Comments on “Wage determination and imperfect competition”

  • Published on

  • View

  • Download

Embed Size (px)


  • Labour Economics 30 (2014) 5961

    Contents lists available at ScienceDirect

    Labour Economics

    j ourna l homepage: www.e lsev ie r .com/ locate / labecoCommentaryComments on Wage determination and imperfect competitionBernd Fitzenberger University of Freiburg, IFS, IZA, ROA, ZEW, Germany 2014 Published by Elsevier B.V.

    Department of Economics, University of Freiburg, 790E-mail address: bernd.fitzenberger@vwl.uni-freiburg.da r t i c l e i n f o likely existence of rents in employment relationships,which the paper re-Article history:

    Received 12 May 2014Received in revised form 14 May 2014Accepted 14 May 2014Available online 6 September 2014

    lates to search and matching frictions, market power in goods markets, aworker's ownership of her/his human capital, human capital formation asa long-term investment, specificity of human capital, or the requirementto be regularly present at the workplace. These are also the reasons whya theoretical model postulating a perfectly competitive labor marketmay not suffice to describe modern labor markets.

    While the emphasis of the paper is on theoretical considerations,there are various references to empirical work. These references areWriting a comprehensive survey paper on Wage Determination is aformidable task, if not a Mission Impossible. To restrict the paper to thestate of theoretical work regarding Imperfect Competition still relatesto an extensive literature, which is impossible to review in a compre-hensiveway. So, Alison Booth is right in her choice to focus on two inter-esting strands of theoretical work as two contrasts to the notion of aperfectly competitive labor market. The two strands are wage determi-nation under trade unions and wage determination under oligopsonis-tic competition. Setting the scene in this way is interesting in its ownright for two reasons. First, wage determination under trade unionsstarts out by arguing thatworkers, as represented by a union, havemar-ket power while wage determination under oligopsonistic competitionstarts out with the assumption that firms have market power. Second,while wage determination under trade unions was a key researchissue in labor economics until the early 2000s, wage determinationunder oligopsonistic competition has become a major research areaduring the last two decades. Alison Booth finds both approaches impor-tant and finds the current lack of interest in wage determination undertrade unions surprising in light of the importance of trade unions inmost OECD countries. In fact, she suggests that the dominance oftop U.S. Journals in the profession is a reason for this, because tradeunions are of little relevance any more for the private sector in theU.S. economy.

    The paper goes beyond the aforementioned simple dichotomymarket power either sides with the workers or the firms and discussesvarious important refinements of the theoretical approaches. One impor-tant refinement relates to thewage bargaining approach postulating a bi-lateral bargaining situation between unions and firms or between singleworkers and single firms. The bargaining situation arises because of the85 Freiburg, Germany.e.more selective in nature and sometimes relate to important ownwork of the author. The purpose is to help the reader understandin a better way the workings of the labor markets in general andwage determination in particular. Furthermore, the paper stressesat various occasions how the predictions of a purely competitivelabor market model or of simple versions of a model of wage deter-mination with imperfect competition may be misleading. The paperprovides a useful reminder that theoretical predictions may not berobust when developing a more general modeling approach. The ul-timate judgment is thus an empirical one.

    This is a very useful and thoughtful paper. I like the focus on tradeunions and oligopsony. I agreewithmost ofwhat is being said, in partic-ular that imperfect competition in labor markets is key to our under-standing of a lot of what is going on in the labor market. Sometimes, Iwould have preferred a more nuanced discussion (see below), butthat is partly a question of taste. I also agree with the tacit assumptionthat a parsimonious model that serves as a blueprint for varioussituations is useful. My subsequent list of specific and often idiosyn-cratic comments can only focus on a subset of what is being coveredin the paper. My comments will stress those points where I thinksome clarification is in order. Before I provide the enumeration of myspecific comments, I would like to make the following three generalpoints. First, ultimately, the paper addresses the issue to what extentthere is a labor market, i.e. a setting where individual workers orfirms respond to given labor market conditions (e.g. when a hedonicwage function describes the distribution of market wages). Althoughthis is implicitly done, I think conceptually the paper could distinguishin a sharper way that the responses of participants in the labor market(e.g. the behavior of these participants could be explained by convention-al labor demand, labor supply, human capital, or search decisionmodels)from the possibility that wage setting and allocation rules at the level ofthe relevant labor market as a whole may differ from the predictions ofa purely competitive model. Second, there does not seem to exist a con-siderable theoretical literature which combines explicitly the two strandsdiscussed in sections 3 and 4 of the paper (this is not a criticism of thepaper!). I think it would be worthwhile to explore this in a realistic

  • 60 B. Fitzenberger / Labour Economics 30 (2014) 5961setting. Take a stylized viewof industrial relations inGermany. At thefirmlevel, a search and matching model with wage posting may explain thewage offers and employment decisions subject to the constraint thatwages must not lie below the union contract wages for the specificgroup of workers. At the market level, unions and employers are in abilateralmonopoly situationwhendecidingupon the schedule of contractwages. Third, workers and firms are heterogeneous which is key for ourunderstanding of wage inequality (see e.g. the recent study by Cardet al., 2013). The paper does not address this issue at length, except forthe discussion in section 4 regarding the impact of minimumwages andthe relationship between training and wage compression.

    Specific comments:

    1) In her paper, Alison Booth argues that labor economics has movedbeyond the notion of competitive labormarkets. Even if the compar-ative statics give the same predictions, one has to analyze a modelunder imperfect predictions first in order to know this. I agreewith these statements, but I still think that a competitive labormarket model with heterogeneity and (possibly intertemporally)optimizing agents is a useful focal point. This is because the modelis well developed formally, possibly providing tractable solutionsand clear empirical predictions, where the latter can be tested. Inaddition to tacit assumptions about production technology, prefer-ences etc., there is a further ambiguity when deciding about whichmodel with imperfect competition to use, and the choice of modeldepends on the specific question analyzed. Thus, while a specificmodel with imperfect competition may seem better suited, one maynot be in a good position to test this model. It would be useful if theliterature were to develop further taxonomy of useful theoreticalmodeling devices. In my reading, the search and matching theoryhas made some progress in this direction (see Burdett andMortensen, 1999), but trade unions andworker orfirmheterogeneitydoes not figure highly in this literature.

    2) I agree with the paper that the decline in the interest in trade unionsis unfortunate and that further economic analysis would be interest-ing and highly needed. It is clear that there has been a decline intrade unionism, which has gone very far in the U.S., but trade unionshave also lost importance in various other OECD countries as well(see Dustmann et al., 2009; Dustmann et al., 2014, or Antonczyket al., 2010 for the case of Germany). And, indeed, there had been alot of research on trade unions up to the early 2000s. Hence, even ifthe U.S. had a smaller share in prestigious journals, I would haveexpected a falling number of studies on trade unions. And, in fact,topU.S. journals still do publish studies,which show that trade unionsplay an important role (see Dustmann et al., 2009; Dustmann et al.,2014 for the case of Germany) or which argue that existing tradeunions play only a little role for our understanding (see Card et al.,2013, again for the case of Germany).1

    3) The paper argues that theoretical model of trade unions are generalizable and are relevant to a broader class of situationswith a bilateral monopoly of workers and management (sixthparagraph in section 3.1). This is an interesting point, and Iwould have liked to see a more explicit and longer discussion.

    4) When models of wage determination under trade unions areextended towage bargaining in a bilateralmonopoly, the bargaining1 Based on reweighting by unionization rates over time, Dustmann et al. (2009) andDustmann et al. (2014) find that the decline in unionization rates (as measured throughcoverage by a collective bargaining agreement) partly explains the rise inwage inequalityin Germany between the 1990s and the 2000s, especially at the lower tail of thewage dis-tribution. In contrast, Card et al. (2013) emphasize the role of rising heterogeneity offirmsand of sorting effects within firms. Antonczyk et al. (2010) find no role for the decline inunionization rates on lower tail wage inequality when holding the distribution of firmcharacteristics fixed. Without going into further details of this ongoing discussion (seeDustmann et al. 2014 for a recent study), these studies confirm the point of Alison Booth'spaper, that an analysis of trade unions is still important.power of each side is typically assumed exogenously. Union mem-bership is a key determinant of the bargaining power of a tradeunion, but by no means the only one, as noted in the paper. AlisonBooth, herself, (see Booth, 1985) has contributed a lot to our under-standing of union membership, emphasizing the role of seeminglynon-economic motives, such as social customs. Models of wagedetermination should account for differences in bargaining power,and ideally explain such differences. In an empirical work, onecould for instance model the bargaining power of a union in aNash bargaining solution as a function of membership. Also interac-tions between membership and coverage by collective bargainingare likely. For instance, the empirical analysis in Fitzenberger et al.(2013) suggests that the coverage wage premium increases withhigher membership.2 The study interprets this finding such thathigher membership involves a stronger bargaining weight of thetrade union thus resulting in higher wages of covered workers.

    5) When modeling the impact of wage determination under tradeunions, it is important to account for the spillovers between theunionized segment and the non-unionized segment of the labormarket (Pencavel, 1991, chapter 6). On the one hand, as the paperemphasizes, there may be union threat effects resulting in higherwages and lower employment also in the non-unionized segment.An alternative mechanism leading to the same result would bethat because of higher wages in the unionized segment, a workermoves from the non-unionized segment to the unionized segment,thus reducing effective labor supply in the non-unionized segment.On the other hand, if trade unions cause unemployment in theunionized segment, then effective labor supply may increase in thenon-unionized segment. This may lower wages and increaseemployment in the non-unionized segment. This is the crowdinghypothesis (see Pencavel, 1991, chapter 6), which Fitzenbergeret al. (2013) find evidence for.

    6) Models of wage determination under trade unions often ignore theheterogeneity of workers. How does a trade union balance off theinterests of different worker types in the face of heterogeneouslabor demand (Hamermesh, 1993)? There does not seem to exist aconsiderable literature on this important issue. One exception ismy study in Fitzenberger (1999, chapter 6), where I estimate astructural monopoly union model (in the spirit of Pencavel, 1991)with two skill groups of labor accounting both for the differencesin averagewages and the dispersion ofwageswithin the two groups.The approach is adhoc in the sense that I just assume that the unioncan set average wages and wage dispersion at the same time.

    7) I think it would be both desirable and challenging to integrate wagedetermination under trade unions and wage determination underoligopsonistic competition. One interesting study along this line isthe paper byDustmannand Schnberg (2009). This paper is stronglyrelated to the discussion in section 4.3 on the relationship betweenoligopsonistic competition and training. Dustmann and Schnberganalyzewhether a trade unionmay compress wages, thus providingan incentive for training. Specifically, the paper argues that collectivebargaining in Germany provides a commitment device to overcomehold-up problems in the provision of apprenticeships, whereapprenticeships are costly for firms and do involve the formationof general human capital.

    8) This last comment is by no means a criticism of the paper given itsfocus and breadth. There are two further issues, which seem impor-tant tome for wage determination and imperfect competition andwhich are not covered by the paper. The first issue relates to thedistinction between employment at the extensive margin (headcounts) and the intensivemargin (hours worked).While themodel-ing of labor supply has kept a focus on this distinction, the issue hasfallen out of fashion (to use a term used in the paper) in the analysis2 Such an argument is also implicit in the analysis in Fitzenberger (1999, chapter 6).

  • 61B. Fitzenberger / Labour Economics 30 (2014) 5961of labor demand. Tomy knowledge, there have been very fewpaperson labor demand looking at this issue over the past two decades.3

    Given the widespread interest of trade unions in controlling hoursand given the interest in the literature on frictions at the firm level(see the analysis of dynamic labor demand [Hamermesh, 1993],or the analysis of frictions in the search and matching theory[Mortensen and Pissarides, 1999]), this is quite surprising. A secondissue, I would like to mention, relates to the intertemporal decisionprocess ofworkers andfirms. As discussed in thepaper, rents are im-portant for our understa...