tisdag 13 oktober 2015

Brandl: lönepakter kan sänka arbetskostnaderna

Bernd Brandl, genom samarbete med Franz Traxler och egna arbeten, är en av de viktigaste lönebildningsforskarna. 2012 publicerade han ett papper om tripartistiska lönepakter vars abstract lyder så här:
"Recent debates on pacts have focused on the prerequisites for their emergence, whereas questions of their efficacy have receded into the background. In particular, systematic analyses of the effectiveness of pacts in terms of their capacity to enhance economic performance are missing. The aim of this article is therefore to assess the economic impact of pacts. As the majority of pacts concern wages, the assessment will concentrate on a comparison of the performance of pacts with alternative governance mechanisms for wage policies, that is, alternative pay-setting modes. The findings show that when wage pacts are endowed with the ability to govern lower-level pay determination, they are better at enhancing economic performance than other forms of coordination."
Och i pappret motiverar han det så här: 
"research on pacts has focused on the prerequisites for their emergence, whereas questions of their effectiveness have clearly receded into the background. Issues of effectiveness have frequently been addressed by case studies that concern one single country (Fajertag and Pochet 1997, 2000). However, case study analysis is scarcely able to examine the specific effect of pacts on intended outcomes, since there is always a large number of other potential factors whose actual impact cannot be isolated in a small number of cases. On the other hand, comparative studies tend to take the efficacy of pacts simply for granted, or merely equate the ‘success’ of pacting with the ability of the relevant actors to reach an agreement, while ‘unsuccessful’ pacting implies failure to arrive at concertation (Avdagic 2010; Culpepper 2002; Hamann and Kelly 2003). Again, such a criterion of success echoes the concern of the debate about the emergence of pacts. Success in concluding a pact does not say anything about its real impact. For this purpose, one has to systematically compare the effects of pacts with alternative governance mechanisms within a quantitative time-series cross-sectional (TSCS) design. There are only a few studies of this kind which concern pacts. Baccaro (2006) examined the impact of bargaining coordination on unemployment and pay restraint in 18 countries between 1960 and 1998. The findings show that coordination has a significantly dampening effect only on pay movements which cannot be directly linked to pacts since pay-setting can be coordinated by mechanisms other than pacts. A further problem is that the period covered is larger than the time which makes pacts distinct from former corporatist concertation, that is, the conversion to monetarism and its consolidation (Hassel 2003; Scharpf 1991). Brandl and Traxler (2005) assessed the impact of pacts on welfare expenditures and unemployment against alternative modes of pay regulation and other industrial relations properties. The time period covered was from 1980 to 1996. Pacts were defined as state-sponsored (i.e. tripartite) pay coordination. It was found that pacts are at neither the top nor the bottom of welfare expenditures. In terms of unemployment, pacts are among the worst performing arrangements if governability (i.e. control over lower-level pay-setting) is lacking. Otherwise, they are in between extremes of unemployment performance.

Since these studies refer to a core regulatory field of pacts, that is, pay policy, they beg questions about the superior performance of pacts, an assumption which explicitly or implicitly permeates the debate. Moreover, theoretical reasoning on public policy and organized interests does not lend support to such assumptions, as will be shown below. Hence, there is good reason to review relevant theoretical accounts and to rigorously test the effects of pacts. This is the aim of this article which tests the effects of pacts on time-series data for 20 countries from 1980 to 2003.1 The time period studied covers the emergence of pacts in the 1980s and the continued conclusion of pacts during subsequent decades." (s 483f)
Jag misstänker att begränsad tillgång på data är en del av motivationen till att pappret handlar om perioden efter 1980, men så här motiverar han det i pappret:
"The time period covered in the empirical analysis covers the emergence of pacts in the 1980s, because of the shift in the early 1980s towards supply-side (i.e. monetaristic) economics in developed countries, which made wage policy necessary to control inflation and to foster international competitiveness (Brandl and Traxler 2011; van Waarden 2003) and the continued conclusion of pacts during the subsequent decades." (s 490)
Så här ser hans kodade dataset, inte helt olikt Brandl och Traxler (2010), ut:

Han använder regressioner där de förklarande variablerna laggats ett år för att undkomma endogenitetsproblem, och där även en laggad beroende variabel används som förklarande variabel.

Så här resonerar han om sina beroende variabler, nominell lönetillväxt och reala arbetskostnader:
"Given that wage pacts have aimed to increase international competitiveness of countries since the 1980s (i.e. Hassel 2000), their performance is understood as their capacity for pay moderation which is captured by two dependent variables: nominal wage growth and real growth of unit labour costs. Hence, this analysis differs from the mainstream of empirical studies which take inflation and unemployment as performance measures. The measures used in this study are preferable for several reasons. Since the focus is on the comparative performance of pacts, one has to proceed from their goals which all give priority to pay moderation as a means of containing inflation and improving international competitiveness (Schulten and Stückler 2000). Reflecting the predominance of supply-side orientation, employment is seen as a function of effective pay moderation. However, this view is challenged by Post Keynesianism, which argues that aggregate demand rather than pay movements affect employment in open economies (e.g. Soskice 2000). If this is the case, then testing the effect on employment of wage pacts which issue guidelines for pay moderation, but not for demand management, would fail to take account of their aims. With respect to international competitiveness, the proof of pay moderation is change in real unit labour costs. To test the nominal effects of pacts, wage inflation (i.e. nominal wage growth) is better than inflation per se because import prices which are beyond the realm of national pay-setting increasingly affect inflation with growing economic openness."* (s 490)
Jag gillar att han konstaterar att det dominerande antagandet i litteraturen, att lönesänkningar alltid ger mer sysselsättning, kan ifrågasättas. Och gillar verkligen att han ifrågasätter deras användande av beroende variabler, som jag har gnällt på utförligt tidigare.

Resultaten ser ut så här:

*I en fotnot konstaterar Brandl: "Baccaro and Simoni (2010) use wage growth in efficiency units which is defined as the ratio of real wages and factor productivity as a dependent variable for their analysis of the economic effects of industrial relations institutions. The analysis in this work controls for factor productivity but uses the direct dependent variables."

Bernd Brandl (2012) "Successful Wage Concertation: The Economic Effects of Wage Pacts and Their Alternatives", British Journal of Indsutrial Relations.

Inga kommentarer: