natural level…at least temporarily when wages and prices are sticky.”  However, rational private actors will anticipate these expansionary “surprises” and compensate for them ex ante, thereby “introduc[ing] an inflationary bias into wage bargaining and price setting at an earlier stage of the game.”  Consequently, the expansionary “surprise” does not raise employment and output, although it does induce “higher inflation.”[2]

          While this time-inconsistency problem may preclude politicians from “engineer[ing] a boost in output,” they at least can reduce inflation by credibly “commit[ting] to refrain from the attempt to stimulate output after wages and prices are set.”  For example, if politicians signal to private actors that they are “[d]elegating monetary policy to an independent central bank staffed with officials that are more averse to inflation than [is] the government,” then they will lower “inflationary expectations,”[3] in turn causing “nominal wage-price settlements [to] be lower than they would otherwise be,”[4] and finally yielding lower “actual inflation.”[5]

          Yet what if these “credible signals” about central bank independence (CBI) do not eliminate each private actor’s “uncertainty” about the “wage-price settlements” that other actors might reach?[6]  In Mixed Signals: Central Bank Independence, Coordinated Wage Bargaining, and European Monetary Union, Hall and Franzese argue that coordinated wage bargaining (CWB) – “the degree to which trade unions and employer organizations actively coordinate the determination of wage settlements” on a national level – helps to resolve such collective action problems.  In their model of low CWB, “each bargaining unit, generally a dyad of employer and union,” independently negotiates a wage settlement, “in the context of considerable uncertainty about what the settlements reached by other bargaining units will be.”  As a result, “the union in each dyad” will demand “an extra ‘inflation increment’ on top of the real wages it desires in order to protect itself from the real-wage losses it will incur if other settlements are more inflationary than its own.”  Each union may understand that its own ‘inflation increment’ contributes to an inflationary “economy-wide level of wage settlements.”  Nevertheless, it will not reduce its demands, for fear that it will “suffer real-wage losses” if other unions do not follow suit.  And when these settlements do induce inflation, the central bank or government “may respond with deflationary policies”[7] that increase unemployment.

          In Hall and Franzese’s model of high CWB, by contrast, the leading national associations of employees and employers anticipate that their settlements are “likely to be copied by other bargaining units.”  This expectation renders them less fearful that other, smaller units will arrive at higher wage settlements.  Instead, they are aware “that the central bank is likely to respond directly” to their own wage settlement.  As a result, they will moderate that settlement in response to the central bank’s signals about future monetary policies, and the central bank will not need “to apply tight monetary policies that induce substantial increases in unemployment” in order to reduce nominal prices and wages.[8]

          To their credit, Hall and Franzese’s economic models expand the set of institutions under examination in “most conventional analyses of [CBI]”[9] by attributing to CWB the theoretically interesting function of resolving mutual uncertainty about wage settlements.  However, they fail to consider that the peculiarities of this collective action problem might render it amenable to a simpler solution than that institutionalized by the national agglomeration of wage bargaining units.  Several such peculiarities merit attention.

          First, we must clarify the nature of this collective action problem.  If union X in a low-CWB context fears that “other settlements [will be] more inflationary than its own,” then it is not because X worries that other unions entertain the possibility that politicians might create an inflationary surprise.  Hall and Franzese’s low-CWB model takes into account that bargaining units are aware of a central bank’s credible signals; it only predicts that these units “are unlikely to be highly responsive” to such “threats from the fiscal or monetary authorities to respond to inflationary settlements with deflation.”[10]  Given sufficiently clear and credible signals, X will be aware, not only (a) that an expansionary surprise is unlikely, but also (b) that each other union does not expect a surprise, and most importantly, (c) that each other union knows that all other unions do not expect a surprise.  Accordingly, if X worries that “other settlements [will be] more inflationary than its own,” then it is only because other unions experience the same fear, because other unions do, too – and so on.

          However, in contrast to other collective action problems such as the prisoners’ dilemma, X cannot simply choose to commit to a collectively optimal outcome or to defect in order to pursue its own self-interest; it cannot raise or lower its wages at will.  Instead, it first must attain its employer Y’s consent.  Hall and Franzese claim that Y will be “more likely to accede to high settlements” if Y “can expect…inflation to erode any nominal-wage concessions [it] make[s]”[11] – yet, reasoning from the above paragraph, Y will anticipate inflation only if other unions can convince their employers to agree to higher wages.[12]

          In further contrast to the prisoners’ dilemma, Y will not expect this suboptimal outcome even if one other actor (or a few) defects; “any one bargaining unit is normally too small to have a noticeable impact of its own on the economy.”[13]  Instead, Y knows that a substantial number of employers must act against their self-interest to award higher wages[14] if they are to establish an inflationary level of wage settlements.  And if union X understands that employer Y will find that prospect unlikely – and therefore that other employers find the prospect unlikely – then it need not fear that other unions will succeed in attaining inflationary settlements.  In short, not only is Y unlikely to accede to inflationary wages; for that very reason, X is unlikely to pursue them.

          The empirical record may reflect this theoretical problem in Hall and Franzese’s models.  The authors write that they find “very strong support” for their “third, and most important, hypothesis – namely, that the unemployment costs of increasing central bank independence” (induced by tight monetary policies) “are not zero but rather depend (negatively) on the degree of wage coordination.”[15]  Indeed, a table depicting “the general patterns in [their] results” lists three columns of unemployment figures, each corresponding to a different “[l]evel of coordinated wage bargaining” (0.00, 0.50, or 1.00).  In each row (corresponding to different levels of “[c]entral bank independence”), these unemployment statistics steadily decrease as the reader moves toward higher CWB levels.[16]  Yet the very steadiness of this numerical decrease presents a problem for Hall and Franzese’s hypothesized collective action problem.  After all, unless a collective action problem can be partially solved, middling CWB levels should represent either a failed or successful attempt to minimize the impact of monetary policy on employment, and the values in the middle column of the table should approximate the values in either one of the other two columns.  That this is not the case – that these values increase linearly rather than exponentially – suggests that the hypothesized collective action problem does not exist.

          How, then, might we account for Hall and Franzese’s finding that “the unemployment effects of increasing the level of central bank independence vary according to the degree to which wage bargaining is coordinated”[17]?  Put differently, how can we explain this correlate of CWB without attributing causality to that bargaining institution?

          We might adopt Adam Posen’s critique of institutional explanations of economic outcomes: such studies – a literature that he and other authors term the “new political economy” – neglect to analyze the perpetual struggle among social interests over “the power and prospects of the institutions themselves.”  Yet “the changing relative strength of interests” is analytically prior to any institutions they establish and any policies they pursue.  Posen uses this theoretical approach to argue that CBI correlates with low inflation rates, not because the former causes the latter, but rather because both variables “are endogenous to the effectiveness of financial sector opposition to inflation.”[18]  Put more concretely, a central bank may dampen inflation not because of its legal autonomy from politics, but instead because it enjoys the backing of a strong anti-inflationary political grouping.  Accordingly, as time passes and “the political situation alters,” this approach predicts that even “[central banks] designed with similar degrees of statutory independence will offer significantly differing degrees of protection from inflation.”[19]

          Similarly, we might theorize that CWB correlates with a specific pattern of inflation and unemployment rates, not because one causes the other, but instead because both are endogenous to the political strength of one or more interest groupings.  In fact, because Hall and Franzese use unemployment – a phenomenon with a multitude of causes and opponents – as a proxy for constrictive monetary policy, we might investigate the explanatory weight of a great variety of interest groups.  One promising set of candidates is ideological or partisan coalitions, “with parties of the right being more inflation-averse than parties of the left,” and the latter presumably being more interested in unemployment.[20]  And as Posen concludes, “interests determine public wants far more than even independent institutions do.”[21]