Robert Lucas on Inflation and Welfare: - A Comment

Robert Lucas's paper is beautifully adapted to the theme of this conference. It takes important policy questions - what is the optimal inflation rate, and what are the costs of deviating from it? - and uses state-of-the-art monetary theory to address them, and to organise relevant empirical evidence. Following Friedman (1969), the optimal rate of inflation is shown to be that which reduces the nominal rate of interest to zero, but if it were achieved we would be in unknown territory. Economic theory, as deployed by Lucas, allows us to extrapolate beyond the boundaries of empirical experience, and warns us when we approach the limits of our established knowledge. Specifically, he shows that when the opportunity cost of holding money is very low, the size of the further gain to be had from pushing it towards zero depends crucially on the form of the demand for money function, and he reminds us that our knowledge here is shaky.

Lucas's approach ensures that, at every stage in the argument, we know exactly where we are, and how we got there. Furthermore, anyone who has lived through a policy experiment which has carried an economy into new territory will appreciate the benefits of trying such things out first on a well specified model economy constructed on the basis of the best available theoretical and empirical knowledge. On reading Lucas's paper, one cannot help but feel that his is the way economics should always be done. Even so, though such work is surely necessary for competent policy analysis, it may not always be sufficient. To use old fashioned vocabulary, the application of economic science to policy questions remains an art, for reasons inherent in the nature of the science itself.

Lucas stresses that the technical skills of its practitioners impose limits upon what economic analysis can achieve at any particular moment, and that it makes progress over time. This implies: first, that today's state-of-the-art formal model is at best the latest, but not the last, word on any topic; and second, that currently available technique will inevitably limit the range of factors which that model takes into account, perhaps to the neglect of some that are important. Let me illustrate the dangers here with a brief, and therefore highly simplified, reference to the not so distant history of macroeconomics.

The early 1930s saw much discussion of the role of endogenous forward-looking expectations in economic life, but that discussion ran well ahead of theorists' capacity to analyse the phenomena formally. In an important sense, the static IS-LM model of the later 1930s was a step forward. It allowed the best of then generally available analytic techniques systematically to be brought to bear on important macroeconomic issues, but one factor permitting this was that model's treatment of expectations as exogenous. In due course, some economists forgot that this simplification underlay what they were doing, and the "Keynesian" economics of the 1960s and 70s was the result. Before we had Robert Lucas's work (eg. 1972), we did not know how to incorporate endogenous forward looking expectations into formal analysis, but in the 1960s and 70s would not economic policy have been better had some attention been paid to what had earlier been informally known about these issues?

Perhaps we are now in a similar situation vis-a-vis the costs of inflation. Lucas emphasises that variations in the opportunity cost of holding money affect the volume of resources devoted to portfolio management in an economy in which markets for goods, services and assets exist and clear with equal ease, regardless of the inflation rate. He is quite right to insist that these are the questions which economic analysis in its present state of development is able to address; but Lucas himself finds it helpful to go beyond putting the services of money into the utility function and to ask just what the nature of those services is. He stresses money's means of exchange role. Money, however, is also a unit of account, and we have extensive empirical evidence that inflation interferes with its capacity to perform this function.

To be sure, the most dramatic and costly effects here - the disruption and, in the limit, the disappearance of longer term capital markets, and of the arbitrage that is needed to maintain a coherent structure of relative prices - occur in conditions of what Daniel Heymann and Axel Leijonhufvud (1995) have termed "high inflation", and Lucas's paper is not directly concerned with such a state of the world. Even so, a substantial literature, surveyed by Peter Howitt (1990, 1997) suggests that such problems also occur in a less dramatic, but still costly, fashion at low inflation rates. Menu costs do exist, and, more important, it seems to be impossible to adapt the accounting procedures upon which the data needed for intelligent business decisions depend, and upon which tax codes are based, to variations in the purchasing power of money. These facts give rise to resource misallocations which probably reduce both the level, and perhaps also the rate of growth, of real income.

To put matters in the language of the Neo-classical theory on which Lucas's analysis rests, a fully anticipated inflation rate is not merely one that is accurately expected by all agents; it is also one to which all contracts and market institutions have been fully adapted. In the current state of knowledge, we cannot be sure that a fully anticipated inflation rate in this sense is anything more than a theoretical construct, and we ought to bear it in mind that ongoing inflation imposes more costs than Lucas's model enables us to discuss. Howitt (1990) has noted that, while neo-classical analysis of the costs of inflation suggests that mild deflation is optimal, arguments about inflation's capacity to disrupt money's unit of account role point to zero as the right number. And now, in the wake of a recent study by Akerlof, Dickens and Perry (1996) of money wage stickiness in the region of zero - surely the result of a particular kind of menu cost - some would argue that mildly positive inflation is desirable. There is, then, a trade-off among costs to be taken account of in fixing an inflation target.

Seen in this light, Lucas's empirical conclusion that the benefits in terms of portfolio management economies of driving the opportunity cost of holding money down from a starting value of 6 per cent per annum are initially large, but shrink, and indeed become problematic, the closer that opportunity cost gets to zero, lend weight to pragmatic arguments against driving inflation into negative territory. I suspect that Lucas would not want to be associated with such an interpretation of his work, and would prefer to confine the discussion to the results yielded by his well organised formal analysis. But perhaps in due course economic theory will advance to the point that inflation's effects on money's capacity to act as a unit of account can be analysed as rigorously as its means of exchange and store of value functions, and he will then change his mind. In the meanwhile, however, policy makers have decisions to make, and in my view those decisions are likely to be better if they are based on all the evidence that economists have amassed, rather than only on the subset that can be incorporated in a formal model.
 

REFERENCES