Worker-level analysis of the minimum wage shocks: effects on wages

The increase of the minimum wage in March 2010 resulted in a remarkable increase in the concentration of low wage workers at the minimum wage. As shown in Figure 9.7, the legislative change caused an increased concentration at the bottom of the wage distribution and, due to a sharp increase of the number of minimum wage earners, created a spike at the minimum. The concentration of workers at or very near the minimum wage persisted also during 2011-2013, with, interestingly, the spike increasing from one year to another. The concentration

Wage distribution - non-agricultural workers in dependent employment, 2008-2013

Figure 9.7 Wage distribution - non-agricultural workers in dependent employment, 2008-2013

Source. Own elaboration based on combined unemployment, employment, and earnings registry data, Statistical Office of the Republic of Slovenia.

Note: Hourly wage refers to the average annual hourly gross nominal wage at a given employer expressed in euros.

of minimum-wage workers at the minimum wage is particularly pronounced in market services, where the share of minimum-wage earners was high also before the legislative change.

In a recent paper with another co-author (Laporsek, Vodopivec, & Vodopivec, 2019), we further examined the spill-over effects in a difference-in-differences framework. We compared changes in wages during the control and treatment periods as experienced by workers in the wage group immediately above the level of the new minimum wage and workers in wage groups higher up in the wage distribution. The estimates showed positive and statistically significant spill-over effects of the minimum wage increase, with these effects being more pronounced for treatment groups with wage levels closer to the new minimum wage. In general, we observed spill-over effects at wage levels that are up to 150% of the new minimum wage - in other words, the wages of workers up to 50% higher than the new minimum wage received pay increases that can be directly attributable to the minimum wage increase. The spill-over effects were higher among young and older workers, especially for wage levels near the new minimum wage.

Worker-level analysis of the minimum wage shocks: effects on employment

Given that the previous section documented that the 2010 minimum wage increase had a considerable effect on the wage distribution in Slovenia, in another recent study (Laporsek, Orazem, Vodopivec, & Vodopivec, 2019) we examined the extent to which the increase affected the probability of staying in employment.2 The results of this study are summarised in Table 9.1. which presents probit estimates of the probability of remaining in employment among individuals in Slovenia who had held a job a year before the minimum wage was increased. The estimation reflects the probability of being employed - possibly at a new

Table 9.1 Employment probability among individuals employed in March 2009, using static firm-level minimum wage shock values from 2009

Dependent variable is probability of employment: Ejt+ lk

before

increase

after increase

2009

2010

2011

2012

2013

2014

Firm minimum wage shock - values from March 2009, a$

  • 0.398***
  • (0.0189)
  • 0.443***
  • (0.0229)
  • 0.139***
  • (0.0179)
  • 0.037**
  • (0.0182)
  • -0.03
  • (0.0187)
  • -0.108***
  • (0.0185)

Ratio of minimum wage to predicted MRP - values from March 2009, a.v

  • -0.132***
  • (0.0021)
  • -0.219***
  • (0.0037)

—0.147*** (0.0039)

  • -0.11***
  • (0.0041)
  • -0.082***
  • (0.004)
  • -0.069***
  • (0.004)

Predicted marginal revenue product, o.w

  • -0.003***
  • (0.0002)
  • -0.002***
  • (0.0004)
  • 0.009***
  • (0.0004)
  • 0.017***
  • (0.0004)
  • 0.024***
  • (0.0004)
  • 0.025***
  • (0.0004)

Output shock, a.

  • 0.027***
  • (0.0033)
  • 0.147***
  • (0.0051)
  • 0.128***
  • (0.0097)
  • 0.193***
  • (0.0111)
  • 0.098***
  • (0.0104)
  • 0.557***
  • (0.0102)

Effect of the MRP on employment

probability, cE‘DLlk d(Ok

0.010

0.019

0.023

0.027

0.031

0.031

Observations

4,633,521

4,633,355

4,632,751

4,632,593

4,632,905

4,632.891

Pseudo R-squared

0.121

0.096

0.083

0.081

0.084

0.094

Source. Laporsek et al. (2019).

Notes: Reported above are marginal effects from the foliowmg equation:

Regressions are estimated on individual-level monthly data. Years are defined to begin in March of each year to coincide with the March 2010 minimum wage increase. Monthly dummies were included in all specifications. Standard errors clustered by individuals are in parentheses. Statistical significance: *** p < 0,01, **p < 0.05, *p < 0.1

employer - conditional on having been employed in March 2009, i.e., before the minimum wage change. We explain in more detail the definition of variables, included in the model, and interpret their effect.

We use the worker's predicted marginal revenue product (MRP) in March 2009 as the primary regressor of interest. Calculated based on a range of individual-level covariates from a regression where log(wages) is the dependent variable, the MRP is our preferred measure instead of an individual's actual wage in March 2009 because it is less prone to underreporting for tax purposes and idiosyncratic factors that do not reflect an individual's options outside their current firm. Furthermore, our regressions assume that the firm minimum wage shocks remain static at their levels in March 2010. Taking the period from March 2009 to February 2010 as the baseline, we estimate the probability of employment with 12 consecutive monthly observations beginning in March of the year 2010 and ending in February of the following year. Therefore, the column labelled 2010 shows the coefficients for the first 12 months after the increase of the minimum wage (March 2010- February 2011). Because the sample is conditioned on March 2009 employment, the coefficients reflect the changing probability of employment since March 2009.

The firm-level minimum wage shock variable (as) measures the change in the cost of employing all the other workers in the firm due to the minimum-wage policy; therefore it is the cross-wage effect on a given worker's employment. The positive coefficients mean that firms facing larger minimum wage shocks are sub- sthuting away from other, now more expensive workers and towards this less expensive worker. The positive effect in 2009 suggests some substitution began in the year before implementation, potentially reflecting anticipation effects. In contrast, the effect dissipates and eventually Uirns negative where only the initial shock is allowed to affect probability of employment. This result is suggesting that in the longer term, workers may be hurt by working in firms with other workers receiving wage increases due to the minimum wage.

The ratio of the minimum wage to the worker's MRP (ctv) always lowers the probability of employment. There is a small effect even in the year before the minimum wage increase was implemented, but the effect becomes larger in the first year of the implementation. The effect dissipates but remains statistically significant five years after the initial increase.

The worker’s predicted MRP in March 2009 appears in two places: in the ratio of minimum wage to MRP and then as a separate independent variable. The total effect of the marginal revenue product on conditional employment probability

8E ( MW '

is —k+l*_= / (.) aw-aM ___f+l , where /(•) is the normal density function

l ‘ (®,r)2

evaluated at sample values. Because the estimated minimum wage effect (o.M) is always negative, the estimated effect of marginal revenue product on employment was positive for the full sample range of values of (b(, as shown in the last row of the table. The impact is largest for the least skilled, meaning that a small increase in skills will have the largest impact on employment prospects for individuals at the bottom of the skill distribution. Although the estimated effect is non-linear, roughly speaking, the effect of the increase in the minimum wage is that an individual at the middle of the skills distribution had a probability of being employed that was 1.9% higher than someone at the bottom of the skills distribution in 2010. By 2014, the probability had increased to 3.1%.

 
Source
< Prev   CONTENTS   Source   Next >