Labour market institutions without blinders: The debate over flexibility and labour market performance

Abstract

The debate over the influence of labour market flexibility on performance is unlikely to be settled by additional studies using aggregate data and making cross-country comparisons. While this approach holds little promise, micro-analysis of workers and firms and increased use of experimental methods represent a path forward. Steps along this path could help end the current ‘lawyer's case’ empiricism in which priors dominate evidence.

Keywords:

• Flexibility
• performance
• aggregate data

Notes

¹This paper deals with the debate over institutions in developed economies. For an analysis of the debate with regard to the developing countries, see Freeman (1993).

²See John Martin (1998).

³IMF (2003: chapter 4, p 129).

⁴IMF (2003: chapter 4, p 131).

⁵IMF (2003: chapter 4, p 129).

⁶Mussa, (2002: 9), the italics are mine.

⁷Krueger (2002), the italics are mine.

⁸This is what one would expect in an ideal competitive economy, where industry shocks alter employment but not wages. See Council of Economic Advisors, (1962) and Salter (1960), for classic statements of this principle.

⁹Nickell & Bell (1996) show that EU–US employment gaps are as large among skilled as among less skilled workers. Card et al. (1994) and Freeman & Schettkat (2001) find little relation between employment and labour costs of workers by industry or occupation across countries. Increased hours worked in the US among the educated and women, whose wages rose relative to the US average, also runs counter to the rigidity story (Freeman, 1995). Evidence that changes in the US minimum had negligible employment effects imply that low wages did not cause high employment (Card & Krueger, 1997)

¹⁰The evolving views of the OECD can be seen in various Employment Outlooks (1995, 1996, 1997, 1999).

¹¹This is true for advanced countries. The situation is more ambiguous in developing countries since unions do not represent workers in the informal sector and rarely represent rural workers, who are paid less than those in the modern sector.

¹²Erickson & Ichino (1995), Manacorda (2004).

¹³Patterns in strikes run counter to a Coase theorem interpretation. If management and workers lose from strikes, they should come to an agreement without strikes or lockouts.

¹⁴The Spring 2004 German IFO institute economics magazine CESifo Forum was devoted to European labour markets. Many economists favoured some aspects of the orthodox view while one (me) argued that European labour markets were not ‘as bad as all that’ (Freeman, 2004). My argument was that no real-world market looks good when compared to the ideal competitive market, but that such a comparison was misleading since no real-world market performs in accord with the competitive ideal; and that it was this comparison of European labour markets with the ideal that led many analysts astray.

¹⁵IMF April (2003: Figure 4.4, p. 134).

¹⁶Charles Ragin (1987) has done the most to analyse the problem of inferring relations from configurations.

¹⁷If we used a high/medium/low categorization, we would have 81 (=34) possible configurations.

¹⁸The highly correlated outcomes imply that having higher frequency data cannot resolve the problem, you do not learn any more by having monthly unemployment rates than from yearly unemployment rates. In developing countries, the Easterly et al.'s (1993) finding that growth rates are not highly correlated creates a different problem in assessing how institutions or policies affect outcomes.

¹⁹Economists, like the general public, also derive priors from ‘anecdotal evidence’, the events that we personally experience or observe others experiencing.