Inflation Targeting: The Delegation and Co-Ordination of Monetary Policy

We begin reviewing the standard Barro-Gordon model of inflation bias and considering the delegation of monetary policy to a more conservative central banker as suggested by Rogoff. We emphasise how inflation bias can be reduced, seemingly at the expense of increased stabilisation cost. We note the tendency to justify delegating monetary policy to an independent central bank based on these highly stylised analytical models of inflation. We emphasise the need to investigate empirically the implications of such a shift in policy regime without excluding important real-world complexities. An obvious omission from the Rogoff model is the possibility of more than one instrument of macroeconomic policy and their coordination. So we recover the model of Nordhaus, who considers the possibility of a non-cooperative game between the fiscal authority and an independent central bank. We note that the predictions from this model seem relevant to the recent experience of the UK. A substantial part of the debate on means available for authorities to reduce inflation concerns the importance expectations and credibility play in determining the output cost of any inflation reduction. With rational expectations, the optimal policy moves immediately to enforce a particular inflation target, an assumption bearing little relevance to reality. So we consider the case where agents "learn" about changes in monetary policy over time. Implementing this learning model is a major break with the empirical literature on the political economy of inflation control that has been dominated by the assumption of rational expectations. Recent theoretical work emphasises the importance of agents' "learning" over time to converge on the rational expectation, but with a lag. We favour this approach. Our empirical work focuses on two issues. First, we consider the implications of delegation monetary policy to a more "conservative" unitary authority with control over both fiscal and monetary policy. Our results support the conclusions of the theoretical models: delegating monetary policy to an institution with a longer time horizon is less likely to lead to an attempt to exploit the inflation-output trade off and so reduces the inflation bias in discretionary policy. Second, we consider the operation of stabilisation policy in a stochastic environment. Using open loop optimal control, we are able to model monetary and fiscal policy in terms of a 'game' played between the central bank and a fiscal authority, where the two 'players' may have very different objectives. The central bank is concerned with inflation while the fiscal authority is concerned with growth and the public sector deficit. We are thus able to consider whether the gain from the reduced inflation bias is outweighed by the inability to coordinate monetary and fiscal policy in response to shocks.