استحکام بازار کار و بهره وری اقتصادی: شواهد از اسپانیا

In the 1990s, Spain approved two labor reforms aimed at reducing the unemployment level and its volatility. Overall, these reforms involved two measures designed to induce firms to meet their labor needs via adjustment of permanent positions: restricting the use of temporary workers and reducing the amount of severance payments. This paper empirically assesses the impact of these reforms on the allocative efficiency of the labor input employing Petrin and Sivadasan's (2011) value of the marginal product-marginal cost gap methodology. We find a statistically significant increase in within-firm permanent labor gaps following the reforms. These results suggest that restrictions on the use of temporary workers (increasing the probability of hiring fragile workers for permanent positions), when coupled with uncertainty about enforcement of reduced severance payments, could more than offset the reduction in severance payments; hence, the net effect of the reforms could be to increase adjustment costs for permanent positions.

Traditional economic theory suggests that firms adjust employment to equate the marginal revenue product of labor with its marginal cost. In practice however, firms are confronted with several rigidities that may prevent them from adjusting employment to its optimal level. Job security provisions, such as dismissal costs and severance payments, raise the firm's adjustment costs; hence firms might find it optimal not to hire workers when a positive shock pushes the marginal revenue product of labor above its wage (Bertola, 1990 and Hopenhayn and Rogerson, 1993). When dismissal costs are high, firms might even choose to retain workers whose wages exceed their marginal product (Blanchard and Portugal, 2001). Moreover, hiring could diminish if job security provisions increased the power of incumbent workers, resulting in higher wages for insiders (Caballero and Hammour, 1997). The effect of employment protection laws (EPLs) on economic performance has sparked a vigorous debate among economists (see Freeman, 2005). Beginning with the seminal work of Lazear (1990), there is a broad and growing literature that explores the consequences of job security provisions on labor market performance and productivity.1Addison and Teixeira (2003) and Heckman and Pagés (2004) summarize some of the findings obtained by different papers that analyze the effect of job security regulations on employment, unemployment, turnover, and so forth.2 Believing that employment protection laws had had a negative effect on economic performance, several countries reformed their labor markets during the 1980s.3 These reforms were mostly at the margin, liberalizing the use of temporary workers (fixed-term contracts, FTCs, with low dismissal costs) but leaving essentially unchanged the laws that applied to permanent workers with indefinite-term contracts (ITCs; see Boeri and Garibaldi, 2007). Following this trend, Spain introduced a two-tier reform in 1984 that relaxed the labor market by allowing employers to hire fixed-term workers with practically no restrictions (Bentolila et al., 2008). In particular, this reform created two types of labor inputs in terms of their flexibility: temporary workers, a nondynamic but totally variable input based on a short-term labor contract (three-year maximum) with no guarantee of renewal and little or no termination costs; and permanent workers, a dynamic input that is costly to adjust because of the significant associated dismissal costs. Despite the reforms, Spain continued to have some of the most stringent regulations for permanent contracts among developed countries.4 A nearly immediate consequence of the 1984 reform was a substantial increase in the use of temporary workers: to fully one third of all workers in the labor market, the highest proportion of the OECD countries. This hike in the use of temporary workers may indicate that firms sorely needed to adjust their total labor input to its desired levels but were unwilling to do so through the use of permanent workers. Moreover, there was a significant increase in the volatility of labor market outcomes, characterized by a practically instantaneous adjustment of temporary workers in response to positive or negative demand shocks. In order to reduce the unemployment level (24% in 1994) and labor market volatility, the Spanish government approved two labor reforms in the 1990s—one in 1994 and another in 1997. These reforms sought to diminish the use of FTCs by inducing firms to fulfill their labor needs through the adjustment of permanent positions (these intentions are evident in preliminary statements of the 1994 and 1997 labor laws; see also Dolado et al., 2002). The reforms increasingly restricted the use of temporary workers—either by eliminating some type(s) of fixed-term contracts, restricting the age or other requirements for using an FTC, or removing the fiscal incentives for using FTCs. Yet in a dual labor market, the adjustment of permanent workers is through transforming temporary workers from FTC to ITC status; for example, highly productive temporary workers may be offered an ITC (Goux et al., 2001 and Portugal and Varejao, 2009). Hence any restrictions on the firm's use of FTCs shrinks its choice set and affects the adjustment of permanent labor. In order to encourage firms to adjust labor needs through permanent positions while extending restrictions on the use of temporary workers, Spain's 1997 reform proposed for all new hires a new ITC with a lower severance payment.5 But the reform also included two other, possibly countervailing, measures: it removed the possibility of hiring new FTC workers; and the labor conditions regulating the new ITCs were to be those currently obtaining in the collective bargaining agreements of permanent workers (see Gomez et al., 2008). The probability of hiring a low-productivity ITC worker increases when the FTC option is removed (or severely restricted). Such workers must be laid off, at a positive firing cost, as soon as it is learned that their productivity is below the firm's threshold. Moreover, the ITC status increases the worker's bargaining power when the firm would prefer not to retain the worker just hired—in response to low worker productivity or even to the arrival of a small negative shock (Caballero and Hammour, 1997, Blanchard and Portugal, 2001, Portugal and Varejao, 2009 and Costain et al., 2010). Finally, the employer's expected adjustment costs depend not only on the nominal severance payments but also on the employer's beliefs about the future path of negative demand shocks (the Spanish economy was recovering from a profound negative demand shock), the possibility of further labor reforms (the new contract structure was proposed to last only four years), and the perceived likelihood that the legislated reductions in severance payments will be strictly enforced (the evidence suggests that most of the unfair dismissals during the fifteen years after the 1997 reform were settled at 45 days of wages per year worked—this despite a reform-approved reduction to only 33 days). As a result, a firm's post-reform expected adjustment costs could actually increase if the above facts dominate the legislated reduction in severance payments. Costain et al. (2010) argue that, unless this reduction exceeds 85% firms will not be induced to adjust labor through permanent positions; however, the actual reduction was about 25%. In sum, these facts suggest that firms' expected adjustment costs probably increased after the two labor reforms. Our aim in this paper is to assess the issue empirically while following the methodology proposed by Petrin and Levinsohn (2011) and Petrin and Sivadasan (2011). That methodology relies on studying the gap between an input's value of marginal product and its price. An increase in labor adjustment costs should affect the allocative efficiency of the permanent labor input—the possibility of reallocating labor from lower-gap to higher-gap firms (i.e., those with greater labor needs). We assess the impact of these reforms on allocative efficiency by studying how they affected the gap between the labor's value of marginal product and its marginal cost; this gap captures the potential output gains if labor were reallocated optimally. Our analysis here basically focuses on the permanent labor input gap because the reforms encouraged firms to adjust labor through permanent positions. That being said, we understand that rejecting the null hypothesis—that the reforms were allocatively efficient in terms of the permanent labor input gap—indicates that the labor reforms introduced some degree of inefficiency into the Spanish labor market. The data used to conduct this analysis are from a 1990–2001 panel of Spanish manufacturing firms provided by the Firms Strategies Survey (Encuesta Sobre Estrategias Empresariales, ESEE). This survey reports the total number of workers by type of worker, the annual hours per worker, the number of overtime hours, and the aggregate labor cost net of dismissal costs; it also reports each firm's output price (which is needed to estimate the production function) and materials price (needed to estimate the materials gap). Our study is restricted to firms of small and medium size—that is, those employing fewer than 200 workers when entering the panel. The share of temporary workers in large firms is especially low when compared to the share in small firms, so adjustment of the labor input via temporary employment is more likely in the latter than in the former. In order to estimate the marginal product of labor, we first estimate a production function that takes into account the duality of the Spanish labor market; that is, the function differentiates between two types of labor inputs, permanent versus temporary workers. To address this duality, we estimate a production function under the framework developed by Olley and Pakes (1996) and Ackerberg et al. (2008). However, we modify their estimation techniques to account for permanent workers as a dynamic input and temporary workers as a nondynamic variable input. Our results suggest a positive and significant impact of these reforms on the permanent labor gap. In particular, the gap increased after the 1994 reform and again after the 1997 reform. Hence these results suggest, in line with Petrin and Sivadasan (2011), that the reforms decreased allocative efficiency of the labor input—for example, the capacity to reallocate labor from lower-gap to higher-gap firms. Of course, there could be many factors beyond the labor reforms (e.g., a tax) that might have affected this gap between the labor value of a marginal product and its price. We isolate the impact of the reforms by using as a control the materials gap, which should not be directly affected by labor market reforms. Introducing this control establishes that our results are robust and that the materials gap is unaffected by the reforms. For additional robustness checks, we split the industries into those with low versus high rates of unionism or voluntary turnover. Again the results are similar; as expected, the impact is greater in industries characterized (respectively) by more unionization and less voluntary turnover. Finally, our production function estimates suggest that, in most industries, temporary workers are no less technically efficient than are permanent workers in small or medium size firms. The rest of the paper is organized as follows. In Section 2 we discuss flexibility in adjusting the two labor inputs. In Section 3 we estimate the production function. In Section 4, we assess the labor reforms by analyzing their effect on the gap. Section 5 concludes.

In the 1990s, Spain enacted two labor reforms intended to diminish the unemployment level and employment volatility. In particular, measures were introduced that aimed at inducing firms to meet their changing labor needs by the adjustment of permanent positions: the use of temporary workers was restricted, and the amount of severance payments was reduced. In the 1994 reform, the government restricted the use of temporary workers but did not modify any regulations affecting permanent labor. In the 1997 reform, the government further restricted the use of temporary workers and also introduced a new type of indefinite term contract, with lower severance payments, to be used for all new hires. In this paper we study whether or not firms, in the wake of these reforms, fully adjusted their permanent positions to the firm's actual labor needs—in particular, we look for evidence of an increase in firms' expected adjustment costs. Toward this end, we analyze the allocative efficiency impact of these reforms by using Petrin and Sivadasan's (2011) methodology for evaluating gaps between the marginal product and the marginal cost. That methodology is based on analysis of the mean absolute gap between the input price and the value of the marginal product. This mean absolute gap corresponds to the mean change in aggregate productivity that would result from making an optimal adjustment to the input. Our findings suggest that the Spanish labor reforms did affect the allocative efficiency of the permanent labor input. We interpret these results as implying that the expected labor adjustment costs increased when the severe restrictions on using temporary workers were not accompanied by a sufficient reduction in the severance payments due to permanent workers.