How do firing costs affect innovation and growth when workers' ability is unknown? Employment protection as a burden on a firm's screening process.

• Author: Berdugo, Binyamin
• Position: Report

1. Introduction

   In recent decades, there has been a dramatic rise in US productivity growth accompanied by a remarkable expansion of US human-capital-intensive industries. These trends, however, have been paralleled by different paths in most European countries of lower growth rates and a tendency to specialize in less human-capital-intensive technologies. (3) Several recent theories suggest that the main difference between Europe and the US that might lead to such productivity differences is their labor market policy. These theories argue that employment protection legislations (such as minimum wage laws, unemployment subsidies and mandatory firing costs) that are rarely applied in the US but are extensively used in Europe might have a diminishing effect on productivity and growth.

   One theory as presented by Hopenhayn and Rogerson (1993) argues that a tax on job destruction, such as a firing cost, can slow down the reallocation of resources from declining industries to growing industries thereby hampering economic growth by reducing productivity. Another theory raised by Bertola (1994) was that labor mobility costs distort income distribution between labor and capital and therefore reduce investments return and decrease the speed of capital accumulation. Davis and Henrekson (1997) have presented evidence on post war Swedish economy and showed that, among other features, labor market inflexibilities (such as employment security laws and centralized wage bargaining system that compresses wage differentials) distort the industrial distribution of employment, and reduce the output and employment share of smaller, younger and labor intensive firms, thereby hampering the efficient allocation of resources and reducing productivity and economic growth. Saint-Paul (1997 and 2002) proposed that due to innovation risks, employment protection legislation might distort the pattern of specialization in favor of mature goods rather than primary innovation, which negatively affects productivity and growth. Other theories emphasize the effect of labor market regulation on delays and barriers to technology adoption (see Gust and Marquez (2004) and Alesina and Zeira (2006)).

   In this paper we attempt to shed light on another channel through which Labor market legislations decrease economic productivity and growth. Namely, this paper focuses on the burden that mandatory firing costs impose on a firm's screening process. We show that high-tech firms with imperfect information about their workers' ability attempt during a trial period to identify those incompetent workers who they will subsequently dismiss. Firing costs stemming from employment protection legislation, however, place a burden on this screening process, thereby motivating innovators to embark on medium-tech projects as they are more flexible in their labor requirements. Employment protection legislation therefore distorts the pattern of specialization in favor of medium-tech firms rather than high-tech firms and consequently slows down the process of economic growth.

   The paper presents a model in which a final good is produced by many intermediate goods that can be upgraded in a quality-ladder fashion (see Grossman and Helpman (1991a, 1991b) and Aghion and Howitt (1992)). These intermediate goods, however, are not identical, since they differ in their productivity rates per quality rank and their labor requirements. Specifically:

   1) High-tech intermediate goods are much more human-capital-intensive than medium-tech goods, and suffer from lower substitutability between skilled and unskilled workers.

   2) Per quality rank, high-tech intermediate goods are much more productive than medium-tech goods and therefore can generate higher economic growth.

   An important assumption of the paper is that the workers' type is unknown, and is revealed only after a certain period of employment (which we henceforth term the "trial period"). Following this trial period, both medium-tech and high-tech firms have the opportunity to dismiss any incompetent workers. However, due to differences in labor requirements, only high-tech firms have the incentive to dismiss incompetent workers, whereas medium-tech firms can continue to keep them with no significant loss of profits. Thus, firing costs affect the profit function of high-tech firms significantly more than medium-tech firms and therefore impinge on the decisions of innovators of whether to embark on a high or medium-tech project.

   The paper has three central results. First, employment protection legislation and various firing costs that stem from them bias the pattern of specialization from human-capital-intensive products toward less human-capital-intensive products. Second, firing costs can negatively affect productivity growth. In closed economies this negative effect is unambiguous, while in open economies the magnitude of this negative effect depends on a firm's adjustment costs. Third, employment protection might trap the economy into adopting inferior technologies that can affect the trajectory of innovation and growth over a long period of time. A major consequence of this latter result is that measures taken belatedly to reduce firing costs might prove ineffective.

   The rest of the paper is organized as follows. Section 2 sets up the basic model. Section 3 relates employment protection legislation to growth. Section 4 extends the basic model. Section 5 discusses the empirical motivation of the model and Section 6 concludes. The mathematical proofs appear in an appendix.

2. The Model

   Consider a small open economy whose activities extend over an infinite discrete time. The economy consists of three types of goods: a final good Y that is used for consumption only, and two types of continuum intermediate goods [x.sub.i] and [z.sub.i] which we denote by "medium" and "high," respectively. The quality of both the "medium" and "high" intermediate goods can potentially be improved over time in a quality-ladder fashion (see Grossman and Helpman (1991a, 1991b) and Aghion and Howitt (1992)), however, these intermediate goods differ in their improvement rate. Formally, the final good is produced by intermediate goods [x.sub.i] and [z.sub.i] in a constant return to scale technology which is given by:

   [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

   where 1/(1-[sigma]) is the elasticity of substitution between the factors of production which is assumed to be higher than one (0

   The final good Y is assumed to be perfectly tradable and its market is perfectly competitive. The intermediate goods, however, are not tradable and their markets are domestic. (4)

   To keep the analysis simple and to highlight the effects of interest, several assumptions are used:

   1) High-tech products are much more productive per quality rank than medium-tech products. Formally, the inequality 1

   2) All products of type x have identical quality rank intervals such that [[lambda].sub.1] is constant across intermediate goods of type x.

   3) Intermediate goods of type z are located in a decreasing order from the highest to the lowest improvement rate such that [[lambda].sub.2]([i.sub.1]) [greater than or equal to] [[lambda].sub.2]([i.sub.2]) for all [i.sub.1]

   4) To reach an equilibrium in which innovators always employ a limit pricing strategy we also assume that [bar.[lambda]]

   The quality rank intervals of high-tech and medium-tech products are shown in Figure (1) below.

   [FIGURE 1 OMITTED]

   2.1 Individuals

   At each period of time, a generation L of individuals is born. All individuals live for one period only and have identical concave preferences denoted by u = u(c). There are two types of individuals in the economy: a portion (1-[mu]) of less-competent individuals and a portion [mu] of highly-competent individuals (0

   Within the population of competent individuals, there exists a measure v=1 of individuals who we henceforth refer to as "innovators." It is assumed that innovators are the only individuals in the economy who have the skill to upgrade existing intermediate goods and subsequently to manufacture them. We further assume that each innovator i [member of] [0,1] is endowed with an idiosyncratic ability parameter A(i) [member of] [[[lambda].bar], [bar.[lambda]]] that reflects his ability to upgrade high-tech products. Each innovator i [member of] [0,1] with the ability parameter [lambda](i) is matched to two brands (prototypes) of intermediate goods: an intermediate good [x.sub.i] of type x and an intermediate good [z.sub.i] of type z whose improvement rate is given by [[lambda].sub.2](i) = [lambda](i). An innovator i [member of] [0,1] can undertake only one project either of type x or type z, but not both.

   One of the most important assumptions of the paper is that members of the economy know, ex-ante, only the probability of being either incompetent or competent (i.e., [mu] and (1-[mu]), respectively), however they do not know their own type nor the types of others. (5) During the production process, however, employers can reveal their workers' type as long as they employ them for at least a 0

   2.2 Production of Intermediate Goods

   All intermediate goods are produced by a linear production function in which labor is the only primary factor. Intermediate goods differ, however, in their labor requirements. We assume that one unit of intermediate good of type x is produced by either one unit of competent workers or 1/[theta] units of less-competent workers (where [theta]>1). Intermediate goods of type z however, differ in their production technology according to whether they were previously or recently upgraded. One unit of an old vintage of the [z.sub.i,j] product can be produced by either one unit of a less-competent or (1/[theta])

   [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

   where [l.sub.u] and [l.sub.s] are labor inputs of less-competent and competent workers, respectively.

   The difference in labor requirements between medium tech and high...