"The uncertainty about inflations costs and benefits raises the possibility that the seemingly high average inflation rates in most industriaUzed countries in recent decades are in fact optimal, [f this is correct, there is not m fact any inflationary bias in monetary policy. We will not pursue this possibility.

"See, for example, ToUey (1957) and Friedman (1969).

9.8 The Costs of Inflation

AU of the analysis so far in this chapter assumes that inflation is costly, and that policymakers know what those costs are and how they vary with inflation. In fact, however, inflations costs are not weU understood." There is a wide gap between the popular view of inflation and the costs of inflation that economists can identify. Inflation is intensely disliked. In periods when inflation is moderately high in the United States, for example, it is often cited in opinion polls as the most important problem facing the country. It appears to have an important effect on the outcome of Presidential elections, and it is blamed for a wide array of problems. Yet economists have difficulty in identifying substantial costs of inflation.

Easily Identifiable Costs of Inflation

In many models, steady inflation just adds an equal amount to the growth rate of all prices and wages and to the nominal interest rate on all assets; it therefore has no effects on relative prices, real wages, or real interest rates. It is this fact that makes it hard to identify large costs of inflation.

The only exception to the statement that steady inflation has no real effects in simple models is that, since high-powered moneys nominal return is fixed at zero, inflation necessarily reduces its real return. This gives rise to the most easily identified cost of inflation. The increased gap between the rates of return on money and on other assets causes people to exert effort to reduce their holdings of high-powered money; for example, they make smaller and more frequent conversions of other assets into currency. Since high-powered money is essentially costless for the government to produce, these efforts have no social benefit. Thus they represent a cost of inflation.

These socially unproductive efforts to conserve on money holdings can be eliminated if inflation is chosen so that the nominal interest rate-and hence the opportunity cost of holding money-is zero. Since real interest rates are typically modestly positive, this requires slight deflation."*

It seems unUkely, however, that this is all there is to the costs of inflation. Most obviously, the shoe-leather costs associated with a positive nominal interest rate are surely small for almost all inflation rates observed in practice. Even if the price level is doubling each month, money is losing value only at a rate of a few percent per day; thus even in this case individuals will not incur extreme costs to reduce their money holdings.

Other Costs of Steady Inflation

There are at least three ways that steady, anticipated inflation may have large costs. First, because individual prices are not adjusted continuously, even steady inflation causes variations in relative prices as different firms adjust their prices at different times. As a result, inflation increases the departures of relative prices from the values they would have under fric-tionless price adjustment. Okun (1975) and Carlton (1982) argue informally that this inflation-induced relative-price variability disrupts markets where firms and customers form long-term relationships and prices are not adjusted frequently. For example, it can make it harder for potential customers to decide whether to enter a long-term relationship, or for the parties to a long-term relationship to check the fairness of the price they are trading at

A second readily identifiable cost of inflation is that nominal prices and wages must be changed more often, or indexing schemes must be adopted. Under natural assumptions about the distribution of relative-price shocks, the frequency of price adjustment is minimized with zero inflation. As Chapter 6 describes, however, the costs of price adjustment and indexation are almost certainly smaU.

The last cost of inflation that can be easily identified is that it distorts the tax system (see, for example, Feldstein, 1983). In most countries, income from capital gains and interest, and deductions for interest expenses and depreciation, are computed in nominal terms. As a result, inflation can have large effects on incentives for investment and saving. In the United States, the net effect of inflation through these various channels is to raise the effective tax rate on capital income substantiaUy. In addition, inflation can significantly alter the relative attractiveness of different kinds of investment. For example, since the services from owner-occupied housing are generally not taxed and the income generated by ordinary business capital is, even without inflation the tax system encourages investment in owner-occupied housing relative to business capital. The fact that nominal interest payments are deductible from income causes inflation to exacerbate this distortion.

In contrast to the shoe-leather and menu costs of inflation, the costs of inflation through tax distortions may be large. Thus it is important for policymakers to account for these effects. At the same time, these distortions are probably not the source of the publics intense dislike of inflation. These costs are quite specific and could be overcome through indexation. Yet the dislike of inflation seems much broader.

Thus it appears that we must look further to understand the popular view of inflation. There are several ways that inflation may have large costs that are more subtle than the costs just described. Some of the potential costs occur when inflation is anticipated and steady; others arise only if inflation is more variable and less predictable when it is higher.

Costs of Variable Inflation

Empirically, inflation is more variable and less predictable when it is higher (see, for example, Okun, 1971; Taylor, 1981; and Ball and Cecchetti, 1990). Okun, Ball and Cecchetti, and others argue that the association arises through the effect of inflation on policy. When inflation is low, there is a consensus that it should be kept low, and so inflation is steady and predictable. When inflation is moderate or high, however, there is disagreement about the importance of reducing it; indeed, the costs of slightly more inflation may appear small. As a result, inflation is variable and difficult to predict.

If this argument is correct, the relationship between the mean and the variance of inflation represents a true effect of the mean on the variance. This implies some potentially important additional costs of inflation. First, since many assets are denominated in nominal terms, unanticipated changes in inflation redistribute wealth. Thus greater inflation variability increases uncertainty and lowers welfare. Second, with debts denominated in nominal terms, increased uncertainty about inflation may make firms and individuals reluctant to undertake investment projects, especially long-term

by comparing it with other prices. Formal models suggest that inflation can have complicated effects on market structure, long-term relationships, and efficiency (for example, Benabou, 1988, 1992; Benabou and Gertner, 1993; Diamond, 1993; Tommasi, 1994; and Ball and D. Romer, 1993). This literature has not reached any consensus about the effects of inflation, but it does suggest some ways that inflation may have substantial costs. This literature also suggests that the immense disruptions associated with hyperinflations may just represent extreme versions of the effects of more moderate rates of inflation.

Second, individuals and firms may have trouble accounting for inflation (ModigUani and Cohn, 1979; Hall, 1984). Ten percent annual inflation causes the price level to rise by a factor of 45 in 40 years; even 3% inflation causes it to triple over that period. As a result, inflation can cause households and firms, which typically do their financial planning in nominal terms, to make large errors in saving for their retirement, in assessing the real burdens of mortgages, or in making long-term investments.

Finally, steady inflation may be costly not because of any real effects, but simply because people dislike it. People relate to their economic envi-romnent in terms of dollar values. They may therefore find large changes in dollar prices and wages disturbing even if they have no consequences for their real incomes. In Okuns (1975) analogy, a switch to a policy of reducing the length of the mUe by a fixed amount each year might have few effects on real decisions, but might nonetheless cause considerable unhappiness. Since the ultimate goal of policy is presumably the publics weU-being, such effects of inflation would represent genuine costs.