Developments in Applied Economic Theory and Econometrics

A one-day workshop on The Costs of Inflation was held on 23 January 1984, under the chairmanship of Oliver Hart. Two survey papers were presented, one by Grayham Mizon and Stephen Thomas on the empirical evidence, and the other by John Moore on the state of the theory. There were also short contributions by Douglas Gale, Linda Hesselman, John Kay and Geoffrey Wood, and a general discussion.

Conventional economic theory tells us that, since consumer and firm decisions depend on relative rather than absolute prices, a doubling of the money supply and all absolute prices should leave the real side of an economy unchanged. It is therefore by no means clear why inflation, at least if it is fully anticipated, should be costly. There are a number of reasons, however, for believing that this view is oversimplified. First, conventional theory tends to downplay the transactions role of money. Yet one cost of inflation may result from the rise of the (opportunity) cost of holding money for transaction purposes in inflationary times, which leads to people making an excessive number of trips to the bank (the 'shoe leather' effect), and to a move from financial to real capital (the Tobin effect). Secondly, conventional theory tends to ignore taxes - but if the tax system is not indexed, it is clear that inflation will have real effects as, for example, workers move into higher income tax brackets.

Such views about the costs of inflation are well known. Moore's survey was concerned with more recent contributions, some of which try to link the level of inflation, the variance of inflation, and the variability of relative prices. Moore began with the price misperceptions model of Barro, Lucas and others, which is based on the idea that agents cannot distinguish between relative demand (real) shocks and absolute demand (nominal) shocks. It suggests that an increase in the variability of inflation due to an increase in the variability of the (unobserved) money supply can lead to inefficiencies in output and employment decisions. The theory does not, however, suggest that an increase in the average level of inflation should have welfare costs or that there should be a relationship between the average level of inflation and the variability of relative prices. Interestingly, it also does not exclude the possibility that an increase in the variability of inflation might decrease the variability of relative prices.

A second line of work assumes that the act of changing nominal prices is itself costly, due, for example, to a firm having to update its sales catalogue. A (monopolistic) firm will then delay making a change until its (relative) price has dropped to some level s, say, which lies below that which would maximise profit in the absence of adjustment costs. Moreover, the new (relative) price S will be above the profit maximising level. Sheshinski and Weiss show that, if the inflation rate rises, s and S fall and rise, respectively, and so the variance (suitably measured) of relative prices in the economy increases. They show also that an increase in the variance of the inflation rate causes a rise in the variance of relative prices. The welfare effects of inflation are not obvious, however, since even without inflation there is a distortion due to monopoly.

Moore also surveyed a number of other theories. These included Deaton's argument that unanticipated inflation leads to involuntary saving, since consumers mistakenly think that relative prices of the goods they are currently buying are high (whereas in fact all prices have risen); and a model in which consumers remember nominal prices rather than real prices, so that during times of inflation consumer information about the distribution of prices is generally poorer (since it becomes out- of-date more quickly) - as a consequence, 'high price' firms charge higher prices, and the difference between prices charged by high price and low price firms becomes greater.

Finally, Moore considered the problems inflation can cause for firms (and also consumers). If loan sizes remain unchanged, the liquidity of firms may be severely reduced as nominal interest rates rise ('front end loading'). This can cause firms to cut back on both output and employment. This effect seems potentially one of the most important considered, but the economic and institutional puzzle is why loan sizes do not increase under inflation to keep real indebtedness constant. Although some work by Gale may provide a clue, this question remains to be answered.

The main conclusions of the theoretical work seem to be that (a) in many cases welfare costs are associated with a variable rate of inflation rather than a constant (possibly high) rate (this is not true of the Sheshinski-Weiss model, however); (b) there is no simple relationship between the welfare cost of inflation and the variability of relative prices; (c) in many cases inflation and deflation are symmetric, so that for welfare cost of inflation, one can substitute welfare cost of deflation.

In surveying the empirical literature, Mizon and Thomas noted that most of it deals with the relationship between a measure of relative price variability and the level of inflation, and that the majority of evidence is for the USA. The measure of relative price variability usually used, the variance of relative inflation rates, is not appropriate for many of the theories, which suggest a link between aggregate inflation and the variability of relative prices. Furthermore, the framework for the empirical analysis of the inflation-relative price variability relationship, the misperceptions model, suggests that relative price variability will depend on unanticipated inflation (or its square) rather than the rate of inflation. Hence the link between the theories and the empirical models is sometimes tenuous (Pagan, Hall and Trivedi).