Abstract Employment protection systems are widely believed to generate distortions in firms' hiring and firing decisions. However much less is known about the impact of these regulations on workers' behavior. In this paper we provide evidence on the latter question using data from a large Italian bank. Our analysis is based on weekly observations for 545 men and 313 females hired as white collar workers between January 1993 and February 1995. These workers begin to be protected against firing only after the twelfth week of tenure and we observe them for one year. We show that -particularly for men -the number of days of absence per week increases significantly once employment protection is granted at the end of probation. This suggests that the provision of employment protection causes the increase in absenteeism. Alternative explanations based on career concerns or on learning about social norms would predict a smooth relationship between absenteeism and tenure instead of the observed discrete jump. This consequence of employment protection seems to have been neglected in European policy debates so far. JEL Classification: J2, D2, D8, M5.

www.ifn.se

This study uses panel data describing about 6,500 employees in a large international company to study the incentive effects of performance related pay. The company uses two performance related remuneration mechanisms. One is an individual "surprise" bonus payment. The other is a more structured system, where part of the salary is determined by individual performance evaluations. We hypothesize that effort is higher in departments where (i) performance evaluation results are more spread out, (ii) person-specific performance evaluations are more flexible over time, (iii) surprise bonuses are used more frequently. These hypotheses are tested using days of absence and overtime work as effort indicators. The tests yield that hypotheses (ii) and (iii) are supported, and that (i) cannot be tested reliably due to possible simultaneity bias in our data. We investigate and confirm the robustness of these findings. They suggest that surprise bonus payments and flexibility in the evaluation of individual performances over time provide effective incentives for employee effort. Keywords: bonus, merit pay, pay for performance, worker effort, social status, mobility, absenteeism, overtime JEL Code: J33, M12, J24, J41, M50, C25 We thank the company which provided the data and Michael Gerfin, Jennifer Hunt, the participants of the meeting of the Ausschuss für Bevölkerungsökonomie in Vienna, of the third Transatlantic Meeting of Labor Economists in Buch, and of the Annual Meeting of the European Society of Population Economists (ESPE) in Bergen for helpful comments. 1 Examples are Groves et al. (1994) investigating the productivity consequences of managerial autonomy in Chinese industries, Jones and Kato (1995) measuring productivity effects of employee stockownership and bonuses for Japanese firms, Lee and Rhee (1996) estimating similar models with South Korean time series on 8 industries, and Morton (1998) who applied quarterly industry data from Taiwan. Cable and Wilson (1989, 1990) provide evidence on productivity enhancing effects of profit-sharing in the United Kingdom and Germany. For further sources see Kahn and Sherer (1990) and 2 Introduction Does performance-related pay help to increase worker effort? Most economists and managers believe so. Yet the empirical evidence remains amazingly sparse even years after Prendergast (1999) pointed out that there has been little empirical assessment of incentive provisions for workers. This is surprising given the increasing reliance of employers on merit pay. Prendergast (1999, p.56) concludes his survey on the provision of incentives in firms: "All in all, the available empirical evidence on contracts does not yet provide a ringing endorsement of the theory." He points to several shortcomings of available studies, among them the weakness of tests, the lack of reliable personnel data, empirical identification and selection problems, and the focus on workers with easily observable outputs. We attempt to overcome these limitations. Our empirical tests are based on panel data describing about 6,500 employees of a large international company. Among these employees are production workers, researchers, administrators, and managers. We analyze objective outcomes using individual-level data. To identify the effect of performance pay on behavior, we compare workers across company departments and over time. Departments differ in the intensity with which the performance pay system is applied: While in some departments performance evaluations fully use the available spectrum of evaluation outcomes, others differentiate to a lesser degree. The validity of this identification strategy hinges on the exogeneity of worker assignment to departments, which we discuss and examine below. This identification strategy, which is based on a comparison within a given firm, is most similar to the one used by Kahn and Sherer (1990). They applied panel data describing a firm's 92 managers and find that managers with a high bonus impact of performance work harder. However, since this result does not hold with respect to merit pay, the authors conclude that only bonus differences affect subsequent performance. Using absenteeism and overtime work as indicators of worker effort we investigate the role of performance pay for richer and more recent data. The longitudinal nature of the data 3 The literature on the pay-performance relation for managers is surveyed by 3 By using objective outcomes as opposed to subjective performance evaluations we avoid the measurement problems that arise when worker evaluations are compared across supervisors (see e.g. Prendergast 1999 or Schwab and Olson 1990 for a discussion) or which are inherent in workers' statements on their attitudes, efforts, and performance. Clearly, objective indicators of individual behavior provide a more reliable reflection of effort than the firm or industry productivity measures applied in the literature. 4 We confirm Kahn and Sherer's (1990) finding of a productivity enhancing effect of bonus payments. Also, worker effort appears to respond positively to the variability of individual ratings over time. Workers in departments where performance ratings vary over time and so ensure a higher degree of competition in the internal labor market provide more effort than workers in departments where individual ratings hardly change. Institutional Background We received personnel data on white and blue collar employees from an international company. The data cover one organizational unit in Switzerland between 1999 and 2002. In this time the unit employed 6,425 individuals for most of whom we have repeated annual observations. 5 6 Depending on employees' salary the bonus can amount to between 10 and 100 percent of a gross monthly salary. 7 The salary rate of merit pay for regular employees are at 10 to 13 percent, they reach 20 percent for middle management, and increase to up to 70 and 85 percent for top management. 5 The firm is organized in departments, of which we observe a total of 78 different ones in the full sample over the years. Due to organizational changes only about half of these departments exist over all four years of our data: 11 exist over three consecutive years, and 21 are observed over two consecutive periods. The remaining 7 units cannot be followed over time. On average there are 106 employees in a department, where the department size varies between 1 and 858. In this company two performance-related remuneration mechanisms are in place: One is an individual "surprise" bonus at the order of about 1,400-3,400 EURO, which is granted for special achievements. 6 This bonus payment is at the discretion of supervisors who can spend a given annual budget for this purpose. On average, they apply this instrument about five times a year, with significant heterogeneity across departments. About one quarter of the departments did not pay out bonuses at all. The others paid on average 7 bonuses per 100 employees per year. The other is a more complex performance pay system, where depending on the hierarchical level of the employee between 10 and 85 percent of the annual salary is determined by the outcome of an annual individual performance evaluation. 7 After a goal-setting session in the beginning of the year, supervisors rate the performance of employees typically on a scale between 0 and 150 percent of the originally envisioned objectives at the end of the year. The ratings are cross-checked by the supervisors' managers. The absolute amount of the payout depends on individual performance, the individual base salary, and the performance of the entire division, adhering to the following formula: 8 If e.g. the individual performance rating is 120 percent, the division performance is 105 percent and the salary rate 10 percent, this worker receives performance pay of 1.2 * 1.05 * 0.1 = 12.6 percent of base salary. For managers with identical personal and divisional ratings but a higher fixed salary rate of 20 percent, the payout would amount to 1.2 * 1.05 * 0.2 = 25.2 percent of the base salary. The workers in our data belong to divisions with at least 1,000 employees such that divisional results should be exogenous for the individual employee. 9 About ten percent of the workers used to be grouped as low achievers, two thirds were in the medium range and about one quarter were in the group of top performers, leaving supervisors substantial leeway in their rating decisions. 10 Taylor and Pierce (1999) and Kahn and Sherer (1999) point out that a system's effects may differ depending on whether it was just introduced or has been in place for several years already. In this respect, we consider our study as an evaluation of an ongoing system where initial employee responses to its introduction already faded. 6 Bonus pay / salary = individual performance rating @ division performance @ salary rate 8 In 2002, the previously required distribution of performance ratings per department was abandoned but management clearly communicated the expectation that future rating distributions should be in the same range. 9 Also, in 2002 mid-year reviews were introduced to complement the set of goal-setting and evaluation meetings. As the general character of the evaluation system did not change by these reforms they are disregarded in our analysis. Performance oriented pay was extended to all employees already in the mid 1990s, several years before our data was gathered. 10 As a share of the firm's base salary cost the expenditures for surprise bonus payments amount to 0.3 percent and those for evaluation-based performance pay to about 15 percent. The company prides itself on paying performance-related remunerations and salaries above industry averages. It also offers an employee stock ownership plan and a stock option plan for executives. Theory and Hypotheses Within this institutional framework we test whether performance-related pay enhances worker effort. The literature shows that beneficial effects on worker behavior are not at all 11 Kahn and Sherer (1990) found manangers to be more responsive to bonus payments than to merit pay. 11 The following determinants of the effectiveness of performance-related pay are discussed in the literature (see e.g. Baron and Kreps 1999): (i) the saliency of a payment matters, where payments that attain more employee attention such as surprise remunerations are more effective than expected payments. (ii) Workers may reduce effort when they perceive inequities, e.g. if current salary differences result from past performances that are no longer known to the individuals. (iii) Risk-averse workers with uncertain tenure value bonuses more highly than long term pay raises. (iv) Payment systems differ in the degree to which payment is determined by individual effort; effort may be lower if the payout formula allows for substitution between individual and divisional performance. (v) The transparency of an evaluation system and its reliability affect its motivational effectiveness. (vi) Tournament theory suggests that higher wage dispersion can motivate higher worker effort. While we cannot test each of these predictions separately, we propose three hypotheses on the effectiveness of the firms' performance-based pay based on these insights. As departments differ systematically in their utilization of the worker evaluation system we observe heterogeneity in the extent to which employees' performance affects their remuneration. If performance pay enhances motivation and effort we would expect to see higher effort levels among those individuals whose performance is more strongly reflected in their pay. To measure the degree to which pay reflects performance we propose two measures: One describes the cross-sectional variance of performance ratings in each department. We expect that individuals who work in a department with little heterogeneity in evaluation outcomes perceive less of an incentive to improve their performance than individuals for whom a change in performance is rewarded with large jumps in ratings and subsequent pay. This is our first 8 hypothesis, which agrees with the predictions of tournament theory, i.e. that effort should be largest when wage dispersion is highest (H1). The second measure describes the variability of person-specific evaluations in a given department over time: In a low variability scenario, workers receive about the same performance rating every year even if their performance varies. In a high variability scenario, individual ratings in one year have little predictive power for next year's rating as they closely match actual performance. Based on the above arguments (ii) and (iv) which favor a high correlation of performance pay to current and person-specific performance we hypothesize that performance incentives are stronger in the second, high variability scenario. Here the performance of every employee is evaluated anew every year as opposed to a situation where established judgements predominate over time (H2). Our third hypothesis builds on numerous of the above listed arguments: we propose that worker effort is higher in departments which over the course of the year provide a larger number of surprise bonus payments per worker. In these departments performance pay is highly salient, it clearly relates to current as opposed to past performance, it depends (directly) on the effort of the individual, and the final payout in these departments is more dispersed than in departments that do not pay surprise bonuses (H3). Empirical Approach and Data Description Empirical Approach: After a descriptive analysis of the correlation between worker effort and performance pay we investigate whether the observed correlations are statistically significant in the framework of multivariate regressions which control for potential composition effects. As we observe individual workers (i) who are associated with departments (j) over several periods (t), we estimate the following baseline specification: In our company we find three types of departments: production, administration, and research. 13 Workers can end a month with less than the contracted number of hours. While on average a share of about 13 percent of all workers end a month with negative overtime its average amount is much less than half of the average amount of positive overtime (-14.4 hours vs. 37.1 hours). 14 For other studies using absenteeism as an indicator of effort see e.g. 15 Landers et al. (1996) discuss scenarios where firms use hours of work as an indicator of employee characteristics. Drago (1991) uses the willingness to work an extra unpaid 20 minutes and the propensity to go to work even if not feeling well as performance indicators. 9 Y represents an indicator of worker i's effort in period t, I measures the pre-period value of the performance pay incentives implied by our three hypotheses, which we consider both, jointly and individually, and X provides background indicators describing worker and department. * controls for department type fixed effects, 12 < reflects year fixed effects, : represents individual-specific unobserved heterogeneity which we assume to be uncorrelated with the explanatory variables, and , is the random error term. We use a least squares estimator for hours of overtime work, which in the case of our firm can be negative, 13 and a tobit model for the always positive number of yearly days of absence. We then simulate the estimated effects to gauge their magnitude. The analysis concludes with a number of robustness tests which address several potential limitations to the reliability of our results such as endogeneous worker sorting between departments and simultaneity bias if supervisors apply the incentive instruments in response to observed worker effort. Dependent Variables: In order to test whether the cross-sectional and the intertemporal variability of departmental performance evaluations or bonus payments affect worker behavior we need indicators of worker effort. We apply two measures which are used frequently in the literature: The first describes workers' days of absence due to health problems, 14 the second indicates how many hours of overtime an individual accumulated at the end of the year. 15 Swiss 16 Workers who e.g. accumulate 140 hours of overtime at the end of one month will start out the next month with a surplus of only 120 hours. 20 hours are deducted from their overtime account.

The present study examines cross-national and sectoral differences in Total Factor Productivity (TFP) in fourteen European countries and ten sectors from 1995 to 2007. The main aim is to ascertain the role of employment protection of temporary contracts on TFP by estimating their effects with a “difference-in-difference ” approach. Results show that deregulation of temporary contracts negatively influences the growth rates of TFP in European economies and that, within sectoral analysis, the role of this liberalisation is greater in industries where firms are more used to opening short-term positions. By contrast, in our observation period, restrictions on regular jobs do not cause significant effects on TFP, whereas limited regulation of product markets and higher R&D expenses positively affect efficiency growth.

The paper uses 15 waves of BHPS data to provide an integrated analysis of the roles of both individual social status and upward mobility relative to own parents on job and life satisfaction, preferences for redistribution, pro-public sector attitudes and voting. Both greater individual social status and greater mobility with respect to parents are associated with higher levels of satisfaction. However, this symmetric effect disappears for political preferences. While greater social status is associated with less favourable attitudes to redistribution and the public sector, greater upward mobility is associated with more Leftwing attitudes. These attitudes translate into actual reported voting behaviour. Upwards social mobility produces satisfied Left-wingers.

Most of the recent literature on labor market institutions draws on reforms rather than on cross-country variation in regulatory levels. This is a significant improvement with respect to the earlier literature which was based on somewhat arbitrary one-dimensional indicators of multi-dimensional institutions. But this literature lacks guidance from a theory of institutional reforms, acknowledging the fact that regulatory changes often create longlasting asymmetries, multi-tier regimes. This chapter provides new evidence on reforms in Europe, where most of the regulatory change in the labor market area have been taking place in the last 20 years. In light of this evidence, it extends a model widely used in analysing the interactions between institutions and labor market flows. Finally, it critically surveys the empirical literature drawing on institutional reforms in Europe.

A well-established finding in the literature is that self-employment enables mothers to accommodate work and family needs better than when they are engaged in organizational employment. With this re-sult in mind, the authors investigate within a dual system of job pro-tection if women under temporary contracts face greater work–family conflicts than those under permanent contracts. The authors use data on women’s work and fertility histories from the Spanish Con-tinuous Sample of Working Histories to analyze whether women under temporary contracts transition to self-employment upon motherhood more than those who are under permanent contracts. Analyses show that being under a temporary contract increases women’s likelihood of transitioning to self-employment upon child-birth. Supplementary analyses show that this is partly the result of voluntary transitions and not an employer’s decision to terminate a temporary contract upon motherhood. Overall, these findings re-veal a hidden cost of temporary contracts: the greater difficulty in balancing work and family. The arrival of a child imposes new family demands on parents who need to ensure that their newborn’s needs are fulfilled. Since women con-tinue to bear the major responsibilities for household production, parent-hood is likely to increase the conflict between employment and family duties for working mothers. Given the increased participation of women in the labor force and in the number of dual-earner families, work–family con-cerns have become an important public policy issue. Consequently, during the past two decades employers and governments have shown growing interest in understanding what policies and practices are effective at reduc-ing work–family conflict. Several papers have investigated the relationship

Abstract: How does job security influence productivity? I create a dataset from the

views of the institute.

Opinions expressed in this paper are those of the author(s) and do not necessarily reflect views of the institute.