هیئت نمایندگی مقاوم با اولویت های سیاست های پولی نامشخص

Recent research has renewed interest in the exploration of the optimal design of monetary policy institutions in the presence of uncertainty. In this paper, we revisit the rationale for delegation to a weight-conservative central banker when the social planner's knowledge about the true preferences of delegates is ex ante ambiguous and he exhibits a preference for robustness. In this context, a robust (worst-case oriented) delegation strategy is intended to minimize the maximum welfare loss over the uncertainty set, when no prior probability distribution for the preference bias is available. We find that both effective over- and underconservatism may emerge with respect to the certainty case, for robust delegation is shown to be model-dependent. Most importantly, under reasonable model's parameterizations, Rogoff's principle is reversed: it is optimal for society to appoint a weight-liberal central banker.

In recent years, several studies have emphasized the importance of model uncertainty for the conduct and performance of monetary policy (e.g. Blinder et al., 2008, Cukierman and Meltzer, 1986, Dennis, 2010, Leitemo and Söderström, 2008, Levin and Williams, 2003, Onatski and Williams, 2003 and Vickers, 1986). As a natural step forward, this strand of research has renewed interest in the exploration of the optimal design of monetary policy institutions under several sources of model misspecification and uncertainty (e.g. Beetsma and Jensen, 1998, Hefeker and Zimmer, 2011, Muscatelli, 1998 and Tillmann, 2009a). The present paper aims at contributing to this literature by addressing the case of robust monetary policy delegation in the presence of ex ante ambiguity about the true preferences of delegates, once in office. 1 In this respect, Tillmann (2008) studies the welfare costs to society of appointing a central bank who exhibits a conservatism-gap with respect to the optimal scheme according to Rogoff (1985). While Tillmann (2008) only allows for an ex post deviation in the central banker's optimal degree of conservatism, the presence of imperfect predictability of monetary policy due to uncertain central bank preferences also has crucial implications from an ex ante perspective, as it might affect the underlying rationale for delegation, i.e. the stabilization bias, and hence modify society's optimal commitment to price stability. It is well-known from the early literature on time-inconsistency of monetary policy (e.g. Barro and Gordon, 1983 and Kydland and Prescott, 1977) that discretionary policy under an expectations-augmented aggregate supply gives rise to an average inflation bias. According to Rogoff's (1985) seminal analysis, delegating monetary policy to an independent central bank with a lower relative weight on output gap stabilization would provide the government with a way to commit itself to a zero inflation policy. While this has the effect to mitigate the average inflation bias, the relative stabilization of inflation and output is no longer optimal. This stabilization bias issue carries over to the recent generation of New Keynesian business cycles models featuring forward-looking inflation dynamics, if the latter are subject to persistent cost-push shocks ( Clarida et al., 1999). In this sense, the rationale for delegating monetary policy to an inflation-averse central banker remains valid. Rogoff's (1985) approach to solving the inflation bias of discretion focuses directly on the preferences of the central bank. To this end, it posits that monetary policy preferences with respect to conflicting objectives are perfectly identifiable ex ante. In this paper, we analyze the optimal delegation arrangement when the social planner's knowledge about the true degree of conservatism of delegates is ex ante ambiguous and he has a preference for robustness.2 The notion of Knightian uncertainty (ambiguity) is employed to capture situations where randomness surrounding a decision-making problem can not be described by a probability measure. We study the optimal degree of commitment of an ambiguity averse social planner who is unable to assign any prior probability distribution to the monetary authority preferences bias. Under this non-Bayesian approach, the ex ante information available to the appointing player simply consists in the knowledge of the (bounded) support of the unknown bias. In order to hedge against this form of ambiguity, he then adopts a minmax approach (e.g. Hansen and Sargent, 2008) to delegation. A robust (worst-case oriented) delegation mechanism is intended to minimize the expected welfare loss to society over the uncertainty set. Following Tillmann (2009b), we refer to such an optimal delegation arrangement as robust delegation. Our most important contribution is to show that society's incentives to delegate monetary policy responsibility to a conservative central bank are dramatically altered by the presence of ex ante preference uncertainty, as conjectured in Tillmann (2008). Specifically, we show that both over- and underconservatism may effectively emerge with respect to the certainty case, for robust delegation is model-dependent. Most significantly, under reasonable parameterizations of the model and a sufficiently high degree of uncertainty, Rogoff's principle is reversed: the optimal commitment to inflation stability entails the appointment of a (supposedly) more liberal central banker than society's.3 The analysis presented here is related to different lines of research. Previous studies on the role of asymmetric information in monetary policy-making, in which the public is uncertain about the central banker's type (among others, Backus and Driffil, 1985, Cukierman and Meltzer, 1986, Vickers, 1986 and Tabellini, 1988), differ from the present one in that they typically focus on the problem of preferences inference on the part of agents endowed with limited information, and how the central bank's decision to preserve its reputation may help to overcome the stabilization bias arising under discretionary monetary policy. Rather, our goal is to identify the optimal appointment scheme in the sense of Rogoff (1985), when the central bank preferences are not perfectly controllable and the social planner is ambiguity averse. Remarkably, in our setting the monetary policy stage remains a complete information game as in Rogoff (1985), since upon conclusion of the delegation process, the monetary authority is assumed to credibly commit to a full transparency regime (e.g. Geraats, 2002). Under these circumstances, once in office the true type of the central banker is common knowledge, and the monetary policy outcome is the usual rational expectations equilibrium with cost-push inflation.4 More closely related to our analysis are the seminal contributions of Beetsma and Jensen (1998) and Muscatelli (1998), who investigate the role of uncertain central bank preferences for the optimal design of monetary institutions. A central prediction of these studies is that a higher degree of conservatism may be advisable for it constrains the volatility of monetary policy behavior in the presence of uncertainty. Our analysis differs from this work along two relevant dimensions. First, from a modeling perspective, the uncertainty introduced by the preference bias in our model is Knightian (ambiguity), while Beetsma and Jensen (1998) and Muscatelli (1998) deal with several forms of stochastic uncertainty. Second, our model's predictions are only in part in line with those derived in the mentioned papers, for we show that (ex ante) preference uncertainty may well drive the delegation process towards a less conservative or even weight-liberal central banker. The paper proceeds as follows. Section 2 briefly reviews standard results about optimal policy delegation under certainty, which are then used as a benchmark. In Section 3 we introduce Knightian uncertainty about monetary policy preferences and discuss the case of robust delegation. Section 4 offers concluding remarks.

In this paper, we have revisited Rogoff's (1985) delegation approach under Knightian uncertainty about central bank preferences. Using a standard New Keynesian framework, we have shown that society's optimal commitment to price stability is dramatically altered by the presence of ex ante uncertainty, and that the familiar Rogoff's result is likely to be overturned when the latter is particularly severe. This finding stands in sharp contrast to the predictions of previous models dealing with the optimal design of monetary policy institutions under uncertain (stochastic) preferences, which rather suggest that a higher degree of conservatism may be beneficial for it inhibits the volatility of the central banker's behavior. Our results might then help to explain why the delegation of monetary policy responsibility to a weight-conservative (independent) central bank remains a highly debated issue on both theoretical and practical levels.