همکاری سیاست های پولی بهینه از طریق قراردادهای مستقل از دولت با اهداف

Simple state-independent monetary institutions are shown to secure optimal cooperative policies in a stochastic, linear-quadratic two-country world with international policy spill-overs and national credibility problems. Institutions characterize delegation to independent central bankers facing quadratic performance related contracts punishing or rewarding deviations from primary and intermediate policy targets.

Among interdependent economies, sovereign policymaking by nationalist policymakers typically involves various inefficiencies that could be internalized to the benefit of all countries. For example, conflicts over the appropriate degree of stabilization towards shocks or permanent conflicts over macroeconomic variables imply that some degree of policy cooperation would be desirable.1 However, the very same incentives that account for inefficiencies in the first place, often make it profitable for one country to deviate from a cooperative scheme in order to pursue self-interests at the expense of other countries. Identification and assessment of mechanisms that sustain cooperative policies are therefore of great importance. In this paper we address this issue with respect to monetary policymaking through a `principal–agent' approach, where governments (the principals) delegate policy conduct to independent central bankers (the agents) whose operational environment is appropriately designed so as to induce optimal, cooperative policies. The design turns out to be remarkably simple, feasible and easily interpretable, and in contrast with those of previous literature it is state-independent even though the world under consideration is stochastic. This may be of theoretical interest in its own right, but the portrayed institution also reveals new results concerning well-known concepts from the related literature such as `conservativeness' ( Rogoff, 1985b) and `inflation targets' ( Leiderman and Svensson, 1995, Persson and Tabellini, 1996a and Persson and Tabellini, 1996b; Svensson, 1997), which are of relevance for the design of real-life monetary institutions in interdependent economies. For example, within European countries participating in the EMU, between the upcoming EMU and non-participants, between EMU and USA, and so forth. Our analysis is conceptually related to Persson and Tabellini (1995), Persson and Tabellini (1996a, b) who examine a two-country, linear-quadratic monetary policy game where policymaking is delegated to independent, benevolent central bankers (CBs). They demonstrate optimality of performance contracts with payments linearly related to some economic variable (typically the inflation rate). The finding draws on the closed economy models of Persson and Tabellini (1993) and Walsh (1995) where linear contracts were shown to eliminate the well-known inflation bias of monetary policy ( Kydland and Prescott, 1977; Barro and Gordon, 1983). But whereas the optimal linear contract in the conventional closed economy case is state-independent, i.e., not contingent upon underlying shocks but only upon realized macroeconomic variables, it is state-contingent in the two-country case. This crucial difference arises since in the closed economy, the linear contract neutralizes the CB's constant inflationary incentives. But as responses towards shocks are efficient, they require no correction and the optimal contract is state-independent. 2 In the two-country case, however, shock stabilization has international externalities that need to be internalized by the contract, which must therefore be state-contingent so as to secure international efficiency for all realizations of shocks. But as noted by Persson and Tabellini, 1993, Persson and Tabellini, 1995, Persson and Tabellini, 1996a and Persson and Tabellini, 1996b themselves, state-contingent contracts do not seem to be very operational in reality as they, for example, are informationally demanding. Furthermore it appears impractical to change the monetary institution each time a new shock occurs (see also the discussions in, e.g., Canzoneri et al. (1997), Herrendorf and Lockwood (1997)). But restricting linear contracts to be state-independent, induces inefficiencies in stabilization policies that must be weighed against average benefits.3 The general message seems then to be that the design of monetary institutions inevitably must take the form of solving a second-best problem. We argue, on the other hand, that one need not resort to second-best analyses. Within a framework similar to the one used by Persson and Tabellini, we demonstrate that state-independent contracts do secure that CBs conduct optimal, cooperative policies for any realization of shocks. This is possible by introducing another class of contracts – just as simple as linear ones – quadratic contracts with targets. They prescribe transfer payments that are constantly related to squared deviations of target variables from respective constant target values. As such, they render the term `target value' meaningful, since deviations in any direction are treated symmetrically (in contrast, a CB facing a linear inflation contract is penalized when inflation exceeds some value, but rewarded when it falls short of the value). Also, institutions specifying various primary and/or intermediate targets in conformity with the quadratic contracts proposed here are observed more and more frequently thus stressing their feasibility. As documented by, e.g., Leiderman and Svensson (1995), many countries have recently adopted inflation target regimes, specifying explicit inflation target values as well as to what extent, and under what circumstances, deviations (in any direction) from target values are accepted. Moreover, various exchange rate mechanisms (for example the ERM in Europe) specify explicit exchange rate targets as well as bands defining how much actual exchange rates are allowed to deviate from target values. Hence, even though we focus on the contract approach to monetary institution building, it should be evident that we consider it a more general one. For example, one can interpret quadratic contracts as broad mechanisms for punishing or rewarding deviations from target values: the loss of the CB's prestige from not meeting the target, operational rules that makes deviations more or less difficult, specification of the width of target bands, and so on. This is in accordance with Rogoff (1985b) who indeed interpreted similar delegation mechanisms broadly as `systems of rewards and punishments'.4 We first demonstrate the achievement of optimal, cooperative policies in the simplest version of the model where countries are only hit by perfectly correlated (positively or negatively) supply shocks, and where a conventional credibility problem vis-à-vis the private sector prevails. Then, a quadratic contract defined over consumer price index (CPI) inflation is sufficient. It is noteworthy that when shocks are positively correlated, international spill-overs in stabilization efforts cause too little inflation variability, and in consequence the optimal contract should actually promote inflation variability. This can be interpreted as `weight-liberalism' in the Rogoff (1985b) terminology, i.e., CBs putting less weight on inflation stabilization than society. A similar result was firstly recognized by Laskar (1989), but he considered the case where the weight attached to inflation stabilization was the only instrument in the delegation scheme. Hence, CBs should only be `liberals' if the gain in terms of better stabilization effort outweighed the loss of higher average inflation (cf. footnote 3). In our model, on the other hand, the choice of weight secures internationally efficient stabilization, and the target embedded in the contract then resolves national credibility problems by imposing a constant marginal penalty on high inflation. The usual trade-off between credibility and flexibility is therefore not present. This also applies when shocks are negatively correlated. But then, international spill-overs cause too much inflation variability, and the appropriate contract should penalize inflation variability, i.e., policymaking should be handed over to conservatives. With respect to the choice of target value for CPI-inflation, the qualitative properties depend upon whether CBs are liberals or conservatives. If liberals, the target value should be above the socially optimal value. The reason is that when the contract promotes inflation variability, the marginal contract gain is decreasing in the target value and will at some point become negative. Hence, the value should be sufficiently high in order to induce an appropriate marginal inflation penalty as required. If, on the other hand, CBs are conservatives, the target value should be below the social optimum for the opposite reasons. Secondly, we consider the general case where shocks are arbitrarily correlated. Hence, a symmetric and asymmetric shock component are present at the same time. Apart from being more realistic, this will reveal that as the set of policy problems widens, the optimal contract must be expanded correspondingly: when two types of shocks prevail, it takes one quadratic contract part to correct incentives concerning policy responses towards the symmetric component of shocks, and another part to correct incentives related to the asymmetric component. More specific, it is shown that a quadratic contract part defined over CPI-inflation, but also one defined over an intermediate target – the cross-country CPI-inflation differential – in combination secure optimal, cooperative policies. The contract part quadratic in CPI-inflation remedies international inefficiencies pertaining to the symmetric component of shocks, and optimality therefore requires a target value for CPI-inflation above the socially optimal and that inflation variability must be promoted; cf. the discussion above. The contract part related to the CPI-inflation differential then secures optimal responses toward the asymmetric component. The model can be interpreted so as to apply to a number of different real-life scenarios as indicated above. For example, the design of monetary institutions within the EU countries contemplating a more coordinated monetary policy stance. Alternatively, it can portray the interaction between two blocs of countries each constituting a monetary union. In any case, we should emphasize that the policy implications of our results are strongly at odds with the existing proposals for monetary institutions in an international setting. Existing proposals building upon equivalent frameworks usually endorse the appointment of conservative CBs (Laskar, 1989Laskar, 1993; Currie et al., 1995Currie et al., 1996);5 inflation target values that are at most equal to the social optimal, but usually less (this is Svensson's (1997), result for a closed economy carrying over in the two-country model of Canzoneri et al. (1997)); price stability in some circumstances as the overriding objective for monetary policy (Persson and Tabellini, 1996a and Persson and Tabellini, 1996b).6 In contrast, we prescribe an institution where CBs are liberals, inflation target values are not necessarily lower than the social optimal, and where policy should not be subordinated to only one objective at all cost. Instead it should exhibit a sufficiently balanced concern for primary and intermediate policy targets. We would therefore expect such an institution to be much more politically feasible in times of low employment in comparison with existing proposals. To paraphrase Goodhart and Viñals (1994), those proposals could, to various extents, be expected to trigger headlines like `Success for monetary policy: zero inflation and 500 000 are thrown out of jobs' and thus cause problems with public presentation. The plan of the paper is the following: in Section 2the stylized, stochastic two-country model and optimal, cooperative policies are presented. Section 3then demonstrates how to attain these benchmark policies through delegation to independent CBs facing state-independent quadratic contracts with targets in the cases where supply shocks are either perfectly symmetric (Section 3.1) or perfectly asymmetric (Section 3.2). The case where shocks are arbitrarily correlated is considered in Section 4where it is demonstrated that this additional dimension in the stochastic nature requires one more contract part in order to attain the benchmark through state-independent means. Section 5examines the robustness of the results when allowing for speculative shocks to the exchange rate, and Section 6discusses how other contracts than those considered may be appropriate. Section 7sums up. The appendix derives optimal, cooperative policies.

We have shown that governments in a stochastic, interdependent world can achieve optimal, cooperative monetary policies by delegating policy conduct to independent CBs facing state-independent quadratic contracts with targets. This is in contrast with existing analyses where delegation is viewed as a second best problem when it cannot be conditioned on the realizations of underlying shocks. This holds for both the linear contract solutions proposed by Persson and Tabellini, 1995, Persson and Tabellini, 1996a and Persson and Tabellini, 1996b and the proposals of appointing more or less conservative CBs; cf. Laskar, 1989 and Laskar, 1993, Dolado et al. (1994), Currie et al., 1995 and Currie et al., 1996. Apart from lifting the delegation design above second best considerations, our proposed contracts also seem more reminiscent of real-life institutions as they – broadly interpreted – portray institutions which make deviations of constant, primary or intermediate targets from particular values more or less difficult. Moreover, their qualitative features seem more politically feasible compared with existing proposals which to various degree promote inflation targeting in a much stricter sense as envisioned here (e.g., by making price stability the more or less sole objective for monetary policy). That our contracts with targets are able to achieve first-best, in contrast with other proposals, reflects a rather general and old principle. In the classic Tinbergen (1952) terminology, a policymaker can only achieve his desired values of target variables if the number of policy instruments is no less than the number of targets. This principle is at work in our model as well. When shocks are arbitrarily correlated, the governments want to remedy three distortions (the inflation bias and the international externalities arising from symmetric and asymmetric components of shocks), and the CPI-inflation contract with a target and the CPI-inflation differential contract exactly offer three instruments in institutional design . Adding more distortions to the model would then require a wider set of contracts so as to accomplish the first-best.13 Thus, even though our contracts remove the usual trade-off between credibility and flexibility present under Rogoff-conservativeness and state-independent linear contracts, this principle suggests that another trade-off may appear in the design of institutions. Namely that between efficient economic performance and the `degree of complexity' of institutions. This trade-off does not seem to be relevant in the simple world envisioned here, but even in more complicated scenarios, where it potentially would, our quadratic contracts with targets may still outperform other forms of delegation. With respect to the plans for monetary unification in Europe it should be mentioned that our proposed contracts do not require a fixed nominal exchange rate. Equivalently, under the interpretation of the model as portraying blocs of countries (each with a common currency), nor do they, e.g., advocate that countries outside the EMU should peg their currencies to the Euro (it is well-known that models of the kind used here leave little room for fixed exchange rates when asymmetric shocks are present; cf. Canzoneri et al. (1997) and Persson and Tabellini, 1995 and Persson and Tabellini, 1996a). But if fixed nominal exchange rates, for some reasons, are advantageous and is going to be a fact of life between countries or blocs, it is of importance to examine whether quadratic contracts with targets of some form are still able to secure first-best policies. With asymmetric shocks, however, such institutions must involve fiscal policy, and we leave the issue for future research. Finally, note that the current approach presumes that governments jointly set up the institution that governs CBs' behavior and that this institution always bind. This ignores potential strategic aspects of institution building, which arises as the incentives causing policymakers to deviate from cooperative policies would also cause governments to deviate from cooperative institutions. Hence, one could argue that the problem about sustainability of cooperative behavior is merely `relocated' to the institution building stage; cf. the critique McCallum (1995) and Jensen (1997) put forward against closed economy models of monetary delegation.14 We do not dismiss such strategic aspects as unimportant, but emphasize that the identification of optimal monetary institutions is of considerably interest in its own right, and that their credible implementation is still an unresolved issue.