4 Optimal Taxation of Financial Savings in Developing Countries Relevance of Supply-Side Tax Policies

Ved Gandhi, Liam Ebrill, Parthasarathi Shome, Luis Manas Anton, Jitendra Modi, Fernando Sanchez-Ugarte, and George Mackenzie
Published Date:
June 1987
  • ShareShare
Show Summary Details
Liam P. Ebrill

This chapter evaluates, for the case of developing countries, the potential impact of supply-side proposals for stimulating savings. Some supply-side economists argue, in the context of developed economies, that the interaction of inflation and high marginal rates of income taxation is such as to imply a serious disincentive to savings. Thus, they recommend sharp reductions in marginal tax rates.1 If applied to developing countries, would such reductions be either desirable or lead to a pronounced increase in savings?

In examining these issues, this chapter concentrates on savings channeled through organized financial markets. This emphasis can be justified on a number of grounds, of which three are worth special mention. First, it is often argued that, while aggregate savings are generally not very sensitive to changes in the net rate of return, that is not the case with individual components such as financial savings in organized markets, since changes in the rate of return that can be earned on any one form of savings can be expected to induce significant shifts into substitute assets. Second, organized financial markets are often central to the efficient allocation of available funds. Third, and particularly important given the supply-side perspective of our analysis, the specific disaggregation of aggregate savings employed here is useful precisely because the components can be distinguished by whether they represent savings in organized as opposed to unorganized financial markets. Since supply-side proposals for reform are typically directed at lowering high marginal rates of income taxation, the most immediate effect of these proposals in developing countries will be on those savings placed in the organized rather than the unorganized money markets.

This chapter first describes the financial markets in developing countries as the backdrop against which the appropriateness of supply-side proposals are evaluated. Particular attention is devoted to the significance of financial repression, an indicator of which is the policy of institutionally setting nominal interest rates at artificially low levels. The low (indeed, negative) real rates of return that result induce asset holders to shift out of assets issued by organized financial markets into investments that are less productive, such as gold and housing. Should this effect be empirically significant, it has important implications for this study given the study’s concentration on savings behavior in organized financial markets.

Given this background, what are the appropriate tax reform policies? The chapter begins with a discussion of the reform proposals of the traditional public finance theorists. It is argued that the existing optimal tax literature does not adequately accommodate the circumstances of developing countries. A modified optimal tax framework is suggested. The results of this exercise are contrasted with the recommendations of the supply-side school. The conclusion is that, while not necessarily incorrect, the proposals of the supply-side school are deficient in that they do not take account of the heterogeneity of developing countries. This is a deficiency they share with the traditional public finance literature. Specifically, the modified optimal tax framework developed in this chapter suggests that, depending on the particular country under consideration, the optimal tax treatment of financial savings runs the gamut from subsidies to substantial taxation.

All of the above treats the problem from a microeconomic perspective. It is also pointed out, however, that the ramifications of financial repression extend to macroeconomic policy, a factor that should be acknowledged in any study of tax policy in this area.

The traditional public finance literature, the supply-side approach, and the modified optimal tax framework developed here do, however, share a common theme: appropriate tax rates should depend on the sensitivity of savings behavior to changes in the rate of return.

I. Institutional Background

Any tax reform recommendations must take account of the economic environment in which the reform takes place. The financial structures of developing countries have a number of features that, although not unique to these countries, are held to be more widely experienced by them. While this is undoubtedly correct, it should also be kept in mind, as these features are introduced and discussed, that the degree to which any feature is relevant varies greatly from country to country. As argued later, this variability has important implications for tax policy.

First, financial institutions in many developing countries tend not to be well developed (cf., Miracle, Miracle, and Cohen (1980)). This has a number of implications. In particular, McKinnon (1973) argued that in developing countries, investors are constrained to using self-finance and that investments are relatively lumpy. As a result, investors must accumulate money balances prior to their investments, leading to the hypothesis that the aggregate demand for money will be greater, the larger is the share of investment in total expenditures. (This is McKinnon’s complementarity hypothesis.)

Second, it has often been pointed out that in many developing countries, interest rates on assets in organized markets are often set at low nominal rates given the rate of inflation (see Galbis (1979)). A large body of literature exists in which it is argued that such institutional restraints can only result in financial repression (Shaw (1973), McKinnon (1973)).2 The existence of these interest rate ceilings, it is held, leads to a reduced supply of financial savings to organized markets, as savers turn to alternative nonproductive investments such as property and precious metals. Depending on foreign exchange market conditions, savings may even be placed overseas. These effects, it is argued, require some type of investment rationing mechanism.

Financial repression may not only result in portfolio reallocations out of organized financial markets but may also influence short-run macroeco-nomic developments in many developing countries. Specifically, Leff and Sato (1980) identify a couple of macroeconomic implications. First, ex ante investment will tend to exceed ex ante savings. Second, and more important for short-run policy, interest rates cannot be used to maintain equilibrium in the savings-investment market. They argue that investment tends to be relatively more buoyant than savings with the concomitant possibility of instability. Price-level movements and foreign capital flows then become central to absorbing shocks. The general conclusion is that inflation and dependency on foreign capital inflows are structurally rooted in many developing countries.

A third distinctive feature of financial markets in some developing countries is that investors enjoy the existence of unorganized curb markets (van Wijnbergen (1983a) and Buffie (1984)).3 By emphasizing the role of curb markets for portfolio allocation decisions in developing countries, van Wijnbergen questions the presumption that the alternative asset to financial holdings in organized markets must consist of nonproductive investments. Instead, it may consist of holdings of financial assets issued in unorganized money markets of the curb market type (van Wijnbergen (1983b)). This possibility alters the implications of financial repression for economic policy. For example, an increase in interest rates in organized markets may no longer result in the “free lunch” of financial flows attracted from nonproductive sources, but may instead lead to a general increase in the working cost of capital faced by firms. Further, the central governments of many developing countries exercise a major claim on savings channeled through organized money markets. If such governments invest less efficiently than the private sector, an increase in interest rates then achieves the ambiguous result of alleviating the distortion faced by savers at the expense of exacerbating the distortion associated with inappropriate government investment decisions.

A fourth and final characteristic is that many developing country governments employ compulsory savings schemes.4 For example, Shome and Squire (1983) point out that some Asian countries such as Malaysia, Sri Lanka, and Singapore have, in varying degrees, funded social security systems.

Given that the above features are not uniformly experienced, they also serve to emphasize the heterogeneity of developing country circumstances. In view of the prominence it has attained in the economic literature, the following assessment of supply-side tax proposals will concentrate on the particular role played by financial repression. An optimal tax framework will be developed that will explicitly allow for such a phenomenon. The robustness of this framework will then be assessed in light of the other structural features of developing country capital markets just discussed. The framework will also form a basis for evaluating the relevance of supply-side tax policies as they apply to savings. Even at this stage, however, there is already the sense that the heterogeneity of the financial systems of developing countries implies that tax changes based on simplistic rules of thumb are unlikely to be appropriate.

II. Evaluation of Reform Proposals

The recent supply-side literature is large. However, perhaps out of a desire to make politically feasible tax policy recommendations, the actual number of recommendations contained in that literature is quite small. Though not bearing directly on the savings issue, the best known are those associated with the Laffer curve, especially that high marginal tax rates should be sharply reduced. This type of recommendation has not been made solely with the United States in mind, however; Wanniski (1983), for example, argues for the relevance of the Laffer curve for a number of developing countries (see Ebrill (Chapter 7)).

More generally, it would appear to be in the spirit of the supply-side approach to say that they favor tax cuts aimed at stimulating (private) savings. This, of course, begs the question of the adequacy of the pre-existing level of savings. Presumably the supply-siders believe that aggregate savings are depressed in the first place due to the existence of the tax system and other distortions. In fact, there is general agreement that savings in developing countries are low in relation to investment needs.

Summary of Theoretical Literature

One way of evaluating the supply-side approach is to contrast its recommendations on savings with those of the more traditional public finance literature. In particular, there are a number of papers that address the issue of how to tax savings.5 The orientation of this literature has been to determine whether the tax base should be income or expenditure (consumption). Among the earlier commentators on this question, Mill (1921) argued for a consumption base, not so much on efficiency as on equity grounds—income taxes tend to favor those who derive their income from wealth rather than work effort. More recently, the consumption base received further support in the work of Kaldor (1955), who argued for an expenditure tax on both efficiency and equity grounds.

Even more recently optimal tax theorists sought to resolve this question by considering solely the efficiency implications. Their solution lay initially in a simple extension of the optimal commodity tax literature. Thus, Feld-stein (1978) gives the static framework associated with that literature an intertemporal flavor by labeling consumption in different periods as different commodities. A two-period life-cycle model is postulated. The consumer is then assumed to allocate his endowment across leisure (which is untaxable), first-period consumption, and second-period consumption. In general, since leisure cannot be taxed, second-best theory suggests that it is optimal to violate all first-order conditions. There are circumstances in which a consumption tax is optimal—as when a compensated change in the wage rate induces the same proportional changes in consumption in the two periods.6 This type of result suggests that in general neither a consumption tax nor an income tax will be optimal—it will in most cases be preferable to have separate tax rates on savings and consumption.

This inconclusiveness in the optimal tax literature is reinforced by more recent work. For example, Atkinson and Sandmo (1980) develop an explicit overlapping generations growth model and demonstrate that the Feldstein type of analysis can be generated within that framework if the government possesses a range of additional instruments, such as an unconstrained debt policy, to achieve the desired intertemporal allocation where the steady-state marginal product of capital equals the discount rate. They then consider the optimal tax solution in a second-best world where the government cannot attain the first-best intertemporal equilibrium. This part of their analysis is of more interest to this paper since, as pointed out above, the supply-side concern with stimulating savings is presumably predicated on a belief that the capital stock is inadequate. The Feldstein approach, however, assumes that capital markets are efficient and asks whether savings should or should not be taxed in an overall package of taxes levied with a view to raising a given amount of revenue with minimum excess burden. Unfortunately, even when it is known that the capital stock is below its “first-best” level, the effect of the recommended tax system remains ambiguous. Indeed, in the specific Cobb-Douglas example they consider, Atkinson and Sandmo (1980) point out that a tax on capital income is desirable, a result that depends on the shape of the aggregate savings function.

In the absence of qualitative results concerning the optimal treatment of savings, some authors have taken the more pragmatic route of parameterizing general equilibrium models and then evaluating the advisability of various tax reform proposals. For example, Fullerton, Shoven, and Whaley (1983) use a dynamic numerical general equilibrium model of the U.S. economy to determine the welfare implications of substituting a progressive consumption tax for the current income tax structure. The initial effect is that individuals consume less. The welfare changes are influenced by both the transition and the steady-state balanced growth paths. Notwithstanding the initial drop in consumption, the net effect of the tax change is welfare-enhancing. Of course, the outcome depends on the precise parameters, as is evident from the range of opinions on the magnitude of the transition costs implied by this type of tax change (Auerbach and Kotlikoff (1983), Seidman (1984)). Nonetheless, there appears to be growing interest in the proposition that, within the context of the models employed, consumption is a more appropriate base for taxation than income.

The main thrust of this chapter is not to evaluate the conclusions of the optimal tax literature concerning the appropriate treatment of aggregate savings but to determine the implications of supply-side policies for financial savings, where the latter are a component of aggregate savings. While the traditional public finance literature has not explicitly addressed the optimal tax treatment of this component, those papers concerned with determining the optimal rate of inflation are relevant. Inflation is viewed as a tax on money balances, which is a component of financial savings. Beginning with Phelps (1973) and Siegel (1978), it has been argued that a positive tax on liquidity (a positive rate of inflation) is in general desirable, on the principle that the demand for all taxable goods ought to be proportionately reduced (the Ramsey Rule). Subsequent writers have pointed out that, along lines analogous to those taken by Atkinson and Sandmo (1980) above, the economy need not be on the golden-rule path. Since inflation can affect the long-run capital intensity of the economy, its impact on this pre-existing distortion should be accommodated (Summers (1981)). Further, it is argued that account should be taken of the fact that money is costless to produce (Drazen (1979)). As might be expected, since the outcome depends both on the degree to which other distortions are recognized and on the empirical magnitude of numerous own- and cross-price derivatives, there are no unambiguous qualitative conclusions. In addition, as has already been pointed out, neither stream of the optimal tax literature (i.e., that pertaining to aggregate savings and that to money balances) is directly applicable to circumstances in the developing world.

An Optimal Tax Framework for Financial Savings in Developing Countries

Given the deficiencies of the existing optimal tax literature, an alternative, also highly simplified, optimal tax model is developed below, which incorporates certain salient features of the financial structures of developing countries as described earlier. In particular, financial repression is explicitly allowed for, in a framework in which the rate of inflation is assumed to be exogenous. The menu of financial assets available to the representative consumer consists of money for transactions purposes, financial savings deposited in organized money markets, and other savings.7 Financial repression is allowed for in the form of pre-existing distortions on both money and financial savings.

The model does not allow for the fact that money for transactions purposes is costless to produce. This is not a serious problem since our purpose is to determine the optimal tax treatment of financial savings other than money (Ml), given a rate of inflation and a degree of financial repression, rather than to determine the optimal rate of inflation. Nor, since it uses a static framework, does the model allow for the possibility of a further distortion due to the typical economy being below the golden-rule path. However, while such an effect will in general alter the magnitudes of the optimal tax changes, it is unlikely to change their direction. This point will be discussed in greater detail.

The important notations used in the model are defined below.

C0 = consumption of leisure;

C1 = consumption of current goods and services;

C2 = consumption stream associated with nonfinancial savings;

C3 = consumption stream associated with financial savings other than financial savings for transactions purposes;

C4 = consumption stream associated with financial savings for transactions purposes (Ml);

B = government’s net revenue requirement;

ti = ad valorem tax rate on Ci;

t¯i = pre-existing proportional distortion on Ci, i = 3, 4;

Pi = market price of Ci;

θi = share of tax in total price of Ci

α = share of financial savings (C3) mediated through the government;

ηij = compensated elasticity of demand for Ci with respect to Pj;

μ multiplier on the government’s budget constraint; and

λ = private marginal utility of income.

The underlying methodology of this type of exercise is well known (cf., Atkinson and Stiglitz (1980)), Accordingly, where possible, the explanation below will concentrate on aspects of the formulation that are unique to the problem at hand. Specifically, there are five commodities, including leisure (Co), where leisure is untaxable. Of the other commodities, C4 is a proxy for the consumption stream a typical consumer obtains from financial wealth held for transactions purposes. The underlying stock that provides this consumption stream is assumed to be M1. C3 refers to the remainder of financial savings. It is assumed that organized money markets act as intermediaries, so that these savings are productively invested. Accordingly, C3 is best viewed as a proxy for the consumption stream that results from that investment. As for the investment itself, a proportion, α, is financed by the government via borrowing. This proportion is assumed to be subject to government manipulation, implying that the private sector is a residual investor. C2 refers to the consumption stream that results from the disposition of the remainder of aggregate savings. It might be useful to view this in terms of the “consumption” of owner-occupied housing, works of art, and so on. Finally, C1 is a variable representing the aggregate of current consumption. Given this structure, it is clear, as pointed out above, that this exercise does not accommodate the intertemporal investment dimensions of the problem or the fact that M1 is costless to produce. It is not a true general equilibrium exercise. Instead, the very fact that the variables should be viewed as proxies indicates the illustrative nature of the framework. However, this illustrative nature can be defended on the grounds that the model may nonetheless yield useful qualitative insights.

The market environment assumed for this exercise, in common with most other optimal tax exercises, presumes competitive behavior on the part of private firms and consumers. The formulation of the consumer’s problem requires some elaboration. Since the paper is concerned with economic efficiency rather than equity, the general equilibrium structure is set in the context of a representative price-taking individual who has a twice continuously differentiate, monotonic, strictly quasi-concave, utility function U(C0, C1, C2, C3, C4). The consumer’s maximization problem is

where C¯0C0 is labor supplied and the other variables are as defined above. The last term in the budget constraint requires explanation. The widespread use of interest rate ceilings is in the present context modeled as a pre-existing tax on C3 (t¯3). Some of the implicit revenue from this tax accrues to elements of the private sector, to the extent that the government only borrows some of the available financial savings. This quantity is gauged by the final term above. However, in order to avoid unnecessary complications, it is assumed that the representative consumer responds to the gross rather than the net price of C3—that is, this income is modeled as lump-sum income.8 Accordingly, the consumer’s first-order conditions are

Consider the government’s maximization problem. The nature of the optimal tax exercise is critically dependent on the constraints placed on the range and flexibility of policy instruments at the disposal of the authorities. In the case under consideration, two constraints are important. The first, already mentioned, is that interest rates on financial assets are institutionally fixed and that the rate of inflation is not readily amenable to control. The net effect of this is modeled as pre-existing taxes on both C3 and C4. The second constraint comes from the assumption that the consumption stream associated with investment in owner-occupied housing etc. (C2) is not taxed. (This could be due to administration costs.) The government’s objective is then to raise a required amount of revenue, some of which is levied by the pre-existing “inflation taxes,” given that the only variables under its control are the tax rates on C3 and C1 the latter being a consumption tax.9 This can be expressed as follows:

where t¯4=t¯3+,>0, constant, in recognition of the fact that the inflation tax on C4 is greater than that on C3 given that cash balances pay a zero nominal interest rate. Differentiating, and substituting for the consumer’s first-order conditions, the following first-order conditions are immediate:

Note the inclusion of a term in ∂α/∂C3 (<0) in recognition of the fact that a reduction in C3 (say) may in general be expected to result in the government increasing its share of financial savings. A more complete exercise in an intertemporal framework would employ a broader government budget constraint incorporating, in addition, the optimal trade-off between borrowing and taxation. This is not attempted here.10

Differentiating the individual’s budget constraint with respect to Pi,i = 1, 3, respectively, and substituting, allows one to re-express equations (2a) and (2b) as follows:

Substitute for the Slutsky equation into total demands ∂Ci/∂Pj where I refers to private income and Sij is the compensated substitution effect of commodity i with respect to a change in the price of commodity j. Further, assume that the general equilibrium income effect associated with ∂C3/∂Pj can be set equal to zero. Equation (3) becomes

These equations are analogous to the first-order conditions of the traditional optimal commodity tax literature with the caveat that the equations here incorporate terms in the pre-existing distortions. t¯i (e.g., Sandmo (1976)). To this point, they lack qualitative content as far as the optimal tax rates are concerned. To alleviate this, some structure can be placed on the magnitude of some of the substitution terms by considering the context of this exercise. Specifically, given the level of aggregation, the utility function is assumed to be separable between C1 and Ci i = 2, 3, 4. The net effect of this assumption is that ,Sij = 0 for j = 3, 4. Using symmetry Sij=Sji) and expressing in terms of elasticities, equation (4) becomes, respectively,


and can be assumed to be negative by the negative definiteness of the substitution matrix, θ1,θ3,θ¯3,θ¯4 are the relevant shares.11

It is clear from equation (6) that the traditional Ramsey Rule, calling for compensated proportional reductions in those markets that are distort-able, must be altered in light of the terms in pre-existing distortions. The Ramsey Rule is a statement about quantities. The concern here is with prices. What are the optima) tax rates t1 and t3? By equation (5), given the zero cross-price effects between C1 and other distortable commodities, the inverse elasticity rule holds—the larger the absolute value of θ11 the smaller the value of η1 (and hence, t1).

The determination of t3, the optimal tax rate on financial savings, is of greater interest for this paper. The magnitude (and sign) of t3 as is clear from equation (b), are influenced by a number of factors in addition to own- and cross-price elasticities. These factors are all a function of the preexisting distortions, θ¯3 and θ¯4. The term in θ¯4 is obvious. To the extent that the cross-price effect between C3 and C4 is nonzero, any change in the price of C3 will interact with the pre-existing distortion on C4. This interaction should be allowed for, and is elaborated on below.

To see how the pre-existing distortions on C3 affect θ3 collect the terms in θ¯3. These are

The first term is a weighted average of the distortion on C3 where the weights are the shares of C3 invested by the private sector and the government, respectively. The weights are valued at λ, the private marginal utility of income, and μ, the social marginal utility of income, respectively. This reflects the fact that the cost of the distortion varies depending on whether it is mediated through the private or the public sector.

The second term reflects the impact of changes in the share of C3 mediated through the government sector. It is negative, since ∂α/ ∂C3< 0 and μ > λ where the latter is due to the excess burden associated with government intervention. Thus, this term mitigates the effects associated with 3. The intuition is that θ¯3 is a pre-existing distortion. The larger in absolute value is the magnitude of ∂α/∂C3, the more will this distortion or tax be used to finance governnment activities, thereby reducing the need on the part of the government to rely on other taxes.12

Some further insight into the value of t3 can be gained by appealing to homogeneity, namely,

Given that η31 = 0 by assumption, and substituting, it follows that equation (6) can be re-expressed as

The value of θ3 (i.e., t3) remains, in general, ambiguous. However, one of the terms in equation (9), namely that in η34, can be assumed to be small. Thus θ¯4 > θ¯3 and, accordingly, the coefficient of η34 may not be large. Further, the absolute magnitude of η34 itself may be small. This is the elasticity of financial savings to a change in the rate of return on transactions money. Some analysts, by modeling the transactions demand for money as being a function solely of the level of income, implicitly assume that this elasticity is zero. If this term is ignored, equation (9) reduces to

This expression shows that the magnitude, indeed the sign, of θ3 depends critically on the magnitudes of θ¯3 and η32.

From the above, it is clear that, even given a number of (plausible) assumptions concerning the magnitudes of some cross-price effects, the optimal tax treatment of financial savings will depend on the circumstances of the individual country under consideration. Specifically, the above limited exercise suggests two obvious polar cases as a mechanism for summarizing and highlighting the possibilities.

Case I: Absence of Financial Repression

In the absence of financial repression, θ¯3 = 0 which, together with the assumption that η34 = 0, implies that equations (5) and (6) can be expressed as

which is a modified Ramsey Rule. To elaborate, as pointed out above, the Ramsey Rule requires that all distortable outputs be reduced proportionately.13 In general, in a world of nonzero cross-price elasticities, this rule does not hold if there are pre-existing distortions. Pre-existing distortions imply that the output of the goods associated with those distortions has already been reduced. Accordingly, when introducing new distortions, account must be taken of how these affect the equilibrium quantities of the predistorted commodities. Ceteris paribus, the size of a new distortion will tend to be larger if the commodity on which it is levied is a substitute for the goods with the pre-existing distortions. The new distortion will tend to cause demand to shift back into the predistorted goods. This argument holds in reverse for goods that are complements. If the new distortion is to be placed on a good with a pre-existing distortion, the new distortion is by this line of reasoning reduced since any good is the perfect complement of itself.

Turning to the case at hand, since θ¯3, the pre-existing distortion on financial savings, is zero, one does not have to be concerned, at least within the context of the model, about prior direct reductions in the equilibrium quantity of financial saving. As for the other pre-existing distortion, that on cash balances, θ¯4, since the cross-price effects between the demand for cash balances and the demand for both financial savings C3 and consumption C3 are assumed to be insignificant, one cannot use changes in the “price” of either of the latter two, that is, changes in θ3 or θ4 to mitigate the effects of the pre-existing distortion on cash balances. It is in this sense that the outcome is a modified Ramsey Rule. There is no guarantee that the demand for all distortable commodities will be proportionately reduced. Rather, the rule is that the demand for both consumption C1 and financial savings C3 be proportionately reduced irrespective of the prior reduction in the demand for cash balances, since the latter is unaffected by the new distortions.

Proceeding from a statement of the optimal outcome in terms of quantities to one in terms of prices, it can be seen from equation (11) that the optimal tax rates depend inversely on the own-price elasticities. Concentrating on the optimal tax treatment of financial savings and using the adding up condition implied by homogeneity, equation (10) becomes

Thus, the statement that the optimal tax on financial savings depends on the own-price elasticity for financial savings has been restated in terms of cross-price elasticities. In particular, the magnitude of the cross-price elasticity between financial savings, C3, and savings mediated through unorganized markets, C2, summarized by η32, is important. For large η32,θ2 should be small.

The intuition of this result is clear. The efficiency losses associated with the taxation of savings in organized markets are the larger the greater is the sensitivity of savings flows to changes in the relative rates of return to be earned in organized and unorganized money markets.

Case II: Existence of Financial Repression

The optimal tax treatment of consumption, C1, is unaffected by the existence of financial repression. Since the cross-price elasticity between consumption and financial saving, η13, is taken to be zero, changes in θ1 have no effect on the demand for financial savings and therefore cannot mitigate the effects of the distortion implied by financial repression.

As for the treatment of financial savings itself, equation (10) now applies. In particular, if the degree of financial repression is large θ3 is large) and if the capacity of the government to borrow at the margin at subsidized rates is limited (∂α/∂C3 is small), then the pre-existing distortion on C3 becomes of paramount importance. As a result, θ3 could easily be negative—a subsidy rather than a tax is appropriate.14 The intuition for this result lies in the fact that the distortion associated with the existence of financial repression with the concomitant shifting of savings into other assets is such as to imply that the government should counteract it in spite of the revenue loss.

This is an important case since it may apply to many developing countries. If the further constraint of precluding the use of subsidies is imposed, this case implies the interesting result that consumption (θ3 = 0) is more appropriate than income as a tax base.

Evaluation of Limitations of Optimal Tax Framework

Some further general comments on the relevance and completeness of the above optimal tax modeling exercise are in order. It has already been noted that the model is not a true general equilibrium intertemporal framework. Neither does the model recognize that cash is effectively costless to produce. However, accommodating these discrepancies within a more complete framework is unlikely to change the qualitative results to a great extent. It remains to be considered whether there are other, more significant, discrepancies between the optimal tax framework presented above and the description of the financial structures of developing countries presented in the previous section.

A first observation is that the model has nothing to say about the public policy implications of whether the alternative asset to organized money market assets consists of precious metals, property, investments held abroad, and so on, or of an unorganized money market asset such as that created by Korea’s curb markets. In analytical terms, this appears to be a re-emergence of the question of whether an economy is or is not on its golden-rule path. What individuals view as savings and what constitutes capital from the point of view of society as a whole are not necessarily synonymous. Investments in such activities as land speculation increase the probability that a given country’s capital stock lies below its golden-rule level. This can be modeled in a more complete optimal tax framework as a pre-existing distortion and, while the outcome is not certain (cf., Atkinson and Sandmo (1980)), it would appear to increase the probability that financial savings should be subsidized. Note that this effect is over and above those direct effects described above associated with financial repression.15

A further institutional feature not incorporated in this exercise is the compulsory savings schemes employed by some governments (Datta and Shome (1981)). Introducing this element into the framework above would excessively complicate matters. Given the short-term illiquidity of assets implied by forced investments in funded social security programs, these savings are not perfect substitutes for other forms of savings. That would require the introduction of a new variable into the utility function. This variable would be quantity-constrained, which has implications for the market behavior of other variables (cf., Neary and Roberts (1980)).

The role of forced savings, on the presumption that they are productively invested, is analogous to the role discussed above of curb markets. Their primary influence may not be on the marginal first-order conditions but rather on the discrete issue of whether the economy is or is not on its golden-rule path. The intuitive implications, though they should be expressed with reservations, again seem clear. For example, a country with a large stock of forced savings and no financial repression is more likely to be a country where financial savings should be taxed.

The above reservations modify the analytical framework. More fundamental is the reservation that financial repression also influences the short-run macroeconomic performance of economies. This is a problem that cannot be solved by recommending a specific tax rate—what is required is greater flexibility of the price system, particularly nominal interest rates, rather than second-best tax reform policies.

Relevance of Supply-Side Tax Policies

What are the implications of the above analysis for supply-side policies in developing country circumstances? This depends on the interpretation of what constitutes supply-side policies. Using the interpretation mentioned earlier, viz., that it concerns, in this context, the use of tax policy to stimulate savings, supply-side policies would appear to have a role to play. For example, recommendations to reduce the top marginal income taxes or to exempt completely interest income from taxation would counter some of the effects associated with financial repression.

There is a more general lesson, however. The model and the reservations presented above imply that the optimal tax treatment of financial savings depends critically on the degree of financial repression, which, as shown in Tables 1 and 2 in the Annex to this chapter, varies widely both across countries and over time. Accordingly, there is no single recommended policy such as reducing marginal tax rates. Each country has to be considered individually. This all implies that global supply-side proposals must be carefully evaluated and tailored to what is desirable. Thus, the appropriate tax policy depends not only on the degree to which financial repression and compulsory savings schemes exist but also on whether these latter distortions, and in particular financial repression, are themselves amenable to elimination.

The chapter has so far examined the merits of proposals for tax reform merely in analytical terms. There were few unambiguous qualitative results, since optimal tax rates depend on the magnitudes of various own-and cross-price elasticities. One must look at the empirical literature to see if the recommended tax reforms, which aim to reduce or eliminate the taxation on savings, will actually stimulate savings. Note that this is a somewhat different issue from the one discussed above. The tax reform discussion goes beyond a concern for stimulating the quantity of savings. Many of the recommended reforms are desirable insofar as they mitigate the effects of pre-existing distortions, such as those associated with financial repression. Such a goal could quite easily be attained without a substantial change in the quantity of savings. The topics are, of course, related since, abstracting from the macroeconomic effects, the optimal tax rates depend critically on the own-price elasticity of financial savings (or, which is effectively the same quantity, on the cross-price elasticities between financial and other forms of saving). The empirical results are summarized in Ebrill (Chapter 3). Given data inadequacies, that literature only allows one to make the general statement that financial savings may be price-sensitive. Point estimates are not possible.

III. Conclusions

The main conclusions of this chapter may be summarized as follows. First, when considering the question of the optimal treatment of financial savings in developing countries, a number of factors must be considered. In particular, the degree of financial repression is an important variable.

Second, having determined the particulars of any given case, in general, a number of policy options must then be evaluated. The best reform proposals within the limited context of tax/price policy will probably involve the abolition of financial repression. Should that be precluded, there remains the second-best type of policy recommendations such as those contained in the specific optimal tax framework developed in this paper. A useful benchmark case is one where financial repression exists and outright subsidization of savings is not feasible. In that case, there are grounds for arguing for a consumption as opposed to an income tax.

Third, the precise magnitude of the optimal tax rates depends on the empirical evaluation of the relevant price elasticities. Given the data inadequacies in most developing countries, the existing empirical literature can only indicate that financial savings may be price-sensitive. It cannot afford useful point estimates of the relevant price elasticities.

Fourth, one must recognize that the existence of financial repression in many developing countries has macroeconomic implications. This modifies the optimal tax results. Specifically, the macroeconomic circumstances in the absence of financial reform call for a flexible tax policy.

Fifth, evaluated against all of the above, the recommendations of supply-side economists appear not so much to be wrong as to be inflexible to local conditions. There are occasions where the recommendations are appropriate, but, as a blanket solution to savings problems in developing countries, they are deficient because they do not allow for the heterogeneity of developing country conditions.

Financial Repression in Developing Countries

Table 1 and 2 document the extent of financial repression in a sample of developing countries. The data in Table 2 were obtained as follows.

RealRate of Return (r). This variable consisted of correcting the nominal rate of return (/’) in the sample countries for expected inflation (x). The nominal rates of return for the selected countries arc defined as fellows.

Table 1.Selected Developing Countries; Estimated Real Rates of Interest on Savings Deposits
Real Interest

From Anand G. Chandavarkar, “Some Aspects of Interest Rate Policies in Less Developed F.conomics: The Experience of Selected Asian Countries.” Staff Papers, International Monetary Fund (Washington). Vol. 18 (March 1971), pp. 48-112.

From International Monetary Fund. Interest Rate Policies in Developing Countries, a Study by the Research Department of the International Monetary Fund, Occasional Paper No. 22 (Washington, October 1983). The real interest rate estimates are derived estimates obtained by the formula r = (1 + i)/(1 + π)—1, wherer is the real rate of interest,i is the corresponding nominal rate, and π is the expected rate ot inflation where this was assumed to equal the actual rate of inflation recorded in the relevant year.

From Vicente Galbis, “Inflation and Interest Rate Policies in Latin America, 1967-76,” Staff Papers. International Monetary Fund (Washington), Vol. 26 (June 1979), pp. 334-66

From Anand G. Chandavarkar, “Some Aspects of Interest Rate Policies in Less Developed F.conomics: The Experience of Selected Asian Countries.” Staff Papers, International Monetary Fund (Washington). Vol. 18 (March 1971), pp. 48-112.

From International Monetary Fund. Interest Rate Policies in Developing Countries, a Study by the Research Department of the International Monetary Fund, Occasional Paper No. 22 (Washington, October 1983). The real interest rate estimates are derived estimates obtained by the formula r = (1 + i)/(1 + π)—1, wherer is the real rate of interest,i is the corresponding nominal rate, and π is the expected rate ot inflation where this was assumed to equal the actual rate of inflation recorded in the relevant year.

From Vicente Galbis, “Inflation and Interest Rate Policies in Latin America, 1967-76,” Staff Papers. International Monetary Fund (Washington), Vol. 26 (June 1979), pp. 334-66

Ghana, Deposit rate on six-month deposits at commercial banks. Source: Bank of Ghana, Annual Reports (Accra), various issues.

Jamaica. Deposit rate on time deposits held for more than six months but less than nine months at commercial banks. Source: National Planning Agency, Economic and Social Survey: Jamaica (Kingston), various issues.

Korea. Deposit rate on six-month time deposits. Source: National Bureau of Statistics, Korea Statistical Yearbook (Seoul), various issues.

Singapore. Deposit rate on six-month deposits. Source: Department of Statistics, Yearbook of Statistics (Singapore), various issues.

Where a change in the institutionally mandated rate occurred within a calendar year, an average rate for that year is presented. Expected inflation is gauged by actual changes in the annual rate of inflation as measured by series 64 in International Monetary Fund, International Financial Statistics Yearbook. 1983. The formula employed is r = (1 + i)/ (1 + π)—1.

Table 2.Selected Developing Countries: Nominal (i) and Real (r) Interest Rates on Time Deposits, 1954-82 (In percent)
Note:… = not available.Sources: As cited in the text of this Annex.
Note:… = not available.Sources: As cited in the text of this Annex.

    AtkinsonA. B. and A.Sandmo“Welfare Implications of the Taxation of Savings,”Economic Journal (London) Vol. 90 (September1980) pp. 52949.

    • Search Google Scholar
    • Export Citation

    AtkinsonA. B. and Joseph E .StiglitzLectures on Public Economics (New York: McGraw-Hill1980).

    AuerbachAlan J. and Laurence J.Kotlikoff“National Savings, Economic Welfare, and the Structure of Taxation,”inBehavioral Simulation Methods in Tax Policy Analysised. byMartinFeldstein(Chicago: University of Chicago Press1983).

    • Search Google Scholar
    • Export Citation

    BuffieEdward F.“Financial Repression, the New Structuralists, and Stabilization Policy in Semi-Industrialized Economies,”Journal of Development Economics (Amsterdam) Vol. 14 (April1984) pp. 30522.

    • Search Google Scholar
    • Export Citation

    ChandavarkarAnand G . “Some Aspects of Interest Rate Policies in Less Developed Economies: The Experience of Selected Asian Countries,”Staff PapersInternational Monetary Fund (Washington) Vol. 18 (March1971) pp. 48112.

    • Search Google Scholar
    • Export Citation

    DattaGautam and ParthasarathiShome“Social Security and Household Savings: Asian Experience,”Developing Economies (Tokyo) Vol. 19 (June1981) pp. 14360.

    • Search Google Scholar
    • Export Citation

    DrazenAllan“The Optimal Rate of Inflation Revisited,”Journal of Monetary Economics (Amsterdam)Vol. 5 (April1979) pp. 23148.

    • Search Google Scholar
    • Export Citation

    EbrillLiam P. and S . M .Slutsky“The Ramsey Rule, Production Efficiency, and Intermediate Goods” (New Orleans, Louisiana: Murphy Institute of Political Economy, Tulane University, 1983, mimeographed).

    • Search Google Scholar
    • Export Citation

    FeldsteinMartin“The Welfare Cost of Capital Income Taxation,”Journal of Political Economy (Chicago) Vol. 86No. 2Part 2 (April1978) pp. S29S51.

    • Search Google Scholar
    • Export Citation

    FullertonDonJohn B .Shoven and JohnWhalley“Replacing the U . S . Income Tax with a Progressive Consumption Tax: A Sequenced General Equilibrium Approach,”Journal of Public Economics (Amsterdam) Vol. 20 (February1983) pp. 323.

    • Search Google Scholar
    • Export Citation

    GalbisVicente“Inflation and Interest Rate Policies in Latin America, 1967-76,”Staff PapersInternational Monetary Fund (Washington) Vol. 26 (June1979) pp. 33466.

    • Search Google Scholar
    • Export Citation

    KaldorNicholasAn Expenditure Tax (London: Allen & Unwin1955).

    LeffNathaniel H . and KazuoSato“Macroeconomic Adjustment in Developing Countries: Instability, Short-Run Growth, and External Dependency,”Review of Economics and Statistics (Cambridge, Massachusetts) Vol. 62 (May1980) pp. 17079.

    • Search Google Scholar
    • Export Citation

    McKinnonRonald I.Money and Capital in Economic Development (Washington: Brookings Institution1973).

    MillJohn StuartPrinciples of Political Economyed. by W. J.Ashley (London: Longmans1921).

    MiracleMarvin P.Diane S.Miracle and LaurieCohen“Informal Savings Mobilization in Africa,”Economic Development and Cultural Change (Chicago) Vol. 28 (July1980) pp. 70124.

    • Search Google Scholar
    • Export Citation

    MunkKnud Jorgen“Optimal Taxation and Pure Profit,”Scandinavian Journal of Economics (Stockholm) Vol. 80No. 1(1978) pp. 119.

    • Search Google Scholar
    • Export Citation

    NearyJ.P. and K. W. S.Roberts“The Theory of Household Behaviour Under Rationing,”European Economic Review (Amsterdam) Vol. 13 (January1980) pp. 2542.

    • Search Google Scholar
    • Export Citation

    NellorD. C. L. “Tax Policy, Regulated Interest Rates, and Saving” (unpublished, International Monetary Fund,August81983).

    PhelpsEdmund S.“Inflation in the Theory of Public Finance,”Swedish Journal of Economics (Stockholm) Vol. 75 (March1973) pp. 6782.

    • Search Google Scholar
    • Export Citation

    PrestA. R. “Compulsory Lending Schemes,”Staff PapersInternational Monetary Fund (Washington) Vol. 16 (March1969) pp. 2752.

    • Search Google Scholar
    • Export Citation

    SandmoAgnar“Optimal Taxation—An Introduction to the Literature,”Journal of Public Economics (Amsterdam) Vol. 6 (July-August1976) pp. 3754.

    • Search Google Scholar
    • Export Citation

    SeidmanLaurence S.“Conversion to a Consumption Tax: The Transition in a Life-Cycle Growth Model,”Journal of Political Economy (Chicago) Vol. 92 (April1984) pp. 24767.

    • Search Google Scholar
    • Export Citation

    ShawEdward S.Financial Deepening in Economic Development (New York: Oxford University Press1973).

    ShomeParthasarathi and LynSquire“Alternative Mechanisms for Financing Social Security,”World Bank Staff Working Paper No. 625 (Washington,1983).

    • Search Google Scholar
    • Export Citation

    SiegelJeremy J.“Notes on Optimal Taxation and the Optimal Rate of Inflation,”Journal of Monetary Economics (Amsterdam) Vol. 4 (April1978) pp. 297305.

    • Search Google Scholar
    • Export Citation

    SummersLawrence H .“Optimal Inflation Policy,”Journal of Monetary Economics (Amsterdam) Vol. 7 (March1981) pp. 17594.

    Tun WaiU“A Revisit to Interest Rates Outside the Organized Money Markets of Underdeveloped Countries,”Quarterly ReviewBanca Nazionale del Lavoro (Rome)Vol. 30 (September1977) pp. 291312.

    • Search Google Scholar
    • Export Citation

    van WijnbergenSweder“Stagflationary Effects of Monetary Stabilization Policies: A Quantitative Analysis of South Korea,”Journal of Development Economics (Amsterdam) Vol. 10 (April1982) pp. 13369.

    • Search Google Scholar
    • Export Citation

    van WijnbergenSweder(1983a) “Credit Policy, Inflation and Growth in a Financially Repressed Economy,”Journal of Development Economics (Amsterdam) Vol. 13 (August-October1983) pp. 4565.

    • Search Google Scholar
    • Export Citation

    van WijnbergenSweder(1983b) “Interest Rate Management in LDC’ s ,”Journal of Monetary Economics (Amsterdam) Vol. 12 (September1983) pp. 43352.

    • Search Google Scholar
    • Export Citation

    WanniskiJudeThe Way the World Works (New York: Simon and Schuster, rev. and updateded.,1983).

    WilliamsonJeffreyG .“Why Do Koreans Save ‘So Little’?”Journal of Development Economics (Amsterdam) Vol. 6 (September1979) pp. 34362.

    • Search Google Scholar
    • Export Citation

This begs the question of whether a more direct response would be to index for inflation.


Table 1 and 2 in the Annex show that financial repression varies both across countries and across time. The phenomenon is not as prevalent today as formerly. However, that only serves to emphasize the heterogeneity of developing country financial policies.


How important are these curb markets? Tun Wat (1977) argues that the ratio of total agricultural or rural indebtedness to the claims of the banking system on the private sector is the best index of the size of unorganized money markets relative to that of the organized money market. By that criterion, he concludes that for India. Nepal, and Pakistan the unorganized markets are larger than the organized markets. (The data he collects refers to 1969/70, 1970, and 1971, respectively.) He finds that, in general, these markets are less important in those Latin American and Middle Eastern countries for which data were available. Chandavarkar (1971) confirmed these results for India, while, of course, the importance of curb markets in Korea has long been recognized (see, for example, Williamson (1979)). On this last case, van Wijnbergen (1982) uncovered evidence, some indirect, to the effect that increased savings are mainly channeled into the curb market over time. Specifically, he discovered that wealth effects were not significant in his M2 demand equations. Since financial assets assuredly have positive wealth elasticities in the aggregate, curb markets must have been the beneficiary of increased wealth.


From the perspective of this study, long-term schemes such as compulsory funded social security systems are the most important. However, as documented by Prest (1969), many countries have also resorted to shorter-term schemes.


This issue is logically different from that of how best to Stimulate savings. However, to the extent that the literature provides qualitative results, it does provide a benchmark against which the desirability of stimulating savings can be weighed.


This is implicitly an implication of the Ramsey Rule (Ebrill and Slutsky (1983)). Under the circumstances laid nut by Feldstein (1978), a labor income tax is optimal and is equivalent to a uniform consumption tax and vice versa. Accordingly, a consumption tax is all that is required to guarantee a compensated proportionate reduction in all distortable commodities where this is precisely a statement of the Ramsey Rule.


Other savings can include savings held abroad.


This treatment avoids the complications associated with the existence of profit income influencing firm behavior at the margin (Munk (1978)). It also avoids complications that might arise from noncompetitive producer behavior.


Note that all variables are defined in real terms. The economy is assumed to be in an initial equilibrium where there is a positive underlying rate of inflation with all relative prices being constant. The tax changes are assumed not to affect the underlying rate of inflation.


Note that such a framework would also make the consumer’s maximization problem more dynamic.


Care should be taken in interpreting θ3 and θ¯3.θ3 = t3/P3, while θ¯3 = t¯3/P3, where P3 is inclusive of the pre-existing distortion. Thus, for the new (derivative) distortions, the shares are in terms of original prices, whereas, for the pre-existing distortions, they are expressed in terms of final prices.


This immediately raises the issues of why the government does not absorb all of C3 from the outset (α = 1). It is assumed that there are some constraints on the government and that, further, the authorities may well view interest rate ceilings as a device for subsidizing private investment.


To be more precise, the Ramsey Rule specifies a reduction along the tangent plane approximation to the compensated demand curve for the good in question due to the distortions. This approximate reduction will equal an actual reduction if the derivatives of the compensated demand curves are constant or if the approximation is taken for small distortions in the neighborhood of zero revenue requirements.


A parallel result can be found in Nellor (1983). In that paper, while considering the positive question of how precisely to counter inflation, he points out that capital income subsidies (or their equivalent in terms of other tax instruments) may well be necessary.


The optimal outcome depends, of course, not only on the nature of the alternative asset but also on its institutional context. In the case where, for example, one of the alternative assets consists of investments held abroad, the attitude of the authorities toward external capital flows is important. For our purposes, the influence of this type of consideration will be captured in the magnitude of the relevant compensated price elasticity.

    Other Resources Citing This Publication