«IZA DP No. 3901 A Behavioral Account of the Labor Market: The Role of Fairness Concerns Ernst Fehr Lorenz Goette Christian Zehnder December 2008 ...»
In particular, if the firm's profit rises, so should the incumbents' wages. Since the studies do not attempt to disentangle shocks that affect firms' profits (for example, productivity shocks) from other shocks (those that only change labor supply), there is no detailed test of this prediction. The study closest to testing this prediction is (Beaudry & DiNardo 1991), who find that current labor market conditions have almost no effect on current wages, but initial labor market conditions are a significant determinant of entry level wages. Support also comes from several case studies of personnel files of firms (Baker et al 1994; Eberth 2003; Treble et al 2001). Compared to data from labor force surveys, such studies, while less representative, show a much clearer picture of how wages change over the course of a career. The evidence of cohort effects is also cleanest in these studies: the picture that emerges shows that entrylevel wages vary widely from year to year. Each cohort then gradually increases from the entry-level wage, thus preserving the initial differences in wages. (Oreopoulos et al 2006) use data on Canadian college graduates to examine the long-run effects of unemployment at the time of graduation on wages. They find very strong and long-lasting effects of the labor market conditions upon graduation on later economic outcomes. If graduating in a boom year (with an unemployment rate five percentage points lower), initial earnings are about 9 percent higher.
One might argue that the timing of graduation is endogenous to the business cycle. However, the results are robust to using the unemployment rate four years after enrollment as an instrument.
There is also strong evidence that employers shy away from wage cuts, and freeze their employees' wages rather than making small wage cuts. There are two noteworthy features in the distribution of nominal wage changes. First, there is a clear drop in the density just below zero. A large fraction of individuals receive a nominal wage change of zero, but almost nobody receives small wage cuts. Second, small wage increases are frequent. Hence, there is a clear asymmetry in the distribution of wage changes: wage cuts occur less often than expected, as the model predicts. Virtually any study that examines the distribution of wage changes finds this pattern (see e.g. (Altonji & Devereux 2000; Fehr & Goette 2005; Wilson 1999)).
There are significant measurement problems when moving to more conventional data sets like the PSID or other labor market surveys; in particular, wages are typically reported with error (Bound et al 1994). This problem is accentuated when looking at wage changes, and may wrongly lead researchers to conclude that there is a substantial amount of wage flexibility.
Indeed, studies that do not control for measurement error find a significant number of wage cuts, though these studies still find a strong asymmetry in the distribution of wage changes (Card & Hyslop 1996; Dickens et al 2007; Kahn 1997; McLaughlin 1994). Several methods have been proposed for correcting for this problem: some rely on parametric modeling of measurement error (Altonji & Devereux 2000; Fehr & Goette 2005), while others are entirely non-parametric (Dickens & Goette 2006; Gottschalk 2005). It turns out that the specific form of the correction has very little impact. All studies find, however, that correcting for measurement error is important: once these estimators are applied, the evidence obtained from the labor force surveys essentially looks like that from personnel files: there are only very few wage cuts.
Nominal wage rigidity or cohort effects could also be caused by forces other than a combination of fairness preferences and money illusion (Malcomson 1999; Malcomson 1997). Suppose, for example, that workers are risk averse, firms are risk neutral and working hours are fixed. Then there is an incentive for the firm to insure its workers against fluctuations in real wages. The real wage is then just given by the current labor market conditions and the optimal insurance condition; thus subsequent labor market events do not affect the real wage. This naturally leads to cohort effects in wages, as the starting wages will differ by year. Yet, these models have difficulties in explaining nominal wage rigidity because they predict that real, not nominal, wages should be insured. Other models (Malcomson 1999,
1997) show that contracts with a fixed wage can induce efficient levels of relationshipspecific investments. These models have the property that incumbents’ wages only respond to exogenous conditions if the outside option becomes binding. Thus as long as the real wage is strictly within the boundaries of the real outside option, the nominal wage will not be adjusted to changes in real outside options. They are adjusted only when one of the outside options becomes binding. With positive inflation, this automatically implies that wage cuts will tend to be rare. However, the model also predicts that during deflation, wage cuts would be frequent, and raises rare, with the asymmetry going the other way. There is no evidence that during the great depression in the U.S., for example, wage cuts became frequent (Akerlof et al 1996). Similarly, Imfeld (1991) reports virtually no wage cuts in five large Swiss firms during the period between 1879 and 1890 when prices in Switzerland declined by 20 percent.
At the more aggregate level, our framework may also help explain some of the business cycle facts that the standard model has difficulty accounting for. First, the model offers a new source of wage stickiness. For example, the model readily predicts that employment should be more volatile than wages. The reason is that raising employment in the face of a positive demand shock lowers average profit (because of diminishing returns to effort). This persuades the workers to work harder for a given wage, because the wage is now higher relative to the average profit the firm makes per worker. This increases the workers' effort, but does not require paying a much higher wage. Therefore, most of the firm's adjustment will come through changes in employment, making the wage relatively unresponsive to changes in demand on the product market (Danthine & Kurmann 2004). On the other hand, the model also predicts a difference between demand shocks and productivity shocks for wage and employment reactions. In contrast to the demand shock discussed above, a positive productivity shock increases the firm's profit directly. The workers will thus lower their effort for a given wage. However, because the workers' effort now becomes more valuable to firms, this reinforces the incentives of the firm to raise wages (Benjamin 2005; Danthine & Kurmann 2004).
4.4 Fairness and the Economic Effects of Minimum Wage Legislations In this section, we illustrate that the psychological forces described in section 2 may be crucial for a better understanding of labor market policy. If people have reference-dependent fairness preferences, policy measures may not only function by affecting outcomes but also by shifting the relevant reference points. We use the case of the minimum wage legislation to illustrate the empirical relevance of this possibility. The minimum wage example is an especially important one because minimum wages are one of the most often-used instruments in labor market policy (see e.g. OECD (1998) for evidence that minimum wage legislation affects most labor markets in the developed world in one way or another).
Despite the remarkable attention that minimum wage laws have received, three empirical findings remain puzzling in light of the standard approach in labor economics. First, a number of papers show that minimum wages have so called spill-over effects, i.e., many firms increase wages by an amount exceeding that necessary to comply with the higher minimum wage (see, e.g., Card & Krueger 1995; Dolado et al 1997; Katz & Krueger 1992; Teulings et al 1998; Teulings 2003). Second, several studies report anomalously low utilization of subminimum wages in situations where firms could actually pay workers less than the minimum (Freeman et al 1981; Katz & Krueger 1991; Katz & Krueger 1992; Manning & Dickens 2002). For example, Katz and Krueger (1991) find that the introduction of the opportunity to pay subminimum wages to youth did not caused a significant decline in teenage workers' wages. Third, there are several cases in which an increase in minimum wages led to zero or even positive employment effects (see, e.g. (Card 1992; Card & Krueger 1994; Katz & Krueger 1992; Machin & Manning 1994; OECD 1998; Padilla et al 1996). This is surprising, because the conventional competitive theory predicts that increases in minimum wages should always reduce employment.
A recent experimental study (Falk et al 2006) suggests that the phenomena mentioned above may be better understood if the labor market is viewed from the behavioral perspective we describe in this paper. In addition to the possibility that workers may have referencedependent fairness preferences, their experiment is based on the idea that labor markets are likely to exhibit imperfectly competitive features. This view is based on a recent line of research in labor economics stipulating that imperfect competition is perhaps the rule rather than the exception in labor markets (Boal & Ransom 1997; Manning 2003). The rationale behind this argument is that labor markets are typically characterized by important frictions (like moving costs, heterogeneous job preferences, or social ties) which prevent the elasticity of an individual firm's labor supply from being close to infinity, a view that also receives empirical support (Barth & Dale-Olsen 1999; Boal & Ransom 1997; Manning 2003).
Therefore it seems reasonable to assume that firms have at least a certain degree of wage setting power.
Falk et al. (2006) thus implement a simple laboratory labor market in which workers' mobility restrictions in combination with heterogeneous fairness preferences give rise to upward sloping labor supply schedules at the firm level. They observe that the minimum wage strongly affects reservation wages, suggesting that it provides an anchor for judging the fairness of the actual wage paid. After the introduction of the minimum wage, subjects in the role of workers strongly increase their reservation wages. While almost all reservation wages were clearly below the level of the minimum wage before its introduction, a substantial share of reservation wages lie above that level after its introduction. The impact of the introduction of the minimum wage on reservation wages is in line with the evidence presented in section
2.2. The mini-ultimatum games of Falk et al. (2003) revealed that changes in the set of available but not chosen alternatives may have important consequences for the perceived fairness of a specific action. The introduction of a minimum wage eliminates a whole range of previously possible wage payments from the firms' strategy set. As a consequence, many subjects seem to perceive a wage payment at the level of the minimum wage, which would have been considered as fair and quite generous before the introduction, as unfairly low after the introduction.
The impact of the minimum wage on reservation wages has important implications for the wage-setting strategy of profit-maximizing employers: they are forced to pay wages above the minimum. Thus, the strong impact of the minimum wage on subjects' reservation wages provides a possible explanation for the spillover effect empirically observed in field studies.
Furthermore, the pattern of reservation wages also shapes the employment effects of the minimum wage. Since firms face upward-sloping labor-supply schedules, a wage increase may lead to higher employment for a firm’s given labor supply schedule. However, since the minimum wage raises reservation wages, the labor supply schedule individual firms face shifts to the left. Accordingly, a potentially positive employment effect of the minimum wage introduction due to imperfect competition is dampened or may be even reversed because of the leftward shift in the labor supply schedule. Whether the introduction of the minimum wage ultimately leads to an increase or a reduction in employment depends on the parameters which shape the relative size of the two counteracting effects. The minimum wage has a positive net effect on employment under the conditions chosen in Falk et al. (2006). However, the effect is much smaller than it would have been, had subjects' reservation wages remained stable.16 In contrast to the experimental settings discussed in the previous section, Falk et al. (2006) implemented a labor market with complete employment contracts. However, gift-exchange experiments by (Brandts & Charness 2004; Kagel & Owens 2006) show that the impact of minimum wages on labor supply also prevails if the labor market suffers from contractual incompleteness. Both papers show that the introduction of a minimum wage has two effects.
Note that when individual firms face highly elastic labor supply schedules in a relatively competitive labor market, the impact of minimum wages on reservation wages exacerbates the employment reduction caused by the rise in minimum wages.
On the one hand, the minimum wage increases average wages, which motivates fair-minded subjects to exert more effort. On the other hand, however, the minimum wage also changes the fair-minded subjects' willingness to provide effort at a given wage level. It seems that a law in place forcing employers to pay at least a certain minimum makes subjects perceive the same wage, once considered fair, to be less fair. As a consequence, the net effect of the minimum wage on effort is ambiguous and depends on the relative size of the two counteracting effects.