The cost of job security regulation: evidence from Latin American labor markets

Tags: employment, job security, employment rates, workers, worker, turnover rates, industrial countries, labor market, Latin American, Labor Market Reform, formal sector, security provisions, labor demand, American labor markets, Latin America, NATIONAL BUREAU OF ECONOMIC RESEARCH, unemployment insurance, LATIN AMERICAN LABOR MARKETS James J. Heckman Carmen Pages, Carmen Pages, market regulations, James J. Heckman, OECD, provisions, formal employment, LAC, JS, United States, unemployment, American Development Bank, developing countries, Economic Studies, unemployment rates
Content: NBER WORKING PAPER SERIES THE COST OF JOB SECURITY REGULATION: EVIDENCE FROM LATIN AMERICAN LABOR MARKETS James J. Heckman Carmen Pages Working Paper 7773 http://www.nber.org/papers/w7773 National Bureau of Economic Research 1050 Massachusetts Avenue Cambridge, MA 02138 June 2000 The views expressed herein are those of the authors and not necessarily those of the National Bureau of Economic Research. 2000 by James J. Heckman and Carmen Pages. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including notice, is given to the source.
The Cost of Job Security Regulation: Evidence from Latin American Labor Markets James J. Heckman and Carmen Pages NBER Working Paper No. 7773 June 2000 JEL No. C20 ABSTRACT This paper documents the high level ofjob security protection in Latin American labor markets and analyzes its impact on employment. We show thatjob security policies have substantial impact on the level and the distribution of employment in Latin America. They reduce employment and promote inequality. The institutional organization of the labor market affects both employment and inequality.
James J. Heckman Department of Economics The University of Chicago 1126 E. 59th Street Chicago, IL 60637 and NBER [email protected]
Carmen Pages Inter-American development Bank 1300 New York Avenue, NW Washington, DC 20577 [email protected]
1. Introduction Labor market regulations are introduced with the stated objective of improving workers' welfare. Mandated benefits and social security programs improve workers' income security in case of sickness, work accidents and old age. Job security provisions are designed to reduce a worker's odds of losing her job and her means of living. But, as is often true in economics, benefits usually come at a cost: mandated benefits may reduce employment; job security provisions may protect some workers at the expense of others. This paper gathers evidence from existing and new sources of information on the costs of job security policies. Latin America has experienced a wide range of labor market policies that provide natural experiments with which to evaluate the impact of these polices. Our evidence challenges the prevailing view (see e.g. Abraham and T4ouseman (1994), Blank and Freeman (1994), Freeman (2000) and the papers he cites) that labor market regulations do not affect employment and have minimal costs. We establish that job security policies have a substantial impact on the level and the distribution of employment in Latin America. The evidence for their effect on unemployment is much weaker but there are good conceptual reasons why this should be so. Our focus on the cost side does not imply we believe the benefits of labor policies for protected workers are small or irrelevant. While the benefits to recipients are well-documented, the costs are often unintended and less well understood. Thus, while the evidence suggests that regulations promoting job security reduce covered workers exit rates out of employment, it also indicates that demand curves are downward sloping, that regulation reduces aggregate employment and that the greatest adverse impact of regulation is on youth and groups marginal to the workforce. Insiders and entrenched workers gain from regulation but outsiders suffer. As a consequence, job security regulations reduce employment and promote inequality across workers. The outline of the paper is as follows. Section 2 describes and quantifies job security regulations in Latin America and the Caribbean. In section 3, we summarize the existing evidence on the impact ofjob security provisions on employment, unemployment and turnover rates in Latin America. Section 4 presents new evidence. In section 5 we summarize the paper and present our conclusions. 2
2. Job Security Regulation in Latin America and the Caribbean In this paper, we define job security legislation (JS) to include all those provisions that increase the cost of dismissing a worker. In this section, we quantify the costs of abiding by the legislation, in terms of wages, in order to address three questions: (1) Row high are the implied costs of JS provisions in Latin America and the Caribbean? (2) Within the region, which countries have costlier termination provisions and which are more deregulated? (3) Row do Latin American and Caribbean countries compare with industrial countries in terms of JS legislation? In Latin An-ierican countries, labor codes based on the civil law system regulate the permissible types, durations and the conditions for termination of labor contracts. In contrast, most Caribbean countries are based on the common law system so the law enforces a contract with which both parties privately agree. As a consequence, in some countries there is not a specific body of law regulating employer-employee relationships, while in others some aspects are regulated while others are left to the courts. In Latin America, labor codes favor full-time indefinite employment over part-time, fixed-term or temporary contracts. These types of contracts not only differ in the length of the employment relationship but also in the conditions for termination. While indefinite contracts carry severance pay obligations, temporary contracts can be terminated at no cost provided that the duration of the contract has expired. In contrast, most Caribbean countries do not regulate the range of admissible contracts. Instead, such decisions are left to the parties involved in collective bargaining. There are important differences as well in the conditions for termination of contracts. In Latin America, the termination of a contract is severely restricted. Thus, labor codes mandate a minimum advance notice period prior to termination, determine which causes are considered "just" or "unjust" causes for dismissal, and establish compensation to be awarded to workers for each possible cause of termination. In some countries, firms must also request permission to dismiss more than a certain fraction of their labor force. Finally, some countries allow the reinstatement of a worker to her post if the dismissal is found to be "unjustified" by the courts, although, this provision has been eliminated in many countries. In contrast, in 3
some of the Caribbean countries, advance notice and severance pay are negotiated as part of collective agreements, so there are no specific laws regulating such provisions. Termination laws (or collective agreements) require firms to incur four types of costs: advance notification, compensation for dismissal, seniority premium for dismissed workers and foregone wages during any trial in which the worker contests dismissal. The period of advance notification should be included in the computation of costs because, in general, the various laws typically allow firms to choose between providing advance notice or paying a compensation equivalent to the wage corresponding to that period. Moreover, since productivity can decline substantially after notice, advance notification should be considered as a part of the dismissal cost even when firms choose to notifj workers in advance. Advance notification periods vary from country to country, ranging from zero in Nicaragua, Guatemala, Pen and Uruguay to three months in Bolivia, Haiti and Venezuela for workers with more than 10 years at a firm (See Table l.A in the Appendix). The second component of dismissal cost is compensation for unjustified dismissal. Since in most Latin American countries the economic difficulties of a firm are not considered a just cause for dismissal, any labor force reductions fall in this category. The formula for calculating this compensation is based on multiples of the most recent wage and the years of service. In contrast, in the Caribbean, under union agreements, severance pay is only awarded to a worker in the case that a firm needs to reduce the work force for lack of work or technological change. In most other cases, employment at will is still the norm provided that the firm gives reasonable advance notice to a worker. Finally, in Belize, Bolivia, Chile and Nicaragua, the law mandates compensation to the worker in case of a voluntary quit2. In some countries, employers are required to make an additional payment, known as a seniority premium, upon termination of the work relationship regardless of the cause or party initiating the termination. In Ecuador, Colombia, Panama, Pen, and Venezuela, this benefit is available to the worker both in the case of unjustified dismissal and in the case of a voluntary quit. If a worker quits, she obtains this payment, whereas if the worker is dismissed she obtains this payment plus the compensation for dismissal. In Brazil, this additional payment is only available in the case of unjust dismissal, and if the worker quits, she receives no pay. In all the above-mentioned countries, firms deposit a certain fraction of 2111 Chile, compensation in case of a quit only occurs after the 7th year of service and if the worker chooses to set up an account. 4
workers' monthly wages in an individual trust fund in order to provide for this payment.3 In Ecuador, Colombia, Brazil and Peru, the worker gains access to the principal plus a yield.4 In Panama and Venezuela, the seniority premium is fixed in terms of multiples of monthly wages and the amount accrued in the fund (Panama) or the fund plus a certain yield (Venezuela) pays for the seniority premium. However, the firm is responsible for covering the difference between the required seniority premium and the amount accumulated in the seniority premium fund. Finally, in some countries, firms are also required to pay a worker's foregone wages during the period of any legal process if a worker brings an action against the firm. This provision increases the overall cost of termination by either increasing the overall compensation due and/or reducing workers' incentives to settle out of court.5 During the nineties, seven countries (Colombia, Guyana, Guatemala, Nicaragua, Panama, Peru, and Venezuela) reformed their labor codes in order to reduce the cost of dismissing a worker. Not all labor reforms reduced JS, however. In Chile (1991) and in Dominican Republic (1992), the amount that a firm has to pay upon dismissal of a worker increased considerably during the nineties. In an attempt to quantify all of these provisions we construct an index of JS encompassing LAC and industrial countries. There have been previous attempts to construct such types of measures. Bertola (1990), Grubbs and Wells (1993) and the OECD (1993, 1999) constructed ordinal measures of JS for industrial countries whereas Marquйz (1998) constructed ordinal measures of job security for a sample of industrial and LAC countries. Also, Lazear (1990) quantified firing costs as the amount (in multiples of monthly wages) owed to a worker if she is dismissed after 10 years of service. These measures, however, are unlikely to accurately reflect the magnitude of dismissal costs. On the one hand, ordinal measures can only state that one country is more regulated than another, but cannot measure how much more regulated it is. On the other hand, JS tends to increase in tenure, which implies that measures conditional on certain level of tenure only measure a given point in the severancetenure schedule. To address these shortcomings, we construct an alternative cardinal measure of firing costs that summarizes the entire tenure-severance pay profile using a common set of dismissal probabilities 3jn Brazil, the fund is called FGTS, in Peru, CTS, in Colombia, Fondo tie Cesantia and in Panama, Fondo de Antiguedad. In Brazil a worker gets access to this fund only if she is dismissed. 5
across countries. This measure computes the expected future cost, at the time a worker is hired, of dismissing her in the future due to unfavorable economic conditions.6 The index is constructed to include only firing costs that affect firm's decisions at the margin and therefore it does not include the full cost of regulation on labor demand. It includes the cost of providing statutory advance notice and severance pay conditional on each possible level of tenure that a worker can attain in the future. The JS index does not include the seniority premium as part of cost because, in most countries, provisions for that payment are regularly deposited in a fund. Thus, because deposits are not directly made conditional on a dismissal they are not likely to alter firing decisions. Rather they should be treated as other labor costs incurred by the firm that do not affect firing decisions and are not included in our index. However, they clearly affect the cost of labor to the firm. The index also does not include the cost derived from foregone wages during trial. Although this component may be a substantial share of the total of cost of dismissal, we do not include it in our index because the information on this cost is not available. Thus we cannot estimate the full cost of resolution of legal costs arising from challenges to dismissals through the courts. Our measure of JS thus reflects the marginal costs of dismissing full-time indefinite workers. However, this measure does not capture the effects of recent reforms that have made temporary and fixedterm contracts widely available in countries like Argentina and Pen. To the extent that fixed-term and indefinite contracts are not perfect substitutes--since temporary workers may be less productive (see the evidence in Aguirragabiria and Borganso, 2000)--our index still captures the marginal cost of firing a tenured worker. However, firms may be at the margin of firing temporary workers and so our index overstates the true marginal cost. Additional information regarding the construction of this index can be found in the Appendix. This measure will be used in Section 3 to quantify the impact of JS on different employment and unemployment measures in a sample of OECD and LAC countries. Graph 1 displays the costs of advance notice and compulsory severance pay in Latin American and the Caribbean for 1990 and 1999 as summarized by our index. This graph reveals that even after many countries have reduced dismissal costs during the nineties, the average cost of dismissing a worker is still Another component of dismissal costs that can be quite important in some countries is given by the specific regulations that govern collective dismissals. Information on those regulations is not available for most countries of LAC and therefore we did not include them in our discussion or measurements. 6
higher in Latin America than in our sample of industrial countries. In comparison, the countries of the Caribbean basin exhibit much lower dismissal costs.
Table 1: Job Security Index across Latin America, the Caribbean and OECD countries. End of the nineties
Country United States
Index Job Security (Monthly wages) 0.000
% Annual wage 0.000
Ranking 1
NewZealand
0.221
1.844
2
Australia Canada
0.443
3.696
3
0.553
4.610
4
Norway
0.912
7.599
5
Germany
1.140
9.498
6
France
1.143
9.526
7
Poland
1.219
10.160
8
Switzerland
1.247
10.395
9
United Kingdom
1.457
12.144
10
Belgium Austria
1.729
14.407
11
1.784
14.864
12
Brazil
1.785
14.871
13
Greece
1.804
15.034
14
Guyana
1.890
15.750
15
Jamaica
1.920
16.003
16
Paraguay
2.168
18.068
17
Uruguay
2.232
18.599
18
Trinidad& Tobago
2.548
21.230
19
Nicaragua
2.563
21.358
20
Panama
2.718
22.652
21
Dominican Republic
2.814
23.454
22
Venezuela
2.955
24.625
23
Argentina
2.977
24.808
24
Costa Rica
3.12 1
26.005
25
Mexico
3.126
26.050
26
El Salvador
3.134
26.116
27
Spain Chile Colombia Honduras Peru Turkey Ecuador Portugal Bolivia
3.156
26.300
28
3.380
28.164
29
3.493
29.108
30
3.530
29.418
31
3.796
31.632
32
3.973
33.110
33
4.035
33.621
34
4.166
34.720
35
4.756
39.637
36
Source: Authors' computations (See Appendix)
6 This measure is based on the index developed in Pages and Montenegro (1999) 7
Looking at the individual countries, it may be surprising that countries like Argentina or Mexico exhibit lower JS than Chile, a country traditionally considered as having a more flexible labor market. This divergence is caused by the fact that our index only measures one component of labor market rigidities. So while Argentina and Mexico have stronger unions than Chile, and therefore are likely to have higher wage rigidity, Chile has higher individual job security provisions. Our index, also discounts penalties that arise far in the future, and so the fact that labor codes in Chile and other countries establish an upper limit in payments is discounted in our measure. Graph 1 shows that four countries in Latin America (Nicaragua, Venezuela, Panama and Pen) undertook substantive reforms in their labor codes. Nicaragua and Venezuela reduced the expected dismissal cost by more than three monthly wages, while Panama and Pen reduced it between one and one and half monthly wages. However, Table 1 also makes clear that even after a decade of substantial deregulation, Latin American countries remain at the top of the JS list, with levels of regulation similar to or higher than those existing in the highly regulated South of Europe. We next consider quantitative estimates of the impact ofjob security regulations. 3. The impact of job security regulations The goal of this section is to quantify the impact of job security regulations on employment and turnover rates. The importance of dismissal costs in Latin America is clear in Graph 1. It is thus important to assess the impact, if any, that such policies have on the labor market. 3.1 Theoretical discussion To analyze the impact ofjob security provisions requires a more complex framework that encompasses dynamic decisions of firms. Bertola (1990) develops a dynamic partial-equilibrium model to assess how a firm's firing and hiring decisions are affected by dismissal costs. In the face of a given shock, the optimal employment policy of a firm involves one of three state-contingent responses: (i) dismissing workers, (ii) hiring workers and (iii) doing nothing, in which case employment in that firm does not change. How are these decisions altered by firing costs? In the face of a negative shock and declining marginal value of 8
labor, a firm may want to dismiss some workers, but it has to pay a mandatory dismissal cost. This cost has the effect of discouraging firms from adjusting their labor force, resulting in fewer dismissals than in the absence of such costs. Conversely, in the face of a positive shock firms may want to hire additional workers but will take into account that some workers may have to be fired in the future if demand turns down, and this is costly. This prospective cost acts as a hiring cost, effectively reducing creation of new jobs in good states. The net result is lower employment rates in expansions, higher employment rates in recessions and lower turnover rates as firms hire and fire fewer workers than they would in the absence of these costs. Bertola's model predicts a decline in employment variability associated with firing costs but the implication of his model for average employment is ambiguous. In particular, whether average employment rates increase or decline as a result of firing costs depends on whether the decline in hiring rates more than compensates the reduction in firings. Indeed, simulations reported in Bertola (1990) and Bentolila and Bertola (1990) suggest that average employment (in a given firm) is likely to increase when firing costs increase. These results, however, are quite sensitive to different assumptions about the persistence of shocks, the elasticity of the labor demand, the magnitude of the discount rate, and the functional form of the production function. Thus, less persistent shocks and lower discount rates are associated with larger negative effects of JS on employment because both factors reduce hiring relative to firing (Bentolila and Saint Paul, 1994). Furthermore, a higher elasticity of the demand for goods implies a larger negative effect of job security on employment rates (Risager & Sorensen, 1997). In addition, when investment decisions are also considered, firing costs lower profits and discourage investment, increasing the likelihood that firing costs reduce the demand for labor (Bertola, 1991). The results just reported analyze employment rates in one firm without considering the impact of firing costs on the extensive margin, that is, on how firing costs affect the creation and destruction of firms. Flopenhayn and Rogerson (1993) develop a general equilibrium model based on the U.S. economy that accounts for entry and exit of firms. In their model, the partial equilibrium framework of Bertola (1990) is embedded in a general equilibrium framework in which jobs and firms are created and destroyed in every period in response to firm-specific shocks. In the context of their model, they find that increasing firing costs in the U.S. would lead to an increase in the average employment of existing firms as a consequence of the reduction in firings. However, they also find that such a policy would result in lower firm entry, and 9
lower job creation in newly created firms. For the parameter values they consider, these two last effects offset the increase in employment in existing firms resulting in a reduction of overall employment rates. Job security may also affect employment through its effect on wages. The insider/outsider literature emphasizes that job security provisions increase the insider power of incumbent workers. This effect results in higher wages for insiders and lower overall employment rates (Lindbeck and Snower, 1987b). Caballero and T4ammour (1997) consider a model in which job security provisions increase the appropriability of capital by labor by increasing capital specifity. That is, a larger part of the capital invested becomes relationship specific and becomes lost if capital separates from labor. While in the shortrun, higher firing costs allow labor to extract higher rents from capital, in the long-run firms invest in less labor intensive technologies, reducing employment demand. Some recent literature has also emphasized the possible impact of job security regulations on the composition of employment. Kugler (2000) proposes a model in which job security regulations provide incentives for high turnover firms to operate in the informal sector. This decision entails producing at a small, less efficient scale in order to remain inconspicuous to tax and labor authorities. In this framework, high job security is likely to increase informality rates. Pages and Montenegro (1999) develop a model in which JS related to tenure biases employment against young workers and in favor of older ones. As severance pay increases with tenure, and tenure tends to increase with age, older workers become more costly to dismiss than younger ones. If wages do not adjust appropriately, negative shocks result in a disproportionate share of layoffs among young workers. Therefore, job security based on tenure results in lower employment rates for the young, relative to older workers, because it reduces hiring and increases firings for young workers. We conclude that higher JS provisions reduce turnover rates and bias the composition of employment against young workers and against employment in the formal sector. The implications for average employment in the economy at large are, however, somewhat less conclusive since they can depend on specific configurations of parameters for the economy. To complicate matters further, by the Coase theorem the impact of job security could be completely "undone" with a properly designed labor contract provided that there are no restrictions on transactions between workers and firms. (Lazear, 1990). Thus, in a world without transactions costs, wages adjust to offset the possible negative impact highlighted 10
in the previous discussion. Given the ambiguity of theoretical models, the magnitude and direction of the impact ofjob security on employment has to be resolved empirically. In the following two subsections, we discuss existing evidence relating JS to labor market outcomes and present some new evidence of our own. 3.2. Empirical Evidence for Latin America and the Caribbean Despite the existence of strict job security regulation in most of the countries of the region, research assessing its impact has been extremely scarce. Fortunately, a recent series of empirical studies assess the impact of job security regulation on employment and turnover rates in Latin America and the Caribbean providing the first systematic evidence of its impact on the labor market.7 Several studies assess the impact of job security on turnover rates in the labor market. Changes in turnover are measured using changes in the duration of jobs (tenure), the duration of unemployment and the exit rates out of employment and unemployment.8 Higher employment exit rates indicate more layoffs (or more quits), while higher exit rates out of unemployment and into formal jobs indicate higher job creation in the formal sector. Other studies examine the impact of job security on employment rates. The definition of employment changes depending on the data considered. In general, most studies focus on employment in large firms, although some also examine more aggregated measures of employment. In addition, a small group of studies also examines the impact of job security on the composition of employment (See Table 2 for an overview of the empirical evidence for Latin America and the Caribbean). A. Turnover Rates The strongest evidence is on the impact of job security on turnover. As predicted by most theoretical models, the empirical evidence confirms that less stringent job security is associated with higher turnover in the labor market. Kugler (2000) analyzes the impact of the 1990 labor market reforms in Colombia. She finds that a reduction in job security is associated with a decline in average tenure and an increase in employment exit rates.9 This decline is significantly larger in the formal sector that is covered by the regulations than in the uncovered or informal sector. In addition, the increase is larger in large firms Most of these projects were developed under the 1DB research network project "Labor Market Legislation and Employment in Latin America" coordinated by J. Heckman and C. Pages. These studies estimate hazard rates. The hazard rate is defined as the probability that a given spell of employment or unemployment ends in a given period conditional on having lasted a given period of time (e.g., one month, one year). 11
and imprecisely determined in the smallest ones. Her results shows similar patterns within tradable and non-tradable sectors, providing a clear indication that the decline in tenure cannot be attributed to contemporary trade reforms. The increasing use of temporary contracts explains only part of the increase in formal sector turnover rates since job stability also declined for workers employed at permanent jobs.1° Her results also indicate that the increase in turnover is larger for those workers who are more protected by high levels ofjob security, that is the middle aged and older men employed in large firms. Kugler also finds a decline in the average duration of unemployment after the reforms. In addition, exit rates out of unemployment increase more for workers who exit to the formal sector than they do for those who exit to informal jobs. Her results show quite similar patterns across sectors and a higher exit rate towards larger firms. Finally, only two-thirds of the increase in the rate of entry into employment can be attributed to higher use of temporary contracts: the rest is explained by increased exit rates into permanent jobs in the formal sector. Her results for different workers suggest that the young and women benefit more from higher exit rates out of unemployment and into the formal sector. The magnitudes of the estimated effects are not negligible. Kugler estimates that after the reform, the increase in probability of exiting employment was 6.4% larger for covered workers than for uncovered ones while the exit rates out of unemployment and into formal jobs increased by 5.9% with respect to exit rates to the informal sector. Saavaedra and Torero (2000) conduct a similar study, evaluating the impact of the 1991 reform in Pen. Like the reform in Colombia, the 1991 reform considerably reduced the cost of dismissing workers. Their analysis shows a consistent decline in average job tenure from 1991 onwards suggesting higher employment exit rates. As in Colombia, the decline is significantly more pronounced in the formal than in the informal sector, but the magnitude of the fall is larger in Pen. Finally, tenure patterns are also quite similar across economic sectors, suggesting that these findings cannot be explained by the far-reaching trade reforms that took place in that country in the early nineties. Finally, Paes de Barros and Corseuil (2000) provide further evidence from Brazil. Their study estimates the impact of the 1988 Brazilian Constitutional reform on employment exit rates. In that year, the In this study tenure is measured by the duration of incomplete spells. 10 In her study, Kugler performs two types of analysis. First, she uses a difference-in-difference estimator to analyze whether changes in average duration of employment (unemployment) are statistically significantly different in the formal than in the informal sector. 12
cost of dismissing workers was raised and therefore a reduction in exit rates would be expected. Their results confirm that aggregate employment exit rates decline in the formal sector relative to the informal sector for long employment spells (two years or more). The credibility of these studies hinges on the validity of the informal sector as a control group unaffected by the reforms. Kugler (2000) shows that while estimates based on formal-informal sector comparisons are likely to be biased, under plausible conditions, such comparisons are still valid, at least as tests of the null hypothesis of no effect of the reform.11 When taken together, these studies provide consistent evidence that dismissal costs and other employment protection mechanisms reduce worker reallocation in the labor market. Unfortunately, these studies do not identifj whether increased worker reallocation is due to increased layoffs, higher quits or a mix of both. Flopenhayn (2000) provides further evidence of the link between JS and worker turnover rates in Argentina. In 1991, the government of Argentina deregulated the use of temporary and fixed-term shortduration contracts. In 1995, additional contractual forms were allowed including a three-month trial period. Such contracts reduced or eliminated the cost of terminating an employment relationship. Flopenhayn (2000) finds that after 1995, employment exit rates increase substantially for short employment duration while they remain constant for long durations. This increase in separations is due to a rise in both quits and layoffs, although the increase in layoffs is higher. Summarizing, the evidence provided in this section indicates that JS regulations protect workers against the risk of losing a job. From this point of view, the recent reforms have reduced the income security of formerly protected workers. However, the evidence also suggests that stringent JS provisions reduce exit rates out of unemployment and into formal jobs, thus prolonging the duration of unemployment. Thus, recent labor market reforms have increased the probability of an unemployed worker finding a job in the formal sector. Second, she estimates an exponential duration model to control for changes in demographic covariates, pooling data from before and after the reform and using interaction terms to assess the differential impact in the formal and in the informal sector. Kugler shows that lower severance pay may induce high-turnover informal firms to move to the formal sector. Under the assumption of no overlap in the distribution of turnover between covered and uncovered firms, or that entry to the covered sector comes from the high-end --or at least from the end that is higher than the formal sector--, this shift results in higher turnover in both the 13
B. Average Employment The available evidence for LAC countries shows a consistent, although not always statistically significant, negative impact of JS provisions on average employment rates. Saavedra and Torero (2000) and Mondino and Montoya (2000) use firm-level panel data to estimate the impact of job security on employment in Pen and Argentina, respectively. Both studies estimate labor demand equations in which an explicit measure of job security appears on the right hand side of the equation, and both find evidence that higher job security levels are associated with lower employment rates. 12 In the case of Peru, Saavedra and Torero find that the size of the impact of regulations is correlated with the magnitude of the regulations themselves. Thus, the impact is very high at the beginning of their sample (1987-1990) coinciding with a period of very high dismissal costs (see Table 1.A). Afterwards, and coinciding with a period of deregulation, the magnitude of the coefficient declines, only to increase again from 1995 onwards, after a new increase in dismissal costs. Their estimates for the long-mn elasticities of severance pay are very large (in absolute value): between 1987 and 1990 a 10% increase in dismissal costs, keeping wages constant, is estimated to reduce long-mn employment rates by 11%. In subsequent periods, the size of the effect becomes smaller but is still quite large in magnitude (between 3 and 6%). In Argentina, the estimated long- run elasticity of a 10% increase in dismissal costs is also between 3 and 6%. 13 formal and the informal sector. Fortunately, higher turnover in the informal sector biases the difference-in-difference estimator downwards. Therefore, a positive estimate still provides substantial evidence of increased turnover in the formal sector. 12 The data for the Peruvian study covers firms with more than 10 employees in all sectors of the economy. The Argentinean study only covers manufacturing firms. Given the nature of these surveys, they are better proxies for formal employment than for employment as a whole. The data used in these two studies does not capture job creation by new firms, since both panels are based on a given census of firms, without replacement. While the estimated job-security elasticity in Argentina is much lower (in absolute value) than the wage elasticity reported in Table 2, in the Peruvian case, this elasticity is larger. This is somewhat surprising since job security reduces job creation and also slows down employment destruction. Therefore, it might be expected that the JS elasticity would be smaller than the wage elasticity in absolute value. One explanation for the seemingly high elasticity found in the Peruvian study is that this measure is upwardly biased by a simultaneity problem arising from the job security measure. Thus, both the Peruvian and the Argentinean studies construct explicit measures ofjob security based on: JSj1=4· TJEPJESPJE Where ) is the layoff rate in sectorj in sector t, 1 is average tenure in sectorj, time period t, 1 is the share of firms in sectorj, time period t, that are covered by regulations and SP1 is the mandatory severance pay in sectorj, given average tenure 1 This measure provides variability across sectors and periods, and therefore it affords a more precise estimation of the impact of job security than before-after types of comparisons. Yet, such measure may also be correlated with the error term in a labor demand equation since the tenure structure of a firm might be correlated with its employment level. The fact that average layoff rates vary by sector may also lead to simultaneity if sectors with higher layoffs have lower employment. Thus, periods or sectors with low employment may be associated with less job creation, high average tenure and, consequently, high measures ofjob security. The Argentinean study shows that fixing tenure to the period average reduces the estimated elasticity of JS. Thus, a JS elasticity between 1/3 and 2/3 of the wage elasticity seems a more realistic estimate of its impact. 14
Table 2: Summary of existing evidence on the impact of job security (JS) in Latin America
A. Studies that analyze exit rates into and out of employment
Study
Country Data
Results
Kugler (2000)
Colombia Household data
Decline in JS leads to reduction in employment and unemployment duration. Also hazard rates out of employment and out of unemployment increase. Some effect due to temporary contracts but not all
Saavaedra and Torero (2000)
Peru
P. de Barros and Corseuil (2000) Brazil
Household data
Lower JS is associated with lower average tenure. Higher decline in formal sector. Hazard rates increase just at the end of
probation period.
Employment Surveys, Administrative data and Higher JS associated with a decline in
Household surveys
employment exit rates in formal in relation to
informal sector.
Hopenhayn (2000)
Argentina Household data
B. Studies that analyze average employment and unemployment
Study
Country Data
Downes et al. (2000)
Barbados Aggregated employment. Annual. It covers
large firms (>10 emp)
Deregulation of temporary contracts leads to increase in hazard rates in short but not in long spells Results Negative effect of JS on labor demand (LD). Coeff. Significant at 10%
Saavedra and Torero (2000)
Peru
Firm and sector-level data. Bimonthy 1986-96. Negative effect of JS on LD when using
Quarterly 1997-98. Formal firms with more sector level-data for whole period. By
than 10 employees. Balanced panel (it does subperiods, JS has a negative effect from
not account for firm creation or destruction) 1987 to 1994, and no effect since then.
Mondino and Montoya (2000) Kugler (2000) P. de Barros and Corseuil (2000) Pages and Montenegro (2000)
Argentina Panel of manufacturing firms. It does not account for firm creation, Colombia Household data on employment.
Negative effect of JS on LD. The coefficient in unbalanced panels is slighly more negative than in balanced ones. Decline in JS in 1990 brings a decline in unemployment rates. Based on computing the net effect of changes in hazard rates, in and out of U induced by the reduction in JS.
Brazil
Monthly establishment-level data. 1985-1998 Two step procedure. First, find parameters for
Manufacturing. Firms employing 5 or more labor demand (LD) function for every month.
workers
Then see whether those parameters change
with labor reforms and other development.
They find no effect of JS on LD parameters.
Chile
Household data on employment. Annual 1960-1998
Negative but not statistically significant effect of JS on aggregated employment.
MarquCz (1998)
Cross- Cross-section data for Latin America, Country Caribbean and OECD countries,
C. Studies that analyze the composition of employment
Study
Country Data
Rank indicator of Job Security. JS is not significantly associated with lower employment once GDP per capita is accounted for. Results
MarquCz (1998) Pages and Montenegro (2000)
Cross- Cross-section data for Latin America, Country Caribbean and OECD countries, Chile Household Survey Data. 1960-1998
Self-employment rates are positively associated with JS even after accounting for differences in GDP per capita. JS is associated with lower employment rates for young workers and higher employment rates for older ones. No significant effect on U for young, middle age or older workers.
15
In a very different type of study, Kugler (2000) computes the net impact of the Colombia 1991 labor reform on unemployment rates. Using unemployment and employment exit rate estimates for periods before and after the reform, she finds that the reforms cause a decline in unemployment between 1.3 and 1.7 percentage points. Thus, as in Mondino and Montoya (2000) and Saavedra and Torero (2000), Kugler's estimates indicate that the positive impact on the hiring margin outweighs the negative impact on the firing margin, resulting in a decline in unemployment rates. Other studies find negative, but not statistically significant, effects of job security on average employment rates. Pages and Montenegro (1999) find that JS has a negative but not statistically significant effect on overall wage-employment rates in Chile. Similarly, Marquйz (1998), using a cross-section sample of Latin American and OECD countries finds a negative but not statistically significant coefficient of job security on aggregate employment rates. Table 3 summarizes the various estimates of job security on employment. (The Fleckman and Pages results are discussed below). Thus, while the theoretical models exhibit some ambiguity regarding the impact of JS provisions on long-run employment rates, the empirical evidence for LAC is consistent across studies. To complement these analyses, we examine two other sources of evidence. First, we review the existing evidence on the impact of JS on employment in OECD countries. Second, in section 4, we provide new evidence combining employment, unemployment and job security measures from a panel of LAC and OECD countries.
Table 3: Summary of Long- Run JS Elasticities Study Saavedra & Torero (2000) Mondino & Montoya (2000) High estimate** Low estimate*** Pages & Montenegro (1999) Heckman & Pages (2000), FE* Heckman & Pages (2000), RE* Heckman & Pages (2000) OLS*
Mean -0.40 6 -0.684 -0.305 -0.1198 -0.0516 -0.0502 -0.0502
LE. 0.06
Employment Rate Employment in Large firms
0.0 145 0.0060 0.2440 0.0318 0.0 168 0.0 168
Employment in Large firms Employment in Large firms Wage-Employment/Population Total Employment/Population Total Employment/population Total Employment/population
Notes: *Estimates for LAC only. **Based on Table 9, Mondino & Montoya (2000), ***Based on Table 10, option B. Mondino & Montoya (2000)
16
The evidence from OECD countries reinforces the results found for LA. Thus, with the exception of Anderson (1993), who finds a positive association between dismissal costs and long-mn employment, the rest of the studies found a negative impact of JS on employment. Using panel data from OECD countries, Lazear (1990) shows that more stringent job security measures are associated with lower employment and labor force participation rates. Grubb and Wells (1993) find a negative correlation between JS and wage-employment rates. Addison and Grosso (1996) reexamine Lazear's estimates using new measures of job security across countries and find similarly negative effects on employment rates. Nickell (1997) finds a negative effect of JS provisions on total employment rates and no effect on primeage male employment rates. Finally, a recent OECD (2000) study finds a negative but not statistically significant effect of JS on total employment rates. In contrast, the evidence regarding the effect of JS on unemployment in OECD countries is ambiguous but there are conceptual reasons for being so. While Blanchard (1998), Esping-Andersen (forthcoming), Jackman et a! (1996) and Nickell( 1997) among others find no effect of JS on unemployment, Lazear (1990), Elmeskov et a! (forthcoming) and Scarpetta (1996) find positive effects. Yet, it should not be a surprise that a negative impact on employment is not necessarily reflected in a positive effect on unemployment. If workers' participation decisions are influenced by JS policies (as shown by Lazear, 1990), a reduction in employment will be associated to a decline in participation rates. This is particularly true for workers with lower attachment to the labor force or with less access to unEmployment Insurance benefits. C. The Composition of Employment Some recent evidence sheds new light on the possible impact of JS on the composition of employment in LAC. Marquйz (1998) constructs a JS indicator for LAC and OECD countries and uses it to estimate the effects of JS on the formal/informal distribution of employment. Re finds that more stringent JS provisions are associated with a larger percentage of self-employed workers. In a study of Chile, Pages and Montenegro (2000) find that more stringent job security is associated with a substantial decline in the wage employment-to-population rates of young workers and an increase in the wageemployment rates of older workers. Their results also suggest that this composition effect is driven by the 17
high costs of dismissing older workers relative to younger ones created by job security provisions related to tenure. 4. New Evidence In this section, we exploit substantial cross-country and time series variability in job security provisions to estimate whether the negative effects of JS encountered in some of the individual-country studies in LAC generalize to a wider sample of countries and reforms. A. The Data We construct a data set that spans industrial and LAC countries. To do so we proceed in two stages. We first collect employment and unemployment data for industrial countries from the OECD statistics. Second, we use the OECD definitions of these variables, to construct the same indicators out of Latin American Household Surveys. Table 4 provides summary statistics for the overall sample, the OECD sample (excluding Mexico, which is included in the LAC sample) and the LAC sample. Table 5 describes the household surveys used to compute the LAC variables. Finally, to characterize job security, we use the index ofjob security described in section 2. The number of countries and the average number of observations per country in our sample varies between 36 and 43 countries and between 1 and 5 observations per country, respectively. Among the countries represented, around 28 belong to the sample of OECD countries, while 15 are from the LAC region. Regarding the period spanned in our sample, for most LAC countries, there are one or two observations from the eighties and one or two from the nineties. The OECD sample only covers the nineties. In relation to the variables used in this exercise, it should be noted that all employment rates are measured as a percentage of working age population and all unemployment rates as a percentage of active economic population (See the Appendix for a definition of the variables used in this study) Table 4 shows some remarkable differences between the OECD and the LAC samples. As noted in section 2, average job security is higher in Latin America and the Caribbean than in OECD countries. In contrast, all employment rates (except for prime-age female employment) are higher and all unemployment rates are lower in the LAC region than in industrial countries. Especially notable are the higher share of 18
self-employment and the much lower share of long-term unemployment (more than 6 months) in LAC. Finally, union density and female participation are both lower in the LAC region. B. Methodology and Results By constructing our own data set from individual household-level surveys, we are guaranteed that all the labor market variables are comparable and reliable. One drawback of our data is that we only have a few time series observations per country (usually three or four), and not necessarily from consecutive years. Given the nature of the data, we decided not to average observations from a given period --as done in most of the OECD studies on job security--and instead control for the state of the business cycle in a given year using GDP growth. We use a reduced form approach to investigate whether countries and periods with more strict job security regulations are associated with lower employment or higher unemployment rates. Thus we estimate an average net effect of JS as it operates through intermediate variables which we do not include in the regression. In this paper, we do not estimate the theoretically more appropriate state-contingent demand functions because we lack the information on the states of demand confronting individual firms. JS costs govern the marginal costs of labor when firms are firing, but they also affect overall labor demand through their effect on expected (across states) labor cost. It is the latter effect that we attempt to identify. Since most of the variation is cross-sectional, we use different types of variables to control for countryspecific factors that may be correlated with job security. First, we use demographic controls such as the share of the population between 15 and 24 and female participation rates. These variables account for the fact that high job security countries in the south of Europe and Latin America tend to have low female participation and a large share of youth population. Since both factors affect overall employment rates, not including them in the specification may lead to substantial biases in the estimates. We protect against common country-specific unobservables that remain constant over time and that may affect both left hand side and right hand side variables by including country-specific fixed effects in a set of regression specifications reported below. Second, we use GDP (measured in 1995 U.S. dollars) to control for 19
differences in development levels across countries. We also include a dummy variable for LAC to control for regional differences not controlled by GDP levels14. Most of the variability in our sample comes from differences across countries and regions, and from some time series variance within the LAC sample. There is very little time-series variability in the OECD sub-sample. Given this variation, fixed effects (FE) estimates are likely to be very imprecise because they only use the time-series variation within the LAC sample. Instead, random effects (RE) or pooled OLS estimates, that use both the cross-section and the time-series variation included in the sample, are likely to produce estimates with smaller standard errors. Yet, the latter estimates will be biased if variables included as controls are correlated with country specific error terms. To protect against the bias that results from using one estimator, we estimate our basic specification by pooled OLS, RE and FE, comparing whether these different methodologies yield similar point-estimates. The results, presented in Tables 6.a to 6.c, are striking. First, the point-estimates for the JS coefficient in the total employment specifications are very similar across estimation methodologies. The three estimates suggest a large negative effect of JS on employment rates. This effect is strongly statistically significant in the OLS and the RE estimates while it is not statistically significant, at conventional levels, in the FE case. One obvious advantage of using a cardinal measure of JS is that we can quantifj the impact of these provisions on employment. The magnitudes of JS elasticities are quite large: an increase in expected dismissal costs equivalent to one month of pay is associated with a 1.8 percentage points decline in employment rates. Given that in Latin America the average dismissal cost in 2000 was 3.04 months (See Graph 1), the estimated loss in employment --as a percent of total working population-due to JS provisions is about 5.5 percentage points. In addition, OLS, FE and RE estimates suggest that JS does not affect the employment rates of all workers in the same fashion. Thus, while the impact on prime-age male employment rates is half the impact on total employment, the impact on young workers' employment rates is almost two times larger. The magnitudes are huge. The OLS and the RE estimates suggest that JS reduces LAC youth employment rates by almost 10 percentage points. This effect is even larger in the FE estimates. Moreover, these magnitudes are consistent with the ones obtained in Pages and Montenegro (1999) for Chile. 14 These specifications should include a measure of labor costs that include wages and other non-wage labor costs. Unfortunately, a 20
Our estimates of the effect of JS on female employment rates, self-employment and unemployment rates are less consistent. The point estimates for female employment rates change from negative to positive across methodologies, but in no case are the estimates statistically significant. These results suggest that women are less negatively affected by JS than men but, as we will show, these results are not robust across regional sub-samples. The estimates of the effect of JS on self-employment also change signs across OLS, FE and RE estimates. Thus, while the pooled estimates suggest a positive and statistically significant association between the strength of JS provisions and self-employment (as found by Marquйz (1998)), the FE estimates show a negative and also statistically significant relationship between both variables. It is clear that more empirical work is required to reach a definitive conclusion on the relationship between JS and selfemployment. Finally, the empirical results on unemployment also greatly depend on the methodology used to estimate the parameters. While OLS and RE yield positive (and often statistically significant) coefficients on JS in all the unemployment specifications, FE yields negative and statistically insignificant results. We do not find a significant relationship between the proportion of workers unemployed for more than 6 months and the strictness of JS provisions. Since there is no a priori relationship between disemployment and unemployment, these results are not surprising, especially given differences across regions in the levels of social insurance. Divergence across estimation methods may result from regional differences in the relationship between JS and some of the variables. This is particularly relevant for our exercise since FE estimates discard practically all of the information for OECD countries. We therefore investigate whether our results are driven by any of the two sub-samples, by estimating separate coefficients for LAC and OECD countries. The results from this exercise are presented in Table 7. While this approach results in small samples and lower statistical significance, the results are still quite remarkable. First, in all the employment specifications, with the exception of female employment rates, the coefficients on job security are negative across regions and estimation methods. In addition, most of the coefficients are highly statistically significant. complete and comparable measure of labor costs across countries and time is not available. 21
Second, with one exception, all coefficients of the effect of job security on unemployment rates are positive both in OECD and in LAC countries. However, the impact on unemployment rates seems much larger in the industrial country sub-sample, in particular for women and youth. It should not come as a surprise that the effect of JS on unemployment rates is smaller in developing countries. In the absence of unemployment insurance or other income support programs, workers either quickly find other (less attractive jobs) or drop out of the labor force.15 The positive and statistically coefficient of GDP level in the unemployment regressions reported in Tables 6a-6c confirms this effect. Third, the ranking of effects between total, male and young workers employment rates is preserved. The point estimates tend to be larger (in absolute value) in the LAC sample. It is very likely that the higher level and variability of JS in this region contributes to these larger (in absolute value) point estimates. It is quite puzzling, however, that the estimates for female employment (and unemployment) rates are so different across regions. Thus, while, JS is negatively associated with female employment rates in the OECD sub-sample, this relationship is actually positive in the LAC sample. The added worker effect is more evident in LAC, where adult female attachment to the labor force is still weak. Understanding Gender Differences in the impact of JS remains one important issue for further research. Finally, the evidence of the impact of job security on the formal/informal composition of employment is not conclusive. A comparison of our estimates for LAC with the elasticities obtained from the individual-country studies (see table 3), suggest that the decline in employment associated with JS is greater in the covered (formal) sectors--such as the manufacturing sector or sectors with large-firms--than in the aggregate.16 This would imply that an increase in job security is associated with a decline in formal employment and an increase--although not enough to compensate the decline in formal jobs--in informal employment. However, the estimates for self-employment--usually considered part of the informal employment--in Table 7 Panel A, indicate an unstable effect of JS on self-employment. While the coefficient resulting from OLS estimation is positive and significant, the coefficient resulting from fixed effect estimation is negative and statistically significant. More research is necessary to understand the relationship between uncovered employment and job security in Latin America. In the case of Chile, Montenegro and Pages (1999) found that the large effects of JS on youth employment rates were compensated with a large decline in participation rates with no significant effects on unemployment. 16 The Heckman and Pages elasticities, reported in Table 3, are obtained from a model identical to the one reported in Table 6, but where job security provisions enter the specification in logs. 22
5. Conclusions In a recent article, Freeman (2000) writes that "the institutional organization of the labour market has identifiable large effects on distribution, but modest hard-to-uncover effects on efficiency." This view is shared by many economists (see Abraham and T4ouseman (1994) and Blank and Freeman (1994)). However, the results summarized in this paper suggest that job security regulations have a substantial impact on employment and turnover rates both in Latin America and in OCED countries and thus substantially affect the efficiency of the labor market. The assertion that job security does not have any impact on employment rates is based on evidence on unemployment, not on employment. However, employment and unemployment are not minor images of each other. In addition, while there is substantial evidence that unions reduce earnings inequality in industrial countries, there is no evidence that job security provisions reduce income inequality. Indeed, given that job security reduces the employment prospects (and possibly wages) of younger and less experienced workers, who bear the brunt of regulation, it is likely that regulation widens earnings inequality across age groups. Thus, there is no trade off between employment and inequality associated with job security provisions. Such provisions worsen both. The choice of labor market institutions matters. What policy lessons can be drawn from these results? Our evidence suggests that job security provisions are an extremely inefficient and inequality-increasing mechanism for providing income security to workers. They are inefficient because they reduce the demand for labor; they are inequality-increasing because some workers benefit while many others are hurt. Their impact on inequality is multifaceted: Job security increases inequality because it reduces the employment prospects of young, female and unskilled workers. It also increases inequality because it segregates the labor market between workers with secure jobs and workers with very few prospects of becoming employed. Finally, job security provisions increase inequality if, as predicted by some theoretical studies and most of the available empirical evidence, they increase the size of the informal sector. In this light, it seems reasonable to advocate the substitution of job security provisions by other mechanisms that provide income security at lower efficiency and inequality costs. However, reducing dismissal costs is a difficult policy to implement in most countries. The persistence of these policies can be 23
explained by a demand for income security for groups with political power (Caballero and T4ammour, 2000). A demand for income security arises because job security lowers flows out of unemployment and into employment. Although job security reduces the probability of exiting employment, conditional on having lost a job, the probability of finding a new one is reduced. This produces a sense of insecurity among protected workers, who exert pressure to maintain high levels ofjob security provisions. A balance of power that favors insider workers helps to sustain job security provisions. Thus, those workers most likely to benefit from such provisions are also more likely to be represented in the political process. Instead, outsider workers are less likely to influence policy. Reform minded policymakers should pursue broad coalitions including representatives of outsider workers --such as young, female, unemployed or discouraged workers-- to obtain support for labor market reforms. 24
References Abraham, K. and Houseman, S. 1994. "Does Employment Protection Inhibit Labor Market Flexibility: Lessons From Germany, France and Belgium," in Rebecca M. Blank, ed., Protection Versus Economic Flexibility: Is There A Tradeoff? (Chicago: University of Chicago Press). Addison, J.T. and Grosso, J.L. 1996. "Job Security Provisions and Employment: Revised Estimates." industrial relations. 3 5(4). Aguirregabiria, V. and Alonso-Borrego, C. 1999. "Labor Contracts and Flexibility: Evidence from a Labor Market Reform in Spain" Manuscript. Department of economics, University of Chicago. Anderson, P.M. 1993. "Linear Adjustment Costs and Seasonal Labor Demand: Evidence from Retail Trade Firms." Quarterly Journal of Economics. 108(4): 1015-42. Bentolila, S. and Saint-Paul, G. 1994. "A Model of Labor Demand with Linear Adjustment Costs." Labour Economics. (1):303-26. Bentolila S. and Bertola, G. 1990. "Firing Costs and Labour Demand: How Bad is Eurosclerosis?" Review of Economic Studies. 57. 38 1-402. Bertola, G. 1990. "Job Security, Employment and Wages." European Economic Review. 34:851-86. Blanchard, 0. 1998 "Thinking about Unemployment" Manuscript. Department of Economics, MIT. Blank, R. and Freeman, R. 1994. "Does a Larger Social Safety Net Mean Less Economic Flexibility?" In: R. Blank and R. Freeman, editors. Working Under Different Rules. New York: Russell Sage. Caballero, R. and Hammour, M. 1997. "Jobless Growth: Appropriability, Factor Substitution and Unemployment." Carnegie-Rochester Conference Series on Public Policy. 48:51-94. 2000. "Institutions, Restructuring and Macroeconomic Performance." Paper prepared for the XII World Congress of the International Economic Association, 25 August 1999. Downes, A. et al. 2000. "Labor Market Regulation and Employment in the Caribbean." Research Network Working Paper R-388. Washington, D.C., United States: Inter-American Development Bank. Elmeskov, J., Martin J.P. and Scarpetta, S. Forthcoming. "Key Lessons for Labour Market Reforms: Evidence from OECD Countries' Experience". Swedish Economic Policy Review Esping-Andersen, G. Forthcoming. "Who is Harmed by Employment Regulation?" In Esping-Andersen, G. and Regini, M. Eds., Why De-Regulate Labour Markets? Oxford University Press, Oxford. Freeman, R. B. 1994. Working Under DUjerent Rules. NY: Russell Sage Foundation. 2000. "Single Peaked vs. Diversified Capitalism: The Relation Between Economic Institutions and Outcomes," NBER Working Paper 7556. Cambridge, United States: National Bureau of Economic Research. Grubb, D. and Wells, W. 1993. "Employment Regulation and Patterns of Work in EC Countries." OECD Economic Studies No. 21. Winter. Hopenhayn, H. 2000. "Labor Market Policies and Employment Duration: The Effects of Labor Market Reform in Argentina." Research Network Working Paper. Washington, D.C., United States: Inter- 25
American Development Bank. Forthcoming. Hopenhayn, H. and Rogerson, R. 1993. "Job Turnover and Policy Evaluation: A General Equilibrium Analysis." Journal of political economy. 10 1(5). Jackman, R., Layard R. and Nickell, 5. 1996. "Combating Unemployment: Is Flexibility Enough". Center for Economic Performance Discussion Paper No. 293. Kugler, A. 2000. "The Incidence of Job Security Regulations on Labor Market Flexibility and Compliance in Colombia: Evidence from the 1990 Reform." Research Network Working Paper R-393. Washington, D.C., United States: Inter-American Development Bank. Lazear, E. 1990. "Job Security Provisions and Employment." The Quarterly Journal of Economics. August. Lindbeck, A. and Snower, D. 1987a. "Union Activity, Unemployment Persistence and Wage-Employment Ratchets" European Economic Review, 31, 157-167. Marquйz, G. 1998. "Protecciуn al empleo y funcionamiento del mercado de trabajo: una aproximaciуn comparative" Mimeo Inter-American Development Bank. Marquйz, G. and Pages, C. 1998. "Ties That Bind: Employment Protection and Labor Market Outcomes in Latin America." Research Network Working Paper 373. Washington, D.C. United States: Inter-American Development Bank. Mondino, G. and Montoya, 5. 2000. "Effects of Labor Market Regulations on Employment Decisions by Firms: Empirical Evidence for Argentina." Research Network Working Paper R-39 1. Washington, D.C., United States: Inter-American Development Bank. Nickell, 5. 1997. "Unemployment and Labor Market Rigidities: Europe versus North America." Journal of Economic Perspectives. 11(3): 55-74. OECD. 1993. Employment Outlook. Paris, France. OECD 1999. Employment Outlook. Chapter 2. "Employment Protection and Labour Market Performance." Paris, France. June Paes de Barros, R. and Corseuil, C. H. 2000. "The Impact of Regulations on Brazilian Labor Market Performance" Mimeographed document. Pages, C. and Montenegro. C. 1999. "Job Security and the Age-Composition of Employment: Evidence from Chile." Working Paper 398. Washington, D.C., United States: Inter-American Development Bank. Risager, 0. and Sorensen, J.R.1997. "On the Effects of Firing Costs When Investment Is Endogenous: An Extension of a Model by Bertola" European Economic Review; 41(7) pp.1343-53. Saavedra, J. and Torero, M. 2000. "Labor Market Reforms and their Impact Over Formal Labor Demand and Job Market Turnover: The Case of Peru." Research Network Working Paper R-394. Washington, D.C., United States: Inter-American Development Bank. Scarpetta, 5. 1996 "Assesing the Role of Labour Market Mobility in Europe: An Empirical Analysis using the EU's Labour Force Survey" OECD Economic Studies, No 26 pp. 43-98. 26
Appendix Construction of the index of job security The job security index is constructed according to the following formula: Index = $S'(1 S)(b+1 + aSP+( + (1-- a)S1) where j denotes country, S is the probability of remaining in a job, f3 is the discount factor, T is the maximum tenure that a worker can attain in a firm, b1,+1 is the advance notice to a worker that has been i years at a firm, a is the probability that the economic difficulties of the firm are considered a justified cause of dismissal, SP( is the mandated severance pay in such event to a worker that has been i years at the firm, and finally, SP+fc denotes the payment to be awarded to a worker with tenure i in case of unjustified dismissal. 17 The constructed index measures the expected discounted cost, at the time a worker is hired, of dismissing a worker in the future. The assumption is that firms evaluate future costs based on current labor law. The index only includes statutory provisions, and thus, it does not include provisions negotiated in collective bargaining or included in company policy manuals. It addition, it does not include dismissal costs that are ruled by a judge if a firm is taken to courts. This assumption explains why dismissal costs-- according to our index--are zero in the U.S., despite the substantial potential costs associated with legal actions. High values of the index indicate periods or countries of high job security, whereas lower values characterize periods or countries in which dismissal costs are lower. By construction, this index gives equal weight to notice periods and to severance pay since both are added up in the calculation of the dismissal costs. This index however gives a higher weight to dismissal costs that may arise soon after a worker is hired--since they are less discounted at the time of hiring-- while it discounts firing costs that may arise further in the future. In computing the index, we assumed a common discount rate and a common turnover rate of 8% and 12%, respectively. The choice of the discount rate is based on the average return of an internationally diversified portfolio. Finally, the choice of turnover rate is based on the fact that real turnover rates are 27
unobservable in countries with job security provisions since the turnover rate, is itself affected by job security. We therefore choose to input all countries with the observed turnover rates in the U.S., the country in the sample with the lowest job security. The minimum tenure at a firm is considered to be one year, and the maximum is assumed to be twenty years. We compute SPJC and SPJUC based on the two different sources. For LAC countries, we use the legal information summarized in Table 1 .A. This information was directly obtained from the Ministries of Labor of the region. In the case of Colombia we consider that severance payment prior to the 1990 reform was one month and V2 per year of work instead of one--as prescribed by law--to include that prior to the 1990 reform, advance withdrawals to the seniority premium fund were accounted in nominal terms. High inflation rates implied that this practice substantially increased overall dismissal costs. For OECD countries, we use the legal information summarized in OECD (1999). In all Latin American countries but Argentina and Chile, economic conditions are not a just cause for dismissal. Consequently, we assumed a=O for those countries. Instead, in Argentina, Chile, economic conditions were a justified cause of dismissal and therefore, a=1. For OECD countries, we used the information summarized in Table 2.A.2 OECD (1999) to parameterize severance payments and advance notice. In all cases, but in Spain, a =1. In Spain, mandatory severance pay in the case of unjustified cause was substantially larger than severance pay for just cause. Consequently most workers fired for just cause appealed to the courts, and there was a high probability that a judge would declare a dismissal unjustified. Based on Bertola, Boeri and Cazes (2000), we assume that prior to the 1997 reform, a=0.2. After 1997, the scope for ambiguity was reduced and a=0.5. For Canada, we used the information relevant to the federal jurisdiction (although JS provisions may vary across states). Finally, in some European countries statutory dismissal costs vary across blue and white-collar workers. To obtain a single measure per country, we compute a separated index for blue and white-collar workers and performed a simple average among the two. (See OECD, 1999 for a description of dismissal costs in OECD countries and the cost divergences between blue-and white-collar workers.) 28
Definition of Variables used in Empirical Section Total Employment. All employed workers between 16 and 65 that declared having a job in the week of reference. It is measured as % of total population 16-65. All measures of aggregate employment include formal and informal workers. They also include unpaid workers. Source: OECD statistics and LAC household Surveys. Prime Age-Male Employment: % of men 25-50 years old employed in the week of reference. Source: OECD statistics and LAC household Surveys. Prime Age-Female Employment: % of female 25-50 years old employed in the week of reference. Source: OECD statistics and LAC household Surveys. Youth Employment: % of people 16-24 years old employed in the week of reference. Source: OECD statistics and LAC household Surveys. Self-Employment: Share of non-agricultural workers in self-employment or as owners of firms. Source: Maloney (1999) Total Unemployment: # of people 16-65 that did not work in the week of reference but are actively looking for a job as a % of total active population in that age group. Source: OECD statistics and LAC household Surveys. Prime-Age Male Unemployment: # of men 25-50 that did not work in the week of reference but are actively looking for a job as a % of male active population in that age group. Source: OECD statistics and LAC household Surveys. Prime-Age Female Unemployment: # of people 25-50 that did not work in the week of reference but are actively looking for a job as a % of female active population in that age group. Source: OECD statistics and LAC household Surveys. Youth Unemployment: # of people 16-24 that did not work in the week of reference but are actively looking for a job as a % of active population in that age group. Source: OECD statistics and LAC household Surveys. 29
Long-term nnemployment: # of people 16-65 that have been without a job, and actively looking for one for more than 6 months as a % of total active population in that age group. Source: OECD statistics and LAC household Surveys. Female Participation: % of total female workers 16-65 that are either employed or actively seeking one. Source: OECD statistics and LAC household Surveys. GDP: Gross Domestic Product measured in 1995 US dollars. Source: World Bank. Population 15-24: Proportion of population in this age group. Source: UN Population Statistics 30
Table 1.A: Legislation Concerning Conditions of Dismissal in 1990 and 1999. X=monthly wages, N=Years of Tenure
Date of I Advance Notice Reform
Seniority Premium
Compensation if worker quits? Compensation for dismissal due to To whom the
economic reasons
reforms apply?
Argentina Bahamas Barbados Belize
I 1990
1999
None 1-2 months 1-2month
None
1/2-1
month
None Negotiable in practice
lmonth, None 1/2-1
month
No changes No changes No changes
1990** I 0 0 0 0
1999 0 0 0 0
1990
I
0
0
0
1/6x*N fN>10
1999
1990
0 2/3x*N, Mm 2
months
0
Negogtiable
0
f0.41 *x*N N> =2
No changes
1/4x*N IfN>5
1999 No changes No changes No changes No changes
Upper lini to compensati for dismiss 1990 Max. urn. Max. x*N Max 42 14
Bolivia Brazil Chile Colombia Costa Rica Ecuador
None 3 months 1988 1 month 1991 1 month 1990 45 days
None None
1 month lmonth
No
0
changes
No Fund (8% wage
changes
+ r)
No
0
changes
No
x*N
changes
Double
retroactivity given lacko) inflationary adjustment of withdrawals
No
0
changes No
Fund (8%
changes
wage+ r)
El Salvador 1994 0-7 days
No
0
changes
Guatemala None
0
0
0
0 Fund (8% wage+ r) 0 Fund (8% wage+r) 0 Fund (8% wage+r) 0 0
1 x*N. fN>=5 0 No Fund
No changes 0 1/2 x*N (2) fN>=7 No changes
1 x*N.
No changes
0.4*FUND 1 x*N. (3)
No changes No changes
x*4.0 fN5 x*4.0 fN5 x*6.6fN10 x*6.6fN10
0 Seniority Premium 0 0
x*16.5fN15 x*21.5fN15
x*21.5 fN--20 x*28.5 fN--20
0
x*N No changes
No changes 0
1/4x*N plus 3*x fN <3 plusx*N fN 3 - 25 plus pension i N>=25 x*N
No changes x*N
0 fbankrupcy Changes in max. x
0
2 days-4 months
No changes
fbankrupcy. x*N
otherwise
All workers Max. x*I All workers
Max. x*
All workers
Max.
w
4 mi wages
32
Guyana
1997 1/2 month lmonth
0
IfN>1
Honduras None lday-2
No
0
months changes
Jamaica None
2-12
No
0
weeks changes
Mexico
None
0- 1
No
0
month changes
Nicaragua 1996
1-2
0
0
months
Panama
1995 lMonth
No
changes
1/4*X*N fN10
Paraguay Peru
None 1996 1995
1-2 months 0
No
0
changes 0 Determined
by judge in legal
0 0 0 0 0 1/4*X*N 0 Fund (8% wage+r)
0
0
Negotiable 1/4*x*Ni
N=1-5
In practice, 1/2 *x*N i
N=5-10
21/2 weeks perN
0
0
x*N No changes
0
0 1/3*x*Nzfx25 No changes
1/2 *X*N if x>5
0
0 2/3 x*N (Mi 3*x) No changes
0 x*NifN13
Negotiated x*NfN13
3x*N + In practice, 2 x*N.
3x*N +
2/3x*N if N>3
2/3x*N if N>3
1/4*X*N fN>10
1/4*X*N
X*NZfN<=1
3/4X*Ni
N<1 0
3*x if N2 7.5*x+1/4*Xi
3*x + 3/4*x*Nj
N> =10
N>2<10
9*x+ ]/4*x*Nj
N>=10
0
0
1/2x*N
1/2x*N
Fund (8% wage+r)
Seniority Premium
3 x*N FUND+1.5*x* N
1991
Proceedings
Rep. Dom. 1992
1/4-1
No
month changes
0
0
0
0
1/2*x*N
Suriname None 1/4-6
0
month.
0
0
Negotiated
Trin. and None 2 months
0
0
0
0 1/3 x*NfN = 1-
Tob.
Uruguay None
0
0
Venezuela 1997 1/4-3
No
x*N Nochanges
x*N
2x*N
0 X*N
4, 1/2x*NifN>5
0
x*N
2x*N
2/32x*N
months, changes
Source: Ministries of Labor in the region **In Brazil, the date refers to 1988 (instead of 1990)
.67*x*Ni N1-4 .74 *x*N i N>=5, Negotiated No changes No changes x*N
All workers Max. x*N New employees 1991 New Max. x*N Employees 1995 All workers 1996 All workers New employees Max. x*N All workers
33
Graphi: Job Security Index (Expected discounted cost of dismissing a worker, in multiples of monthly wages)
Behze Guyana Jamaa ____________ Barbados ________ Tnnidad & Tobago _______________
Brazil ___________ Paraguay Uruguay Nicaragua _____________ Panama ________ Domican Republic Venezu&a _________ Argentina ____________ costa Rica __________ Mexo ____________ El Salvador ________________ Chde _____ coloma ________ Honduras _________ Porn _____ Ecuador _________ BolMa ________
0.00
1.00
Indu trial countries Aver ge, 1999 I Latin Americ I Average, 199
________
________ _________ ________ _________ _________ _________
____________ E
Caribbea, Average, 1 99
___ ________________ _______ ________ ________ j _________ fl\ _____ J _________ ________ ________ ________ ________ ________ ________ ________
Diggo
2.00
3.00
4.00
5.00
6.00
7.00
8.00
Monthly
Table 4: Summary Statistics
Avenge Statistics for the overall sample
Variable
Observations
# countries
Total Employment
221
43
Prime-Age Male Employment Prime-Age Female Employment
139
43
139
43
Youth (15-24) Employment Self-employment
140
43
84
40
Total Unemployment
221
43
Prime-Age Male Unemployment
221
43
Prime-Age Female Unemployment
139
43
Youth (15-24) Unemployment
139
43
Unemployed> 6months/Total U.
140
40
Job Security
205
36
GDP(US dollars 1995) GDP growth
212
42
179
41
Proportionpoplsto24
221
43
Female Participation
221
43
Union density
47
39
Average Statistics for Latin America and the Caribbean
Variable
Observations
# countries
Total Employment
59
15
Prime-Age Male Employment Prime-Age Female Employment Youth (15-24) Employment Self-employment Total Unemployment
59
15
59
15
59
15
59
15
59
15
Prime-Age Male Unemployment Prime-Age Female Unemployment Youth (15-24) Unemployment Unemployed> 6months/Total U.
59
15
59
15
59
15
42
15
Job Security
108
16
GDP(US dollars 1995) GDP growth
66
20
59
17
Proportionpoplsto24 Female Participation
71
17
59
18
Uniondensity
21
17
Average Statistics for OECD Sample (Excluding Mexico) Observations
# countries
Total Employment
162
28
Prime-Age Male Employment
80
28
Prime-Age Female Employment
80
28
Youth (15-24) Employment
81
28
Self-employment
25
25
Total Unemployment
162
28
Prime-Age Male Unemployment
162
28
Prime-Age Female Unemployment
80
28
Youth (15-24) Unemployment
80
28
Unemployed> 6months/Total U.
81
24
Job Security
97
16
GDP(US dollars 1995)
146
25
GDP growth
120
24
Proportionpoplsto24
150
25
Female Participation
162
28
Union density
26
22
# per country 5.1 3.2 3.2 3.3 2.1 5.1 5.1 3.2 3.2 3.5 5.7 5.0 4.4 5.1 5.1 1.2 # per country 3.93 3.93 3.93 3.93 3.93 3.93 3.93 3.93 3.93 3.93 2.69 5 3.88 3.47 3.94 1.23 # per country 5.79 2.86 2.86 2.89 1.00 5.79 5.79 2.86 2.86 3.38 6.06 5.84 5.00 6.00 5.79 1.18
Mean 66.09 89.19 56.88 53.05 26.92 8.01 8.01 4.99 6.25 13.42 2.62 5.E+11 2.90 0.16 55.64 26.52 Mean 7 1.950 91.746 47.191 63.662 32.742 7.404 3.881 4.666 10.881 14.548 3.512 1.24E+11 3.3 12 0.197 44.255 18 Mean 63.96 87.31 64.02 45.33 13.17 8.22 8.22 5.80 7.43 15.28 1.63 6.25E+11 2.70 0.15 59.79 33.43
Std. Dcv. 8.44 4.93 14.85 15.47 11.87 4.15 4.15 3.09 4.39 7.71 1.74 9.E+11 3.30 0.03 13.34 17.79 Std. Dcv. 4.222 3.157 10.699 11.078 8.269 3.296 2.578 3.134 4.670 7.262 1.567 1.99E+11 3.837 0.016 10.526 11.37 Std. Dcv. 8.59 5.16 13.39 13.54 6.47 4.41 4.41 3.19 4.81 8.90 1.36 1.07E+12 3.00 0.02 11.77 19.18
35
Country Bolivia Brazil Chile Colombia Costa Rica Dominican Republic Ecuador El Salvador Honduras Mexico Nicaragua Panama Paraguay Peru Venezuela
Year 96 97 81 83 86 88 92 93 95 96 87 98 92 94 96 95 97 81 83 85 87 89 91 93 95 97 96 95 95 89 92 96 98 84 89 92 94 96 93 79 91 95 97 95 85-86 91 94 96 97 81 86 89 93 95 97
Table 5: Description of Household Surveys
Name of the survey
sample size Households
Encuesta Nacional de Empleo Encuesta Nacional de Empleo Pesquisa Nacional por Amostra de Domicilios
8,311 8,461 183,193
Pesquisa Nacional por Amostra de Domicilios Pesquisa Nacional por Amostra de Domicilios Pesquisa Nacional por Amostra de Domicilios Pesquisa Nacional por Amostra de Domicilios Pesquisa Nacional por Amostra de Domicilios Pesquisa Nacional por Amostra de Domicilios Pesquisa Nacional por Amostra de Domicilios Encuesta de Caracterizaciфn Socioeconфmica Nacional Encuesta de Caracterizaciфn Socioeconфmica Nacional Encuesta de Caracterizaciфn Socioeconфmica Nacional Encuesta de Caracterizaciфn Socioeconфmica Nacional Encuesta de Caracterizaciфn Socioeconфmica Nacional Encuesta Nacional de Hogares - Fuerza de Trabajo Encuesta Nacional de Hogares - Fuerza de Trabajo Encuesta Nacional de Hogares - Empleo y Desempleo Encuesta Nacional de Hogares - Empleo y Desempleo Encuesta Nacional de Hogares - Empleo y Desempleo Encuesta de Hogares de Propфsitos Multiples
113,599 65,277 68,833 78,188 88,854 85,167 84,862 22,719 25,793 27,666 45,379 33,636 18,255 32,442 6,684 7,132 7,351 7,518
Encuesta de Hogares de Propфsitos Multiples
7,637
Encuesta de Hogares de Propфsitos Multiples
8,882
Encuesta de Hogares de Propфsitos Multiples
8,696
Encuesta de Hogares de Propфsitos Multiples
9,631
Encuesta de Hogares de Propфsitos Multiples Encuesta Nacional de Fuerza de Trabajo Encuesta de Condiciones de Vida
9,923 5,548 5,818
Encuesta de Hogares de Propфsitos Multiples Encuesta Permanente de Hogares de Propфsitos Multiples Encuesta Permanente de Hogares de Propфsitos Multiples Encuesta Permanente de Hogares de Propфsitos Multiples Encuesta Permanente de Hogares de Propфsitos Multiples Encuesta Nacional de lngreso Gasto de los Hogares Encuesta Nacional de Ingreso Gasto de los Hogares Encuesta Nacional de Ingreso Gasto de los Hogares Encuesta Nacional de lngreso Gasto de los Hogares Encuesta Nacional de lngreso Gasto de los Hogares Encuesta Nacional de Hogares Sobre Medicion de Niveles de Vida Encuesta Continua de Hogares - Mano de Obra Encuesta Continua de Hogares - Mano de Obra Encuesta Continua de Hogares Encuesta de Hogares Encuesta de Hogares - Mano de Obra Encuesta Nacional de Hogares sobre Mediciфn de Niveles de Vida Encuesta Nacional de Hogares sobre Mediciфn de Niveles de Vida Encuesta Nacional de Hogares sobre Mediciфn de Niveles de Vida Encuesta Nacional de Hogares sobre Niveles de Vida y Pobreza Encuesta Nacional de Hogares sobre Niveles de Vida y Pobreza Encuesta de Hogares por Muestra Encuesta de Hogares por Muestra Encuesta de Hogares por Muestra Encuesta de Hogares por Muestra Encuesta de Hogares por Muestra Encuesta de Hogares por Muestra
8,482 8,727 4,757 6,428 6,493 4,735 11,531 18,538 12,815 14,842 4,455 8,593 8,867 9,875 9,897 4,667 5,188 2,388 3,623 16,744 3,843 45,421 129,713 61,385 61,477 18,782 15,948
Month when
Individuals
Survey was Held
35,648 June
36,752 November
481,488 September
511,147 September
289,533 September
298,831 September
317,145 September
322,811 September
334,186 September
331,142 September 97,844 December
185,189 November
118,555 November
178,857 November
134,262 November
79,812 September
143,398 September
22,178 July
23,449 July
23,968 July
34,591 July
34,368 July
35,565 July
37,783 July
48,613 July
41,277 July
24,841 February 26,941 August to November
48,884 1995
46,672 September
24,784 September
33,172 September 32,696 March
23,985 Third quarter
57,289 Third quarter
58,862 Third quarter
68,365 Third quarter
64,916 Third quarter
24,542 February to June
24,284
38,888 August
48,328 August
39,786 August 21,918 August to November
26,323 July 1985 to July 1986
11,587 September-November
18,662 May-August
88,863
19,575 September-November 239,649 Second semester
682,636 Second semester
315,658 Second semester
386,629 Second semester
92,458 Second semester
76,965 Second semester
36
LAC JobSecurity GDP growth GDP level Female part. Pop 15to24 Constant
Total Emp. (1) 16.04*fl
Table 6.a: OLS Estimation. Full Sample
Male
Female
Youth
Self-
Total
Male
Prime-age Prime-age
Prime-age
Female Prime-age
Youth
Proportion of Unemp.
Emp.
Emp.
Emp.
EmpI. Unemployment Unemployment Unemployment Unemployment >6 months
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
47Ш*** -1137
2847
11.67***
_2.12**
_2.75***
_4.23***
_7.16***
_4414***
(133) _1.37***
(91) _0.81***
(322) -146
(329) _354***
(321) 1.37**
(32)
(258)
(90)
(3.97)
(58)
-108
-005 -0124
008
.50**
(133)
(.110)
(387)
_3E_12*** -1.97E-12 2.45E-12
(36) -3.5E-12
(23) -3.OIE-12
(1.28e-12) (1 .39e-12) (4.86e-12) (4.58e-12) (3.33e-12)
Ш399***
-
-
334***
.240***
(115) 0.83*fl (28) 006 (.116) 3.51E-12 (1.1 le-12) _.108***
(70) .87*** (19) -004 (08) 2.91E_12*** (1.06e-12) -
(111) .833*** (31) 10 (13) 3.6E_11** (1 .68e-1 1) -
(257) .87* (.53) 0083 (21) 2.55E-12 (2.69e-12) -.186
(376) .86 (89) -016 (036) 6.71E_12* (3.88e-12) _.65***
(0.047)
(.12)
(.084)
(.04)
1156
-
-
-
115.26**
34.49
-
(.078)
(0.14)
-
-69.89
-96.57
(27.08) 41.63*** (5.21)
89.95*** 62.81*** 33.19***
(1.21)
(4.27)
(8.32)
(52.12) -1935 (10.59)
(23.53) 1743 (5.07)
3.24*** (.93)
509 (1.47)
(48.85) 36.21** (10.12)
(17.28) 104.7*** (17.25)
N. observations
114
77
77
78
65
114
77
77
78
64
R-square
073
033
029
053
057
023
032
026
030
85
Notes: Standard errors reported within parenthesis. * indicates significant at 10, ** significant at 5% and *** significant at 1%
LAC Job Security GDP growth GDP level Female part. Pop 15to24 Constant
Total Emp. (1)
Table 6.b: Random--Effects (RE) Estimation. Full Sample
Male
Female
Youth
Self-
Total
Male
Female
Prime-age Prime-age
Prime-age
Prime-age
Youth
Proportion of Unemp.
Emp.
Emp.
Emp.
EmpI. Unemployment Unemployment Unemployment Unemployment >6 months
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
15.26*** (2.15) -1 .84***
4.62** (1.82) 1 .04"
-11.05" 29.99***
(5.47) 526
(5.23) _3.28***
14.56*** (3.90) .35
-2.24 (1.93) .69
_2.36* 1.26 77**
-3.79 (1.92) 1.06**
-7.29 (3.81) .99
_48.61*** (6.35) .95
(505)
(.48)
(1.33)
(1.38)
(.87)
(.45)
(.34)
(.515)
(.86)
(1.49)
-0.001
.054
.218
0.164
393***
-.04
.016
.12
-.084
-0.171
(.073) -4.14E-12
(.091)
(.199)
(.278)
-2.68E-12 1.31E_11* -7.18E-12
(.166) -5.36E-12
(.06) 4.23E_11*
(.07) 3.13E_12*
(.09) 4.72E_12*
.135 -5.36E-12
(.246) 9.49E-12
(2.51e-12) (2.42e-12) (7.03e-12) (6.87e-12) (4.39e-12)
0.33***
-
-
0.63***
.036
(2.24e-12) .021
(1.71e-12) -
(2.57e-12) -
(4.39e-12) .037
(6.80e-12) _.304*
(0047)
(13)
(08)
(04)
316
-
-
-
40.22
29.98
-
077
(161)
-
41.98
115.79
(26.84) 4777***
90.37*** 54.06***
16.80*
(54.40) 6.95
(25.22) .53
3.36**
4.23**
(46.25) 4.95
(115.28) 50.7"
(5.74)
(1.89)
(5.34)
(9.43)
(11.13)
(5.38)
(1.36)
(2.01)
(9.81)
(22.22)
N. observations
114
77
77
78
65
114
77
77
78
64
R-square
0.72
.32
.23
0.50
.57
.13
.31
.25
.17
0.82
HausmanTest
5.46
3.90
2.17
9.43
53.56
9.53
4.87
3.75
8.78
8.06
(.36)
(.27)
(.57)
(0.05)
(0.00)
(0.08)
(.18)
(.28)
(.11)
(.15)
Notes: Standard errors reported within parenthesis. * indicates significant at 10, ** significant at 5% and *** significant at I %.
37
JobSecurity GDP growth GDPlevel Female part. Pop 15to24 Constant
Total
Table 6.c : Fixed --Effects (FE) Estimation. Full Sample
Male
Female
Youth
Self-
Total
Male
Female
Prime-age Prime-age
Prime-age
Prime-age
Youth
Proportion of Unemp.
Emp. (1) -155
Emp.
Emp.
Emp.
EmpI. Unemployment Unemployment Unemployment Unemployment >6 months
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
-0.013
327
_6.04* _8.43***
-187
-106
0021
-116
151
(107)
(1183) (229)
(3.55)
(173)
(.99)
(96)
0049
143
145
278
111
-009
-005
(128) 0024
(162) .25*
(464) -017
(078)
(101)
(19)
(303)
(150)
-1.92E-11 _2E_11*** 5.5E_11** _6.7E_11** -3.01E-12
(07) 1.6E_11***
(08) 2.IE_11***
(.11) 2.4E_11**
(13) 3.9E_11***
(28) 3.90E-11
(8.84e-12) (9.97e-12) (1 .93e-1 1) (3.25e-1 1) (3.74e-12)
Ш34***
-
-
1.00*fl
240
(8.le-12) 07
(8.15e-12) -
(1 .08e-1 1) -
(1.48e-12) 08
(4.55e-1 1) -07
(005)
(19)
(104)
(05)
-5.93
-
-
-
11526
5603*
-
(09)
(23)
-
6071
529.05**
(3120)
(5113)
(2863)
59.67*** 9594*** 27.14*** 42.15*** -1935
-905
300
-008
(4910) _7.12**
(21891) _63.79***
(721)
(3.37)
(654) (1135) (1037)
(662)
(276)
(366)
(1163)
(45.53)
N. observations
114
77
77
78
65
114
77
77
N. countries
28
28
28
28
27
28
28
28
R-square
009
005
005
003
030
003
003
008
Notes: Standard errors reported within parenthesis. * indicates significant at 10, ** significant at 5% and *** significant at 1%
78
64
28
25
001
004
Table 7: The impact of job security in the regional sub-samples
A. Latin America and the Caribbean
Dependent Variable TotalEmployment Maleprime-ageEmployment Femaleprime-ageEmployment YouthEmployment Self-employment
II Ohs. 53 53 53 53 53
OLS Coefficient _9*** _Ш3*** 0.78 4.21fhc 1.09*
OLS SE. (0.36) (0.30) (1.11) (0.94) (0.63)
RE Coefficient _1.62*** _1.44** 3.15** _433*** -0.58
RE S.E (0.59) (0.58) (1.52) (1.30) (0.98)
FE Coefficient -1.83 -0.48 3.10 _75Ш* _8.34***
TotalUnemployment Male prime-age Unemp. FemalePrime-ageUnemp. Youth Unemployment %Long-termUnemp.
53
0.34
(0.35)
.06
(0.04)
53
Ш94*** (0.24) 0.91*** (0.43)
53
0.27
(0.33)
0.51
(0.52)
53
0.35
(0.47)
-0.22
(1.60)
30
0.13
(0.98)
-0.11
(1.36)
0.13 -0.74 0.06 -0.22 0.42
FE SE. (1.34) (1.24) (2.59) (3.70) (1.73) (1.26) (1.02) (1.42) (1.60) (5.31)
B. OECD Countries (Excluding Mexico)
Dependent Variable II Ohs.. OLS
OLS
RE
RE
FE
FE
TotalEmployment
Coefficient SE. Coefficient SE. Coefficient SE.
61
-0.82
(0.57) _33Ш*** (1.16)
-
-
Maleprime-ageEmployment
24
-0.06
(0.66)
-0.07
(1.13)
-
-
Femaleprime-ageEmployment
24
_5.80*** (1.69) _6.16***
(2.38)
-
-
YouthEmployment
25
1.32
(2.81)
-4.41
(4.58)
-
-
Self-employment
Not enough observations
TotalUnemployment
61
1.14**
(.56)
2.27**
(1.10)
-
-
Male prime-age Unemp.
24
0.50
(0.49)
0.48
(0.77)
-
-
FemalePrime-ageUnemp.
24
2.23*** (0.85)
2.04*
(1.19)
-
-
Youth Unemployment
25
.586
(1.98) 470*
(2.93)
-
-
%Long-termUnemp.
35
2.003
(1.85)
3.31
(3.62)
-
-
Note: standard errors between parenthesis. The specifications for the two sub-samples include the same repressors than in the overall sample. * indicates significant at 10, ** significant at 5% and *** significant at 1%.
38

File: the-cost-of-job-security-regulation-evidence-from-latin-american.pdf
Published: Tue Sep 17 11:47:23 2002
Pages: 39
File size: 0.3 Mb


A history of bestiality, 140 pages, 1.01 Mb

García Márquez, Gabriel, 2 pages, 0.05 Mb

, pages, 0 Mb
Copyright © 2018 doc.uments.com