Edwin LE HERON1 & Emmanuel CARRE
Sciences Po Bordeaux – CRECQSS -ADEK Sciences Po Bordeaux – ADEK
Since WW2, three monetary policy regimes have succeeded one another: the Keynesian regime until the mid 1970s, the monetarist regime until 1982 and finally a New Consensus gradually forced itself upon central banking. Inside this New Consensus, inflation targeting became the mainstream during the 1990s. This evolution was often explained through the debate rules versus discretion. This would thus give first discretion with the Keynesianism, then rules with the monetarism and the ‘New’ Classical Economy (NCE) and finally constrained discretion with the ‘New’ Keynesians.
We shall develop the opposition credibility versus confidence as a more relevant and enlightening way to understand this evolution. In particular for the last 25 years, a period in the course of which anticipations became fundamental in the explanation of inflation and in the implementation of the monetary policy. We are moving from a credibility strategy to a confidence strategy.
By a strategy of credibility, we mean a central bank that adopts a model of behaviour and then follows it. It says what it does and does what it says. With monetarism and the NCE, the model of behaviour issues from the common knowledge of natural laws of economy. To respect these laws, the real model imposes a rule of behaviour to be followed by economic agents and the central bank. Then credibility is connected to respect for ‘truth’ (the natural laws). The simplified chain of the credibility strategy is R-C-C: Rule - Commitment - Credibility. The monetary policy simply is an economic and technical problem. Credibility appears as the evolution from the NCE to propound an alternative to the quantity theory of money. The objective is also to preserve the image of a quite powerful central bank able to control inflation alone and perfectly. If uncertainty is studied, rational expectations make it possible to avoid the defects of co-ordination between economic agents and the central bank, which undertakes to respect the rule.
With confidence, natural laws and the underlying real model do not exist any longer. Knowledge is the result of an interaction between the central bank and the agents, which implies a high level of communication. There is confidence when there is a mutual understanding between the central bank and the economic agents. This common understanding should be understood on two levels: in the first place one takes into account the other and secondly the strategy and the conventions of the central bank correspond to those of the other economic agents.
Therefore, it is possible to have credibility without confidence. The central bank should not only account for, but also take into account the economic model, which the agents hold at the moment. The simplified chain of the confidence strategy is C-C-C: Communication - Common Understanding - Confidence. The strategy of confidence is a return to Keynes's philosophy, which does not mean to Keynesianism: the monetary policy is above all a political problem and not the simple respect for an economic rule. The accountability of a central bank, particularly if it is independent, is the cornerstone of the strategy of confidence. A single or predetermined equilibrium does not exist and strong uncertainty makes the management of expectations necessary. Good co-ordination between the central bank and economic agents is therefore essential. This implies a high level of communication and a definition of the objectives of monetary policy accepted by all.
We shall point out five conflicts between a credibility strategy and a confidence strategy:
Credibility versus Confidence,
Independence versus Governance
Responsibility versus Accountability
Common knowledge versus Common understanding
Transparency versus Openness
In the debate credibility versus confidence, inflation targeting (IT) carries an ambiguity. IT is often the end of the credibility strategy: the target replacing the rule. But sometimes, it is the beginning of the confidence strategy: the target as a focal point, i.e. part of a communication strategy. The absence of a specific theoretical model explains the numerous definitions and the varied practices of IT. The opposition credibility versus confidence can clear up the evolutions and the contradictions inside the New Consensus. The virulent current debates within the FED, notably between Greenspan, Kohn, Gramlich (opposing IT) and Bernanke, Mishkin, Goodfriend, Svensson (favourable to IT), can be explained by this opposition.
We shall be interested only in the countries that mostly pursue domestic stability and not those worried mostly with the exchange rates. We shall analyse first the history of credibility, and second the framework of confidence, before finally clarifying the ambiguous status of IT.
1. The credibility framework
1.1. A survey of the credibility framework
At the beginning of the 1970s, Milton Friedman proposed an alternative policy framework to Keynesianism, in which inflation was only a monetary phenomenon. Monetarism believes that a long-term equilibrium exists. A rule is deduced from this presumption. An unconditional rule has to ensure the conditions of this long-term equilibrium, notably through the quantitative relation. Money is neutral in the long term and must be neutralised in the short term. Monetarism supposed that market forces are able to come to equilibrium naturally, over the long run. It frames the ‘true model’ of the economy.
The credibility framework is a synthesis of Monetarism and the NCE, especially with the rational expectations hypothesis. The credibility literature usually starts with the model of Kydland and Prescott (1977)i. Its main contribution is the introduction of a game-theory approach, i.e. to consider monetary policy as endogenous. ‘Economic planning is not a game against nature but, rather, a game against rational economic agents’ (Ibid, p.473). The credibility framework wanted to go beyond the lack of a convincing demonstration of the monetarist argument for rules. Time-inconsistency is presented as a new argument in favour of rules over discretion. Under the rational expectations hypothesis, discretion leads to an average inflation bias, which is related to the augmented Phillips curve and to the natural rate of unemployment. The solution to this problem is that policymakers abandon the optimal control theory (OCT), and adopt ‘policies such as constant growth in the money supply’ (Ibid, p.487). In a context of high inflation, these authors defined price stability as zero per cent, to establish the condition of the long-term equilibrium. Central bank credibility relies on its ability to respect a rule. Thus policymakers would stop pursuing an impossible inflation/growth trade-off and so eliminate the inflation bias. Then, the credibility literature will generate a set of solutions to this inflation bias. The scheme of the credibility literature is: Natural model – Rule – Announcement – Commitment – Enforcement.
Barro and Gordon (1983a-b) really started the credibility framework. Problems arise from the lack of credibility of Kydland and Prescott’s solution to time inconsistency. The solution is to find a commitment technology that makes the announced rule credible to the public. Barro and Gordon introduced reputation in monetary policy as a solution to the inflation bias. They (1983a) developed a game theory approach. Concluding on the impossibility of an enforced commitment (rules), they advocate a reputational equilibrium. Repeated games between agents and central bank should make it possible to reveal the ‘type’ (Backus, Drifill, 1985) of central bank, and the construction of its reputation. In a second paper (1983b), Barro and Gordon preserve the ‘naturalist’ theoretical framework. A zero per cent inflation rate is now rejected, because the authority has too many temptations for fooling. The conclusion is that a formal anchor to a rule is less important than institutional anchorage of expectations (institutional set-up).
The next contribution is the strategic delegation (Rogoff, 1985). Like Barro and Gordon, he puts the emphasis on institutional arrangements, moving from rules to the independence principle. Full independence is presented as the optimal enforcement that makes credible the commitment to price stability. Already, Nordhaus (1975) and Hibbs (1977) had underlined the necessity of isolating the monetary policy from government elections. It was necessary to avoid the political business cycles dependent on opportunist behaviour. For Rogoff (p.1177), a rigid rule is impractical for three reasons:
a cost in terms of unemployment exists in case of unexpected disturbances
it is ‘difficult to alter after it becomes outmoded’
uncertainty over the ever-changing structures of the economy ‘complicates the problem of designing a permanent monetary rule’.
He tries to reduce the inflation bias under discretion. Rogoff’s proposition consists in the delegation of the monetary policy to a conservative central banker. Nevertheless Rogoff is concerned by the credibility/flexibility dilemma of the rule. He focuses on flexible targeting, adding that it is not optimal to constrain the central bank ‘to hit their target exactly’ (Ibid, p.1186). Contrary to previous authors, he defends a hierarchical mandate, not a unique mandate. Rogoff’s solution suffers from democratic problems:
a central banker has different preferences from those of the general public
monetary policy needs public support since it operates in the long-run.
Lohmann (1992) accepts the conservative central banker. Following Rogoff, she insists on the credibility - flexibility dilemma, especially in the presence of large shocks. Her solution is a commitment to an override procedure. The government is allowed to take control of monetary policy in case of a large shock. The knowledge of this institutional arrangement invites the conservative central banker to respond more actively to shocks. This institutional design gives support to the idea that credibility comes from democratic insurance, but also from performance of the central bank.
The weaker was the case for the natural model of the economy and its rule, the more credibility emphasised the optimal institutional design (Persson and Tabellini, 1993). Generally associated with the New-Zealand central bank, Walsh’s model represents a new stage in the credibility framework. Central bank fooling is understood in a different conceptual framework: information asymmetry. Consequently, in a principal-agent model, the author presents an optimal incentive contract. Walsh develops the hypothesis submitted by Barro and Gordon (1983b, p.607): commitments are ‘long term contracts with the public’. Contracts have replaced rules. A contract is seen as the best institutional design for credibility; an institution is considered as an incentive structure. The contract eliminates any inflationary bias and ‘credibility and flexibility are simultaneously achieved’ (Walsh, 1995, p.153). Walsh criticises the reputational equilibrium. Information is imperfect, unobservable and unverifiable. He also points out a limit of Rogoff’s solution. He argues that preferences are not favourable to improvements in terms of the institutional design of central banking.
Svensson (1997) has definitely abandoned a perfect and naturalist model of the economy with its rules. He remains in the delegation perspective with a contract and inflation targeting. His proposal is a criticism of the previous models. Because of enduring unemployment, Rogoff’s solution produces a restriction of output bias. For the same reason, Walsh’s solution encounters both practical and political difficulty and becomes a third best equilibrium. Finally, Svensson defends a mix of ‘weight-conservative and inflation-target-conservative targets’. That’s why Svensson pinpoints the central question of accountability of the central bank. This is aimed at reducing the stabilisation bias.
Woodford (2003) is the last major proponent of the credibility school. He develops a rule-based policy-making strategy with history-dependent variables. He looks for ‘optimal interest rate feedback rules by a commitment to inertial behaviour’ or to predictable responses to shocks. His new Wicksellian model is based on three equations: an expectational IS curve, an inflation adjustment equation and a specification of monetary policy, i.e. an objective function for nominal interest rate. In the credibility literature, the current debate is targeting rules (monetary and inflation targeting of Svensson) versus instrument rules (McCallum, Woodford, Taylor). His proposal of interest rate inertia (optimal rule) uses the Wicksell natural rate of interest. So he gives a Wicksellian theoretical basis for the rule.
To summarise, a central bank is judged ‘credible’ when economic agents think that the central bank will continue to follow the same rule, making its reaction function stable, despite the necessary adjustment to temporary economic conditions (Le Heron, 2003, p.12). As summarised by Blinder (2000, p.1422): ‘A central bank is credible if people believe it will do what it says’. The credibility apparatus can be reduced to RCC: Rule-Commitment-Credibility.
The natural model of the economy. The fundamental assumption of the monetarist and the NCE framework is the existence of a single long-term natural equilibrium (with exogenous and neutral money). The central bank knows this ‘true’ model of the economy and its credibility comes from the respect for this.
Rule and commitment. The rule is founded on the belief in a natural equilibrium model. A fixed rule can be interpreted as a commitment to respect the ‘true’ model of the economy. As agents are representative and hold rational expectations, everyone knows the same model. They can observe if the central bank is respecting it. That is why there must be an automatic and unconditional commitment to the rule. Monetary policy is, therefore, a technical and economic problem.
There is a vertical and hierarchical relationship between the central bank and economic agents, because there is no interaction. They both refer to the same model, which is common knowledge. There is no communication: the common knowledge is sufficient for expectations’ co-ordination and anchorage. It builds expectations; it is a benchmark for policymakers and agents. Full independence is the enforcement of the natural price stability-oriented and long-term oriented strategy, and its related culture of stability and neutrality of money.
Natural model of the economy
Real model with a stable long term equilibrium, dichotomy, neutrality of money
Credibility Commitment to the rule
Figure 1: The credibility strategy in the NCE
1.2. Credibility framework: critical appraisals
(1) New historical period. According to Goodhart (1991, pp.275-277), the macroeconomic conditions that generated a consensus around Monetarism disappeared after 1982. The demand for money was increasing and unstable. The velocity of money became unpredictable. The main target of money supply could no longer be calculated. As a Bank of Canada manager remarked in 1982: ‘We did not abandon M1, it was M1 that abandoned us’. This was the end of the perfect automatic transmission mechanism, via M3. New channels of transmission appearing in the literature and made monetary policy more complex: a bank lending channel and a balance sheet channel were proposed. Furthermore, reaction lags of 18-24 months became a commonplace in monetary policy discussion. The implications of delays, according to central banking theory, are that monetary policy has a long-term horizon and forecasting is fundamental. Exogenous money (vertical curve) and monetary targeting should therefore be abandoned.
In the 1980s, credibility and independence first had to be built with respect to the financial market. Inflation remained the leading economic objective. Once established, a low and steady inflation would favour creditors (or ‘rentiers’). The real interest rates would be higher and inflation expectations would decrease. But the long-term equilibrium was not being established. As growth was slowing and inflation falling, the imperative for employment grew. The new topic became whether the central banks should respond to shocks, i.e. adopt ‘stabilisation’ policy. With the credibility strategy, the danger was a deflationary bias.
While the automatic monetary targeting rule disappeared, the underlying theoretical scheme of credibility faced criticism obliging it to evolve. Thus it moved from automatic rule to strategy.
(2) Time inconsistency and the inflation bias. According to McCallum, even without any absolute commitment, there is no reason to think that the central bank will act in a discretionary way that produces an inflation bias. It usually decides not to play the fooling game. McCallum (1995) believes that there is no credibility problem and there is no necessary trade-off between flexibility and commitment. Blinder (1997, p.13) goes further and claims that time inconsistency does not exist: ‘Let me begin with a non confession: during my brief career as a central banker, I never experienced this temptation. Nor do I believe my colleagues did. I firmly believe that this theoretical problem is a non problem in the real world’ii. He shows that the credibility scheme is considered outdated in order to to build credibility in practice by both central bankers and academics.
(3) No natural model. Discrepancies between ‘natural’ assumptions of the economy and reality are easily identified. The case of the NRE or NAIRU doesn’t appear to be a robust hypothesis. These strong theoretical weaknessesiii necessitate the modification of monetary policy rule design. To follow a fixed rule could be problematic. A fixed rule, such as the Taylor rule, cannot be an ex-ante rule, but only an ex-post rule corresponding to the central bank convention. This rule only works under standard economic conditions. A monetary policy has to work especially when the economic situation is out of the ordinary. So this kind of rule is useless to implement monetary policy.
(4) An other pillar of the credibility strategy, rational expectations, was also confronted with theoretical criticisms. Even for the orthodox economists, it was established that RE was leading to multiple points of equilibriumiv. Co-ordinating this co-ordination failure requires something more than market forces alone. Stabilisation policy is needed since there is no automatic long-term equilibrium.
(5) From rule to institutional design. The credibility literature always alluded to an institutional theme, but in every case eluded it. In the credibility framework, a monetary institution is an incentive structure that tries to fill the incompleteness of the economy. As institutions are social structures, this raises the question of legitimacy. However incentive contracts poorly mimic institutional design. Logically, time inconsistency would lead to full independence. Nevertheless, the credibility framework does not defend a free mandate for the central bank. Independence is usually limited, since government is the principal of the delegation. This is one of the contradictions of the credibility approach: defending the idea that the inflation bias comes from the political vulnerability of the central bank, while not recommending total insulation from political pressure and short-sightedness. Goodhart and Vinals (1994) insist on the political difficulty for a central bank to pursue preferences significantly different from the society (price stability rather than growth).
To sum up, the credibility literature appears to be an elegant way to solve the problem in the purely orthodox theory, but it is not really relevant to reality. For in the day to day practice of central bankers, the time-inconsistency problem is a marginal issue. Credibility faces growing criticism in the literature. It appeared that the credibility literature had failed. By the end of the 1990s, commitment, delegation, incentive contracts and probably most of the credibility literature, were no longer the basis of a monetary policy strategy. A New Consensus on monetary policy appeared in the 1990s.