RBA Annual Conference – 1992 Introduction Adrian Blundell-Wignall
Inflation has been a major issue in OECD countries in the past couple of decades. Inflation rose sharply during the 1970s. This was followed by disinflation during the 1980sandearly 1990s. The costs of this disinflation differed significantly between countries, and some succeeded in getting their inflation rates down more than others. Despite considerable intellectual effort, however, there has been surprisingly little consensus on the causes of short-run inflation, the costs of permitting it to continue, the costs of reducing it, and how monetary policy should be conducted to control it. Australia has been no exception in this regard. The papers in this volume attempt to address some of these questions. This introduction aims to set out some of the important and unresolved issues in order to provide a context for those papers and the discussion to which they gave rise.
1. The Causes of Inflation
With regard to its causes, there is little dispute that, in the longer run, inflation is essentially a monetary phenomenon or, more precisely, is determined by the overall stance of monetary policy. In the short run, however, inflation is generally thought to be related to two things:
- the degree of slack in the real economy – the so-called Phillips curve trade-off; and
- inflation expectations, which influence price setting and wage negotiations.
The Phillips curve suggests (for given inflation expectations) that an easing of monetary policy leads temporarily to faster economic growth and tighter conditions in labour markets, but with the cost of permitting inflation to increase. Tighter monetary policy, on the other hand, reduces demand relative to supply and damps inflation pressure, but at the cost of higher unemployment. Inflation expectations influence the position of the short-run Phillips curve. Thus, for example, if tighter policy succeeds in reducing inflation expectations quickly, the overall trade-off with economic activity during a period of disinflation could, in principle, be improved.
This framework is widely accepted within the economics profession, and is certainly reflected in most of the papers in this volume. However, it is a very broad framework within which there are diverse opinions about the emphasis that should be given to factors influencing the Phillips curve trade-off and the expectations formation process. These alternative views have quite different implications for policy.
One area where important differences arise concerns explanations of the universal empirical observation that the current inflation rate appears to be strongly correlated with inflation rates in the recent past – often referred to by economists as inflation ‘persistence’. While opinions differ on how inflation expectations are formed, it is reasonably clear that they are an important influence on this persistence. For example, inflation expectations may be very forward-looking. That is, they are responsive to current and expected future monetary policies. However, inflation may not adjust quickly to a tightening of policy because of staggered wage and price setting, which is fully reflected in expectations. Slow downward adjustment may also result from forward-looking expectations if there is a lack of credibility concerning the low-inflation resolve of the monetary authorities. Alternatively, people may simply be much less sophisticated, basing their inflation expectations on extrapolations of past trends in prices. Here, inflation persistence results from backward-looking expectations.
If persistence is related to credibility, then the way monetary policy is conducted might influence the trade-off between inflation and unemployment. For example, there might be a ‘credibility bonus’ from ‘tying one's hands’ with a fixed exchange rate. Such a bonus would be less important if expectations were backward-looking. These issues are important, therefore, not only in explanations of inflation, but also for determining the costs of disinflation (see below) and for discussing monetary policy itself.
Another complication is that, over the short run, monetary policy is by no means the only influence on inflation. The price level adjusts as part of the normal equilibrating process in response to a variety of ‘real’ shocks. Real wage bargaining, terms of trade shifts, oil price shocks, changes in the stance of fiscal policy and the structure of taxation, investment booms unconnected with monetary policy (e.g. mineral discoveries, ‘catch-up’ investment in eastern Germany), and so forth, will all affect the price level to a greater or lesser extent. If these effects are misinterpreted as monetary policy-induced inflation, they may become built into expectations, generating wage-price cycles and requiring rising unemployment to return inflation to its longer-run trend.
Economists differ on the issue of whether expectations are forward- or backward-looking, and on the relative importance of real versus monetary influences on inflation in the short run. Most agree that there is no long-run trade-off between activity and inflation, in the sense that easier monetary policy ultimately leads to faster growth of nominal demand and inflation, and not faster real growth and lower unemployment. Equilibrium growth and unemployment are determined by supply-side endowments and the structural characteristics of the labour market.[1] There are a range of alternative views on factors that determine these equilibrium rates. However, some economists consider that the long-run Phillips curve may not be vertical if equilibrium growth and employment are dependent on the past history of inflation. The long-run Phillips curve may be positively sloped if there are important costs to steady inflation.
2. The Costs of Inflation
If inflation is costly, and gradually undermines the productive potential of the economy, economic efficiency will be impaired and unemployment could be pushed higher in the presence of labour market rigidities in the longer run. Some of the costs of inflation are as simple as requiring firms and households to carry out activities which would not otherwise be required. For example, changing prices on shelves and invoices more frequently. Aside from these, however, there are three very broad areas where inflation may be much more costly.
First, inflation makes it difficult to arrive at mutually advantageous contractual arrangements. Here the distinction between anticipated inflation, which can be taken into account, and unanticipated inflation, is often quite important. High inflation is probably more variable than low inflation, and thus creates greater uncertainty about the real value of implicit or explicit contracts in the future. Theoretical examples abound. There is an implicit contract between the monetary authority charged with protecting the stability of the currency and the holders of that currency. The real value of the currency is undermined by a bout of unanticipated inflation. Inflation essentially taxes the holders of currency. Within labour markets, excessive real wage claims may result in an uncertain high-inflation environment, because individual or collective negotiators try to compensate for the possibility of unanticipated inflation. Within financial markets, there are real costs associated with inflation uncertainty premia in interest rates. Within the corporate sector, depreciation is allowable for tax purposes only at nominal historical costs. Inflation therefore undermines corporate balance sheets in favour of the government's balance sheet.
A second major cost of inflation is that it distorts relative price signals. Inflation never involves the equi-proportional increase of prices in all sectors of the economy. This makes it difficult to distinguish relative price shifts justified by fundamentals from those due to inflation, and resource misallocation may result. Inflation differences between countries also interact with national tax systems to distort real interest rate differentials, giving rise to exchange rate pressures.
Finally, inflation creates an environment in which speculation is often more rewarding than production. For example, since asset prices adjust more quickly than goods prices, inflation is often associated with excessive investment in assets such as real estate, where quick capital gains seem more profitable than investing in real capital.
While it is relatively easy to list the costs of inflation, it is much more difficult to quantify them. While it is generally accepted that low inflation is a ‘good’ thing, there has been surprisingly little agreement amongst economists concerning the empirical magnitude of its costs. Yet this is a key issue. It can be very costly in terms of unemployment to reduce inflation. If the costs of inflation are low, it may not be worth moving the economy in the direction of price stability. If the costs are relatively high, then low inflation would be highly desirable, even if there are short-run costs of disinflation.
3. The Costs of Disinflation
Many of the costs of disinflation arise from the problem of persistence referred to earlier. The less responsive is current expected inflation to the announcement of tighter monetary policy, the greater is the rise in unemployment necessary to achieve a given reduction in inflation. In this sense, the ‘credibility’ of monetary policy is potentially an important factor. In principle, the more people believe the authorities are serious about their intention to reduce inflation, the more quickly inflation should adjust to any given degree of slack in the economy. However, problems of ‘time consistency’ may arise. The announcement of a lower objective for inflation will not be credible, if people believe the authorities will be tempted to ease monetary policy in the future, in response to the short-run rise in unemployment. This belief will be reflected in inflation expectations, and will increase the cost of reducing inflation.
Another important determinant of the costs of disinflation is the degree of wage rigidity in the labour market. If nominal and real wages are flexible, tighter monetary policy will see wage inflation slow quickly. Inflation itself will be more responsive to the degree of slack in the economy, and the tendency for real wages to decline will cushion the impact of declining activity on the rate of unemployment. Conversely, less flexible wages increase the costs of disinflation.
Under some circumstances, it is also possible that the costs of disinflation will not be transitory, but permanent. The current natural rate of unemployment may depend (positively) on past unemployment. For example, persons who leave employment for too long a period of time may lose their skills and become less employable as a result. Such people essentially leave the labour force, and exert little downward pressure on wages. Successively higher levels of unemployment therefore become necessary to achieve a given reduction in wage inflation. This phenomenon – often referred to as ‘hysteresis’ in labour markets – greatly complicates the interpretation of the Phillips curve. Inflation comes to depend not only on the level of unemployment, but also on the change in unemployment. On this view, the position of the long-run Phillips curve shifts in response to the current level of unemployment. Thus, using tighter monetary policy to reduce inflation tends to increase the natural rate of unemployment. Hysteresis, if it is of any empirical relevance, would partly counter-balance the resource allocation costs of higher inflation, raising questions about whether it is in fact sensible to disinflate.
The way in which the other arms of economic policy support the thrust of monetary policy may also impact on the cost of disinflation. Incomes policy, for example, may be a useful ‘circuit breaker’, helping to reduce wage inflation more quickly than relying on demand pressure alone. A fiscal tightening, in line with firmer monetary policy, may help to reduce demand quickly, minimising the extent of upward pressure on interest rates. At the same time, tight fiscal policy could help by reinforcing the notion that the authorities would not be tempted to reinflate later on, to reduce the real burden of the public debt.
Finally, there may be important non-linearities in the cost of disinflation. Thus, it is relatively costless for countries with very high inflation to reduce the current rate by a few percentage points. But countries with moderate inflation rates seem to find it much more costly to squeeze additional inflation out of the system.
A crude measure of the cost of disinflation is the so-called sacrifice ratio. The period from when monetary policy is tightened until inflation reaches its new trough is selected, and the decline in activity or rise in unemployment (compared to some notion of their underlying rates) is expressed as a ratio of the induced fall in inflation. What is confusing about the behaviour of such sacrifice ratios, however, is that they appear to differ significantly between countries, and over time within countries, in ways that are often difficult to explain. There is a clear need to improve our understanding of why experiences differ so markedly.
4. Monetary Policy to Achieve Low Inflation
The difficulty of measuring the costs of inflation, and of balancing these against the costs of disinflation, goes some way towards explaining why economists do not always appear to agree on a best method for conducting monetary policy.
Some economists have argued that policy makers should ignore the short-run trade-off between activity and inflation, focusing instead on the need to provide a nominal anchor for the economy via some rule that ties the authorities’ hands. This view rests on the notion that influences on inflation are too numerous, and lags too long and variable, for monetary policy to be used effectively for counter-cyclical stabilisation. If the transmission of monetary policy is mainly via the money supply, for example, then inflation should be tied down with a monetary target. If the country is ‘small’ and has close trading links with a larger low-inflation country, then fixing its exchange rate to that country would also serve to tie down the inflation rate in the longer run. International linkages would ensure that the smaller country's inflation rate would converge to that of the larger one, as in the Exchange Rate Mechanism (ERM) of the European Monetary System (EMS).
But these rules have their own sets of difficulties. With regard to monetary targets, it is difficult to define some measure of ‘liquidity’ that is closely tied to nominal demand and inflation. Demand functions for traditional definitions of the money supply (base money, M1, M2, M3) have been notoriously unstable, not least as a consequence of the liberalisation of financial markets in many OECD countries (including Australia). Charles Goodhart has gone so far as to suggest that even if a good relationship between money and prices could be found, any attempt to target the aggregate would immediately undermine this relationship. Financial market innovations would produce close substitutes for the controlled aggregate. This phenomenon is often referred to as ‘Goodhart's Law’.
Exchange rate targets are effective if the anchor country can be relied upon to solve the problem of low inflation, and if the only source of pressure on prices in the domestic economy is inflationary monetary policy itself. But tying one's hands with an exchange rate target is highly inefficient, if some of the real disturbances which drive prices (referred to earlier) begin to affect countries with fixed exchange rates asymmetrically. The exchange rate is a key relative price that acts as a shock absorber in response to real shocks. Fixing the exchange rate in these circumstances can result in quite inferior inflation/unemployment trade-offs.
An alternative to these kinds of rules is the pre-emptive adjustment of monetary policy to achieve a broad objective for the inflation rate itself or, some would argue, for nominal income. Given the problem of inflation persistence, this approach requires the authorities to forecast likely inflation pressures in advance, and to set policy to ensure that future outcomes remain within tolerable limits. Since bad forecasts risk monetary policy settings which themselves generate cycles in the economy, it is important to be aware of which inflation indicators are currently giving the most accurate picture of the momentum of price increases in the economy. Given the range of potential influences on inflation, and structural changes which may undermine its relationship with any one variable, this approach requires some discretion in the choice of indicators most pertinent to forecasting inflation in the current cycle. There is, of course, much less room for discretion about the ultimate objective of achieving a low-inflation outcome, if the approach is to remain credible in the eyes of the private sector.
One indicator that has proven particularly difficult to interpret in a number of OECD countries in recent years is asset price inflation. Some have argued that asset price inflation is a legitimate objective of monetary policy that might, for example, be included as a component of a more general price index. Alternatively, since asset prices are a key element of the monetary transmission mechanism, they might be thought of as useful leading indicators of goods price inflation, but should be excluded from the definition of inflation. Some perspective on these issues is important for deciding how monetary policy should respond in circumstances of asset price inflation or deflation.
5. The Papers
This introduction has sought to set out some of the key issues concerning the causes of inflation, its costs, the costs of disinflation and approaches to the conduct of monetary policy that might achieve a low-inflation outcome in a reasonably efficient manner. The issues are far from settled at the theoretical level, and their practical relevance differs from country to country. To improve our understanding of the collective wisdom of the economics profession on the subject of inflation and monetary policy generally, and on how we should think about the issues here in Australia, the Bank commissioned six papers covering the issues outlined above.
John Taylor's paper covers the causes of inflation, the trade-off between inflation and economic activity and the relative efficiency of alternative approaches to the conduct of monetary policy. The paper outlines recent thinking on all three issues, and brings evidence to bear on them derived from the experience of the three largest OECD countries.
Doug McTaggart sets out the main conceptual issues on the costs of inflation, and presents empirical evidence for Australia.
Having reviewed the causes of inflation and the costs of persistent inflation, the problem of disinflation is then addressed. Pallé Andersen of the Bank for International Settlements was asked to write a paper on the costs of disinflation in a wide range of OECD countries. By comparing and contrasting widely divergent experiences, and the factors which contributed to them, he provides some general insights into key determinants of the costs of reducing inflation.
To focus more explicitly on Australia's case, Glenn Stevens provides a comprehensive survey of the causes of inflation and the costs of disinflation in post-War cycles in Australian inflation.
Following the papers on inflation and disinflation, Adrian Blundell-Wignall, Philip Lowe and Alison Tarditi, explore how leading indicators should be used to set monetary policy pre-emptively, in order to achieve efficiently a low-inflation outcome for Australia. Taking note of the liberalised financial markets and large terms of trade shocks in Australia, they examine which leading indicators are more useful in different circumstances, including how asset price inflation should be thought about in this context.
Finally, Charles Goodhart provides a wide-ranging survey of key issues concerning the conduct of monetary policy. His paper outlines recent thinking on the rules versus discretion debate, including monetary targets, exchange rate objectives and direct inflation goals. He also provides views on the interrelationships, if any, between monetary policy and prudential supervision, including how policy should respond to banking system crises.
Max Corden, acting as moderator for the conference, provides his own perspective on the issues. The rich general discussion stimulated by the papers and discussants is also summarised to give the reader a broad perspective on the subjects of inflation, disinflation and monetary policy.
Footnote
In some papers included in the volume, the equilibrium unemployment rate is referred to as the Non-Accelerating Inflation Rate of Unemployment (NAIRU). [1]