On Growth and Inflation in Developing Countries

International Monetary Fund. Research Dept.
Published Date:
January 1984
  • ShareShare
Show Summary Details

The association between inflation and economic growth in developing countries has been the subject of interest and debate for several decades. There appears, by now, to be widespread agreement that continuously increasing inflation is bad for economic development and growth. But is some inflation always bad for growth in the developing countries? Is some inflation inevitable in the growth process in those countries? Why does a given level, or change in, inflation seem to have different growth consequences in different countries? Finally, suppose an attempt is made to reduce inflation, but associated social costs dampen the speed with which that can be accomplished, how can any adverse growth consequences of living with inflation be reduced? To throw some light on some of these issues, this paper examines the nature and economic content of the hypothesis that inflation, up to a point, can be beneficial to growth in the developing countries; some of the theoretical criticisms that have been levied against optimistic versions of the hypothesis; and the empirical implications, as well as some cross-country evidence.

Even the hypothesis as it stands raises a number of issues, and it is necessary to be more specific in at least two respects. In the first place, the paper is not concerned with the relation between the long-term average rates of inflation and growth. Rather, it focuses on the relation between changes in the rate of inflation and changes in the rate of growth. In the second place, the paper is concerned with annual changes. One interesting issue ignored by these two restrictions is whether a particular country that has experienced two periods—each, say, of three to ten years in duration—marked by significantly different average inflation rates also realized significantly different growth rates between the two periods, and, if so, whether the difference in growth rates was related to the difference in average inflation rates.

The case for positive inflation to enhance growth in the developing countries has traditionally been made in the context of structuralism, and of inflationary financing of government investment and of private investment. In Sections IIII the theoretical arguments and some of the limitations and crucial implicit assumptions of these approaches are given. Section IV discusses the empirical implications of the approaches and the conclusions of several authors who have looked especially at the cross-country data. The final brief section contains some concluding remarks.

I. Structuralism

According to the structuralist view of inflation, a positive relation between inflation and growth, at least up to a certain rate of inflation, is an inescapable one, barring radical and fundamental changes in the structure of the economy. This relation is not the consequence of economic policy toward inflation but, rather, a constraint on what is achievable through policy measures. There is, in short, an inevitable trade-off between inflation and growth.

The argument of the structuralists is that various rigidities and inelasticities in the economic environment of the developing countries lead to a buildup of inflationary pressures from the real side during the growth process, and that the banking system must validate such pressures if potential economic growth is to be fully realized—indeed, in some cases, if losses in output are to be avoided.1 There are three important aspects to this argument. First, because of factor immobility and downward rigidity of factor prices, upward movements of wages and prices are required to reallocate labor and other resources; this upward movement is particularly necessary to attract labor out of traditional (including subsistence) sectors and into developing (or so-called modern) sectors of the economy. Second, because of supply inelasticities in agriculture, the increase in demand for food resulting from development in other sectors, as well as from population growth, will tend to push up the relative price of food. Again, downward price rigidities in the economy will tend to translate such relative price changes into absolute price increases at given levels of employment and output in the modern developing sectors. Third, because of inelasticity of demand for (traditional) exports of the typical developing country, foreign exchange earnings from such exports cannot rise fast enough either to finance the growing requirements for food imports or to satisfy the demand for imports of intermediate goods by producers of export-oriented and purely domestic goods. Consequently, import substitution is required, and, in light of the low degree of comparative advantage of the country in some of these import substitutes, costs of production will be quite high, thus creating inflationary pressures.

In the structuralist framework the price level in the industrial sector is typically set according to some markup over the nominal wage. The nominal wage is based on labor’s desired real wage and on the aggregate price level. These factors lead to an equilibrium relative price between agricultural and industrial goods that is consistent with labor’s required real wage (see Cardoso (1981)). In such a framework, growth in the industrial sector, which alters the relative price between agriculture and industry and raises the absolute price level, will influence the wage rate in the industrial sector, leading to an upward rise in industrial prices and giving further impetus to the rise in the aggregate price level.

A good part of the structuralist analysis has been concerned with formalizing the relation between the relative price of food and the inflationary process. Now assume two sectors, agriculture (A) and industry (I). Let the aggregate price level (P) be a linear homogeneous function (or geometric average) of the price level in agriculture (PA) and in industry (PI). That is,

0 < α < 1.

Taking percentage changes (indicated by circumflexes), we have

where P* = PA/PI Thus, the rate of inflation can be specified as a function of the rate of change of the relative price of the agricultural good and of the absolute price in the industrial sector.

A fundamental hypothesis of structuralism, especially as it is applied to Latin America, is that the relative price of the agricultural good (A) must increase with growth, at least in a certain type of developing country. This is so for at least two reasons. One is that growth in the agricultural sector is relatively slow; the other is that the price reaction to excess demand is much greater in agriculture than in industry, as regards both the speed of price response to the emergence of excess demand (positive or negative) and the elasticity of price with regard to the change in excess demand.

Now equilibrium in the agricultural sector requires equality of supply and demand. Assuming that autonomous demand grows at rate ψ and supply at rate ϕ, then in equilibrium

where η and ε are the price elasticities of demand and supply, respectively. Thus

Consequently, equation (2) can also be written as

as shown by Canavese (1982).2 In other words, inflation is accompanied and affected by increases in the relative price of the agricultural good.

Alternatively, suppose that kA and kI represent the coefficients of price reaction to excess demand in agriculture and industry, respectively; these coefficients are, in effect, indicators of both the speed and elasticity of response of price to a 1 percent change in excess demand. Then

where Di represents excess demand in sector i. Consequently,

so that equations (2) and (5) can also be written as

It is hypothesized, of course, that kA>kI.

Consider now a situation in which excess aggregate demand does not exist, so that αD^A=(1α)D^I. If prices behave according to equation (6), we then have

so that substituting for P^I in equation (8) yields

Equation (9) is consistent with Wachter’s result (1976)—that the inflation rate depends on the distribution of excess demand between the two sectors, even in the absence of excess aggregate demand, and this result depends crucially on the differential price-reaction coefficients emphasized by Wachter. This can be seen by setting kA = kI in equation (9).

In principle, we can divide the growth in excess demand in agriculture as emanating from two forces: growth in excess supply in industry, from productivity changes or autonomous investment or consumption, matched by excess demand for the agricultural good, D^A0; and growth in excess demand, Dg, from credit creation to finance budgetary deficit or private sector activity. Thus

where x is the income elasticity of demand for the agricultural good. In that case

so that

Equation (5) then becomes

It is clear from equation (12) that, even within the structuralist framework, inflation can be attributable to purely structuralist factors (ψ - ϕ), to excess demand pressures (D^g), and to cost-push factors in industry (P^I).3 To observe that a given inflation rate is proceeding hand in hand with growth is not to establish that the inflation rate is the result of only structural factors in the growth process.

In general, the structuralist view suffers from serious theoretical and empirical shortcomings and overstatements. Although a full discussion of the weaknesses in the structuralist explanation is not a purpose of this study, it is useful to note that some of the rigidities and inelasticities emphasized by the structuralists may have been “induced”—that is, are the consequence of a history of inflation stemming from other sources, inappropriate exchange rate policies, and price and exchange controls. Moreover, many believe that, stripped of its overstatement and inaccuracies, the argument may not, for the typical developing country, “justify” an inflation rate of more than 4 to 10 percent (for example, Harberger (1964) and H.G. Johnson (1966)). It must be realized, though, that such assessments implicitly assume a zero rate of “world inflation,” a process that even such critics agree exists (for example, Harberger (1978)). In that case, the 4 to 10 percent may be said to be over and above whatever is considered to be “the” world inflation rate.

II. Inflation Tax

Despite the lively debate that structuralists have engendered in this area, it is not the issue of whether inflation is inevitable in the growth process that has made the topic of inflation and growth so interesting to many economists. It is, rather, the possibility that inflation can be used as an instrument to finance investment in developing countries.

inflationary financing and growth

An argument of the literature on inflationary financing and growth4 is that government expenditure can have substantial real effects. In particular, a good fraction of the investment expenditure with the highest social marginal returns, taking into account external economies generated, will in many developing countries have to be made by the government (including the public enterprises). To the extent that the normal channels of taxation are “undeveloped,” “inadequate,” or “inelastic” (to choose expressions that have been frequently used), under zero inflation the share of national resources available to the government to expend on socially profitable investment projects may remain far less than optimal. In such cases the well-known “inflation tax,” consequent upon the inflation produced by relatively rapid base (or high-powered) money creation by the monetary authorities, will make additional real resources available for use in the development process, even when the inflation is fully anticipated. In this framework, therefore, the relation between inflation and growth is a consequence of economic policy, and inflation is a direct policy instrument.

For instance, suppose the government issues additional high-powered money (H) in order to make investments GI. Then

where K is real capital. Note that private capital in this model is not at all affected by government investment (that is, crowding out does not occur). Suppose that y = kK, where k is the output-capital ratio and y is real output. Then

Also, using the definition ŷ ≡ (dy/dt)(1/y), the result is obtained that

where Ĥ is the rate of change of high-powered money.

Let Vh be the income velocity of high-powered money, and assume that it is constant. Also note that, in equilibrium and with constant velocity, we have

with M^ = H and where M^ is the rate of change of the money stock. The above implies that, in equilibrium,


From equation (17) the rate of growth in real gross domestic product (GDP) is seen to vary directly with the inflation rate, given Vh and k. But, in general, velocity will tend to increase with the rate of inflation (P^). If it is hypothesized that a log-linear relation exists—such that Vh=aeαP^, where a and α are constants—substituting for Vh(= Py/H) in equation (15) will indicate that

The above results are obtained by assuming that the government invests the proceeds of the real high-powered money it issues, and that real savings (and investment) increase by the amount of the real high-powered money. Thus, the savings-income ratio is increased by the inflation tax.5

In general, of course, inflationary financing of government deficits cannot be said to be due to current or capital expenditures of the government, for what is being financed is simply an excess of government expenditure (current and capital) over government (normal) tax and nontax receipts. In addition, it is not always the case that, from the viewpoint of contribution to growth, some expenditures classified as current are less productive than certain expenditures called capital. A typical case in point would be one in which investments in health and schooling—investments in so-called human capital—are more productive in their contribution to real GDP growth than are many prestige projects (particularly in construction) included in capital expenditure. Furthermore, starting from some level of the government budget deficit (say, relative to GDP) considered to be noninflationary, such deficit can increase, not because of domestic investment or apparently growth-generating “current” expenditure but because of government subsidies promoting certain distributional goals. Or, the deficit can increase because of an inflationary surprise—say, because of an external shock—while the government tries to maintain its expenditure in real terms. The increase in government expenditure then perpetuates the deficit and the inflationary process, since government revenues can fall behind in real terms in many developing countries because of collection lags, the structure of expenditure and revenue, government response to inflation, and other factors (Aghevli and Khan (1978), Goode (1984, pp. 221-24), and Heller (1980)).

Taking all these caveats into consideration, the inflationary finance and growth thesis can be said to imply that the greater is the change in the ratio of government investment or capital expenditure to total government expenditure (GKT), the greater is the marginal efficiency of government investment, and the smaller is the crowding out of private investment ensuing from an increase in government investment relative to GDP, the greater also will be the increase in growth ensuing from a marginal increase in inflation through inflationary financing of government investment (that is, the greater will be y^/P^ or y^/H^).

inflation and real tax revenue

Tanzi (1978) has emphasized that the inflationary finance and growth thesis must incorporate the effect of inflation on the existing tax system.6 Abstracting from discretionary changes in tax rates, he argues that inflation lowers the real value of normal tax receipts and that the net contribution of inflationary finance to government revenue must incorporate this reduction. What is important is the ratio of inflation tax revenue plus normal tax revenue relative to GDP, and an interesting analytical question is the rate of inflation at which this total revenue is maximized. Tanzi (1978) showed the importance of the built-in elasticity of the tax system and the length of the collection lag for the revenue-maximizing rate of inflation; he also showed that such a rate was not necessarily identical with that at which the pure revenue from inflationary finance was maximized relative to GDP.

Of course, discretionary changes may be made in the tax system (Heller (1980)), so that what is ultimately relevant is a total elasticity that incorporates such changes. But it is not always clear that such discretionary changes will have a net effect of increasing the ratio of tax revenue to GDP. Some changes will have such an effect, as when specific tax rates are modified sufficiently and on a timely basis to prevent deterioration in the real values of their yields. But other changes may have just the opposite effect, as when income tax rates are adjusted to reduce the real burden of taxation on certain income groups.

Because government expenditure financed from domestic sources will come from normal tax and nontax revenue, as well as from inflationary and noninflationary sources of deficit financing, it is quite likely that, within any given time period, increases in inflation that are due to inflationary financing result in increases in the ratio of government expenditure to GDP only up to a certain rate of inflation.7 That is,

where GE is government expenditure, Y is GDP, and h0 and h1 are constants. The above relation (19) incorporates the effects of collection lags, built-in elasticity of the tax system, discretionary changes, and inflationary financing. The ratio h0/h1 will indicate the rate of inflation beyond which further increases lead to a drop in the GE/Y ratio.

What had been said earlier about inflationary financing, government investment, and growth would, therefore, require some modification. The hypothesis can be restated as follows: as long as (GE/Y^)/P^>0 and the ratio of government investment to total government expenditure increases with government expenditure, growth will tend to be positively related to the inflation consequent upon inflationary financing of government expenditure. Economic policy can therefore influence the relation between inflation and growth by ensuring that the actual rate of inflation does not go beyond h0/h1 in equation (19); that is, the rate of inflation at which (GE/Y^)/P^ becomes zero.

open economy considerations

It is recognized that conclusions about inflationary finance and growth must be modified in an open economy,8 for at least two important reasons. First, the typical developing country imports a substantial fraction of the intermediate and capital goods required for production and investment. Consequently, a government that is able to obtain significant resources from the inflation tax would still need to obtain foreign exchange to purchase intermediate and capital goods to implement its investment projects. In the absence of external loans and grants to cover all such expenditures in foreign exchange, the proceeds from the inflation tax must be spent partly or wholly in foreign exchange. Second, when the inflation rate is rapid relative to inflation rates of trading partner countries, the prices of nontraded goods are increasing relative to those of traded goods, thereby tending to induce a shift in productive resources toward nontradables and away from tradables. In contrast, domestic households and firms are encouraged to shift demand away from nontraded commodities and toward traded commodities and factors of production.

For our purposes, the implication is that, given a relatively high domestic inflation rate, the need to avoid serious losses in foreign exchange, and the desire to prevent relative price changes between traded and nontraded goods that are not dictated by fundamental changes in factor endowments, in factor productivity, or in tastes, then downward adjustments of the exchange rate would need to accompany the inflationary process. The smaller is the appreciation in the real effective exchange rate accompanying inflation, the greater will be the increase in growth resulting from a marginal increase in inflation.

inflation tax, real interest, and growth

It has been shown by von Furstenberg (1983) that, under conditions which (he argues) normally prevail in the typical developing country, anticipated increases in the rate of growth of high-powered money resulting from inflationary government financing will tend to raise the real rate of interest, and that this increase in the real interest rate will adversely affect real output (income) growth through its effect on the desire to accumulate wealth in the private sector.

In the von Furstenberg framework, the key variables are the real rate of interest, r, equivalent to the real rate of return required by private decision makers on claims to business and consumer capital, and Tobin’s q, defined as the ratio of the market value of ownership claims on physical assets to the replacement costs of those assets. A rise in q indicates a decline in required rates of return on private capital relative to actual rates of return (and vice versa for a decline in q); a decline in q is associated with a rise in the real interest rate. An increase in the rate of growth of high-powered money to finance government expenditure, which also raises the rate of growth of money, is assumed in this model to lower q because of the increased risk of increased instability of future government policies and of rates of return on private physical capital. The decline in q in turn increases r. In short, the following holds (with mM^, the rate of growth of money):

such that dr/dm = (∂q/∂m)(∂r/∂q) > 0, where ∂q/∂m < 0 and ∂r/∂q < 0.

Desired private wealth (or net worth), Wd, is specified as a positive function of income (output)9 and as a negative function of the real rate of interest. That is,

such that ∂Wd/∂Y > 0 and ∂Wd/∂r < 0.

There is also a private savings function stating that private saving varies positively with the excess of desired wealth over actual wealth (W*), where the rate of adjustment of W* to Wd, which is symbolized by s in equation (22), is between zero and unity. Hence,

where 0 < s < 1.

Wealth of the private sector is defined, in equation (23), as the sum of (1) high-powered money, H; (2) net debt of the public sector to the private sector, B; (3) the replacement cost of nationally owned business physical capital (net fixed assets plus inventories), defined as total business capital, KB, less the fraction, f, owned by foreigners; (4) the replacement cost of physical capital owned by households, which is assumed here as bearing some fixed relation ρ to business capital; and (5) the market value of assets such as land that are nondepreciating. Business capital, KB, is adjusted by q to take account of any difference between the market value of claims on physical assets and the replacement cost of those assets. In short,

The actual and desired business physical capital (KB and KBd, respectively) are related to Wd by the coefficient sk (=skd), which measures the ratio KBd/Wd desired by private holders of wealth. An increase in the rate of money creation is assumed to engender an increase in skd. This can be summarized by the following equation, where the parameter skd0 takes account of other (exogenous) factors affecting skd:

such that 0 < sk < 1 and ∂skd/∂m > 0.

The system is completed below by equation (25), which defines the savings-investment equilibrium, and by equation (26), which relates output (income) to the stock of physical capital (including, for completeness, the physical capital of the government, KG). S denotes private savings in equation (25) whereas, as before, the output-capital ratio is expressed by k in equation (26). Equation (25) in effect states that savings is the sum of the change in business and household capital stock; that is, (1 - f) (1 + ρ)ΔKB plus the change in high-powered money (ΔH) plus the change in public debt holdings of the private sector (ΔB). That is,

Equation (26), the output equation, is stated as

Assume now that, with regard to equations (20) through (26), a situation of equilibrium exists such that money is growing at a constant rate and the rate of inflation is zero. In that case, the change in the rate of growth of money is zero, and equation (20) is not relevant. Changes in nominal and real values are also the same. Suppose now that the government announces and implements a policy of financing part of its expenditure through inflation tax, thereby increasing the rate of growth of money. Consider a once-for-all but permanent increase in this rate of growth of money. How does real GDP in the new steady-state compare with the original real GDP? The argument of the von Furstenberg type of analysis, as can be seen from equations (20) through (26), is simply that the once-for-all permanent increase in high-powered money growth, in this fashion, will tend to lower q, raise r, diminish desired real wealth Wd, and raise sk. If the negative relation between Wd and r is rather strong, it could lead to a decline in skd Wd such that real desired and actual private physical capital (KBd = KB) falls in the new long-run equilibrium. The result would be that Y declines from what it would have been.

In the above model, government physical capital is not financed by any of the receipts from inflationary finance; hence, the decline in KB is not counterbalanced by sufficient increase in KG. This assumption is considered realistic by von Furstenberg, implying that inflationary finance is for the purpose of government consumption expenditures. But even if KG increases with the receipts, he assumes that the “measured” productivity of public capital is only a fraction of that of private capital. Further still, if the B/W ratio desired by the private sector is constant, the decrease in W lowers B and hence KG, given that KG is affected by B (all measured in real terms, of course).

III. Inflationary Financing of Business Investment

In the framework of inflation tax, creation of high-powered money is typically for the purpose of financing government investment. But there is also another tradition, implicit or explicit particularly in the discussions of selective credit controls as instruments of development policy (see O.E.G. Johnson (1974)), in which high-powered money is created for the purpose of financing investment in the private sector. More specifically, the central bank could make funds available to commercial banks for on-lending to private businesses for investment; or the central bank could rediscount commercial bank paper to improve the liquidity position of the commercial banks, permitting the latter to make additional loans for investment purposes; or both.

Where there are idle or underemployed resources that can be quickly and efficiently mobilized by those who have been privileged to get additional bank loans, no inflation need ensue from the creation of additional high-powered money. The question of inflation arises when resources must be enticed away from consumption goods and forced saving must occur. In particular, the additional lending to business is assumed to increase the desired investment in the private sector. The increased aggregate demand raises the price level. In this model, it is also assumed that wage increases lag, and are smaller than, the price increases ensuing from the above process. This lag of wages behind prices then raises the profit share in the private sector, lowers real consumption by households, and increases real saving, since the marginal propensity to save of profit earners is assumed to be greater than that of wage earners. The process must continue until the in creased real saving equals the increased real investment desired by business.

An implication is that the savings-income ratio is, in the aggregate, increased by this type of inflationary financing policy. In other words, as stated above, inflation is assumed to transfer income from wage earners as a group to profit earners as a group; it is then assumed that the savings-income ratio of the wage-earning group declines while the savings-income ratio of the profit-earning group increases. For the aggregate savings-income ratio to increase, it follows that, at any given level of real income, the increased saving of the profit earners must be greater than the decrease in saving of the wage earners. A Kaldorian situation is therefore implicit in this sort of analysis.

In short, the inflationary financing of business investment supposedly permits an effective redistribution of income from wage earners (and consumers) to profit earners, as well as an effective increase in the savings ratio S/Y, where S is saving and Y is GDP. As an example of models in this area of analysis, consider the following framework of Thirlwall (1978). Using a savings function of the Kaldorian type (S = sw W + srR), Thirlwall specifies the following savings function at time t:

where sw and sr represent the marginal propensity to save, respectively, out of wages (W) and profits (defined as the difference between output and wages); W0 and Y0 are the initial wage bill and income (GDP) level, respectively; P^ is the rate of inflation; and w is the rate of increase of the wage bill. If we divide equation (27) by Y0eP^t we get

Suppose now that wages do change with prices. In particular, suppose that w = w0 + w1(P^), where w0 is the autonomous rate of change in wages and w1 is the wage-price coefficient. Then

Following Thirlwall, we differentiate equation (29) with respect to P^ to get

The inflationary financing of business investment in the sense we have been discussing implicitly assumes that (S/Y)t/P^>0.

This, as will be seen from equation (30), happens only if w1 < 1 and if sw < sr. That is, income redistribution must occur between wage earners and profit earners, and the marginal propensity to save (and to invest) of the profit earners (who gain real income) must be greater than the marginal propensity to save of the wage earners (who lose real income).

In the view of this author, the assumption that sw < sr could be accepted more easily than the assumption that w1 < 1. In general, one may hypothesize that w1 would tend to approach unity as P increases. In short, at very low (and nonchronic) rates of inflation, the effects of inflationary financing of business investment could be favorable to growth because wage earners may behave as if the rate of inflation will be constant. But as inflation rates increase, the savings-income ratio, hence real GDP growth, is not likely to be increased by inflation because w1 becomes virtually unity.

In terms of von Furstenberg’s analysis, the foregoing argument would take the following form. The bank financing of business enterprise leads to a rise in the market value of ownership claims in respect of physical assets relative to the replacement cost of those assets; that is, q would rise. This rise in turn lowers the real rate of interest, raises desired wealth of the private sector, and, given sk, increases the desired capital stock of businesses and households. The end result is increased growth. But the rise in inflation, if it continues, eventually tends to increase uncertainty, thereby reducing q and reversing the process just outlined.

Apart from the view that, beyond modest rates, such inflationary financing could not be used effectively to redistribute real income from wage earners to profit earners, the quality of investment in the private sector is expected to deteriorate as the rate of inflation increases, so that the rate of return to investment, and hence real GDP growth, would be adversely affected.10

IV. Empirical Aspects

We shall briefly present the conclusions of some authors who have looked at the international evidence relating inflation and growth in developing countries, as well as the empirical implications of the models and approaches that we have been discussing.

the general evidence

At least since the late 1950s, attempts have been made to examine the international data to see if differences in GDP growth rates or in the changes in these rates have had any association with differences in inflation rates or in their changes. Tun Wai (1959), in a study of 31 developing and industrial countries over the period 1938-54, found that for developing countries the data were inconclusive; for most of a small number of countries for which available statistics covered periods in which the rates of price increase differed significantly, however, the evidence indicated that the rate of growth was higher when the rate of inflation was lower. It is interesting that the countries for which Tun Wai’s intracountry study could be done were primarily Latin American countries with relatively high average rates of inflation.

Dorrance (1966) studied 48 countries, also both developing and industrial, over the period 1953-61 and found that his data supported the argument for mild inflation except for the low-income countries. In particular, declining prices or “unduly low price increases” tended to be associated with low rates of growth. Once inflation exceeded a certain level, however, rising prices tended “to discourage economic development.”

Wallich (1969), studying 43 countries (18 industrial and 25 developing) and using five-year averages for the period 1956-65, found that the effect of (current) inflation on growth was significantly negative. He concluded that “[the] size of the coefficients indicates that only the smaller part of the effect of inflation operates via the effect of inflation on investment. The larger part operates through other channels—which may include factors such as the quality, distribution, and cost of investment” (p. 300).

Thirlwall and Barton (1971), in a study of 51 countries (17 of which had annual per capita incomes in excess of US$800 in 1963, and the rest below that) over the period 1958-67, concluded that, in general, the protagonists of mild inflation had the evidence on their side. For countries with annual per capita incomes exceeding US$800, there was a clearly positive association between inflation and growth. In the income group below US$800 a year, no significant relation between growth and inflation emerged. But when the authors divided this group into two subgroups—countries with inflation of less than 10 percent a year and those with inflation above 10 percent—they found that for the first subgroup no particular relation was discernible, but that for the second subgroup a significant negative association was discernible. They concluded: “It would appear that once the rate of inflation exceeds 10 percent per annum the negative aspects of the effects of inflation on growth tend to come to the fore” (p. 271). In a later study Thirlwall (1974) also found that, in a cross section of 15 countries in Latin America over the period 1958-68, there was evidence of a definite negative relation between inflation and growth. But, given the very high inflation rates experienced by many of the countries concerned, Thirlwall cautioned against jumping to the conclusion that the Latin American experience provides evidence that inflation is detrimental to growth. Evidently there is no dispute that, after a point, inflation is detrimental to growth.

A Fund staff study (International Monetary Fund (1982)) of 112 “non-oil” developing countries over the period 1969-81 concluded that, for the most part, “relatively low inflation rates have been associated with relatively high growth rates and that reductions, or at least relative reductions, in inflation have been associated with an improvement, or relative improvement, in growth rates” (p. 134).

Empirical testing of structuralism per se has tended to take up the issue of the role of structural factors versus monetary or excess aggregate demand factors in explaining inflation in Latin America. For the analysis in this paper, a recent work by Wachter (1976) is of special interest. Using data for Chile, Brazil, Mexico, and Argentina, she found support for the view that, on the basis of quarterly data, inflation was positively correlated with the percentage change in the relative price of food, with the latter defined as the price of the food component of the consumer price index (CPI) divided by the total CPI. In these equations changes in the stock of money (lagged and unlagged) were also important parameters explaining inflation. Thus, an important structuralist hypothesis was generally supported without completely negating “monetarist” claims. But the results were not favorable for the structuralist variable when annual data were used. Wachter argues that one cannot expect the structuralist variable (P* in our model) to be significant in the long run, whether a passive or an active money supply is assumed, even if this variable is significant in the short run. Assuming money supply is passive (and she does find evidence of this using Sim’s test of causality), a higher P*, which induces a higher inflation rate, will “cause” the money stock to increase. Given inflationary expectations and the level of real income, Wachter explains, the rate of monetary expansion will then fully mirror any changes in the inflation rate. Even if the money supply is not passive, she argues (in conformity with general monetary theory) that if the inflation rate varies with P* in the short run it will not in the long run because of the real balance effect. That is, in the long run the money stock will expand in response to higher prices so that the effect of P* on the general price level will no longer be perceptible; alternatively, absolute prices will return to their initial level despite a higher P* if the money stock remains unchanged. Wachter’s point is that, if prices respond quickly to the real balance effect, or if the money stock responds quickly enough to inflation, or both, then the “long-run” effects may come within a year. For instance, using Almon lags, she found that the bulk of the price response to changes in the money stock in Chile occurred within the year.

empirical implications of the models

The models discussed in Sections IIII can be interpreted as stating the conditions under which changes in inflation and changes in growth will be positively correlated. Or they may be interpreted as stating that changes in inflation and changes in growth will be positively correlated because of certain conditions which tend to exist at least in some developing countries.

In the case of structuralism, the empirical arguments are that increases in growth lead to increases in the relative price of the agricultural good and that, because of price rigidity in the industrial sector, increases in inflation will tend to accompany increases in the relative price of the agricultural good. These arguments lead to the conclusion that increases in growth must be accompanied by increases in inflation for the set of developing countries for which the hypothesis holds. In short, remembering that PA is the price of the agricultural good, PI the price of the industrial good, and P* the ratio of PA to PI, structuralism implies that

The arguments for both inflation tax and inflationary financing of business investment require that the investment-GDP ratio increase with inflation and that growth increase with the investment-GDP ratio. If these requirements hold, then there will be a positive correlation between changes in inflation and changes in growth. In the case of the inflation tax, it must be true empirically that the ratio of government expenditure to GDP increases with inflation—that is,

that government capital expenditure increases relative to total government expenditure when the latter increases relative to GDP—that is,

that government investment does not crowd out private investment, so that the investment-GDP ratio increases with the ratio of government capital expenditure—that is,

and that the productivity of government expenditure is such that growth can be increased by increasing government investment relative to government total expenditure—that is,

In short, for the case of inflation tax to finance government investment we have that:

For the inflationary finance of business investment to generate increased investment relative to GDP, private saving must increase relative to GDP. For this to happen, as has been shown, it must be empirically true that increased inflation results in a transfer of income from wages to profits and that the propensity to save out of the income of profit earners is greater than that of wage earners. If these requirements hold, then the investment-GDP ratio (I/Y) will increase with inflation. In short,

Strictly speaking, the above empirical implications assume only a closed economy. Once an open economy is considered, an additional restriction implicit in the above approaches is that the change in inflation must not lead to an appreciation of the real effective exchange rate (RER) with negative output effects large enough to dominate any positive relation that would otherwise exist between inflation and growth. That is, it would be especially valuable if it is the case that

The question arises whether the nature of the relation between inflation and growth remains invariant, irrespective of whether the average rate of inflation over an extended period of time is low, medium, or high. It is only fair to say that many (for example, Dorrance (1963, 1966) and Thirlwall (1974, 1978)) who have argued that there is a positive correlation between changes in inflation and changes in growth state explicitly that the positive relation holds only up to a certain rate of inflation. In short, the hypothesis is that, up to a certain level of inflation, increases in growth will be positively correlated with increases in inflation. That is, the relation between inflation and growth will take the following form:

where a and b are constants. This simple differential equation then implies that

where α, β, and γ are constants.

In the form of equations (34) and (35), the hypothesis is not inconsistent with the view that, in the developing world on average, low-inflation countries will tend, other things being equal, to have higher rates of economic growth than high-inflation countries. Up to a point, increases in inflation will be accompanied by increases in growth, but the latter will be declining for marginal increases in inflation. Indeed, at the point at which the marginal gain in growth from inflation becomes zero, P^ will be equal to a/b in equation (34), or to β/2γ in equation (35).

some additional findings

The development literature has, perhaps understandably, been concerned with the output-inflation trade-off because inflation may be aggravated by certain structural features important in developing countries or because inflation may be used as a tool for financing investment. But even in the absence of the structural features stressed by the structuralists or the desire to use inflation as an explicit tax, there could still exist an output-inflation tradeoff because of the role of relative prices in influencing capacity utilization rates and factor supplies in the economy.

In this regard, although they have not featured prominently in the literature on growth and inflation in the developing countries, the natural rate hypothesis and the rational expectations approach can, no doubt, be used with possibly beneficial results to analyze the output-inflation trade-offs in developing countries. The natural rate hypothesis posits that changes in inflation will stimulate changes in real output if and only if such changes succeed in getting suppliers of factor services and of commodities to believe that relative prices are moving in their favor in line with inflation. Lucas (1973), for instance, has shown that the larger is the variance of inflation, the smaller will be the increase in output accompanying a given percentage increase in inflation within any given time period. In other words, the greater is the variance of inflation, the less likely it is that decision makers will be deceived into believing that changes in general price movements constitute relative price changes.

Hanson (1980) has also used a rational (or what he calls a quasi-rational) expectations approach to study inflation and growth in five Latin American countries over the period 1950-74 and has found positive correlation between unexpected inflation and growth. Edwards (1983) has, however, shown that Hanson’s results must be substantially modified11 for the five countries in view of Hanson’s neglect of the role of fiscal deficit in the money-creation process in developing countries and in view of the fact that these countries are open economies (for example, terms of trade changes affect growth). For instance, with his empirical methodology Edwards finds no evidence of a significant effect of unexpected money on output growth for Brazil and Chile, but there was some evidence of significant positive effects of unexpected money on growth for Colombia, Mexico, and Peru, with coefficients that varied widely in magnitude and lags. Edwards also found, for the high-inflation countries of Brazil and Chile, that inflation tends to have a negative effect on growth.

V. Conclusion

This paper has been concerned with the nature of the economic arguments relating inflation and growth as they have been generally advanced in the literature of development. In structuralism, the presumption is that the rise in the relative price of “the” agricultural commodity increases with growth and that this price increase in turn induces increase in the rate of inflation. For the case of inflation tax as an instrument for financing government investment, the implicit assumptions are that increases in inflation are beneficial to raising government expenditure relative to GDP; that government capital expenditure rises relative to total government expenditure when the latter rises relative to GDP; and that increases in government capital expenditure relative to total expenditure raise domestic investment relative to GDP, engendering increases in (real GDP) growth. Any positive relation between changes in inflation and changes in growth ensuing from these factors will tend to be augmented if the real effective exchange rate does not increase in appreciation with increases in the inflation rate. The analysis involving inflationary finance of business investment assumes implicitly that the investment-income ratio increases with inflation, which in turn entails the dual condition that the wage-price coefficient relating change in money wage to change in prices be less than unity and that the marginal propensity to save of wage earners be less than that of profit earners.

The findings of past empirical work on the subject were reviewed. It was seen that, in general, such work has concluded that the cross-country data have been consistent with the view that, other things being equal, low-inflation countries tend to perform better than high-inflation countries as far as (real GDP) growth is concerned. Some of the literature, however, has also produced evidence showing that, up to a certain relatively low level of inflation, increases in inflation tend to be positively correlated with increases in growth. It was also seen that there is some evidence supporting the basic structuralist thesis only in the very short run. More complete empirical investigation of the theories under discussion should in principle involve two steps. The first would look at the overall relation between inflation and growth, and the second would investigate which of the preconditions (or implicit assumptions) for a positive correlation between changes in growth and changes in inflation have been satisfied and which have not.


Mr. Johnson, Assistant Chief of the Developing Country Studies Division of the Research Department, holds degrees from the University of California (Los Angeles).

For a discussion of the structuralist view on inflation, particularly in the Latin American context, see Campos (1964), Herrick and Kindleberger (1983), H.G. Johnson (1966), and Wachter (1976).

Olivera (1971) discusses some aspects of the economics of passive money in this general context.

Note that, in general, P^I will also change because of excess demand pressures. This, of course, is not denied by structuralists.

See, especially, Aghevli (1977) and Mundell (1971, chapter 4).

Much analysis of the Mundellian type is concerned with finding the rate of inflation or monetary expansion at which growth is maximized (for example, Mundell (1971, chapter 4) and Calvo and Peel (1983)), or at which the discounted flow of total consumption is maximized (see Aghevli (1977)).

See also Tanzi (1982) for a summary of the issues relating to fiscal disequilibrium in developing countries.

Note that government expenditure is equal to government revenue plus government borrowing.

See, for example, O.E.G. Johnson (1976), where it is argued that in an open economy the choice for inflationary financing becomes that for inflationary financing cum currency depreciation.

In this section, Y is used to represent both the nominal and real income (output or GDP).

For a general analysis of these and related issues in the cost of inflation, see Bailey (1956), Dorrance (1963), Fischer and Modigliani (1978), as well as Kessel and Alchian (1962).

But see Hanson (1983) for a response to Edwards’s criticism.

Other Resources Citing This Publication