Economic Effects of a Minimum-Wage Increase in Japan: Panel Data Analysis Yoshio Higuchi, Kazuma Sato and Toru Kobayashi

Transcription

1 Discussion Paper No Economic Effects of a Minimum-Wage Increase in Japan: Panel Data Analysis Yoshio Higuchi, Kazuma Sato and Toru Kobayashi

2 Economic Effects of a Minimum-Wage Increase in Japan: Panel Data Analysis Yoshio Higuchi, * Kazuma Sato, ** Toru Kobayashi *** Abstract Since 2007 in Japan, region-specific minimum wages have been raised by much larger amounts than before mainly in major urban areas that have been economically vibrant. Taking this change as a change in a control variable in a quasi-experiment and using data from the Keio Household Panel Survey for the period , this paper analyzes how a minimum-wage increase affects the wage level of non-regular workers (hiseiki rodosha) and their wage gap with regular workers (seiki rodosha), and whether a minimum-wage increase causes an employment loss. The following four conclusions are reached as a result. (1) A minimum-wage increase had a significant negative effect on the size of the wage gap: with a minimum-wage increase, the wages of female non-regular workers rose, the income gap with regular workers narrowed, and, as found in an analysis of different wage classes, the 20th percentile wage increased substantially. (2) Although a minimum-wage increase had no significant effect on the wages of male nonregular workers and regular workers, the analysis of different wage classes showed that the 20th, 40th, and 60th percentile wages increased significantly with a minimum-wage increase. (3) A minimum-wage increase led neither to loss of employment among existing female and male nonregular workers nor to reduced employment among female and male job seekers. (4) A minimum-wage increase was not associated with a reduction in the average weekly hours worked by existing female and male non-regular workers. 1. Introduction The issue of minimum wages is attracting increasing interest in Japan. As stated in Article 1 of the Minimum Wages Act, the purpose of the minimum wage is to promote stable living among the workforce, improvement of labor quality, fair business competition, and sound development of the nation's economy by improving working conditions, by setting guaranteed minimum wages for low-wage workers according to regions and types of businesses and occupations. With the increasing poverty rate being emphatically pointed out, not only government officials and This research was supported by the Grant-in-Aid for Specially Promoted Research NO The data for this analysis, Keio Household Panel Survey, was provided by the Global Center of Excellence Program at Keio University. * Faculty of Business and Commerce, Keio University, Tokyo, Japan ** Faculty of Economics, Meikai University, Urayasu, Chiba, Japan *** Doctoral Program, Graduate School of Business & Commerce, Keio University, Tokyo, Japan

3 researchers, but also businesses and the public have become interested in whether the minimum wage is actually achieving its purpose. There are three types of minimum wage: the regional minimum wage to which a labor agreement is extended and applied, the region-specific minimum wage, and the industry-specific minimum wage. The regional minimum wage program is designed such that when a labor agreement with a minimum wage applies to most of the workers and employers in a certain industry in a region, that minimum wage can be extended to all workers in the region upon petition by either the labor or management side. The program, however, is hardly used at present. The region-specific minimum wage is set in each prefecture and applies, in principle, to all labor and management regardless of their occupation and industry. The industry-specific minimum wage is set in certain industries by relevant labor and management groups when it is deemed necessary to institute a minimum wage for core workers that is higher than the relevant region-specific minimum wage. A total of 250 minimum wages are being applied in prefectures as a whole. In terms of determining the region-specific minimum wage specifically, which is the topic of this paper, the Central Minimum Wage Council of the national government, which is comprised of labor, management, and public-interest representatives, decides on the suggested amounts for changes in the prefectures minimum wages upon requests for advice from the Minister of Health, Labor and Welfare. The Minimum Wage Council for each prefecture then sets its region-specific minimum wage based on the suggestion each year. While taking into account the situation of the entire Japanese economy, the Central Minimum Wage Council is said to focus on differences among the prefectures in terms of standard of living, employers ability to pay, and the influence of the minimum wage on employment, and that there has been a tendency to keep the minimum wages of major urban areas low relative to their economic strength in order to prevent interregional wage gaps from increasing. As a result, a reverse phenomenon has been seen in some prefectures, whereby the amount of income from full-time work with a minimum wage is smaller than the amount of (regionally determined) public welfare assistance. This has prompted the criticism that the region-specific minimum wage could cause a moral hazard. At the same time, opinions have been voiced in the Minimum Wage Councils, especially from employers, that productivity improvement is needed at small and medium-sized firms in order to raise minimum wages, that if minimum wages are increased without productivity improvement, firms will go bankrupt and employment will decline, and that increases in minimum wages, which are under jurisdiction of the Ministry of Health, Labor and Welfare, should be examined in combination with support for productivity improvement at small and medium-sized firms, which are under jurisdiction of the Ministry of Economy, Trade and Industry. In order to implement such combined measures, in 2007 the government launched the Round-Table Conference for Promotion of Economic Growth Strategies, which consisted of government, labor, management, and public-interest representatives, under the Kantei (Prime Minister s Official Residence) and started to increase minimum wages and provide support for productivity improvement at small and medium-sized firms. Even after a change in the ruling party, this approach has been continued in Employment Strategy Dialogues in which government, labor, management, and public-interest representatives participate.

4 As a result of these strategies, significant changes have been observed in region-specific minimum wages; double-digit increases were seen in 2007, compared to national average increases of around one percent previously. In particular, to solve the problem of the aforementioned reverse phenomenon, region-specific minimum wages have increased significantly mainly in urban areas, where employers have a greater capacity to pay. Questions then arise as to how such minimum-wage increases affect the wage level of non-regular workers and their wage gap with regular workers, and whether significant minimum-wage increases cause employment losses. Here, regarding such changes in minimum wage policy as a social experiment and using panel data that follows individuals behavior during the relevant period, this paper analyzes the effect of minimum-wage changes on wage level, wage gap, and employment. The remainder of this paper is structured as follows. Section 2 discusses recent trends in minimum wages in Japan based on data. Section 3 provides a survey of earlier studies on minimum wages and considers the questions to be addressed in this paper. Section 4 describes the panel data and estimation method used in this paper, and Section 5 discusses the estimation results. Section 6 concludes the paper by examining policy implications of the estimation results and issues for future research. 2. Changes in the Minimum Wage System: Major Changes and Relationship with Public Welfare Assistance As mentioned earlier, since 2007, region-specific minimum wages have continued to be increased nationwide, and the increases have been large in major urban areas especially. As factors behind these increases, this section considers issues involving previous minimum wages and debates on such increases. The situation in the periods before and after the start of the minimum-wage increases is also discussed. Based on discussions held, for example, in meetings of the Round-Table Conference for Promotion of Economic Growth Strategies, the following three issues will be mainly discussed regarding the minimum wage. First is the notion that the minimum-wage level in Japan may be too low. The minimum wage amount in Japan in absolute terms is indeed low compared to other countries. The Kaitz index, which indicates the level of a minimum wage relative to the average market wage, is 36.5 percent on average for Japan as of 2006, which is the lowest among the OECD countries for which the index is available. In response to this fact, and alongside attention also being paid to the problem of income gaps, the Round-Table Conference proposed that minimum wages be raised to a certain percentage of the average wage. The second issue concerns what criteria should be used in determining the rate of that minimumwage increase. In 1954 when the Minimum Wages Act was implemented, the amount of minimum-wage increase was determined based on the salaries that junior high school graduates received. Although this method has no longer been used outwardly since 1978 when minimum wages began to be determined based on the aforementioned Central Minimum Wage Council's suggestion, a similar method was still used it was pointed out to be outdated when the high

5 school enrollment rate became high. After discussions, consensus was reached that the criterion should be changed to the salaries of high school graduates and minimum-wage increases were proposed. The third issue involves the reverse phenomenon between minimum wages and public welfare assistance. The amount of minimum-wage increase in each prefecture is determined by its Minimum Wage Council based on the suggested amounts for the 47 prefectures which are obtained by dividing the prefectures into ranks A through D based on income, consumption, firms business performance, and other economic factors and considering the situation in each prefecture. This approach has remained the same before and after However, it is considered that before 2007, when considering the regional situation, emphasis was placed on avoiding increased regional gaps, which eventually tended to keep minimum wages low in major urban areas, that is, economically vibrant rank A regions. As Figure 2 shows, the pre-revision inadequacy rate (miman ritsu), post-revision inadequacy rate (eikyo ritsu), and Kaitz index are all low especially before 2007 in the rank A regions that are highly urbanized. 1 Taking into account regional differences in terms of economic level and cost of living, it can be seen that minimum wages in major urban areas are low compared to other regions. While the minimum wages in the rank A regions do not properly reflect the cost of living for the regions, the amount of public welfare assistance is determined by considering the cost of living for different regions. As a result, the reverse phenomenon between minimum wages and public welfare assistance has become clear, causing concerns about the emergence of a moral hazard involving employment choice. A quick inspection of the minimum wage and public welfare assistance in the prefectures for 2006 reveals that the level of public welfare assistance surpasses the minimum wage amount in Tokyo, Kanagawa, Osaka, Chiba, Saitama, Kyoto, Hyogo, Hiroshima, Hokkaido, Aomori, and Akita, with the difference being particularly large in Tokyo, Kanagawa, Osaka, and Chiba. After discussions on these factors, the labor, industry, expert, and government representatives participating in the June 2008 Round-Table Conference for Promotion of Economic Growth Strategies reached a final agreement that government, labor, and management would make cooperative efforts to raise minimum wages over the next five years, taking into account the socioeconomic conditions, consistency with the criteria used for public welfare assistance, and a balance with the lowest level of income earned by high school graduates working at small-scale firms. Since then, minimum wages have been significantly raised nationwide, but mainly in the highly urbanized, rank A regions. Moreover, the pre-revision inadequacy rate, post-revision inadequacy rate, and Kaitz index for the rank A regions have all risen substantially since 2007, as shown in Figure 2. The policy shift was maintained after the change in ruling party, and in the fourth meeting for Employment Strategy Dialogues held in June 2010 a goal was proposed to raise the minimum wages to at least 800 yen as soon as possible and then, with consideration given to the country's economic situation, to a nationwide average of 1,000 yen. 1 The pre-revision inadequacy rate (miman ritsu) is the proportion of workers whose wages are below the prerevision minimum wage; the post-revision inadequacy rate (eikyo ritsu) is the proportion of workers whose wages are to be below the post-revision minimum wage. These rates are calculated by each prefecture and indicate what proportion of workers is directly affected by a minimum-wage increase, and the rate of change for the minimumwage is discussed in the Minimum Wage Council with consideration given to these rates and other data.

6 3. Previous Literatures and the Role of This Paper The substantial revision of minimum wages raises expectations that the wage level will rise and the wage gap narrow, but there is a concern that a minimum-wage increase will potentially reduce employment at the same time. This is because, theoretically speaking, employment falls and the number of unemployed rises when the minimum wage exceeds the market equilibrium wage in a perfectly competitive labor market. However, as the management side pointed out in the Minimum Wage Councils, employment does not necessarily fall as the minimum wage rises if the equilibrium wage is kept above the minimum wage due to a simultaneous increase in productivity or if the labor market is not perfectly competitive because of, for example, the existence of monopsony power. In such cases, a minimum-wage increase can raise the wage with little change in employment and can contribute to a wage gap reduction without employment losses. It is also possible that a rise in the minimum wage will stimulate the willingness to work of suppliers of labor and thus increase the labor force participation rate. What is the actual effect of the minimum wage on employment and wage level? Many empirical studies have tried to answer this question, but no consistent conclusion has been reached as results vary depending on the analytical method used or the target of analysis. The remainder of this section summarizes the results and issues of earlier studies on the effect of a minimum-wage increase on the wage, a reduction in the wage gap, and employment loss and discusses the significance of this paper Effect of the Minimum Wage on a Rise in the Market Wage and a Reduction in the Wage Gap The leading empirical studies analyzing the effect of the minimum wage on the wage are those by Card (1992), Abe (2001), and Abe and Tanaka (2007), using data for the 1990s. For lowwage regions and high-wage regions, Card (1992) analyzes the impact of the minimum-wage increase of the same amount that was implemented in all states in the United States in April Using individual-level cross-sectional data, he finds that an increase in the average wage was greater in low-wage states where the effect of the minimum-wage increase was large. Abe (2001) analyzes the effect of the minimum wage on the wage earned by female part-time workers, using data for 1990 and 1995 from the Comprehensive Survey of Part-Time Workers in Japan. She finds that the minimum wage raises the wage of part-timers in regions where their wage is relatively low, that the effect of a minimum-wage increase is small in major urban areas where the wage of part timers is high, and that an increase in the wage of part timers is not as much as a minimum-wage increases in major urban areas. Abe and Tanaka (2007) also analyze the effect of the minimum wage on the wage of part-timers, using Japanese data for 1990 through 2001 from 2 The survey studies on the minimum wage include a study on the case of the United States by Asuyama (2006) and a study on the case of Europe by Ohashi (2007), as well as studies by Kawaguchi (2009) and Hori and Sakaguchi (2005). Parts of this paper are based on these studies.

7 the Basic Statistical Survey on the Wage Structure. They too find that the effect is observed in low-wage regions, but is weak in major urban areas. Dickens and Manning (2004) use UK data and find that a minimum-wage increase does not raise wages even if focus is put on the lower end of the wage distribution, which is different from the results of the analyses using data for Japan. As for research that uses Japanese data for the 2000s, Kambayashi, Kawaguchi, and Yamada (2008) perform a fixed-effect analysis to examine the effect of the minimum wage in keeping wages from falling during a recessionary period, using a prefecture-level dataset created from data for 1994 through 2003 from the Basic Survey on the Wage Structure. They find that the minimum wage kept the wages of low-wage workers from falling, although wages in general declined during the recessionary period from the mid-1990s through early 2000s. As these studies show, the effect of the minimum wage on the wage level varies for different regions and wage classes of workers, and the minimum wage raises wages or keeps wages from falling in regions or for workers that are likely to be affected by the minimum wage. 3.2 Effect of the Minimum Wage on Employment Losses Analyses of the effect of the minimum wage on employment losses have been conducted more actively than analyses of its effect on wage level. Card and Krueger (1995) conducted an influential study that sparked the interest of other researchers. Their analysis uses data collected from fast food restaurants located in states with a minimum-wage increase and neighboring states without a minimum-wage increase and takes advantage of the natural experimental setting. More specifically, they compare employment at two neighboring fast food restaurants on different sides of a state border for the periods before and after a minimum-wage increase. The pair of fast food restaurants faces the same market attributes and economic fluctuations, and the two restaurants differ only in terms of whether the minimum-wage was increased or not. This enables analysis of the pure effect of the minimum wage on employment. The results show that employment rose at restaurants with a minimum-wage increase and fell at restaurants without a minimum-wage increase. The existence of monopsony power is explained as being behind this result. Their study has been criticized, however. Pointing out that the data used in the study by Card and Krueger (1995) are based on a telephone survey and contain large observational errors, Neumark and Wascher (2000) created their own dataset for the same period based on payroll records and conducted an analysis. They find that the decline in employment was larger in states with a minimum-wage increase. Actually, before the study by Card and Krueger (1995) was published, Neumark and Wascher (1992) had conducted an analysis with state-level panel data and identified a negative effect of the minimum wage on employment. Neumark and Wascher (2007) also argue that long-term panel data should be used to identify fixed effects and year fixed effects and to consider time lags for the emergence of a negative effect on employment because a comparison of years immediately before and after a change in the minimum wage may not capture a negative effect on employment due to its time lag. With regard to research on Japan, Tachibanaki and Urakawa (2006) examine the effect of the minimum wage on the employment of females in their 20s with a cross-sectional analysis based on data from the 2002 Employment Status Survey. They find that there is no negative effect on

8 employment even if the minimum-wage is raised and the Kaitz index consequently rises. In contrast, several studies suggest the existence of a negative effect of the minimum wage on employment: Yugami (2005) identifies a positive correlation between the minimum wage and unemployment rate in prefecture-level data; Ariga (2007) shows that the job openings-toapplicants ratio for high school graduates is low in prefectures with a high minimum wage; and Kawaguchi and Mori (2009) find that employment of teenage males and middle-aged married females falls with a minimum-wage increase. These studies, however, are not based on panel data that Neumark and Wascher (2007) recommend be used and therefore do not consider fixed effects, year fixed effects, or time lags for a negative effect on employment. Studies using panel data for Japan include a study by Kawaguchi and Yamada (2007) that is based on individual-level data on females for the period from the Japanese Panel Survey of Consumers of the Institute for Research on Household Economics. They focus on workers earning wages that are close to the minimum wage and conduct an analysis assigning those who were affected by a minimum-wage increase to the treatment group and those who were not to the control group. The results show that those who were affected by the increase were less likely to keep their employment in the following period than those in the control group. In sum, the effect of the minimum wage on employment is negative for certain groups such as teenage males, middle-aged females, and high school graduates in several Japanese studies except a study by Tachibanaki and Urakawa (2006), whereas U.S. studies show mixed results. 3.3 The Significance of the Analysis in This Paper To summarize the results of earlier studies, a minimum-wage increase may raise wages or lead to employment losses in certain subgroups such as low-wage classes or regions. What kinds of effects result then if segment-specific measures are taken, such as raising minimum wages mainly for high-wage classes or regions and curbing the amount of minimum wage increases for low-wage classes or regions? Put differently, this question asks about the effect of the minimum wage policy after the system was changed in 2007, as discussed in Section 2. Data used in earlier studies are for periods prior to 2006 when the rates of minimum-wage increases were small and varied by only small amounts across regions. Therefore, the question has not been analyzed. Also, while many studies are designed to see the effect of minimum wages itself, few studies have examined the entire labor market, the low-wage labor market, and the high-wage labor market individually and analyzed the effect of the whole minimum-wage system by taking into account effects on these markets comprehensively. This paper therefore uses micro data for the period from a panel survey and analyzes in a multifaceted manner the effect of a minimum-wage revision on the wage and employment losses. Empirical studies on minimum wages in Japan have traditionally used prefecture-level aggregate data or cross-sectional micro data. The panel data used in this paper make it possible to see changes in the wage and employment situation for each individual as the survey follows the same individuals over several years. The use of this panel data therefore help us not only to understand whose wages are affected by a revision of the region-specific minimum wages, but also examine changes in hours worked and the effects on transitions between employment and

9 unemployment or non-employment. Taking advantage of the characteristics of the panel data, this paper comprehensively examines the effect of changes in minimum wages, which occurred both before and after the shift in the system, on people s wages, employment situation, and hours worked. 4. Data and Empirical Method The data used for estimation come from the Keio Household Panel Survey (KHPS). The survey is conducted annually with 4,005 males and females who were aged 20 to 69 at the time of the first survey (January 31, 2004). At the time of writing, the newest survey is the 2010 survey, and therefore seven years worth of data have been accumulated. All observations for the seven years are used in the analysis of this paper. 3 In the following, the data for 2004, 2005, and so on are denoted as KHPS 2004, KHPS 2005, and so on, respectively. Also, KHPS 2007 includes an additional sample of 1,419 survey respondents, which is also considered. The KHPS includes a wide range of variables such as those concerning the respondents education, employment, household structure, assets, residence, and health. In the first survey, each respondent s history of education and employment was recorded for each year from the year when the respondent was 18 years old through to the survey year. Therefore, the data can also be used as retrospective panel data. In addition, in the 2005 survey, data on each respondent s education or employment for ages 15 through 17 were recorded. As a result, by combining KHPS 2004 and KHPS 2005, we have data on each respondent s education and employment for each year from the time the respondent was 15 years old. If the respondent has a spouse, the same survey questions are also asked to the spouse. In this paper, the following four estimation models are considered in our analysis of how a minimum-wage revision affects the wage level of non-regular workers and their wage gap with regular workers, and whether a significant minimum-wage increase causes employment losses. 4.1 What Effect Does a Minimum-Wage Revision Have on Market Wage? To see the effect of a minimum-wage revision on the wage gap between non-regular workers and regular workers, a wage function for a worker is estimated. The analysis considers non-regular and regular workers (both males and females), the sample is divided into four categories, and estimation is conducted for each category. The goal with this estimation is to provide answers to the questions of (1) whether a minimum-wage increase raises wages of non-regular workers and regular workers and (2) whether a minimum-wage increase narrows the wage gap between nonregular workers and regular workers, making the wage distribution a more equalized distribution. 3 For details on the data, see Higuchi, Seko, and the Twenty-First Century COE Program at Keio University (2004).

10 Based on the estimation model used by Neumark et al. (2004), this paper uses the following estimation equation β γ (1) In the above equation, i denotes the individual, and t denotes the time of observation. The explained variable W it - W it-1 is the difference in the per-hour wage rate between t-1 and t. X' it-1 is a vector of individual attributes, including education dummies, years in the labor market and its square, tenure and its square, a temporary/contract worker dummy (used only for analysis of non-regular workers), the prefecture unemployment rate, industry dummies, occupation dummies, firm size dummies, urban size dummies, and year dummies. 5 Note that the time index of these variables is t-1. ΔMW it is the difference in the region-specific minimum wage between t-1 and t and is the variable of focus in the analysis with equation (1). The minimum wage variable has the individual index i because the minimum wage amount that individual i receives differs depending on the prefecture where the individual works. 6 This variable is used to examine the effect of a change in the region-specific minimum wage on a change in the wage of non-regular workers or regular workers. If the sign of the coefficient of the variable is positive, this can be interpreted as an increase in the region-specific minimum wage leading to a rise in the wage. Conversely, if the sign is negative, this can be interpreted as an increase in the region-specific minimum wage leading to a fall in the wage. Particular attention is paid to the sign of this coefficient after the estimation of the equation. Also, α i represents the time-invariant fixed effect and ε it is the disturbance term. Fixed-effect estimation and random-effect estimation, which take into account α i, are performed. For the purpose of comparison, pooled OLS estimation is also performed. In the case in which a minimum-wage increase led to employment loss, estimation of the wage function with a sample including only those who maintained their employment could result in overestimation of the effect of a minimum-wage increase. Therefore, the wage function is estimated using Heckman's two-step procedure. With the two-step procedure, it is possible to handle the sample selection bias resulting from employment losses caused by a minimum-wage increase. 4.2 Which Wage Class Is Affected by a Minimum-Wage Revision? 4 Using data for 1979 through 1997 from the Current Population Survey, Neumark et al. (2004) examine the effect of a minimum-wage increase on wages, hours worked, employment, and labor income. For each wage class of workers, they analyze how the rate of change of the minimum wage is associated with the rates of change of the wage, hours worked, employment, and labor income. In estimating the wage function for regular and non-regular workers, this paper analyzes the effect of a minimum-wage increase, using differences instead of rates of change. 5 The data on the prefecture unemployment rate come from the Labor Force Survey published by the Statistics Bureau of the Ministry of Internal Affairs and Communications. 6 The individual index i is used in the other estimations to show that the amount of the minimum wage varies depending on the prefecture where the worker is employed.

11 To determine which wage class a minimum-wage revision affects, another wage function is estimated using quantile regression. The analysis considers non-regular workers (both males and females). The goal with this estimation is to provide answers to the question of whether a minimum-wage increase has a particular impact on workers in a low wage quantile. The following are used in the estimation. y ε π Z θ γ MW ε (2) β R k N Min q y i X i β q i:y i X i βq q y i X i β q i:y i X i βq (3) N Equation (2) is the wage equation. The parameters for each wage quantile are estimated by solving optimization problem (3) which uses the wage equation. The subscript q denotes a wage quantile and takes a value between 0 and 1. ε is the disturbance term. The explained variable y in equation (2) is the log of the per-hour wage rate. Z is a vector of individual attributes, including education dummies, years in the labor market and its square, tenure and its square, a temporary/contract worker dummy, the prefecture unemployment rate, industry dummies, occupation dummies, firm size dummies, urban size dummies, and year dummies. ΔMW in equation (2) is the difference in the region-specific minimum wage between t-1 and t and is the variable of focus in the analysis with this equation. This variable is used to examine which wage quantile for non-regular workers is affected by an increase in the region-specific minimum wage. Tachibanaki and Urakawa (2006) show that a rise in the wage due to a minimum-wage increase is greater for lower quantiles. If a similar result is obtained from this estimation, it implies that a minimum-wage increase contributes to making the wage distribution a more equalized distribution and thus improves the employment condition for non-regular workers. The question of whether such a trend exists is examined with this estimation. Here, estimation is performed for the 20th, 40th, 60th, and 80th percentiles, and each estimation result is examined. 4.3 Does a Minimum-Wage Revision Lead to Employment Losses? The results of the first two estimation models are analyzed from the standpoint of what effect a minimum-wage increase has on the wage. However, not only the wage, but also employment is affected by a minimum-wage increase. In the labor market, if the minimum wage is raised above the equilibrium wage, it is possible that demand for labor decreases along the labor demand curve, and that involuntary unemployment arises as a result. Since a minimum-wage increase affects both the wage and employment, analyzing only one of them does not lead to a proper examination of the effect of the minimum wage on the labor market. To see whether a minimum-wage revision leads to firing of existing employees or to reduced new hiring, a job separation function and a new hire function are estimated using logit models.

12 Non-regular workers (both males and females) are considered in the analysis on employee firing; the unemployed/non-employed (both males and females) are considered in the analysis on reduced new hiring. The goal with this estimation is to clarify whether a minimum-wage increase leads to employment loss through firing of existing employees or reduced new hiring. The estimation involves equations (4) through (7). Both the job separation function and the new hire function are estimated using a logit model and random-effect logit model. 7 it 1 y β γ (4) Pr 1 Pr y it 0 Pr β γ F β γ (5) it 1 z δ γ (6) Pr 1 Pr z it 0 Pr δ γ F δ γ (7) In the above equations, i denotes the individual, and t denotes the time of observation. Equation (4) is the job separation function. The explained variable y it in equation (4) is a dummy variable which equals 1 if the individual is a non-regular worker at t-1 and is separated from his/her job at t, or 0 if he/she is a non-regular worker at both t-1 and t. 8 X it-1 in equation (4) is a vector of individual attributes, including education dummies, a spouse dummy, age, tenure, a temporary/contract worker dummy, the prefecture unemployment rate, industry dummies, occupation dummies, firm size dummies, urban size dummies, and year dummies. The time index of these variables is t-1. ΔMW it in equation (4) is the difference in the region-specific minimum wage between t-1 and t and is the variable of focus in the analysis with the job separation function. This variable is used to examine whether an increase in the region-specific minimum wage leads to job separation of non-regular workers. If the sign of the coefficient of the variable is positive, this can be interpreted as an increase in the region-specific minimum wage leading to a rise in the job separation rate of non-regular workers. Conversely, if the sign is negative, this can be interpreted 7 Estimation was also performed using the fixed-effect logit model according to Chamberlain (1980), but the estimates for the parameters could not be obtained due to a lack of variation in the explained variable. Therefore, as the second best method, a random-effect logit model, which takes into account unobservable individual attributes, is used. 8 The workers who are separated from their job due to retirement are excluded.

13 as an increase in the region-specific minimum wage leading to a fall in the job separation rate. Particular attention is paid to the sign of this coefficient after the estimation of the equation. Equation (6) is the new hire function. The explained variable z it in equation (6) is a dummy variable which equals 1 if the individual is unemployed or non-employed at t-1 and is hired as a non-regular worker at t, or 0 if the individual is unemployed or non-employed at both t-1 and t. R it-1 in equation (6) is a vector of individual attributes, including education dummies, age, total years in the labor market, the prefecture unemployment rate, a spouse dummy, urban size dummies, and year dummies. The time index of these variables is t-1. ΔMW it in equation (7) is the difference in the region-specific minimum wage between t-1 and t and is the variable of focus in the analysis with the new hire function. This variable is used to examine whether an increase in the region-specific minimum wage prevents employment of nonregular workers. If the sign of the coefficient of the variable is positive, this can be interpreted as an increase in the region-specific minimum wage leading to a rise in the employment probability of non-regular workers. Conversely, if the sign is negative, this can be interpreted as an increase in the region-specific minimum wage leading to a fall in the employment probability. Particular attention is paid to the sign of this coefficient after the estimation of the equation. 4.4 How Does a Minimum-Wage Revision Affect Hours Worked? The result of the third estimation model is analyzed from the standpoint of whether a minimum- wage increase leads to employment losses. The effect of a minimum-wage increase, however, can be observed in the form of not only job separation of workers and other types of reduced employment opportunities, but also a reduction in hours worked due to a fall in labor demand. It is considered that when the minimum wage is raised firms reduce their demand for labor and adjust their workforce because of an increase in cost. In such a case, firms can reduce the number of workers or the number of hours worked, and the choice between them depends on the relative size of the cost associated with each adjustment method. If reducing hours worked is less costly than reducing the number of workers, firms are considered to choose the former. Few studies exist in Japan that analyze this issue, although studies have been conducted by researchers overseas including Zavodny (2000), Couch and Wittenburg (2001), Neumark et al. (2004), and Stewart and Swaffield (2008). To see the effect of a minimum-wage revision on hours worked, an hours function for a worker is estimated using regression. The analysis considers non-regular workers (both males and females). The goal with this estimation is to clarify whether a minimum-wage increase leads to a reduction in hours worked. Based on the estimation model used by Neumark et al. (2004), this paper uses the following estimation equation. 1 1 β γ (8)

14 In the above equation, i denotes the individual, and t denotes the time of observation. The explained variable L it - L it-1 in equation (8) is the difference in the average weekly hours worked between t-1 and t. 9 The average weekly hours worked include overtime hours. X' it-1 in equation (8) is a vector of individual attributes, including education dummies, years in the labor market and its squared, tenure and its square, a temporary/contract worker dummy (used only for analysis of non-regular workers), the prefecture unemployment rate, industry dummies, occupation dummies, firm size dummies, urban size dummies, and year dummies. The time index of these variables is t-1. ΔMW it is the difference in the region-specific minimum wage between t-1 and t and is the variable of focus in the analysis with equation (8). This variable is used to examine the effect of a change in the region-specific minimum wage on a change in the average weekly hours worked by non-regular workers. If the sign of the coefficient of the variable is positive, this can be interpreted as an increase in the region-specific minimum wage leading to a rise in the average weekly hours worked. Conversely, if the sign is negative, this can be interpreted as an increase in the region-specific minimum wage leading to a fall in the average weekly hours worked. Attention is paid to the sign of this coefficient after the estimation of the equation. Also, α i in equation (8) represents the time-invariant fixed effect and ε it is the disturbance term. Fixed-effect estimation and random-effect estimation, which take into account α i, are performed. For the purpose of comparison, pooled OLS estimation is also performed. In addition, to handle the sample selection bias resulting from employment losses, Heckman's two-step estimation procedure is also used. The results for the above estimation models are analyzed in this paper. The analysis aims to clarify how a minimum-wage revision affects the wage level of non-regular workers and their wage gap with regular workers, and whether a large minimum-wage increase leads to employment losses. In passing, in each estimation, observations with residential relocation during the period considered are excluded. The exclusion is done in order to avoid a situation where the pure effect of a minimum-wage revision cannot be examined due to the possibility that inter-prefectural relocation causes a significant change in the worker's minimum wage. Descriptive statistics for the variables used in the estimation models are listed in Tables Emprical Results 5.1 How Does a Minimum-Wage Revision Affect Market Wage? Table 4 analyzes the effect of a minimum-wage increase on the wage of non-regular workers. With regard to the result for females, the coefficient of the change in the minimum wage is positive and significant, as seen in columns (1) (4) in Table 4. This result shows that a minimum-wage increase leads to a rise in the wage of female non-regular workers. It can be said 9 The average weekly hours worked is calculated based on the response to the following survey question: On average, how many hours do you work per week to earn income? (Please include overtime hours.)

15 then that a minimum-wage increase contributes to improving the employment conditions of female non-regular workers. In contrast, for males, the coefficient of the change in the minimum wage is not statistically significant, as seen in columns (5) (8) in Table 4. This result shows that even if the minimum wage is raised, the wage of male non-regular workers does not rise. This difference between males and females in terms of the impact of a minimum-wage increase can be attributed to a difference between males and females in terms of the proportion of workers earning a wage close to the minimum wage. Tachibanaki and Urakawa (2006) point out that the number of workers whose wage is less than or close to the minimum wage is greater for females than for males in a statistically significant manner. Therefore, females are more likely to be affected by a minimum-wage increase than males. Based on the above analysis, it can be said that a minimum-wage increase leads to a rise in the wage of female non-regular workers. Now, let us consider the effect of a minimum-wage increase on the wage of regular workers. Since the wage level of regular workers is higher than the minimum wage, the effect of a minimum-wage increase is considered to be limited or unobservable. If a minimum-wage increase has no effect on the wage of regular workers, a minimum-wage increase potentially contributes to reducing the wage gap between regular and non-regular workers and making the wage distribution a more equalized distribution because a minimum-wage increase raises the wage of non-regular workers. This possibility is examined with a estimation of the wage function for regular workers. Table 5 analyzes the effect of a minimum-wage increase on the wage of regular workers. For females, the coefficient of the change in the minimum wage is not statistically significant, as seen in columns (9) (11) in Table 5. This result shows that even if the minimum wage is raised, the wage of female regular workers does not increase. As for males, the coefficient of the change in the minimum wage is also not statistically significant, as seen in columns (12) (14) in Table 5. This result shows that even if the minimum wage is raised, the wage of male regular workers does not rise. Based on the above analysis, it can be said that a minimum-wage increase does not affect the wage of male and female regular workers. It is considered that the wage of regular workers is not influenced by a minimum-wage increase because regular workers wage is higher than the minimum wage. Combining this result and the result for non-regular workers, a minimum-wage increase potentially contributes to reducing the wage gap between female regular workers and female non-regular workers and making the wage distribution a more equalized distribution because while a minimum-wage increase raises the wage of female non-regular workers, it does not affect the wage of regular workers. In the case of males, a minimum-wage increase does not contribute to making the wage distribution a more equalized one because it does not influence the wage of regular and non-regular workers. In sum, the following two points are revealed by analyzing the effect of a minimum-wage revision on the wage with estimation of the wage function for non-regular and regular workers. First, a minimum-wage increase raises only the wage of female non-regular workers and does not affect the wage of female and male regular workers and male non-regular workers. Second, a minimum-wage increase reduces the wage gap between female regular workers and female non-

16 regular workers and makes the wage distribution a more equalized distribution; however, there is no such tendency in the case of males. Given these results, it can be said that a minimum-wage increase improves the employment condition of female non-regular workers. 5.2 Which Wage Class Is Affected by a Minimum-Wage Revision? Table 6 analyzes the effect of a minimum-wage increase on different wage quantiles for nonregular workers. For females, the coefficient of the change in the minimum wage is positive and significant, as seen in columns (15) (18) in Table 6. This result shows that a minimum-wage increase leads to a rise in the 20th, 40th, 60th, and 80th percentile wages for female low-wage workers. As for the size of the coefficient, the coefficient of the change in the minimum wage and the coefficient of the rate of change in the minimum wage are largest for the 20th percentile, which shows that a minimum-wage increase particularly affects workers in a low wage percentile. Based on this result, it can be said that, in the case of females, a minimum-wage increase leads to a rise in low percentile wages especially and contributes to making the wage distribution a more equalized distribution. As to the result for males, the coefficient of the change in the minimum wage is positive and significant in columns (19) (21) in Table 6, but is not statistically significant in column (22). This result shows that a minimum-wage increase leads to a rise in the 20th, 40th, and 60th percentile wages for male low-wage workers. As for the size of the coefficient, similarly to the case of females, the coefficient of the change in the minimum wage and the coefficient of the rate of change in the minimum wage are largest for the 20th percentile, showing that a minimumwage increase particularly affects workers in a low wage percentile. It can be said therefore that, in the case of males, a minimum-wage increase leads to a rise in low to middle percentile wages (as seen in the 20th, 40th, and 60th percentile wages) and contributes to making the wage distribution a more equalized distribution. Taking the results in Table 4 into account, we can see that although a minimum-wage increase does not raise the wage of all non-regular workers, it improves the employment conditions of male non-regular workers through a rise in low to middle percentile wages. In sum, with the estimation of the wage function for non-regular workers using quantile regression, the following two points are revealed by analyzing which wage class is affected by a minimum-wage revision. First, a minimum-wage increase raises the quantile wages for female non-regular workers, particularly the wage of workers in a low wage percentile. A minimumwage increase is thus considered to contribute to making the distribution of the wage of female non-regular workers a more equalized distribution. Second, a minimum-wage increase raises low to middle percentile wages for male non-regular workers, especially the wage of workers in a low wage percentile. A minimum-wage increase is therefore considered to contribute to making the distribution of the wage of male workers a more equalized distribution. 5.3 Does a Minimum-Wage Revision Lead to Employment Losses?

17 Table 7 analyzes the effect of a minimum-wage increase on employment loss for non-regular workers. For females, the coefficient of the change in the minimum wage is not statistically significant, as seen in columns (23) (26) in Table 7. This result shows that a minimum-wage increase does not lead to a rise in the probability of job separation for female non-regular workers. Thus, a minimum-wage increase does not cause employment losses for female nonregular workers. As for males, the coefficient of the change in the minimum wage is not statistically significant, as seen in columns (27) (30) in Table 7. This shows that a minimumwage increase does not lead to a rise in the probability of job separation for male non-regular workers. That is, a minimum-wage increase does not cause employment losses for male nonregular workers. Based on these results, it appears that a minimum-wage increase does not cause employment losses for non-regular workers. How should this be interpreted? On the one hand, the Japanese labor market can be seen as monopsonistic rather than perfectly competitive. Card and Krueger (1995) argue that when employers have monopsony power in an imperfectly competitive labor market, a minimum-wage increase does not necessarily cause employment losses. This is consistent with the estimation result of this paper. However, Ohashi (2009) points out several issues with this monopsonistic labor market model, raising questions about its theoretical basis. 10 On the other hand, it can be considered that although employment losses occur due to a minimum-wage increase as Japan's labor market is competitive, the negative effect of the minimum-wage increase on employment is offset by an increase in the value of the marginal product of labor and thus in labor demand resulting from government support for productivity improvement at small and medium-sized firms. If the government support is effective, the employment-reducing effect of a minimum-wage increase is considered to be offset. It is possible that the estimation result reflects this scenario. Further examination is needed to determine which of the two interpretations is appropriate as it is not possible to make a judgment based on the estimation results obtained in this paper. This is a question to be analyzed in future research. Let us now examine the estimation results for the new hire function. Table 8 analyzes the effect of a minimum-wage increase on new hiring of the non-employed/unemployed as non-regular workers. For females, the coefficient of the change in the minimum wage is not statistically significant, as seen in columns (31) (32) in Table 8. So, for females, a minimum-wage increase does not lead to a rise in the probability of being newly hired as a non-regular worker, and thus a minimum-wage increase does not hinder females employment as non-regular workers. As for males, the coefficient of the change in the minimum wage is also not statistically significant, as seen in columns (33) (34) in Table 8. Thus, for males, a minimum-wage increase does not lead to a rise in the probability of being newly hired as a non-regular worker. Similarly to the case of 10 Ohashi (2009) raises three issues: (1) since firms with monopsonistic power in a regional labor market are usually large and offer wages that are generally higher than wages at small and medium-sized firms, the effect of the minimum wage can be limited; (2) types of employment are diverse even at medium, small, and micro-sized firms, and not all their workers earn wages close to the minimum wage; (3) a positive effect of the minimum wage on employment increases with the wage elasticity of labor supply, but, according to Cahuc and Zylberberg (2004), the effect is not large on average in cases of Europe and the United States.

18 females, a minimum-wage increase does not hinder males employment as non-regular workers. 11 The above results show that, for both males and females, a minimum-wage increase does not hinder their employment. Although a minimum-wage increase is considered to curb new employment of non-regular workers as it increases the cost of hiring them, it turns out that this is not the case. In sum, with the estimation of the job separation function and the new hire function based on a logit model, the following two points are revealed by analyzing whether a minimum-wage revision leads to firing of existing workers or hindering of new employment. First, a minimumwage increase does not affect employment losses of female and male non-regular workers. Second, a minimum-wage increase does not hinder employment of males and females as nonregular workers. 5.4 What Effect Does a Minimum-Wage Revision Have on Hours Worked? Table 9 analyzes the effect of a minimum-wage increase on hours worked of non-regular workers. For females, the coefficient of the change in the minimum wage is not statistically significant, as seen in columns (35) (38) in Table 9. A minimum-wage increase does not affect the average weekly hours worked by female non-regular workers, and thus a minimum-wage increase does not reduce the hours worked by female non-regular workers. As for males, the coefficient of the change in the minimum wage is also not statistically significant, as seen in columns (39) (42) in Table 9. This result shows that a minimum-wage increase does not influence the average weekly hours worked by male non-regular workers. Similarly to the case of females, a minimum-wage increase does not reduce the hours worked by male non-regular workers. With the estimation of the hours function using regression, the following two points are revealed by analyzing the effect of a minimum-wage revision on hours worked. First, a minimum-wage increase does not affect the average weekly hours worked by female non-regular workers. Second and similarly, a minimum-wage increase does not affect the average weekly hours worked by male non-regular workers. The above four estimation models are analyzed to reveal the effect of a minimum-wage revision on the wage and employment. Based on the results, it can be said that a minimum-wage increase contributes to improving the employment conditions of non-regular workers and to making the wage distribution a more equalized distribution through raising their low percentile wages, without causing employment losses. The change in hours worked is also analyzed in this paper, but no tendency for a minimum-wage increase to reduce hours worked is identified. Although employment losses caused by a minimum-wage increase have been emphasized by some 11 Based on a survey conducted by the Japan Institute for Labor Policy and Training, Sakamoto (2009) points out that many enterprises in Japan do not correctly recognize the region-specific minimum wage itself, and that the proportion of enterprises that have refrained from hiring new workers due to an increase in the region-specific minimum wage is extremely low.

19 previous research, the estimation result does not show the presence of such an effect and reveals improvement of the employment conditions of workers through a rise in their wage. 6. Policy Implications and Issues for Future Research Since 2007, region-specific minimum wages in Japan have been raised by much larger amounts than before mainly in major urban areas which have been economically vibrant. Taking this change as a change in a control variable in a quasi-experiment and using panel data that follows individuals behavior during the relevant period, this paper has examined what kind of impact a minimum-wage increase has on individuals wages and employment situation. The following conclusions are reached as a result. (1) A minimum-wage increase had a significant negative effect on the size of the wage gap: with a minimum-wage increase, the wages of female nonregular workers rose, the income gap with regular workers narrowed, and, as found in an analysis of different wage classes, the 20th percentile wage increased substantially. (2) Although a minimum-wage increase had no significant effect on the wages of male non-regular workers and regular workers, an analysis of the different wage classes show that the 20th, 40th, and 60th percentile wages increased significantly with a minimum-wage increase. (3) A minimum-wage increase led neither to loss of employment among existing female and male non-regular workers nor to reduced employment among female and male job seekers. (4) A minimum-wage increase was not associated with a reduction in the average weekly hours worked by existing female and male non-regular workers. Needless to say, the above results are obtained for the range of minimum-wage increases observed in the prefectures for the period considered in this study (i.e ). For a minimum-wage increase that goes beyond this range, there is no guarantee that the same results are obtained, or that employment does not fall. To what extent can the minimum wage be raised without reducing employment? This is one of the questions to be examined in future research. It is also necessary to investigate the reasons behind the results of this paper, or, put another way, the kinds of conditions that led to the results. There are two main reasons why employment does not fall even if the wage rises. In one case, a small wage increase does not lead to a decrease in employment because the wage is kept low relative to labor productivity with the employer s monopsony power in an imperfectly competitive labor market. In the other case, an increase in the wage does not result in reduced employment as the wage increase is accompanied by improvement in labor productivity. In considering future policies, it is important to know which case is more applicable to the little effect on employment observed in this paper, and discerning it is another challenge left to future research.

20 References Abe, Y. (2001). Effects of the Region-Specific Minimum Wage on Part-Timer Wages. In T. Inoki and F. Ohtake (Eds.), Economic Analysis of Employment Policy (pp ). Tokyo: University of Tokyo Press. Abe, Y. and Tamada, K. (2009). Thoughts on Regional Differences in Minimum Wage and Public Welfare Assistance. Japanese Journal of Labor Studies, 593, December. Abe, Y. and Tanaka, A. (2007). Wage Gap between Regular Employees and Part-Timers and the Role of the Region-Specific Minimum Wage ( ). Japanese Journal of Labor Studies, 563, June. Asuyama. Y. (2006). Debate in the Unites States on Raising the Minimum Wage. Institute of Developing Economies Country Report ( Cahuc, P. and Zylberberg, A. (2004) Labor Economics, The MIT Press. Card, D. (1992). Using Regional Variation in Wages to Measure the Effects of the Federal Minimum Wage Industrial and Labor Relations Review.Vol.46, No.1, pp Card, D. and Krueger, A. (1995). Myth and Measurement. Princeton University Press. Chamberlain, G (1980). Analysis of Covariance with Qualitative Data, The Review of Economic Studies, Vol. 47, No. 1. Claudio E. Montenegro and Carmen Pagés (2008) Who Benefits from Labor Market Regulations? Chile, Law and Employment: Lessons from Latin American and the Caribbean, James J. Heckman and Carmen Pagés, University of Chicago Press, ISBN: Couch, K.A. and Wittenburg, D.C. (2001) The response of hours of work to increases in the minimum wage. Southern Economic Journal, 68(1): Dickens, R. and Manning, A. (2004) Spikes and Spill-overs: The impact of the national minimum wage on the wage distribution in a low-wage sector The economic Journal, Vol. 114, No. 494, Conference Papers (Mar., 2004), pp. C95-C101. Hori, H. and Sakaguchi, N. (2005). Economic Analysis of the Minimum Wage in Japan. JILPT Research Report, 44. Japan Institute for Labor Policy and Training (2011). Theory and Analysis on the Effect of a Minimum-Wage Increase on Employment. JILPT Document Series, 90. Kambayashi, R., Kawaguchi, D. and Yamada, K. (2008). Minimum Wage in Deflationary Economy:Experience of Japan, mimeograph Kawaguchi, D. (2009). Minimum Wage and Employment. In I. Ohashi (Ed.), Economics of Labor Demand (Ch. 8). Kyoto: Minerva Shobo. Kawaguchi, D. and Mori, Y. (2009). Attributes of the Workers Earning Minimum Wages and the Effect of a Minimum-Wage Increase on Employment. Japanese Journal of Labor Studies, 593, December. Kawaguchi, D. and Mori, Y. (2009). Is Minimum Wage an Effective Anti-PovertyPolicy in Japan? RIETI Discussion Paper Series 09-E-032. Kawaguchi, D. and Yamada, K. (2007). The impact of minimum wage on female employment in Japan, Contemporary Economic Policy, Vol.25, No.1, pp , Neumark, D. and Wascher, W. (1992). Evidence on employment effects of minimum and subminimum wage: Panel data on state minimum laws. Industrial and Labor Relations

21 Neumark, D, and Wascher, W. (2000) The Effect of New Jersey s Minimum Wage Increase on Fast-Food Employment: A Reevaluation Using Payroll Records. American Economic Review. Vol. 90, No. 5 (December), pp Neumark, D. and Wascher, W. (2004). Minimum Wages, Labor Market Institutions, and Youth Employment: A Cross-National Analysis. Industrial and Labor Relations Review, Vol. 57, No. 2, pp Neumark, D. and Wascher, W. (2007). Minimum Wages and Employment. IZA DP No Neumark, D, Schweitzer, M. and Wascher, W. (2004). Minimum Wage Effects throughout the Wage Distribution. Journal of Human Resources, Vol. 39, No.2 (Spring), pp Ohashi, I. (2009). Japan's Minimum Wage System with Reference to Actual Situations and Debates in Europe and the United States. Japanese Journal of Labor Studies, 593, December. Ohashi, I. (2007). EU Labor Market and the Minimum Wage. Expansion and Deepening of the EU: Issues Arising after the Currency Unification (Report of the Research Group on Issues Facing the Expanding EU; Ch. 4). Japan Center for Economic Research. Sakaguchi, N. (2009). Minimum Wager from Companies' Perspective. Japanese Journal of Labor Studies, 593, December. Stewart, M. B. (2002). Estimating the impact of the minimum wage using geographical wage variation. Oxford Bulletin of Economics and Statistics, 64, Stewart, M. B. and Swaffield, J. K. (2008). The Other Margin: Do Minimum Wages Cause Working Hours Adjustments for Low-Wage Workers? Economica, 75, Tachibanaki, T. and Urakawa, K. (2007). Empirical Analysis on Poverty and Labor in Japan. Japanese Journal of Labor Studies, 563, June. Tachibanaki, T. and Urakawa, K. (2006). Research on Poverty in Japan. Tokyo: University of Tokyo Press. Tamada, K. (2009). How is the Minimum Wage Determined? Japanese Journal of Labor Studies, 593, December. White, H. (1980) A Heteroskedasticity-consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity. Econometrica, Vol. 48, pp Yugami, K. (2005) Analysis of Regional Labor Markets with Prefecture-Level Data: Examination of Regional Differences in Unemployment and Non-employment. Japanese Journal of Labor Studies, 539, April. Zavodny, M. (2000). The effect of the minimum wage on employment and hours. Labour Economics, 7,

22 Figure 1: Comparison of the Kaitz Indices for Different Countries Ratio of the National Minimum Wage to the Average Wage Ireland Luxemburg France Greece Belgium Netherlands Malta Slovenia Portugal Cyprus Slovakia UK Bulgaria Latvia Lithuania Japan Czech Brazil Poland Hungary Estonia Spain USA Romania Note: The data for 2002 are used for Belgium, Luxemburg, the Netherlands, and Portugal; 2003 for Brazil and France; 2005 for the U.S.A; 2004 for the other countries. Source: European Foundation for the Improvement of Living and Working Conditions* (2006) "Key themes in global industrial relations: Minimum wages and relocation of production" (* An independent agency of the European Commission)