Impact of Minimum Wages on Wages, Wage Spillovers and Employment in China

Info: 18559 words (74 pages) Dissertation
Published: 18th May 2020

Share this: Facebook Twitter Reddit LinkedIn WhatsApp

The Impact of Minimum Wages on Wages, Wage Spillovers and Employment in China: Evidence from Longitudinal

Individual-Level Data

Abstract

We utilize the substantial variation in both the magnitude and frequency of minimum wage changes that have occurred in China since its new minimum wage regulations in 2004 to estimate their impact on wages, wage spillovers and employment. We use county-level minimum wage data merged with individual-level longitudinal data from the Urban Household Survey for the period 2004-2009, spanning the period when the new minimum wage regulations were in place. Our results indicate that minimum wage increases raise the wages of otherwise low-wage workers by a little less than half (41%) of the minimum wage increases. Depending upon the specification, these wage effects also lead to a 2 to 4 percentage point reduction in the probability of being employed, with a 2.8 percentage point reduction being our preferred estimate. We also find statistically significant but very small wage spillovers for those whose wages are just above the new minimum wage, but they are effectively zero for those higher up in the wage distribution.

Introduction

There is a growing literature on the impact of minimum wages in China, most of which focuses on the impact on employment. Recent examples, that also review that literature, include Long and Yang (2016) and Wang and Gunderson (2018) as well as studies written in Chinese including Guo and Zhang (2018) and Zhang and Yang (2016).

In contrast to most previous studies that use aggregate level data to examine the minimum wage effects in China,[1] our study uses nationally representative data at the individual level and provides evidence on the impact of minimum wages in China on wages, wage spillovers and ultimately on employment. The main contributions of the paper are: (1) Our empirical analysis spans the years 2004 to 2009[2] so we are able to take advantage of the substantial variation in both the magnitude and frequency of minimum wage changes that have occurred in China since their new minimum wage regulations in 2004. (2) Our data is longitudinal at the individual level so we are able to use individual and year fixed-effect panel estimation procedures to better control for unobservable factors that can otherwise contaminate the estimates of minimum wage impacts. (3) We merge county-level minimum wage data (which is the level where minimum wages tend to be set) with individual-level data (which is the level where wage and employment effects occur). (4) The individual-level data enables us to use both the “at-risk” and wage gap methodologies (outlined subsequently) for estimating wage and employment effects. (5) We estimate both the wage and employment effects of minimum wage increases, which provides corroborating evidence since positive wage effects and negative employment effects would go hand-in-hand. (6) We estimate wage spillovers to see if minimum wages have ripple effects on the wages of those who are above the new minimum wage and who may be indirectly affected.

Theoretical Expectations of Minimum Wage Effects on Wages, Employment and Spillovers

Basic economic theory predicts that minimum wage increases will reduce employment as firms substitute other inputs for the workers whose wages have increased, and as firms reduce their output in response to the higher costs. Such conventional adjustments may not occur in situations of monopsony (Manning 2003; Bhaskar, et al., 2002; Dong and Putterman 2000, 2002) or when there are cost offsets from such factors as management being shocked into more efficient practicesor because of improvements in employee commitment and loyalty to the firm and reductions in turnover (Cooke, 2005; Hirsch, Kaufman and Zelenska 2005; Metcalf, 2008). Cost offsets can also occur through firms cutting back on fringe benefits and non-wage components of compensation (Long and Yang, 2016; Metcalf, 2008; and Wang and Gunderson, 2012).

Basic economic theory also highlights the complementarity between wage and employment outcomes. If wages do not increase (perhaps because of non-compliance or minimum wages being a non-binding constraint) then there also should be no adverse employment effect.

The theoretical literature also has implications for wage spillover effects, especially for workers just above the minimum wage. Campolieti (2015) provides a thorough review of the literature on such spillover effects and highlights the mechanisms through which they can work. Positive wage spillovers can occur if employers substitute some higher-paid employees for the now more costly low-wage workers affected by the minimum wage, and they may raise wages of those above the minimum wage to restore former wage relativities, as emphasized in the industrial relations literature. In contrast, if such higher wage workers are complements in production to the lower wage minimum wage workers, then any reduction in the employment of minimum wage workers would also lead to reductions in the employment and wages of workers above the minimum wage.Clearly, the spillover effect from minimum wage increases onto the wages of those above the minimum wage is ultimately an empirical proposition, although it likely would be to increase those wages. The effect should clearly be largest for those just above the new minimum wage and then declining rapidly with no substantial effect on higher wage groups.

Minimum Wage Legislation in China

The evolution of minimum wages in China is well documented in studies such as Fang and Lin (2015), Long and Yang (2016), Wang and Gunderson (2012, 2018) and Xing and Xu (2015). For the purpose of our study, the main changes involved “The Minimum Wage Regulations” that came into effect on March 1, 2004. The regulations became much more stringent and coverage was extended to part-time workers and to those in towns and villages, in state-owned enterprises, private enterprises, private non-enterprise units, as well as employees in self-employed businesses. The new laws set up a monthly minimum wage and an hourly minimum wage separately for fulltime and non-fulltime workers respectively.

Table 1 shows the variation of minimum wages by presenting average real minimum wages for the 31 provinces and autonomous regions across the country over the 2004-2009 period. Importantly for research purposes, the changes in minimum wages and their magnitudes both over time and across counties provide substantial variations from which to identify minimum wage effects. Specifically, over the period 2004-2009 of our survey data for 16 provinces, there were 124 minimum wage increases at the county level, with mean real minimum wages rising by 63% from 346 RMB to 563 RMB (converted to 2009 units using the urban CPI). Such variation is crucial for estimating the impact of policies such as minimum wages. In his presidential address to the Society of Labor Economists, Hamermesh (2002) called for more international evidence on the impact of policy initiatives and he specifically singled out minimum wage legislation as benefiting from evidence from countries where there is considerable variation across jurisdictions and over time. China post 2004 certainly fits that bill.

Data

Our empirical analysis is based on individual-level longitudinal survey data from the Urban Household Survey (UHS) for the years 2004-2009, covering the period when the new minimum wage regulations were put in place in 2004.[3] The UHS is a continuous, large-scale social-economic survey conducted by China’s National Bureau of Statistics (NBS) aiming to study the conditions and standard of living of urban households, which include agricultural and non-agricultural residents or non-residents who live in the urban areas for at least six months.[4]

Note that the individual-level data is an unbalanced panel. The UHS handbook (Wei, 2006) indicates that the sample households are followed and replaced after three years; in practice, however, local authorities have the flexibility to retain some households above the term and often less than the indicated three years. We use several individual characteristics (gender, age, educational attainment, year when an individual began to work, and length of stay in the current city) along with household ID numbers, to carefully match with the same individuals over time, and then create a longitudinal identifier for all observations. We report the panel structure of the data in Table 2. We then merge-in the county-level data on minimum wages (over 2000 counties each year at the 6-digit code) as well as variables to control for economic conditions such as GDP per capita and the CPI that can affect wage and employment outcomes. We also include a measure of city-level foreign direct investment (FDI) to control for possible omitted variable bias in that such investment can be correlated with minimum wages (cities may restrain minimum wages to attract FDI) as well as wage and employment outcomes.

Most minimum wage studies for China use aggregate level data. Our use of individual-level data, however, involves the level of aggregation where wage and employment decisions in response to minimum wage changes are actually made. It also enables estimating spillover effects on wages just above the minimum wage. The longitudinal aspect of the individual-level data enables following the same individual over time and hence controls for unobserved heterogeneity that is fixed within the individual as well as over time. We use minimum wage data at the county-level rather than the provincial-level data that is conventionally used in China. This allows for a more accurate measure of the minimum wage at the level where it is set, and it enables controlling for local labor market conditions. The large number of minimum wage changes at the county-level also provides more variation in the “treatment” to help identify the effect of minimum wages. Our minimum wage data for each county was compiled by carefully recording the minimum wage data from every local government website for approximately 2,300 counties every year from 1994 to 2012 (we used only the 2004-2009 data in the empirical analysis in order to match with the survey data after the 2004 minimum wage reforms and prior to the most recent data available to us, 2009).

Enforcement of labour laws is often regarded as weak in China (Deng and Li 2012; Rawski, 2006).[5] Fang and Lin (2015), however, provide evidence that enforcement of minimum wage laws has increased over time, especially after the 2004 reforms. As such, we may expect wage and employment effects in the 2004-2009 period of our data.

As discussed, minimum wage information at the county level is important given that minimum wages are effectively set at that level of aggregation, and they can vary by counties within the same province, even for geographically contiguous neighbors within the same province. To address the potential issue that counties in a province within the same year may experience different adjustment dates of the minimum wage, we use the time-weighted method as in Rama (2001) to obtain the mean minimum wage.[6] For our empirical analysis, the minimum wage data is then merged into the 16-province UHS data over the 2004-2009 period, with individuals matched to their county-level minimum wage. The minimum wages and individual wages are adjusted for inflation and converted into 2009 RMB using the urban resident CPI.

Table 2 gives the descriptive statistics for our data for those who are “at-risk” and “not-at-risk” of being affected by a minimum wage increase in that their wages fall between their old and new minimum wages. The ratios of the at-risk and not-at-risk workers to total employment are 0.022 and 0.978, respectively. For our preferred control group (MinWage + 80 RMB) specification, the ratio of at-risk workers to total workers including those in that comparison group (at-risk and not-at-risk) is 0.512 (i.e., the at-risk treatment group and the preferred comparison groups are of about equal size). The at-risk individuals and not-at-risk individuals are quite similar, respectively, in age (41, 42) Han Ethnicity (98%, 97%), Local Hukou (97%, 97%), years of residence (33.7, 31.7), and years of work experience (23.8, 23.2). Not surprisingly, not-at-risk individuals are more likely to be male (55%, 38%), married (90%, 85%), to have more years of schooling (12.7, 11.2), and earn significantly higher monthly wages (1781, 489), than at-risk individuals. The average monthly minimum wage over the period was 544 RMB, and the average monthly wage for those at risk in that their wage fell between the old and the new minimum wage was 489 RMB. The average wage gap, defined as the difference between the individual’s monthly wage and the monthly minimum wage was 55 RMB. As indicated previously, a comparison with the 2005 Census indicates that our data is fairly representative of the workforce in those provinces.

Estimating Equations

Wage effect equations

The effects of minimum wage increases on the wages of those who should be affected by such increases are of importance in determining if minimum wage increases have their intended effect of increasing the wages of low-wage workers. As well, if there is no effect on the wages of those who should be affected, then this suggests that enforcement is lax or that the minimum wages are a non-binding constraint in that they do not “bite” into the wage distribution.

We use fixed-effects panel regressions at the individual level to estimate the effect of minimum wages to see whether changes in the minimum wage affect the observed wages of the at-risk individuals whose wages should be directly affected by minimum wage increases and who remain employed after the minimum wage increase. Minimum wage studies that have used variants of the “at risk” methodology include: Ashenfelter and Card (1981), Campolieti, Fang and Gunderson (2005), Currie and Fallick (1996), Draca, Machin and VanReenen (2011), Egge, Kohen, Shea, and Zeller (1970), Fang and Gunderson (2009), Linneman (1982), Yuen (2003) and Zavodny (2000).

An individual is at-risk (i.e., bound by the change) if he/she was working at a wage between the old () and the new minimum (); that is, . Our wage equation is:

(1)

where is the log of the wage for individual i in county j in year t; is the log of the minimum wage (in levels) of individual i received in county j he/she works in year t; if the monthly wage of worker i from county j in year was between the old and the new minimum wages when there is an increase in the minimum wage in county j of year t (treatment group); otherwise, if the worker is not affected by a minimum wage increase (control group)[7]; For control groups, we use MinWage + XXX (current minimum wage plus XXX Chinese dollars RMB) to mimic the counterfactual increase for these groups who do not experience a minimum wage increase, and XXX is between 10 and 200 RMB; is a set of individual characteristics that exhibit within variation in our data such as years of schooling, marital status, work experience, work experience squared, occupation, and industry; is a set of individual fixed-effects; is the city specific linear time trend; and is a set of year fixed-effects; is the error term.

The treatment group consists of low-wage workers (defined by those whose wages are bound by the old and new minimum wages) in the counties where minimum wage changes took place. The control or comparison group includes those low-wage workers in the counties with no minimum wage changes, whose wages are bound by the current minimum wages and current minimum wages plus the hypothetical minimum wage change, which was measured as the average minimum wage changes in the counties where such changes actually occurred. As such, the coefficient for the interaction terms between and captures the counterfactual for the wage changes of those who were affected by minimum wages changes relative to those who were not affected in the provinces without minimum wage changes but otherwise would have been affected if similar minimum wage increases were introduced.

Wage spillover equation

As discussed previously, examining the spillover effects on the wages of others in the wage distribution whose wages may be indirectly affected by minimum wage increases is important to determine if minimum wage increases also give rise to an indirect ripple effect beyond those who are directly affected by minimum wage increases. A large ripple effect could exacerbate the cost increases for employers, and it could mean that the wage effects are spilling over into higher wage groups.

The spillover effects are measured by wage changes for those whose wages were slightly above the new minimum wages in those jurisdictions that increased their minimum wage. As such, the coefficient for the interaction terms between and captures the counterfactual for the wage changes of those whose wages are slightly above current minimum wages as the result of minimum wage changes (treatment group) relative to those who were not affected in provinces without minimum wage changes but otherwise would have been affected if similar minimum wage increases were introduced (control group).

We estimate the wage spillover effects as:

(2)

where is the log of the wage for individual i in county j in year t; is the log of minimum wage levels in county j for worker i in year t; is a set of dummy variables equal 1 if the wage of worker i in county j falls into the range and is in a jurisdiction that experiences a minimum wage increase (treatment group) and equal 0 if a worker’s wage is in the same range but is in a jurisdiction that does not experience a minimum wage increase (control group); XXX is between 0 and 200 RMB and with increments of 20 RMB. For example, the first spillover category (XXX is zero) is for those whose wage falls between the new minimum wage and 20 RMB above the minimum wage; the second spillover category (XXX is 20) is for those whose wage falls between 20 RMB above the new minimum wage and 40 RMB above, and so forth. is a set of individual characteristics as indicated before; is a set of county fixed- effects; is the city-specific linear time trend; is a set of year fixed-effects, and is the error term. We estimate the model using OLS, separately for each of the 10 wage spillover categories, indicated in Table 4.

Employment effect equations

We use individual wages and employment status to identify those employed workers who were most likely to be directly affected (“bound”/ at risk) by the changes in the minimum wage in those years.[8] We then examine whether these individuals have a lower probability of being employed one year later. As stated in Currie and Fallick (1996), this procedure does not introduce any selection bias into our estimates since initially employed individuals constitute the appropriate population for estimating the effect of the minimum wage on the transition out of employment.

As conventional in this literature, we use a Linear Probability Model to estimate the effect of minimum wage changes on the probability of remaining employment in the subsequent period using (1) an at-risk methodology and (2) a wage gap methodology.[9] The at-risk methodology estimates the probability of a worker being employed in year conditional on the worker being employed in year . The wage gap methodology uses the magnitude of the increase in the worker’s wage necessary to bring it to the new minimum wage as the key independent variable. The gap methodology exploits the considerable variation in the magnitude of the wage adjustment in the data. It also controls for within-group heterogeneity arising from workers who may have a large gap or wage adjustment necessary to bring their wage up to the new minimum wage, disproportionately in which unobserved characteristics may lead to employment instability. This is so because the larger the gap, the lower the initial wages, and lower employment stability of the low-wage workers. In contrast, a larger wage gap also implies that employers need to pay more to bring the wages up to the new minimum wages; such high costs of adjustments thus increase the probabilities of layoffs. The at-risk and wage gap methodology as empirical equations are shown below, respectively.

(3)

(4)

where if the worker i in county j, who was employed in year , was employed in year t, and otherwise. if the monthly wage of worker i from county j in year was between the old and the new minimum wages when there is an increase in the minimum wage in county j of year t; otherwise, if the worker is not affected by a minimum wage increase. The AtRisk variable captures the effects of the minimum wage increase on the conditional probability of employment for the at-risk group. X is the vector of individual characteristics as before.

is defined as the difference between the minimum wage in year t and the worker i’s wage from county j in year if he/she is affected by the minimum wage change in year t; for workers in the control group counties, the Wagegap is the hypothetical wage adjustment that would be necessary to bring the worker up to the minimum wage bound for that particular control group. As stated in Campolieti, Fang and Gunderson (2005), the gap adjustment for both treatment and control groups essentially controls for within-group heterogeneity for both groups. The coefficients from the wage gap methodology are converted to employment effects by multiplying them by the average minimum wage increase of 78 RMB.

To address the issue of unobserved heterogeneity and to control for possible business-cycle effects and for the aging of the sample, we also include an individual fixed-effect and a year fixed-effect in Equations (3) and (4) by exploiting the panel nature of our data:

(5)

(6)

where represents constant, individual-specific heterogeneity and is the year fixed-effect. As indicated, we estimate Equations (5) and (6) using Linear Probability Models for both the pooled cross-section regressions and the random-effects panel regressions, respectively.[10]

As common in this literature, various low-wage control groups are used to reflect those workers who are in a jurisdiction where there was not a minimum wage increase, but who likely would have been affected by one if it had occurred (i.e., they were working at a wage that was above their own minimum wage but below the hypothetical wage that would result if they received the typical minimum wage increase that occurred in jurisdictions with an increase). For the control groups we provide estimates of these hypothetical wage increases ranging from 10 RMB to 200 RMB above their current minimum wage, which approximates the actual minimum wage increases that occurred in the treatment jurisdictions that increased their minimum wage (listed subsequently in Table 5). As indicated, our preferred control group is those with a hypothetical minimum wage increase of 80 RMB which is close to the 78 RMB that is the average minimum wage increase in our data, and also close to the modal increase of 70 RMB.

Results

Direct wage results

Table 3 indicates how the wages of at-risk individuals (i.e., whose wage fell between the old and the new minimum wage and who remain employed after the minimum wage increase) are affected by increases in minimum wages. The wage effects are all statistically significant and of substantial magnitude.

The direct wage effects increase in going from the pooled estimates of column 1 to the panel estimates of columns 2-5 which control for unobservables that can otherwise affect the estimates. This suggests that those at risk who are affected by minimum wages have unobserved characteristics that make them less productive relative to their wage and hence they obtain a large wage gain from the minimum wage once that unobserved productivity is controlled for by the individual fixed-effects. Within the panel regressions, the direct wage effects drop slightly in moving from controlling only for individual fixed-effects (column 2) to also adding year fixed-effects (column 3), city trends (column 4) and macroeconomic controls (column 5). This highlights the importance of controlling for these factors that can otherwise bias the effect of minimum wages on the wages of those affected by a minimum wage increase.

As broader control groups are added (going down the rows) the direct wage effect decreases (e.g., from 0.626 for the MinWage+10 control group to 0.396 for the MinWage+200 control group in column 5). This is expected since when netting out the wage increases that are occurring naturally in the control group, those wage increases are of larger magnitudes for the higher wage groups and this reduces the magnitude of the pure direct wage effect for the treatment group that experienced a wage increase.

Our preferred specification is the panel estimates of the last column 5 (since they included the broader controls for unobservables) and our preferred control group is the MinWage + 80 RMB (because that is the group that did not experience a minimum wage increase but is given a hypothetical minimum wage increase close to the average increase of 78 RMB). The coefficient estimate from that preferred specification (column 5) and control group (MinWage+80) is 0.409. Since the double-log specification is an elasticity estimate this indicates that a 10% increase in the minimum wage gives rise to a 4.1% total increase in their wages. The average minimum wage increase over that period of 14% would therefore be associated with a 5.7% increase in the wages of those potentially affected by the increase.[11] In essence, slightly less than half (41%) of the percent increases in minimum wages is reflected in wage increases for those who are expected to be affected by minimum wage increases in that their wage falls between the old and the new minimum wage increase. The fact that 100% of the minimum wage increase is not reflected in the actual wage increases of those at risk reflects the fact that many of those persons had wages that were above the old minimum wage and hence their wage adjustment would be less than the minimum wage increase. It can also reflect non-compliance.

Overall, minimum wage increases do seem to have a substantial effect on the wages of those whose wages are potentially affected by the minimum wage increase, raising their wages by slightly less than half (41%) of the minimum wage increases. This suggests that minimum wage increases have their intended effect of raising the wages of otherwise low-wage workers who retain their jobs. We next turn to see if they also have indirect spillover effects on the wages of workers just above the minimum wage, followed by seeing if minimum wages have the unintended consequence of reducing the employment probability of workers who are directly affected by minimum wages.

Wage spillover results

Table 4 gives our wage spillover results based on estimating equation (2). Since these are double-log specifications, the coefficient on the minimum wage term is an elasticity. It indicates that a 10% increase in the minimum wage would give rise to a statistically significant 0.28% (i.e., about one-quarter of 1%) increase in the wages of those in the spillover category who earn between the new minimum wage and 20 RMB above the minimum wage and who are in a “treatment” jurisdiction that experienced a minimum wage increase, compared to those in that same spillover category but who are in a control jurisdiction that did not experience a minimum wage increase. Since the average minimum wage increase was 14%, this suggests that the average minimum wage increase gives rise to a 0.39% increase in wages for those in that first spillover category. Similar size spillover effects prevail for the next two spillover categories, after which the effects drop substantially and are statistically insignificant. This confining of wage spillover effects to those just above the new minimum wage is expected for reasons discussed previously.

Overall, statistically significant but quantitatively very small positive wage spillover effects prevail up until those who earn 60 RMB above the new minimum wage. The spillover wage effects are so small, however, that the reasonable conclusion is that minimum wage effects are largely confined to the wages of those whose wages are directly affected by the minimum wage increase in that they fall between the old and the new minimum wage.

Effects on the probability of remaining employed

Table 5 illustrates the employment effect of the expected wage increase for the at-risk group whose wages are potentially affected by the minimum wage increase, based on estimating equations (3) to (6). As outlined previously, the gap measure is the difference between the individual’s actual wage and the new minimum wage if they are at-risk in that their wage falls between the old and new minimum wage, and is set to zero for those not bound by the new minimum.

The results of Table 5 indicate that minimum wage increases are associated with statistically significant reductions in the probability of remaining employed across all specifications and across all control groups. Our preferred specification is the column 4 panel estimates based on the wage gap methodology because the wage gap methodology takes account of the actual wage increase induced by the minimum wage increase and the panel estimates control for individual and year fixed-effects as well as city-specific time trends and macroeconomic factors. Our preferred control group is the MinWage + 80 RMB because that is the group that did not experience a minimum wage increase but is given a hypothetical minimum wage increase close to the average increase of 78 RMB. Based on that preferred specification and control group, the minimum wage increases that occurred over that time period are associated with a 0.028 reduction in the probability of being employed in the subsequent period.

This preferred estimate of the adverse employment effect is fairly similar across the different estimation procedures and it is in the mid-range of the adverse employment effects across the different control groups. Based on the different specifications and control groups, the adverse employment effects from the minimum wage increases tend to range from a 2 to 4 percentage point reduction in the probability of being employed, with a 2.8 percentage point reduction being our preferred estimate. Such adverse employment effects also corroborate our previous wage increase estimates since there should be no adverse employment effect if there were no wage increases.

Conclusions

Our empirical work contributes to the minimum wage literature on China in a number of important ways. First, the empirical analysis spans the years 2004 to 2009 so we are able to take advantage of the substantial variation in both the magnitude and frequency of minimum wage changes that have occurred in China since its new minimum wage regulations in 2004. Second, the data is longitudinal so we are able to use individual and year fixed-effect panel estimation procedures to better control for unobserved heterogeneity that can otherwise contaminate the estimates of minimum wage impacts. Third, we are able to merge county-level minimum wage data (which is the level where minimum wages are set) with individual-level data (which is the level where wage and employment effects occur). Fourth, the individual-level data enables us to use both the “at-risk” and wage gap methodologies for estimating wage and employment effects. Fifth, we estimate both the wage and employment effects of minimum wage increases which provide corroborating evidence since positive wage effects and negative employment effects would go hand-in-hand. Sixth, we estimate wage spillovers to see if minimum wages have ripple effects on the wages of those whose wages are above the new minimum wage and who may be indirectly affected.

The wage results indicate that minimum wage increases have their intended effect of raising the wages of otherwise similar low-wage workers, raising their wages by a little less than half (41%) of the minimum wage increases. They do not raise them by the full minimum wage increase because many were already working at a wage above the old minimum wage but below the new minimum. Also, there may not be full compliance.

The employment results indicate that minimum wage increases also have the unintended consequence of reducing their probability of being employed. Based on a variety of different specifications and control groups, the adverse employment effects from the minimum wage increases that occurred over the period are all statistically significant and fairly substantial in magnitude. They range from a 2 to 4 percentage point reduction in the probability of being employed, with a 2.8 percentage point reduction being our preferred estimate. Such adverse employment effects also corroborate our previous wage increase estimates since there should be no adverse employment effect if there were no wage increases.

We find statistically significant but quantitatively very small positive wage spillover effects that are confined to groups only slightly above the new minimum wage. The spillover wage effects are so small, however, that the reasonable conclusion is that minimum wage effects are largely confined to the wages of those whose wages are directly affected by the minimum wage increase in that they fall between the old and the new minimum wage.

Clearly, China faces a trade-off that is typical in economics. Minimum wages can raise the wages of low-wage workers in China, but at the expense of reducing their employment probability. There is no such thing as a free lunch for raising minimum wages in China.

REFERENCES

Ashenfelter O., Card D. (1981). Using longitudinal data to estimate the employment effects of the minimum wage. Princeton University, Princeton, NJ.
Bhaskar ,V., Manning, A., and To, T. (2002), “Oligopsony and monopolistic competition in labour markets”, Journal of Economic Perspectives, Vol. 16 No. 2, pp.155-74.
Cameron, A. C., & Trivedi P. K. (2005). Microeconometrics: Methods and Applications. Cambridge university press.
Campolieti, M. (2015). Minimum wages and wage spillovers in Canada. Canadian Public Policy, 41(1): 15-34.
Campolieti, M., Fang, T. & Gunderson, M. (2005). Minimum wage impacts on employment transitions of youths: 1993-99. Canadian Journal of Economics, 38(1): 81-104.
Cooke, F. L. (2005). HRM, Work and Employment in China. Routledge.
Currie J., & Fallick B. C. (1996). The minimum wage and the employment of youth evidence from the NLSY. The Journal of Human Resources, 31 (2): 404-428.
Deng, Q. & Li, S. (2012). Low paid workers in urban China. International Labour Review, 151(3): 157-181.
Ding, S. (2009). An analysis of minimum wage effects on the employment of rural-urban migrant: evidence from the survey of 827 rural-urban migrants in Beijing. China Rural Survey, 4: 26-36 (in Chinese).
Ding, S. (2010). Employment effects of minimum wage regulation and cross effect of the Employment Contracts Law. Social Science in China, 31(3): 146-167.
Dong, X. & Putterman, L. (2000). Pre-reform industry and state monopsony in China. Journal of Comparative Economics, 28(1): 32–60.
Dong, X. & Putterman, L. (2002). China’s state-owned enterprises in the first reform decade: An analysis of a declining monopsony. Economics of Planning, 35(2): 109–39.
Draca, M., S. Machin & J. Van Reenen. (2011). Minimum wages and firm profitability. American Economic Journal: Applied Economics, 3(1): 129-51.
Egge K., Kohen A., Shea J., & Zeller F. (1970). Changes in the federal minimum wage and the employment of young men, 1966-67. GPO, Washington, D.C.
Fang, T. & Gunderson, M. (2009). Minimum wage impacts on older workers; Longitudinal estimates from Canada. British Journal of Industrial Relations, 47(2): 371-388.
Fang, T. & Lin, C. (2015). Minimum wages and employment in China. IZA Journal of Labor Policy, 4 (22)1-30.
Guo, F. and S. Zhang. (2017), “The Impact of Minimum Wages on Rural Workers’ Employment and Working Hours”, Statistics and Decision, 21(19): 111-115. (In Chinese)
Hamermesh, D. (2002). International labor economics. Journal of Labor Economics, 20:709–32.
Heckman J. J. (1981). Statistical models for discrete panel data. In: Manski C. F., McFadden D. (eds) Structural analysis of discrete data with econometric applications. MIT Press, Cambridge, Mass, pp 114-178.
Hirsch, B., Kaufman, B. & Zelenska, T. (2015). Minimum wage channels of adjustment. Industrial Relations, 54 (2): 199-239.
Linneman P. (1982). The economic impacts of minimum wage laws: A new look at an old question. Journal of Political Economy, 90 (3): 443-469.
Long, C & J. Yang. (2016). How do firms respond to minimum wage regulation in China? Evidence from Chinese private firms. China Economic Review, 38: 267-284.
Manning , A. (2003), Monopsony in Motion. Princeton University Press, Princeton, N.J.
Metcalf, D. (2008). Why has the British national minimum wage had little or no impact on employment? Journal of Industrial Relations, 50: 489-512.
Ni, J. Wang, G. &Yao, X. (2011). Impact of minimum wages on employment: Evidence from China. The Chinese Economy, 44: 18-38.
Rama, M. (2001). The consequences of doubling the minimum wage: The case of Indonesia. Industrial and Labor Relations Review, 54(4), 864-881.
Rawski, T. (2006). Recent developments in China’s labour economy. In: Katsuji Nakagane & Tomoyuki Kojima. (eds.) Restructuring China. Tokyo: Toyo Bunko, 18-47.
Wang J. & Gunderson, M. (2011). Minimum wage impacts in China: Estimates from a pre-specified research design, 2000-2007. Contemporary Economic Policy, 29: 392-406.
Wang J. & Gunderson, M. (2012). Minimum wage effects on employment and wages: dif-in-dif estimates from Eastern China. International Journal of Manpower, 33: 860-876.
Wang J. & Gunderson, M. (2015). Adjustments to minimum wages in China: cost-neutral offsets.” Relations industrielles/Industrial Relations. 70 (3): 510-531.
Wang, J. and M. Gunderson, (2018) “Minimum Wage Effects on Low-Skilled Workers in Less Developed Regions of China”, International Journal of Manpower, 39 (3): 455-467.
Xing, C. and Xue, J. (2016). Regional variation of the minimum wage in China. IZA Journal of Labor and Development, 5(8) 1-22.
Yuen, T. (2003). The effect of minimum wages on youth employment in Canada: A panel study. Journal of Human Resources, 38: 647-672.
Zavodny M. (2000). The effect of the minimum wage on employment and hours. Labour Economics, 7 (6): 729-750.
Zhang, S. and Z.Yang. (2016). “The Employment and Wage Effect of Minimum Wage Hikes
- on Migrant Workers”, Finance and Economics, 28(10):100-106. (in Chinese).

Endnotes

Table 1. Minimum wages across regions in China (2004-2009)

Province	2004		2005		2006		2007		2008		2009
	mean	s.d	mean	s.d	mean	s.d	mean	s.d	mean	s.d	mean	s.d
East
Beijing	586.3	0.0	624.3	0.0	667.2	0.0	717.1	0.0	758.5	0.0	800.0	0.0
Tianjin	565.4	7.4	616.3	9.7	705.7	9.5	751.3	9.1	792.0	2.2	820.0	0.0
Hebei	423.5	31.7	511.2	31.2	509.8	32.0	505.0	41.5	573.8	52.2	610.8	62.6
Shandong	400.9	45.2	489.4	64.2	495.8	66.5	513.2	79.2	589.8	95.2	594.9	96.1
Shanghai	679.4	0.0	735.3	0.0	776.4	0.0	816.2	0.0	922.1	0.0	960.0	0.0
Jiangsu	479.1	81.7	507.7	87.7	582.7	106.3	637.9	94.1	670.3	103.3	678.5	107.4
Zhejiang	552.5	51.9	618.1	65.3	677.2	66.0	730.1	75.3	780.5	80.5	851.8	89.5
Fujian	364.9	52.9	400.6	61.0	471.3	75.2	541.1	80.9	564.9	79.5	569.8	80.2
Guangdong	415.4	72.9	476.8	93.2	513.7	100.2	557.0	99.2	592.7	94.0	620.9	98.8
Hainan	403.1	64.1	427.3	67.6	464.8	66.6	491.0	63.8	510.6	60.4	515.0	60.9
Northeast
Liaoning	324.8	60.5	405.7	45.6	442.4	53.8	499.5	64.7	567.4	71.4	573.4	71.5
Jilin	363.6	23.1	357.9	22.7	443.0	32.3	618.3	39.3	585.6	37.2	590.6	37.5
Heilongjiang	324.6	34.7	319.5	34.2	417.7	50.8	450.5	58.6	470.3	61.4	474.3	61.9
Central
Shanxi	451.0	70.5	494.3	45.5	495.3	44.8	513.3	42.9	553.4	43.2	626.7	51.2
Henan	289.4	32.2	309.1	32.2	376.2	43.1	400.0	41.1	492.2	57.0	496.5	57.5
Jiangxi	284.8	23.2	352.6	25.1	352.8	24.8	460.7	27.2	474.7	34.3	478.8	34.6
Anhui	350.6	29.1	367.0	32.3	381.8	35.9	431.8	48.6	433.6	49.1	437.3	49.5
Hubei	312.5	50.3	355.1	53.3	359.5	53.8	432.7	55.7	466.3	60.6	527.3	72.5
Hunan	389.9	33.5	416.4	34.0	449.2	33.1	476.5	32.8	507.7	32.1	547.5	37.9
West
Inner Mongolia	387.6	17.8	432.0	18.1	437.4	29.0	452.3	62.0	530.3	65.0	534.9	65.6
Guangxi	371.8	16.9	425.3	42.4	429.8	43.3	446.0	45.7	504.3	51.4	554.7	57.4
Chongqing	384.6	27.8	405.9	30.2	445.6	38.8	514.9	58.5	572.2	63.9	577.1	64.4
Sichuan	304.0	43.5	314.7	62.7	360.7	56.6	458.7	50.9	493.2	61.8	497.5	62.4
Guizhou	348.1	26.9	384.4	35.5	416.2	37.2	522.3	41.9	578.0	39.7	583.0	40.0
Yunnan	342.4	28.0	405.3	34.2	440.1	34.3	460.1	34.9	543.8	47.1	548.5	47.5
Tibet	343.1	3.5	537.5	21.8	529.7	21.5	507.0	20.6	702.5	39.0	708.6	39.3
Shaanxi	309.8	26.1	394.2	28.3	493.9	35.5	489.4	38.8	523.0	36.8	527.5	37.1
Gansu	343.2	13.5	337.8	13.3	352.2	20.9	371.3	40.9	484.6	40.4	535.9	41.4
Qinghai	297.4	17.0	384.3	16.7	436.1	12.7	472.4	8.7	539.0	8.3	591.3	8.3
Ningxia	385.5	26.5	382.4	26.5	436.7	30.2	458.5	29.7	516.0	28.1	520.5	28.4
Xinjiang	365.7	37.9	374.4	37.9	408.3	40.8	479.8	43.1	561.4	55.8	566.3	56.3

Notes: Minimum wages have been calculated as time-weighted average values based on county level minimum wage data. Values have been accounted for inflation to the price level in 2009 using the urban CPI.

Table 2. Summary statistics of at-risk and not-at-risk employed individuals, 2004‒2009

	At-risk	Not-at-risk
Variable	Mean	Mean	Diff.
Age	41.11	41.89	-0.78^***
	[10.22]	[9.31]	(0.00)
Male	0.38	0.55	-0.18^***
	[0.48]	[0.50]	(0.00)
Years of schooling	11.24	12.71	-1.47^***
	[2.67]	[2.78]	(0.00)
Married with spouse present	0.85	0.90	-0.04^***
	[0.35]	[0.31]	(0.00)
Han Ethnicity	0.98	0.97	0.00
	[0.15]	[0.17]	(0.14)
Local hukou	0.97	0.97	-0.00
	[0.17]	[0.16]	(0.23)
Work experience (year)	23.84	23.16	0.68^***
	[10.99]	[10.31]	(0.00)
Years of residence	33.69	31.73	1.96^***
	[14.58]	[14.90]	(0.00)
Wages (monthly RMB)	488.51	1781.09	-1292.58^***
	[142.48]	[1431.72]	(0.00)
Minimum wage (monthly RMB)	543.50	543.50	–
	[146.61]	[146.61]	–
Wage gap (monthly RMB)	54.99	-1237.59	-1292.58^***
	[46.87]	[1332.54]	(0.00)
Educational attainment (%)
Elementary school or no schooling	4.49	2.54
Junior high school	34.01	21.96
High school	35.53	25.32
Vocational school	10.04	12.71
Junior college	12.19	24.11
College or above	3.73	13.37
Occupation (%)
Public sector	4.90	11.86
Technical job	2.82	14.94
Clerical and related staff	21.63	30.61
Production, transportation, operators	9.91	15.52
Business or service job	39.07	9.77
Agricultural job	14.21	6.84
Other	7.46	10.46
Industry (%)
Mining	1.80	3.17
Manufacturing	21.35	22.60
Power production and supply	1.28	3.54
Construction	2.81	3.12
Transportation and postal service	4.49	7.33
Information technology	1.71	2.14
Wholesale and retail sales	19.02	9.66
Hotel and restaurant	4.90	2.43
Banking and finance	1.35	2.95
Real estate	2.14	1.88
Leasing and commercial service	1.55	1.53
Scientific Research	0.69	2.08
Environment and public facility	1.03	1.33
Housekeeping	21.14	8.86
Education	2.73	7.20
Health care	2.54	4.67
Sports and entertainment	1.15	1.69
Public service	8.30	13.82
Percent of individuals with
1 obs.	45.27	53.04
2 obs.	38.15	33.89
3 obs.	13.48	10.19
4 obs.	2.99	2.70
5 obs.	0.08	0.12
6 obs.	0.04	0.07
Number of observations	2,879	130,261
Note: ^*** statistically significant at the 1% level. At-risk individuals are workers whose monthly wages in the previous year (t-1) are less than the new minimum (t) but no less than the old minimum in the year t-1., i.e., Wages and minimum wages have been adjusted for inflation (2009 base year) and accounted for the differing living costs among provinces by applying the PPP-adjusted deflator developed by Brandt and Holz (2006). The means of wages and wage gaps are calculated at the individual level, whereas the average of minimum wages is calculated using the new minimum at the county level from the 16 provinces. Standard deviations are in brackets and standard errors are in parentheses. The ratios of the at-risk and not-at-risk workers to total employment are 0.022 and 0.978, respectively.

Table 3. Effect of minimum wage changes on wages of at-risk individuals, 2004‒2009

Dep. variable: log wage		Pooled OLS	Fixed-effects panel regressions
Control group	N	(1)	(2)	(3)	(4)	(5)
MinWage+10	5238	0.427^***	0.840^***	0.825^***	0.741^***	0.626^***
		(0.043)	(0.023)	(0.029)	(0.045)	(0.048)
MinWage+20	5678	0.345^***	0.782^***	0.764^***	0.651^***	0.529^***
		(0.035)	(0.028)	(0.034)	(0.049)	(0.046)
MinWage+30	6223	0.291^***	0.747^***	0.715^***	0.605^***	0.481^***
		(0.032)	(0.030)	(0.038)	(0.056)	(0.050)
MinWage+50	7208	0.246^***	0.698^***	0.652^***	0.567^***	0.445^***
		(0.029)	(0.030)	(0.040)	(0.063)	(0.055)
MinWage+70	8403	0.224^***	0.646^***	0.597^***	0.538^***	0.419^***
		(0.027)	(0.031)	(0.041)	(0.074)	(0.066)
MinWage+80	8990	0.220^***	0.616^***	0.572^***	0.524^***	0.409^***
		(0.026)	(0.031)	(0.041)	(0.078)	(0.070)
MinWage+100	10140	0.222^***	0.591^***	0.558^***	0.514^***	0.401^***
		(0.026)	(0.032)	(0.040)	(0.080)	(0.073)
MinWage+120	11503	0.223^***	0.577^***	0.542^***	0.504^***	0.393^***
		(0.026)	(0.031)	(0.039)	(0.081)	(0.074)
MinWage+150	13250	0.229^***	0.580^***	0.535^***	0.506^***	0.397^***
		(0.026)	(0.030)	(0.039)	(0.084)	(0.077)
MinWage+180	15215	0.240^***	0.577^***	0.537^***	0.505^***	0.398^***
		(0.025)	(0.030)	(0.038)	(0.085)	(0.080)
MinWage+200	16436	0.247^***	0.584^***	0.536^***	0.504^***	0.396^***
		(0.026)	(0.029)	(0.037)	(0.084)	(0.080)
Individual fixed-effects Year fixed-effects City trends Macroeconomic controls			Yes	Yes	Yes	Yes
		Yes		Yes	Yes	Yes
		Yes			Yes	Yes
		Yes				Yes

Note: ^*** statistically significant at the 1% level; ^** at the 5% level; ^* at the 10% level. Cluster-robust standard errors at the county level are in parentheses. Clustering at the individual-level does not alter the significance level of the estimates are hence not reported here. The pooled OLS regression has controlled for individual characteristics—gender, years of schooling, marital status, work experience, work experience squared, ethnicity, hukou status, occupation, and industry—as well as year fixed effects, province fixed-effects, city trends and macro controls for GDP per capita and foreign direct investment at the city level. The fixed effects panel regressions do not include variables that are time invariant such as gender and ethnicity. The full estimation results are available on request. For control groups (which do not experience minimum wage increases), MinWage+XXX denotes minimum wage (MinWage) plus XXX Chinese dollar (RMB) above the minimum. For our preferred control group (MinWage + 80), the ratio of at-risk workers to total workers (at-risk and not-at-risk) is 0.512, indicating that the treatment and control groups are of about equal size.

Table 4. Wage spillover estimates by wage category, 2004-2009

Dep. variable: log wage
Treatment and control groups	Total obs.	Coefficients (St. Errors)
	898	0.028^***
		(0.009)
	876	0.030^**
		(0.015)
	1005	0.025^*
		(0.013)
	933	0.008
		(0.014)
	986	0.016
		(0.011)
	1022	0.010
		(0.015)
	894	0.014
		(0.014)
	1118	0.016
		(0.010)
	961	-0.003
		(0.006)
	1002	0.015
		(0.010)

Table 5. Marginal Effects of minimum wage changes on employment probability of at-risk individuals, 2004‒2009

Dep. variable: log wage		At-risk Methodology		Wage Gap Methodology
		Linear Probability Model		Linear Probability Model
		Pooled	Panel random-effects	Pooled		Panel random-effects
		(1)	(2)	(3)		(4)
Control Group	Total obs.	Empl. Effect	Empl. Effect	Estimated Coefficient	Empl. Effect	Estimated Coefficient	Empl. Effect
MinWage+10	2,158	-0.039^***	-0.036^***	-0.000526^***	-0.041	-0.000510^***	-0.040
		(0.010)	(0.008)	(0.000067)		(0.000066)
MinWage+20	2,443	-0.033^***	-0.028^***	-0.000497^***	-0.039	-0.000479^***	-0.037
		(0.009)	(0.008)	(0.000064)		(0.000063)
MinWage+30	2,776	-0.030^***	-0.027^***	-0.000475^***	-0.037	-0.000460^***	-0.036
		(0.009)	(0.008)	(0.000063)		(0.000061)
MinWage+50	3,397	-0.023^**	-0.021^***	-0.000424^***	-0.033	-0.000410^***	-0.032
		(0.009)	(0.007)	(0.000059)		(0.000058)
MinWage+70	4,144	-0.020^**	-0.018^***	-0.000392^***	-0.031	-0.000374^***	-0.029
		(0.008)	(0.007)	(0.000055)		(0.000054)
MinWage+80	4,557	-0.021^***	-0.019^***	-0.000382^***	-0.030	-0.000365^***	-0.028
		(0.008)	(0.006)	(0.000054)		(0.000053)
MinWage+100	5,268	-0.021^***	-0.019^***	-0.000363^***	-0.028	-0.000350^***	-0.027
		(0.007)	(0.006)	(0.000053)		(0.000052)
MinWage+120	6,089	-0.020^***	-0.017^***	-0.000352^***	-0.027	-0.000336^***	-0.026
		(0.007)	(0.006)	(0.000051)		(0.000051)
MinWage+150	7,198	-0.021^***	-0.018^***	-0.000341^***	-0.027	-0.000326^***	-0.025
		(0.006)	(0.005)	(0.000050)		(0.000049)
MinWage+180	8,377	-0.023^***	-0.019^***	-0.000332^***	-0.026	-0.000319^***	-0.025
		(0.006)	(0.005)	(0.000049)		(0.000048)
MinWage+200	9,150	-0.023^***	-0.019^***	-0.000327^***	-0.026	-0.000314^***	-0.024
		(0.006)	(0.005)	(0.000048)		(0.000048)

Note: ^*** statistically significant at the 1% level; ^** at the 5% level; ^* at the 10% level. Cluster-robust standard errors at the county level are in parentheses. Clustering at the individual-level does not alter the significance level of the estimates are hence not reported here. All Models have controlled for individual characteristic, GDP per capita and FDI at the city level. MinWage+XXX denotes minimum wage (MinWage) plus XXX Chinese dollar (RMB) above the minimum. The increase of the minimum wage ranges from 5 to 260 RMB between 2004 and 2009 and the mean increase is 78 RMB. The employment effect for the at-risk methodology is the estimated coefficient of the at-risk dummy variable (AtRisk) from the linear probability model (equation 5); whereas the employment effect for the wage gap methodology is the estimated coefficient times the mean increase of minimum wages (78 RMB) from equation 6.

Figure 1. Average Minimum Wage (Panel A) and Frequencies of Increases (Panel B) in China, 1995-2012

Average minimum wages have been calculated as time-weighted average values based on county-level minimum wage data. Real minimum wages have been adjusted for inflation and shown at 2009 Chinese dollar values using the official urban CPI which applied to the 1990 urban basket (priced at 1990 urban prices) developed by Brandt and Holz (2006). The updated indices are obtained from http://carstenholz.people.ust.hk/SpatialDeflators.html. The number of counties ranges from 1850 to 2370 each year over the 1995-2012 period in our data. There were no minimum wage increases over the country in 2009 due to the Great Recession.

[1] For example, Ni, Wang and Yao (2011) and Wang and Gunderson (2011, 2012) use provincial-level data from the Chinese statistic year books; Fang and Lin (2015) use aggregated county-level data calculated from household surveys; Ding (2009) uses an employment survey in Beijing and Ding (2010) uses a firm survey in Fujian and Guangdong provinces.

[2] As described subsequently, our survey data is from 2002 to 2009 and our minimum wage data is from 1994 to 2012. For our empirical analysis, we use the data from 2004-2009, after the 2004 Minimum Wage Regulation in China.

[3] The UHS is not publicly available. The National Bureau of Statistics of the People’s Republic of China, however, allows limited access to the microdata for up to 16 provinces under certain conditions for academic research. Due to increased concerns on confidentiality and social stability (for instance, using the UHS to calculate sensitive indicators such as unemployment rates), the bureau has stopped granting access to the microdata. As a result, the UHS data are no longer accessible starting 2010. The 16-province sample (Heilongjiang, Liaoning, Gansu, Beijing, Shanxi, Shanghai, Shandong, Jiangsu, Anhui, Jiangxi, Hubei, Henan, Guangdong, Chongqing, Sichuan and Yunnan) includes most economically important provinces in China, containing 65% of the total population covering 60% of the counties and 35% of GDP in the country (National Bureau of Statistics of China 2010). A comparison of the descriptive statistics from the UHS with 2005 Census data for all provinces, indicate that the UHS sample is representative (results available on request).

[4] Therefore, the UHS contains some migrant households with local residency; however, most migrants working in urban areas without an urban household registration are not included in the surveys. Consequently, our analysis focuses on workers with urban residency and the results should be interpreted with caution.

[5] Although such violation of the minimum wage laws is possible, the non-compliance rate in our data is 7.63% (i.e., those below the old minimum wage). Such low wages can also be the result of measurement error. In particular, if the low-wage workers were not in compliance in the old regime, they are unlikely to be in compliance in the new regime. As a result, we have excluded those whose wages were below the old minimum wage from the estimations.

[6] For example, if the adjustment in a particular county and a particular year occurs on June 1st, the figure for that year and county is the average of the old and the new minimum wages, with weights of 5/12 and 7/12, respectively.

[7] We define the control group as both (1) workers whose wages fall in the same range as the treatment group but live in a different county without a minimum wage increase and (2) workers who work in the same county as the treatment group but earn MinWage + XXX. (Note: (1) applies to estimations of the wage and employment effects; (2) only applies to the estimation of wage spillover effects.)

[8] The at-risk methodology can assess the effects of the minimum wage increases only on the transition from employment to non-employment, referred to as the dis-employment effect. It cannot estimate the effects of the minimum wage on transitions from non-employment to employment because there is no wage information on non-employed persons to define an at-risk group. To the extent that a minimum wage increase will also reduce the probability that a non-employed person can obtain a job and become employed, our estimates would be a conservative estimate of the dis-employment effect.

[9] The results are similar to the marginal effects from a logit or probit model.

[10] Heckman (1981) shows that the fixed-effects logit (or probit) estimator is inconsistent when the number of observations per person are small as in our case. Therefore, we use Linear Probability Estimates for the fixed- and random-effects models and chose the random-effects specifications based on the cluster-robust Hausman tests (Cameron and Trivedi 2005). All cluster-robust Hausman tests with 400 bootstrap replications do not reject the null hypothesis that the random-effects model provides consistent and efficient estimates. The results are available upon request.

[11] The average minimum wage increase is 78 RMB (close to the mode of 70) which is a 14% increase over the monthly minimum wage of $543 as indicated in Table 2.