V. (2013). The impact of minimum wages on employment of low-wage workers. Evidence from Vietnam. Economics of Transition 21(3): 583-615. Vietnam Note: We did not take into account the minimum wage effect on self-employment as we do not consider it as an employment measure in our analysis. 39 Rani, U., Belser, P. and Ranjbar, S. (2013). Role of minimum wages in rebalancing the economy. World of Work Report 1: 45-74. Brazil, Costa Rica, India, Mexico, Peru, Vietnam 40 Bhorat, H., Kanbur, R. and Stanwix, B. (2014). Estimating the impact of minimum wages on employment, wages, and non-wage benefits: The case of agriculture in South Africa. American Journal of Agricultural Economics p. 1-18. South Africa Note: We did not include in the meta-sample the minimum wage impact on employment presented in table 4 of that study, because the dependent variable is a dummy variable equal to 1 if the individual is employed in the agriculture sector and equal to 0 otherwise, suggesting probability (probability of being employed as a farm worker after introducing the minimum wage law). 41 Cadena, B. C. (2014). Recent immigrants as labor market arbitrageurs: Evidence from the minimum wage. Journal of Urban Economics 80: 1-12. USA 42 Even, W. E. and Macpherson, D. A. (2014). The effect of the tipped minimum wage on employees in the U.S. restaurant industry. Southern Economic Journal 80(3): 633- 655. USA Note: We had to exclude one elasticity from table 2 of that study, since its standard error is zero and in the REML estimation (in the meta-regression analysis) the command metareg requires the standard errors not to have zero value. 43 Hoffman, S. D. (2014). Employment effects of the 2009 minimum wage increase: New evidence from state-based comparisons of workers by skill level. The B.E. Journal of Economic Analysis & Policy 14(3): 695-721 USA Note: We are unable to calculate the size of the sample of the estimates. 44 Neumark, D., Salas, J. M. I. and Wascher, W. (2014). Revisiting the minimum wageemployment debate: Throwing out the baby with the bathwater? Industrial and Labor Relations Review 67(Supplement): 608-648. USA Note: We are unable to calculate the size of the sample of the estimates in some estimates. 45 Sabia, J. J. (2014). The effects of minimum wages over the business cycle. Journal of Labor Research 35: 227-245. USA