The effect of employment protection legeslation on unemployment

(1)

The Effect of Employment Protection Legislation

on Unemployment

Casper Burik

14 August 2014

Abstract

In this thesis the effect of employment protection legislation on the unemployment rate and the percentage of unemployed that are long-‐term unemployed is

estimated using ordinary least squares. Data on 25 OECD countries is used for the period 2002-‐2006. There is no significant effect found on the unemployment rate, but there is a significant effect found on the percentage of long-‐term unemployed. Employment protection legislation for regular workers and protection against Casper Burik (10001420)

Economie en Bedrijfskunde, richting Economie FEB UvA

Beoordelaar: Dr. K. Vermeylen

(2)

1. Introduction

One of the greatest challenges for politicians in Europe at this time is the fight against unemployment. After the recent financial crises and with the economic downturn that followed, it has been suggested to change employment protection legislation to create more flexible labour markets in order to reduce

unemployment. This is not the first time rigid labour markets are blamed for high unemployment. High unemployment and slow job creation in Europe has previously been called ‘Eurosclerosis’ and has been blamed on rigid labour markets before. A challenge for economists is to explain unemployment and research the effects of labour market institutions. In this thesis the effect of employment protection legislation on unemployment rates is researched empirically.

There has been a lot of research on the effect of employment protection legislation on unemployment or employment rates, together with research on other labour market rigidities. However, the effect on the unemployment rates is still unclear. The results of these studies may have political implications as the employment protection legislation laws, while protecting employees, may have negative side effects.

Some former studies find increasing effect on unemployment or decreasing effect on employment. Other studies find no results and a single one finds a decreasing effect on unemployment. Although there has been research on this area, there has been very little research on the effects of employment protection legislation on the percentage of unemployed that are long-‐term unemployed.

The goal of this thesis is to estimate the effect of employment protection legislation on both unemployment rates and the percentage of long-‐term unemployed.

Section two will give an overview of the literature. It contains a subsection on theoretical models and a subsection on former empirical studies. The effect of employment protection legislation on unemployment rates is ambiguous in the theoretical models; this will be explained in section two. The effect on the length of employment is clearer, since employment protection legislation decreases job

(4)

in-‐ and outflow in these models. In former literature different results have been found on the effect on unemployment rates. There has been substantially less research on the length of unemployment or the amount of long-‐term

unemployed. Research on this topic indicates an increasing effect on the length of unemployment.

In this thesis ordinary least squares regression is used on data collected by the OECD. Unemployment rates and the percentage of long-‐term unemployed are regressed on indices describing the strictness of employment protection and control variables. The third section explains this method and describes the data that is used in this thesis.

The results will be discussed in the fourth section. No significant effects have been found for the effect of employment protection legislation on unemployment rates. There is a significant effect of employment protection legislation on the percentage of long-‐term unemployed.

The fifth section gives a conclusion; employment protection legislation does not affect unemployment rates, it does however affect the percentage of

unemployed that are long-‐term unemployed.

2. Literature Review

2.1. Theoretical Models

Several theoretical models have been used to determine the effect of

employment protection legislation on unemployment or employment rates and job turnover rates.

The starting points to model the effects on unemployment rates are the job separation rate and the job finding rate. These two rates can be used to calculate the natural rate of unemployment (Mankiw, 2010, pp. 164 – 166). Firing and hiring rates of employees are part of these job separation and finding rates. As firing costs make it more costly for employers to fire employees, it will be less likely that employers will fire them. So it can be expected that firing rates will decrease. They may also decrease hiring rates, as employers can expect future firing costs and may be less likely to hire new employees. The net effect of

(5)

these two rates is ambiguous as it is unknown which effect is bigger. Below are some theoretical models that have different outcomes.

Bertola (1990) finds that the employment protection legislation does not affect the average employment. The reasoning behind this is as follows:

employment protection legislation increases costs for dismissals of employees and firms take these costs into account while hiring employees. Firms subject to positive economic shocks therefor hire less, firms that face negative shocks will however fire fewer employees. These two effects will average each other out in the long run. It does however, affect job turnover. Less hiring and less firing leads to longer periods of employment and unemployment for workers.

Bentolila and Bertola (1990) present a partial equilibrium model where the effect of firing costs on firing employees exceeds the effect of firing costs on hiring employees. They find that labour demand by a firm is more stable and on average higher if firing costs are large. In this case stricter employment

protection legislation would lead to a decrease in unemployment.

Hopenhayn and Rogerson (1993) use a general equilibrium model of a job reallocation process, with job creation and destruction, that also accounts for entry and exit of firms. In their model they create firing costs by imposing a tax on the destruction of jobs, which leads to a size-‐able decrease in employment. This would indicate that firing costs, like employment protection legislation, significantly decrease employment.

Different theoretical models can lead to different assumptions on the effect of employment protection legislation on unemployment rates. However employment protection legislation is assumed to decrease job turnover, leading to a longer duration of unemployment.

2.2. Empirical Studies

There have been many studies on the effect of employment protection legislation on unemployment or employment rates. This subsection gives a summary of the results of nine empirical studies on this subject. Four studies find that

(6)

unemployment. Four other studies find no significant effects and one study finds that employment protection legislation decreases unemployment rates.

Most studies use an index to measure employment protection legislation strictness; a lot of them are based on the index created by the OECD. The

measurement of employment protection legislation is mentioned in the summary of each study.

Blanchard and Wolfers (2000) find significant results regarding the effect of employment protection legislation on unemployment rates; it increases unemployment. They use an index based on the index of the OECD on employment protection legislation strictness and data on 20 countries from 1960-‐1996. A study on employment protection legislation in Latin America has been done by Heckman and Pagés-‐Serra (2000). They collect their own data using the same definitions as the OECD for the employment protection legislation strictness. They conclude that employment protection legislation decreases employment substantially. Botero et al. (2004) review all labour regulation laws of 85 countries using data from 1997; they find that countries with heavier labour regulations have significantly higher unemployment. Di Tella and MacCulloch (2005) have done a study on employment protection legislation using their own survey that has been send out to several employers and other experts regarding the flexibility adjusting employment levels and compensation. They find that countries with higher flexibility have higher employment rates. The study covers the period 1984-‐1990 for 21 OECD countries.

Nickel (1997) uses the index of the OECD to rank 20 countries on how strict their employment protection legislation is. Estimating the effect of the ranking on unemployment rates Nickel concludes that there is no significant effect of employment protection legislation on unemployment. Investigating 17 OECD countries for the period 1960-‐1999, Belot and Van Ours (2004) only found significant results for employment protection legislation with an equation

excluding country and time fixed effects and without cross terms with other variables. Employment protection has a decreasing effect on unemployment in this case. To measure employment protection they use a self-‐created index in from one of their previous studies. Which is build up in a similar fashion as the OECD index (Belot and Van Ours, 2001). Using the same data set as Belot and Van

(7)

Ours (2004), Garibaldi and Violante (2005) estimate several different equations in which most of them they do not find significant effects for employment protection legislation. They find significant in equations with country and time fixed effects; and cross-‐terms with other variables. Here employment protection legislation leads to a higher employment rate. Baccaro and Rei (2007) use several dynamic and static models with country fixed and time varying effects and find no significant effect of employment protection legislation in any of the equations. To measure employment protection they use the same index as

Blanchard and Wolfers.

The overall result of empirical studies on the subject is somewhat ambiguous. Across studies different datasets and different methods have been used. Most studies use data for OECD countries. An increasing effect on

employment or decreasing effect on unemployment is only found using the data from Belot and van Ours (2004) and including cross-‐terms. There is no common method found for the studies that do find a significant positive effect on

unemployment rates. This use of different methods, datasets and variables strengthens the evidence that employment protection legislation increases unemployment, as different methods find the same results. However, neither is there a common method found for the studies that do not find an effect. The same logic can be applied here, and overall results may still be unknown.

There have been very few empirical studies on the effect of employment

protection legislation on the length of unemployment, job turnover rates or the percentage of long-‐term unemployed.

Kugler and Pica (2008) have done a study on the effect of dismissal costs, like employment protection, on job turnover rates. They study this effect using data at firm level, before and after a legislative reform in Italy. They conclude that the higher dismissal costs associated with the legal reform decreases the flow of workers in and out of unemployment significantly. A similar study has been done by Autor, Kerr and Kugler (2006). They use micro data from the United States and also conclude that employment protection legislation

decreases job in and out flow. Both these studies are in line with the theoretical implications mentioned in section 2.1

(8)

3. Empirical Research

In the first part of this section gives a description of the data that is used in this thesis. The second part contains an explanation of the method with which the data will be analysed.

3.1. Data

All data used in this paper comes from the OECD. It covers the period 2002-‐2006 and is averaged over these five years (A list of the most important averaged data can be found in the appendix). This way economic upswings and downswings should be averaged out. This period is after the recession caused by the dot-‐com bubble and before the financial crisis starting in 2007. While the latter had its obvious effects on unemployment rates, it can be argued that the dot-‐com bubble had little effect. One must also be careful with averaging data over a longer period of time. In this five year time span, little was changed in the employment protection legislation overall, although some countries had somewhat bigger changes in employment protection for temporary workers. This aim for this thesis is to look at the effects of employment protection legislation in ‘normal’ times.

The definition of all variables is given in table 1. The three indices on strictness of employment protection legislation are constructed by the OECD on a yearly basis. Each index consists of a certain amount of items, for each item a score from 0 to 6 is assigned, 6 being the strictest legislation. The indices are weighted averages of the items.

The first index, EPR, is the index for regular workers, it includes the following eight items: Notification procedures; delay involved before notice can be start; length of notice period; severance pay; definition of justified and/or unfair dismissal; length of the trial period; and compensation following unfair dismissal.

The second index, EPC, is an index for protection against collective dismissals, which includes the following four items: definition of collective dismissal; additional notification requirements for collective dismissal;

(9)

additional delay involved before notice for collective dismissal can start; and other special costs to employers for collective dismissals.

Table 1: Definition of variables

Variable Definition

Unemployment The unemployment rate in a country

Longterm The percentage of unemployed that is long-‐term unemployed. Long-‐ term is defined as longer than six months.

EPL The index for strictness of employment protection legislation for regular worker, composed by the OECD.

EPC The index for strictness of employment protection legislation for collective dismissals, composed by the OECD.

EPT The index for strictness of employment protection legislation for temporary workers, composed by the OECD.

Active_LMP Public Expenditure on active labour market programmes as percentage of GDP.

Benefits The amount received in unemployment benefits in US Dollars for single adults without children previously earning the average wage.

Min_wage The minimum wage in US Dollars for adults.

Taxes The tax wedge between employers and employees for single adults without children with the average wage.

Union The percentage of workers in a country that is part of a trade union.

The third index, EPT, is the index for temporary workers, it includes the following eight items: Valid cases for use of fixed-‐term contracts (FTC);

maximum number of successive FTC; maximum cumulated duration of successive FTC; types of work for which temporary work agency (TWA)

employment is legal; restrictions on number of renewals; maximum cumulated duration of TWA assignment; does the set-‐up of a TWA require authorization or reporting obligations; and do regulations ensure equal treatment of regular and agency workers at the user firm.

The range of EPR is from 0.26 (United States) to 4.48 (Portugal). EPC is ranged from 0 (New-‐Zealand) to 5.13 (Belgium). EPT is ranged from 0.25 (Canada and United States) to 4.88 for Turkey. There is almost no correlation between EPR and EPC. EPT has some correlation with EPC and more correlation

(10)

with EPR, which can be seen in Table 2. Scatter plots of EPR, EPC and EPT can be seen in figure 1.

Table 2: Correlation between indices

EPR EPC EPT

EPR 1,000 0,047 0,437

EPC 0,047 1,000 0,256

EPT 0,437 0,256 1,000

Figure 1: Scatter Plots of EPR, EPC and EPT

3.2. Method

In order to study the effect of employment protection legislation on

unemployment the following two regression equations are estimated with the ordinary least squares method and variations on the two equations:

𝑈𝑛𝑒𝑚𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡 = 𝛼_!+ 𝛼_!𝐸𝑃𝑅 + 𝛼_!𝐸𝑃𝐶 + 𝛼_!𝐸𝑃𝑇 + 𝛽!_{𝑋 + 𝜀} ₍₁₎

𝐿𝑜𝑛𝑔𝑡𝑒𝑟𝑚 = 𝛾!+ 𝛾!𝐸𝑃𝑅 + 𝛾!𝐸𝑃𝐶 + 𝛾!𝐸𝑃𝑇 + 𝜃′𝑋 + 𝜂 (2)

Variations will include estimation with less control variables where the least significant control variables will be taken out and single variable regressions.

Here unemployment is the unemployment rate, longterm is the percentage of unemployed that has been unemployed for six months or longer. EPR is the index that represents strictness of employment protection legislation for regular workers, EPC for collective dismissals, EPT for temporary workers and X is a

(11)

vector with control variables (A list with the definition of variables is given in table 1). α, β, γ and θ are constants and coefficients, and ε and η are the error terms. The results of the regression estimations are given in section 4.

The results for α1, α2, α3, γ1, γ2 and γ3 are the most important; they indicate

the effect of employment protection legislation. The first equation will test what the effect of employment protection legislation on the unemployment rate and the second equation will test the effect on the percentage of long-‐term

unemployed.

The control variables include other labour market rigidities, they are defined in table 1. A variable for the height of unemployment benefits is added, as this decreases the effectiveness of the unemployed in their search for jobs and increases unemployment (Nickel, 1997).

Secondly, there is controlled for minimum wages. Theoretically minimum wages can increase unemployment, though there is only empirical evidence to suggest an effect on youth unemployment (Nickel, 1997).

A third control variable has been added for the labour tax wedge between the employer and employee. This also has been suggested to have a negative effect on unemployment (Nickel, 1997).

The fourth variable that has been controlled for is the unionization. As unionization increases the bargaining power of employees this may increase labour costs and with that, increase unemployment (Nickel, 1997).

The last variable that has been controlled for is public spending on active labour programmes. These programmes, designed to get the unemployed employed again may decrease unemployment for participants (Kluve, 2010). It could also shorten the duration of unemployment of participants (Carling and Richardson, 2004).

4. Results

The results of the OLS estimation of equation 1 and variations on it can be seen in Table 3. Neither the coefficients of the indices nor the coefficients of the control variables are significant at any level in the first regression. In the second regression with less control variables only the control variable on the labour tax

(12)

rate has a significant coefficient on the 5 per cent level. In the rest of the regressions the variables of interest also have insignificant coefficients. This would lead to the conclusion that there is no evidence for employment protection legislation to have an effect on unemployment rates.

Table 3: Results for unemployment rate

Variable Coefficients (standard error)

All control variables Less control variables Only EPR Only EPC Only EPT

C 4,207 1,421** 5,217** 4,478* 6,497*** (4,985) (2,925) (2,017) (2,388) (1,206) EPR 0,462 0,192 0,898 (1,223) (0,972) (0,874) EPC 0,857 0,439 0,878 (0,941) (0,762) (0,747) EPT -‐0,209 -‐0,313 0,389 (1,081) (0,632) (0,573) ACTIVE_LM -‐1,874 (2,865) BENEFITS -‐0,036 (0,059) MIN_WAGE -‐0,0001 (0,000) TAXES 0,143 0,157** (0,103) (0,073) UNION -‐0,031 (0,059) Number of observations 25 25 25 25 25 R-‐squared 0,328 0,240 0,039 0,050 0,017

S.E. of regression 3,680 3,425 3,621 3,600 3,662

* Significant at the 10 per cent level ** Significant at the 5 per cent level *** Significant at the 1 per cent level

The results of the OLS estimation of equation 2 can be seen in Table 4. The coefficients of EPR and EPC are significant at the 1 per cent level in the first two regressions. The coefficients of 12.5 and 9.5 for respectively EPR and EPC in the first regression and 13,1 and 10,2 in the second regression indicate large effects on the percentage of long-‐term unemployed. The coefficient for EPT is only significant at the 10 per cent level. The coefficient of EPT is negative, which is different than expected. It might be the case that employment protection for

(13)

regular workers has a different effect on unemployment than employment protection for temporary workers. All the control variables are insignificant in the first regression and in the second regression the variable depicting the height of unemployment benefits is significant at the 5 per cent level. In the regressions with only the variables of interest EPR is only significant at the 5 per cent level and EPC at the 10 per cent level. EPT is insignificant in the single variable regression.

The results of the estimation of the two equations indicate that employment protection legislation does not increase unemployment; it does however affect the percentage of long-‐term unemployed. Employment protection for regular workers and protection against collective dismissals increase the percentage of unemployed workers that are long-‐term unemployed. The effect of employment protection for temporary workers might decrease the percentage of long-‐term unemployed, although the found coefficient is only significant at the 10 per cent level.

(14)

Table 4: Results for percentage of long-‐term unemployed

Variable Coefficients (standard error)

All control variables Less control variables Only EPR Only EPC Only EPT

C -‐3,261 -‐2,786 13,444 10,334 33,035 (15,471) (9,551) (8,562) (10,334) (5,510) EPR 12,523*** 13,125*** 8,249** (3,795) (3,035) (3,708) EPC 9,523*** 10,199*** 3,235* (2,920) (2,306) (3,235) EPT -‐6,379* -‐5,185* -‐0,971 (3,353) (2,534) (2,616) ACTIVE_LM -‐1,159 (8,890) BENEFITS -‐0,289 -‐0,368** (0,184) (0,133) MIN_WAGE 0,0000 (0,000) TAXES 0,210 (0,319) UNION -‐0,119 (0,183) Number of observations 25 25 25 25 25 R-‐squared 0,665 0,647 0,160 0,137 0,005

S.E. of regression 11,419 10,369 15,372 15,581 16,727

* Significant at the 10 per cent level ** Significant at the 5 per cent level *** Significant at the 1 per cent level

5. Conclusion

The effects of employment protection legislation on unemployment rates and the percentage of long-‐term unemployed have been estimated with OLS regression using data from the OECD. To measure the strictness of employment protection legislation indices of the OECD have been used.

As can be seen in section four, no significant effects have been found for employment protection legislation on unemployment rates. Which could indicate that there is no effect, however only one method has been used to analyse this effect. There were significant effects of employment protection legislation on the percentage of long-‐term unemployed. Employment protection for both regular

(15)

workers and collective dismissals has been found to increase the percentage of long-‐term unemployed significantly and the coefficients indicate large effects. The sign of the coefficient for employment protection for temporary workers is surprising. However it is only significant at the 10 per cent level, which is too weak to draw any conclusions from.

Together the results imply that the unemployment rate does not go up, but the unemployed will stay unemployed for a longer period of time. The

employment protection legislation indeed protects the employed, but has negative side effects for the unemployed.

The result on unemployment rates is in line with some former research, but opposite to others. The result on the percentage of long-‐term unemployed is in line with former research.

There are some limitations to this thesis, as only 25 countries have been included and only for one time period. Furthermore, to draw stronger conclusions, more data should be used. Also only OLS has been used as

estimation method. As high unemployment can be a big problem it is important that a lot of research is done on this topic to explain the dynamics behind it. If the dynamics can be explained there is a broader foundation for political decisions, after which the society may benefit.

(16)

Appendix

Table A: List with items

Countries EPL _Regular EPL _Collective EPL _Temporary Unemploy-‐_{ment
rate} Percentage of long-‐term unemployed Australia 1,417 2,875 0,875 5,507 22,937 Austria 2,445 3,250 1,313 4,643 25,029 Belgium 1,810 5,125 2,375 8,172 48,102 Canada 0,921 2,969 0,250 7,103 10,254 Czech Republic 3,306 2,125 0,750 7,724 50,754 Denmark 2,135 3,475 1,375 4,721 21,495 Finland 2,167 1,875 1,563 8,638 27,244 France 2,443 3,375 3,625 8,400 38,571 Germany 2,793 3,625 1,300 9,970 50,800 Hungary 2,004 3,375 0,925 6,562 46,252 Ireland 1,403 2,900 0,475 4,624 39,458 Italy 2,762 4,125 2,075 8,130 51,903 Japan 1,702 3,250 0,875 4,775 39,289 Korea 2,369 1,875 2,125 3,540 1,491 Netherlands 2,885 3,000 0,938 3,885 35,402 New Zealand 1,560 0,000 1,000 4,346 13,866 Norway 2,333 2,500 2,800 4,203 10,714 Poland 2,230 3,075 1,250 18,027 48,185 Portugal 4,483 2,875 2,663 6,677 41,733 Slovak Republic 2,289 3,850 0,925 16,749 64,475 Spain 2,357 3,750 3,250 10,353 24,747 Sweden 2,607 2,500 1,438 6,509 21,240 Switzerland 1,595 3,625 1,125 3,766 28,906 United Kingdom 1,198 2,875 0,350 4,952 25,357 United States 0,257 2,875 0,250 5,401 11,668

(17)

References

Autor, D. H., Kerr, W. R. and Kugler, A. D. (2007). Does employment protection reduce productivity? evidence from US states, The Economic Journal, 117 (521), 189-‐217.

Baccaro, L. and Rei, D. (2007), Institutional Determinants of Unemployment in OECD Countries: Does the Deregulatory View Hold Water?, International Organization, 61 (3), pp. 527 – 569

Belot, M. and Van Ours, J. C. (2001). Unemployment and labor market institutions: An empirical analysis, Journal of the Japanese and International

Economies, 15 (4), 403-‐408.

Belot, M. and Van Ours, J.C. (2004), Does the recent success of some OECD countries in lowering their unemployment rates lie in the clever design of their labor market reforms?, Oxford Economic Papers, 56 (4), pp. 621-‐642 Bentolila, S. and Bertola, G. (1990), Firing Costs and Labour Demand: How Bad is

Eurosclerosis?, The Review of Economic Studies, 57 (3), pp. 381-‐402 Bertola, G. (1990), Job Security, Employment and Wages, European Economic

Review, 34 (4), pp. 851–879

Blanchard, O. and Wolfers, J. (2000), The role of shocks and intitutions in the rise of European unemployment: the aggregate evidence, The Economic Journal, 110, pp. 1-‐33

Carling, K. and Richardson, K. (2004), The relative efficiency of labor market programs: Swedish experience from the 1990s, Labour Economics, 11 (3), pp. 335–354

Di Tella, T. and MacCulloch, R. (2005), The consequences of labor market flexibility: Panel evidence based on survey data, European Economic Review, 49 (5), pp. 1225–1259

Garibaldi, P. and Violante, G.L. (2005) The Employment Effects of Severance Payments with Wage Rigidities, The Economic Journal, 115 (506), pp. 799-‐ 832

Heckman, J.J. and Pagés-‐Serra, C. (2000), The Cost of Job Security Regulation: Evidence from Latin American Labor Markets, NBER working paper, http://www.nber.org/papers/w7773.pdf (last visited on 27-‐1-‐2014) Hopenhayn, H. and Rogerson, R. (1993), Job Turnover and Policy Evaluation: A

General Equilibrium Analysis, Journal of Political Economy, 101 (5), pp. 915-‐938

Kluve, J. (2010), The effectiveness of European active labor programs, Labour Economics, 17 (6), pp. 904–918

(18)

Kugler, A. D. and Pica, G. (2008). Effects of employment protection on worker and job flows: Evidence from the 1990 italian reform, Labour Economics, 15 (1), 78-‐95.

Mankiw, N. G. (2010). Chapter 6: Unemployment, Macroeconomics (7th ed., pp. 163-‐188) Worth Palgrave Macmillan.

Nickell, S. (1997), Unemployment and Labor Market Rigidities: Europe versus North America, The Journal of Economic Perspectives, 11 (3), pp. 55-‐74