Liam P. Ebrill
The widespread emergence of inflation during the 1970s led to much interest in indexation mechanisms, particularly in those countries with relatively high inflation rates. Such mechanisms, while differing in form and coverage across countries, essentially involve the automatic linking or indexing of the nominal values of certain economic variables to movements in price indices.
While indexation mechanisms can, in principle, be applied to all payments or values denominated in terms of money, such arrangements have, in practice, proved more popular in some contexts than others. Most common is the use of cost-of-living adjustment clauses in labor contracts. In financial markets, indexed bonds have been issued in some countries: such bonds provide a link between nominal returns and changes in prices. In some countries, personal income tax brackets have been automatically adjusted for inflation, so as to avoid undue rises in tax liabilities due solely to inflation. Other elements of tax systems, such as the treatment of depreciation allowances, have also in some cases been indexed.
The underlying motivations for indexation arrangements are clear. Such arrangements reflect a desire to insulate the real variables of economies from the effects of inflation, and to avoid some of the costs of frequently renegotiating contracts when prices are changing.
It is possible to conceive of situations in which indexation mechanisms can be beneficial. Consider, for example, the case of tax indexation schemes which have been implemented so as to ensure that the real impact of tax systems is unaltered by changes in price levels when inflation rates are very high. However, more generally, the experience has been that indexation mechanisms can have adverse real and monetary consequences. On the real side, such mechanisms have frequently been too rigidly applied, obstructing necessary adjustments in relative prices. As practiced, indexation schemes have all too frequently been based on past changes in prices leading to inappropriate changes in relative prices. Further, when inflation rates are changing, and such mechanisms have only been applied to particular classes of payments, significant changes in relative rewards have sometimes occurred. On the monetary side, the widespread adoption of indexation has sometimes been felt to introduce significant inertia into the inflation process and to lead to difficulties in eventually bringing inflation under control.
In the remainder of this article, these themes are covered and elaborated upon in the context of three forms of indexation: the indexation of taxes, wages, and bonds, particularly government-issued bonds. Given the limited space available, the treatment of topics is brief and selective. For example, the topic of exchange rate indexation is not explicitly discussed.
Indexation of taxes provides an example of the purpose of most indexation regimes, namely, to insulate payments against unintended variation caused by inflation. Many tax systems were set up at a time when rates of inflation were low. When inflation subsequently increased, a number of unintended allocative effects emerged in some countries.
The nature of these effects can vary greatly, depending on the provisions of individual countries’ tax systems. However, a couple of general observations can be made. First, many personal income taxes are progressive in terms of nominal income, with the result that inflation tends to drive taxpayers into higher tax brackets. While such an effect enhances the elasticity of the tax system, it tends to be arbitrary and capricious and, clearly, increasing marginal tax rates can have serious implications for work effort. This effect can, in principle, be avoided by indexing the personal income tax structure.
More serious is the fact that, in the absence of indexation provisions, many tax systems do not treat the incomes from labor and capital symmetrically. Even a proportional tax on capital income tends to discriminate against capital, if the tax is based on nominal as opposed to real capital income; such a system fails to recognize that some proportion of the yield on a capital asset may be compensation for inflation. As a result, increases in the expected rate of inflation may adversely affect the real returns earned on typical assets, often resulting in negative real rates of return. This problem is compounded if other assets exist whose nominal returns are untaxed. In some countries, an example of such an asset is owner-occupied housing. In those countries, not only may the implicit return to owner-occupied housing be untaxed but mortgage interest may also be deductible. If mortgage rates rise in line with inflation, increases in the expected rate of inflation will tend to subsidize additional investments in housing. As a result, changes in inflation may encourage capital owners to move capital in response to inflation-induced changes in relative asset yields rather than in response to movements in the underlying productivity of competing investment opportunities.
Indexing, in principle, again offers a solution to this type of problem. For consistency, it should occur on both sides of the balance sheet. (For example, in many corporate tax systems, interest on debt is deductible. As a result, any move to insulate those tax systems from inflation by indexing depreciation allowances and so on should also recognize that deductible interest payments may be influenced by inflation levels.) In fact, the solutions implemented in practice tend to be partial, with only some elements, such as depreciation allowances and inventories, being indexed.
The above discussion glosses over a number of problems. There are many practical difficulties associated with the implementation of tax indexation. For example, what is the appropriate price index to use? Further, the introduction of tax indexation schemes may have a negative impact on a government’s budget, aggravating an imbalance that may have had a role in the inflationary process in the first place. (This effect ignores any influence inflation may be having on government expenditures.) The result is that any decision to implement a tax indexation scheme must find a balance between these considerations, on the one hand, and the fact that tax indexation may ameliorate some of the arbitrary effects of inflation on rewards, on the other. Tax indexation tends to be introduced in countries with rates of inflation so high as to preclude the repeated use of ad hoc solutions, such as statutory tax bracket changes or accelerated depreciation allowances based on historical cost. This has been the case with countries such as Argentina, Brazil, Chile, and Israel.
While indexation can, in principle, be applied to incomes from all sources, it has often been the case that the income generated by labor has been the subject of the most extensive indexation arrangements. This has reflected both the significance of the share of labor in national income, and that fact that the protection of the rewards earned by labor has been a matter of high social and political priority in some countries. Wage indexation arrangements have, in some instances, been the outcome of agreements between employers and employees, while, in others, they have come about as a result of social contracts involving the public as well as the private sectors.
The reasons why wage indexation arrangements might tend to emerge during periods of inflation are well known. They include the avoidance of the costs of frequently renegotiating contracts when prices are changing and the reduction in uncertainty about the future purchasing power of income that indexation can afford. The interests of both employers and employees are relevant. Under certain conditions, wage indexation may be a natural response to both sets of interests in light of the uncertainty implied by inflation. Thus, nonindexed contracts that involve fixing nominal wages for lengthy periods of time, perhaps on the basis of expectations about the course of inflation, impose risks for both employers and employees. For employers, there is a risk that the level of inflation will turn out to be lower than that which was expected, implying a lower level of profits. For employees, there is the risk that inflation will turn out to be higher than expected, leading to a reduced purchasing power of income.
To the extent that the risks faced by employers and employees are equal and offsetting, there is a gain to negotiating indexed contracts; such contracts, by tying the evolution of wages to the movements in an agreed-upon price index, imply that both employers and employees can insure themselves against the consequences of unexpected movements in inflation. In practice, however, the gains and losses to employers and employees are unlikely to be equal and offsetting. While employees will be most concerned about the value of their incomes in terms of a diversified basket of goods, employers will be most immediately concerned with the evolution of the prices of the products they produce. The risk of changes in these prices relative to prices generally implies that the risks faced by employers and employees typically do not precisely cancel each other. Under these conditions, optimal wage indexation will, in general, involve some provisions to take account of movements in relative prices; as such, the optimal contract involves clauses which allow for changes in real wages when economic conditions dictate that such changes are appropriate.
In reality, either due to the complexity of negotiating optimal wage indexation contracts or on account of the difficulties in determining the appropriate form such contracts should take, actual wage indexation arrangements have typically only been able to accommodate a limited range of circumstances; in their simplest form, the evolution of wages depends only upon the course of a single aggregate price index. Accordingly, such contracts have all too frequently not allowed for, or have slowed down, necessary adjustments to real wages in response to factors such as shifts in the terms of trade, productivity changes, and so on. Furthermore, such arrangements have often been based on the past behavior of inflation, rather than its expected evolution, over the life of contracts. In periods when inflation rates have changed dramatically, such contracts have often led, therefore, to large changes in real wages, even when economic conditions would not dictate that such changes were appropriate.
These difficulties associated with the practical implementation of wage indexation mechanisms are important. Some have argued that they are sufficiently serious to warrant a re-evaluation of the appropriateness of permitting such mechanisms. Indeed, some countries have taken steps either to limit the use of wage indexation mechanisms or to change their provisions so as to make them more responsive to movements in the underlying real variables of the economy.
A useful approach to the merits of government provision of indexed bonds is to begin by considering the rationale for privately issued indexed debt. Analogous to the case of wage indexation, bond indexing can be seen as an attempt on the part of individuals to pool the risks of unexpected price changes. Indexed bonds will tend to come into being when one person’s gain is another’s loss, thereby offering an incentive for mutual insurance. The fixed yield on a nonindexed bond will reflect, among other elements, some perception concerning future inflation. Should the rate of inflation subsequently be greater than anticipated, the borrower will gain at the expense of the lender. The reverse holds true when the rate of inflation is below the anticipated level. An indexed bond can avoid these kinds of effects, and, provided that the gains and losses to lenders and borrowers offset each other, the yield may not incorporate an inflation risk premium. In the absence of this balancing of risks, indexed bonds would tend, in general, to include some allowance for such risk.
An interesting extension on the risk-pooling or insurance perspective to indexation involves recognizing that wage and bond indexation may be inversely related—if the real wages an individual earns, say, are protected by an indexed contract, that individual will typically be less willing to pay for bond indexation.
Now consider the role of government-issued indexed bonds. Governments issue them to attain some objective. In particular, governments may feel that they are better able to absorb, or have different attitudes toward, the risks associated with price variability. Seen in this light, the issuing of indexed bonds by governments may be justified as a response to the need for individuals to obtain insurance against inflation.
The substitution of indexed for nonindexed debt implies changes, which depend on the subsequent course of inflation, in future government tax payments. This raises the possibility of additional distributional and allocational effects. One way of evaluating these is to consider a benchmark case in which these effects may be negligible. Specifically, consider the case of an individual who is both taxpayer and bondholder. Assume that the switch to indexed debt results in an increase in tax payments and that the individual in question is aware of this effect. If it is further assumed that the prospective increase in the liabilities is precisely balanced by the change in the yield on his holdings of government debt, then a case can be made for arguing that the individual will be indifferent as between the two forms of debt. Should these circumstances not hold, then a wide range of effects is possible. In particular, when taxpayers and bondholders are not the same individuals, indexed debt may become an instrument for income redistribution either within generations or across generations (to the extent that that debt is retired by the tax payments of subsequent generations.)
Of course, the other side of this scenario is that the use of nonindexed debt raises the possibility of an inflation tax. In the nonindexed case, unanticipated increases in future rates of inflation reduce the cost of repaying outstanding debt. Related to all of these budgetary effects is the possibility that the existence of government-issued indexed bonds may alter the conduct of economic policy. For example, some have argued that their existence might encourage governments to pursue more vigorous anti-inflation policies so as to reduce future repayments if for no other reason than to hold down future tax revenue requirements. Finally, note that by permitting the use of dollar-denominated deposits so as to afford hedges against both domestic inflation and exchange rate risks, some Latin American countries have effectively been permitting indexed financial assets.
In an ideal world, the indexation of nominal values can help to reduce significantly the costs of inflation. With real shocks occurring, such indexation will generally allow for necessary changes in relative prices, and, with inflation changing in a known and predictable way, indexation formulae would be based on current and not past price changes.
In practice, however, indexation mechanisms have tended not to have these desirable attributes. In addition, they typically have not been uniformly applied across sectors. The result may be significant changes in relative rewards over time. This may be an additional undesirable feature, though any conclusion in this respect should be made with reservations. Thus, as was pointed out earlier, a case can be made for arguing that the optimal amounts of wage and bond indexation may be inversely related. Taken together, the discussion of the three examples of indexation suggests that attempting to adjust automatically for the effects of inflation, while in principle beneficial, can in practice frequently end up generating a new set of problems precisely because actual indexation contracts are not as complete as the optimal contracts of the frictionless world. There are additional issues associated with indexation. Thus, much of the above has addressed topics that arise in association with particular kinds of indexation. As such, it has not dealt with the macroeconomic implications of the widespread use of indexation, and therefore has not fully considered the issue of whether indexation arrangements that have been introduced in response to inflation may themselves have implications for inflation. In this context, concern has been expressed that economies in which a large proportion of nominal variables are indexed may become inflation prone. Some Latin American countries, which have made widespread use of indexation techniques, have had to consider whether the continuance of such arrangements is ultimately consistent with the control of inflation. The emerging consensus is that the adoption of broadly based indexation schemes may often be associated with a range of adjustment problems. It is more preferable to adopt policies which control inflation than indexation policies that are aimed at allowing one to live with inflation.