Because official credit programs offer more lenient terms to borrowers than are available in the market, or in many cases than those at which the government itself borrows, they contain a pure loan component, reflecting the government’s role as a financial intermediary, and a pure grant component, reflecting the government’s role as a distributional agent. Because interest paid and received both appear above the line in the unified cash budget, the annual cost of these subsidies is reflected in the observed fiscal balance but is nowhere identified overtly.1 But because these subsidies are spread over the entire lifetime of the loans, their total magnitude is easily overlooked and usually grossly underestimated, and they contribute to the creation of long-term structural deficits that restrict the short-term flexibility of fiscal policy.

Because official credit programs offer more lenient terms to borrowers than are available in the market, or in many cases than those at which the government itself borrows, they contain a pure loan component, reflecting the government’s role as a financial intermediary, and a pure grant component, reflecting the government’s role as a distributional agent. Because interest paid and received both appear above the line in the unified cash budget, the annual cost of these subsidies is reflected in the observed fiscal balance but is nowhere identified overtly.1 But because these subsidies are spread over the entire lifetime of the loans, their total magnitude is easily overlooked and usually grossly underestimated, and they contribute to the creation of long-term structural deficits that restrict the short-term flexibility of fiscal policy.

This paper examines the role of credit subsidies in government direct lending. The emphasis is on the presentation of a simple technique to measure the actual financial cost to the government of these subsidies.2 The discussion focuses on an important way by which governments intervene in the financial intermediation process—namely, through direct provision of loans—and the fiscal implications of this type of intervention. The scale of such lending operations is not widely appreciated. Even in a strongly market-oriented economy, such as that of the United States, the federal government is the largest financial intermediary—with outstanding loans at the end of 1987 of $234 billion.3 Governments can, however, and do, intervene in more or less pervasive ways, with greater or lesser fiscal implications. The central purpose of this introductory section is to describe some of the other forms of intervention, which are beyond the scope of this study, and how some of them are related to subsidized official lending.

Examples of more pervasive interventions include a government’s ownership of the banking system, or its imposition of differential deposit and lending rate ceilings throughout the financial system to the relative advantage of certain sectors. Depending upon how these credit subsidy schemes were financed, they would generate either a larger or smaller supplementary fiscal deficit, which should be added to the conventionally measured deficit to capture the true aggregate demand impact of fiscal operations.4 As this is a far more intractable issue operationally, it will not be addressed here. Similarly, government loan guarantees, which do not affect the cash deficit unless default occurs but which do involve substantial subsidies and governmental intervention in credit markets, are beyond the scope of this study.5

Finally, it should be emphasized that the method employed below to separate official lending into its “pure loan” and “cash grant” equivalents is limited to the expenditure side of the budget. This subsidy mechanism on the outlay side is, in principle at least, completely interchangeable with the operation of so-called tax expenditures on the revenue side that result from the tax deductibility of interest income and payments.6 The two methods, however, are neither equivalent in practice nor in their budgetary treatment. In particular, the relative size of the government sector in the economy would always be smaller if tax expenditures were used. An example may clarify the differences; for present purposes, the complications of maturities and discounting are dispensed within the tax expenditure case, since the effect of these is the same from both the tax and expenditure perspectives.

First, assume that the government, which pays 7 percent on its borrowings of similar maturities, makes a direct loan of $100 million to a final borrower at 3 percent interest for 15 years, while the best market rate available to this borrower would have been 10 percent on a similar loan. The total amount of the subsidy is 7 percent annually. This can be further divided into two parts. The explicit portion (4 percent a year) represents the differential amount the government must actually pay; the implicit portion (3 percent a year) includes the additional benefit received by the borrower, compared with his opportunity cost.7

The focus in this paper is on the explicit portion of the subsidy. Assuming a simple annuity structure, the direct cost to the government of this arrangement would involve an interest outlay of just over $39 million in excess of interest income over the period of the loan. Stated in terms of present value of the total debt service (discounted at 7 percent), such a credit subsidy would be worth about $23.7 million, which is defined as the subsidy (or grant) value of the loan.8 Alternatively stated, at current government interest rates, the loan could be sold by the government to private sector lenders in a secondary market for $76.3 million. In this case, the government is agreeing to a “concealed” subsidy of $23.7 million when it enters into the loan contract. It is committing itself to a deficit in this loan operation in the first year, and for the subsequent 14 years. Of course, the government would not have to finance the implicit subsidy directly, but it would be offset economically by efficiency losses owing to the associated misallocation of resources in the economy.

Alternatively, the same subsidy operation could have been financed by forgone tax revenues. That is, the tax system could have been adjusted so that interest income to the lender was tax exempt. Lenders in the 70 percent tax bracket or above (in the example) would then have found it profitable to lend at 3 percent tax exempt, even though the borrowers’ opportunity cost rate was 10 percent. Whether such lending would actually occur depends on the other alternatives available to the prospective lenders. In any case, the direct cash cost to the budget in this case would be the amount of tax revenue forgone, and, even more than in the previous case, this is not clear under the usual accounting conventions. Once again, the real economic cost of such a program would usually be greater than the realized financial cost to the government. These additional costs are inherently more difficult to measure directly because they depend on the borrowers’ opportunity costs, which are usually not observable, and, because they are funded by resource misallocations and inefficiencies, they cannot readily be measured indirectly from the financing side.

In fact, most governments operate both types of direct lending and tax expenditure schemes simultaneously, so that a comprehensive measure of official credit subsidies should take both mechanisms into account. However, because this paper focuses on official lending and its terms, credit subsidies based on tax expenditures are beyond the scope of this study.

Computation of Budgetary Credit Subsidies

No generally accepted objective method exists for estimating the subsidy value in official direct lending programs, mainly because of the difficulty of precisely establishing the private rates that would have been paid by borrowers in private markets without government intervention. Moreover, for some programs—for example, those addressing a “total” market failure—there may be no alternative private rate at all. Still, it is the premise of this paper that the concealed subsidies are usually so great that an attempt at estimation must be made, even if the resultant measure is not precise. By focusing on the explicit portion of the total subsidy in applications, the most intractable of the operational obstacles can be avoided at the final-borrower stage. The resultant estimate is, of course, biased downward in terms of the value received by the borrower because it omits the implicit portion of the subsidy. In practice, it may also be preferable to consciously underestimate the calculated explicit subsidy so that the direction of bias in the answer is known.

Measurement of the Subsidies

A method for estimating the value of the credit, or interest, subsidy is first presented in the simplest of cases, then complications are introduced, and finally issues relevant to the operational use of the technique are discussed.

Conceptual Framework

First, assume that a loan of amount A is made at interest rate i; it is disbursed in full immediately and amortized over N years, to be repaid in equal annual installments of principal plus interest. If P is defined as the value of this annuity, then






The variable DN is defined as the discount factor that gives the present value of one unit (if the loan above is for A dollars, then it is one dollar) payable yearly for N years. Alternatively, 1/DN is the annual payment necessary to pay off a loan of one dollar over N years.

Analogously, if the same loan were made at market rate i*, then under market conditions the equation would be


Since the presumption is that i* > i, the subsidy element each year is


If both loans, however, were made in terms fixed until maturity, then 5, which gives the annual subsidy, grossly understates the present value of the total subsidy involved in making the loan at the lower rate because this would accrue every year, for N years. Thus, it is necessary to calculate the capitalized value of this stream of annual differences in payments, where the discount rate used is based on the market (or “true” opportunity) cost of capital:


where Δ = the grant or subsidy value in the loan.

The proper interpretation of Δ is that the two situations are identical in the following sense: it is precisely equivalent either to grant the N-year loan at rate i when the opportunity cost rate is i* or to grant the N-year loan at rate i* and to provide a cash grant of Δ. Thus, provision of a “low-interest” loan is equivalent to providing a “pure loan” combined with a “pure subsidy.” Alternatively, one can think of the reduction of i below i* as inducing an excess expansion of credit in the amount of Δ, which is defined as the credit subsidy.

Another form of presentation is a table of cash flows (Table 1) designed to obtain the present value of the total debt payments.9 This is useful because it allows the structure of the loan to be changed easily, so that the new subsidy value can be computed. Such a technique is necessary when attempting to sum subsidies across a wide variety of lending programs, because lending terms frequently differ dramatically. Education or construction sector loans, for example, often involve grace periods, whereas housing loans rarely do.

Table 1.

Cash Flow of Simple Annuity Loan

article image

For example, if A = $10,000,000, i = 10 percent a year, N = 20 years, and i* = 12.5 percent a year, then Δ = $1,494,334, which is 14.9 percent of the loan’s face value.

Now, if the above loan were structured so that the payback period were preceded by a grace period (of n years) during which nothing were payable, then the cash flow table would appear as shown in Table 2.

Table 2.

Cash Flow of Loan with Full Grace Period

article image

The subtraction of the discount rates may appear confusing in years n to n + N. In the former example, DN* was used to obtain the present discounted value of the stream of net payments for the simple loan. The problem now is just that this stream occurs n years in the future, so it must be discounted again by 1/(1 + i)n:






Continuing the previous example, if n = 5 years, the new subsidy value is $5,280,002, or 52.8 percent of the loan’s face value. Thus, the addition of a “pure” grace period can be seen to affect the subsidy calculation dramatically, raising the subsidy element by 38 percent of the loan’s face value.

Consider next a loan with a grace period of n years, but only with interest payable during the grace period at rate in. In this case, the cash flows would be as shown in Table 3.

Table 3.

Cash Flow of Loan with Interest-Only Grace Period

article image

To continue the example, if in = iN = 10 percent a year, then the subsidy value would be $1,719,391, or 17.2 percent of the loan’s face value. Alternatively, if in < iN, say in = 8 percent a year in this example, then A becomes $2,431,505, or 24.3 percent of the loan.

Note that the results in this extended example (Table 4) conform to an intuitive a priori impression about the relative “softness” of the loan terms.10

Table 4.

Comparative Subsidies in Different Loan Structures1

article image

Based on the extended example in the text.

Finally, consider a case in which the loan is set up like a bond—that is, interest only is payable throughout the life of the loan and the principal is repaid in one lump sum at the maturity date (Table 5).

Table 5.

Cash Flow of Bond-Type Loan

article image

Although a fairly simple loan structure has been retained for exposition, the principles involved remain the same, even if loan structures are different and more complex. Most generally stated, the subsidy value of a loan is the difference between the present value of its disbursements and the present value of its service payments, discounted at the market rate of interest. The grant element is defined as the value of the subsidy or grant as a percentage of the present value of the disbursements. According to these definitions, therefore, a loan made at the market rate of interest carries a subsidy value and grant element of zero, while a pure grant has a subsidy element of 100 percent. For a “soft” or “concessional” loan, the grant element lies somewhere between these extremes.

The Development Assistance Committee (DAC) of the Organization for Economic Cooperation and Development (OECD) and the External Debt Division of the World Bank regularly compute the degree of concessionality in their own and other foreign lending by computing the grant element (based on a fixed discount rate of 10 percent a year) and simply defining any loan for which this value is greater than 25 percent as a concessional loan.11 Most lending covered is assumed to be structured around equal principal repayments.

The major difficulty with the DAC and World Bank procedure is that the unchanged discount rate of 10 percent a year is too arbitrary and rigid. Other rates would have been (and will be) more appropriate at various times, depending on the conditions prevailing in world capital markets. For example, in the early 1980s, the 10 percent rate clearly was much too low and presumably resulted in significant underestimates of concessional lending. Moreover, the massive volume of loans analyzed by these organizations requires that simplifying assumptions be made regarding the uniform structuring of loans, particularly that all loans are disbursed immediately. Calculations done by the author suggest, however, that grant elements—and particularly, grant equivalents (the subsidy values)—are quite sensitive to the way loans are structured. Therefore, elements of error (of unknown magnitude) are introduced when the analysis is not done on a disaggregated basis in order to retain a high degree of accuracy. Since the emphasis of these organizations is on the grant element, the problem is less critical than when the focus is on calculating subsidy values. Both of these difficulties, however, introduce an element of uncertainty into the published concessional lending figures and make unambiguous interpretation of them rather difficult.

Before turning to a discussion of the problems that arise in practice, mention should be made of the general relationship between the subsidy element of a loan and the interest spread between its quoted rate and the opportunity cost interest rate. The regularity and shape of this relationship have important implications for applications of the subsidy estimation technique and for interpretation of the results. Chart 1 shows the subsidy (or grant) element as a function of this spread for a bond, where A = 100, i = 15, and n = 20. This function crosses the zero axis when the opportunity cost rate of interest (i*) is set equal to the effective rate of interest (ERI) on the loan (see below), which in this case also equals the quoted rate. When i* is below the loan’s ERI, then profits can be made in the government’s borrowing-cum-lending operations. This is indicated in the figure by negative subsidy rates at these levels.12 Alternatively, when i* rises above the ERI on lending, then positive subsidy rates are implied; the higher the i*, the greater is the grant element of the loan. Note, however, that the relationship is highly nonlinear. Even a relatively small interest spread (5 percent) gives a reasonably large (25 percent) subsidy element. Similarly, once a certain spread has been attained (say, 20 percent, i.e., when i* equals 40 percent), then increases in the spread do not give rise to much additional subsidy flows. In the limit, of course, when i* is infinitely high, the subsidy element approaches its maximum limit of 100 percent.

Chart 1.
Chart 1.

Subsidy Element and Interest Spread

Chart 2, which presents the same relationship for different types of loan structures, shows that this shape is not unique to the bond structure assumed. Specifically, this figure plots the examples given earlier in this section: for A = 10,000,000; i = .10; N = 20; and i* = .125; A1 plots the simple amortized loan, and A3 the bond. The same general terms are assumed in plotting A4 but a grace period of five years is added and the equal payments structure is used. Amortized loans with a grace period are shown in A2.A, A2.B, and A2.C; the same general terms are assumed, but the five-year grace period added is characterized differently: it is assumed in A that full interest (0.10 percent a year) is paid during the grace period; in B that reduced interest (0.08 percent a year) is paid; and in C that nothing is paid.

Chart 2.
Chart 2.

Comparative Patterns of Subsidy Element and Interest Spread

Complications in Applications

The preceding section glosses over a number of difficulties that need to be addressed when the method is applied. It is necessary to know a great deal about specific program borrowers to properly estimate the income transfers, the impacts on credit allocation, and the total interest subsidies involved. As Break has noted, official lending operations conducted with a fixed supply of total credit result in income transfers to inframarginal borrowers of the difference between the interest expense they would have incurred on private credit and that paid to the government, while the amount transferred to submarginal borrowers is whatever income is yielded by their projects after repayment of their government loans. Moreover, loans to this latter group create a real transfer of credit, while lending to the former involves no reallocation of credit at all, unless these borrowers incur more debt than they otherwise would have in private markets. Therefore, precise quantification of the economic role of official lending operations requires reliable information about the credit status of the borrowers, the alternative interest rates they would have paid in private markets, the interest rate elasticity of their credit demands, and the profitability of their projects.13

Of course, most of these data do not exist, and their estimation over a wide spectrum of groups at the microlevel is not practical. Furthermore, even if available, their usefulness would be limited by several factors that will be discussed in the following section (e.g., different stages and lag patterns among programs, etc.). Thus, it seems inevitable that if any useful descriptive figures are to be generated at the aggregate level, while minimizing the data requirements and assuring some simplicity in the analysis, a less ambitious goal than estimating the total subsidies must be accepted. In these circumstances, it makes sense to focus on the explicit subsidy transfer alone and forgo the desirable aim of measuring the larger flows and real impacts.

This focus solves some, but not all, of the difficult operational problems. If the method is applied directly to government lending, taking the variables A, i, N, and n, from observable official loans and using the government’s marginal borrowing rate for i*, then the major assumption required is that the borrowers are inframarginal relative to the government’s borrowing rate. This seems to be true for most capital-scarce countries.

Chart 3 helps to set the context of this discussion, depicting the explicit subsidy measure as well as the inframarginality assumption. If dD and SS are the market demand and supply schedules, then G would be the free market solution without government intervention; OC of credit would be extended at rate iFM. If the government decides to make D’D of credit available and selects recipients at random from among all those who demand credit at the government’s subsidized lending rate (iLG), then the fraction D′D/iLGD of all those demanding credit at any rate of not less than iLG receive official subsidized loans, with the remaining demand satisfied in the unassisted private market along D’d. However, D’d determines only the residual quantity of credit supplied in the market. To fix the rate that unassisted borrowers must pay to private lenders, the total demand for credit must be obtained by adding back in the quantity channeled to the subsidized borrower, D’D. The intersection of this combined schedule (d’D) with the original supply schedule determines the rate for unsubsidized borrowers (iFM) who demand OA of credit, out of a total of OE, where AE equals D’D. In most applications, iFM would not be directly observable, but it would be some weighted average of rates in various markets—for example, in many developing countries on the unorganized money market and in the banking system. The value of the total interest subsidy (explicit plus implicit) is represented by the area IGDD’.

Chart 3.
Chart 3.

The Explicit Subsidy Measure

Source: Adapted from von Furstenberg (1976), Figure 1.

If iBG represents the government’s marginal borrowing rate, then YBDD’ represents the value of explicit credit subsidies received by government loan recipients. In most applications, the present value of this area would be approximated by the present value of the slightly different area ACDD’ (equals TAD’X) on the assumption that all borrowers are inframarginal to the government’s borrowing rate iBG (i.e., all lie on demand schedule dD above B).

In terms of practical applications, therefore, selection of the government’s marginal borrowing terms remains as the final major operational issue, and this issue is far from trivial because the results are very sensitive to variation in the i*’s.14 The major alternatives available are the terms on official external and internal borrowing. In general, the internal rate, for example, a government bond yield, should be preferred because it entails the same currency in which the lending occurs, thus minimizing errors that might be introduced through incorrect estimates of exchange risk.15

In applications, however, economists may have to use an external borrowing rate because a domestic counterpart is unavailable. But even though governments still issue a large amount of fixed term foreign bonds,16 this market may be effectively closed to some countries. They may have access only to floating rates, or to adjustable rate markets (e.g., the Eurodollar market). In this case there are three alternatives available, none of which is entirely satisfactory. First, the observed rate could be used, assuming that it would not change, at least on average, over the period of analysis. Second, forecasts could be made of expected changes in the rate, perhaps by using different values to determine a range of subsidy values. Finally, if there seemed to be too much uncertainty in either of the former strategies, it might be preferable to forgo estimation of the capitalized value of the subsidy stream and simply to do the analysis on an annual basis. Computationally, this is much easier as it involves only the calculation of the service payments on outstanding loan balances at the mean lending rate and again at the mean borrowing rate, and then the subtraction of one from the other. The difference is the subsidy value for that particular period.17

It is well known that unanticipated inflation can result in a shift in real income from lenders to borrowers as the real value of debt contracted in fixed nominal terms is eroded. However, if the government is both borrowing and lending at fixed rates over similar periods then, as a first approximation, it can be assumed that there are no net inflationary effects on the subsidy calculations because what is lost on the lending side is gained on the borrowing side. Similarly, if it is assumed that nominal interest rates are adjusted to incorporate fully actual inflationary developments under flexible rate loans, then more rapid rates of amortization will be implied than would have occurred under comparable fixed rate loans.18 But again, if the government is both borrowing and lending in this way, then the net effect on the subsidy transfer should largely net out. Only if there were some mixed combination of structures would there seem to be a significant impact on the real value of the subsidy transferred through the government. The more common configuration probably would be the case of foreign borrowing at variable rates, but domestic lending on fixed terms. In this case, an increase would be expected in the real value of the subsidy transferred through the government due to inflation, and calculations based solely on nominal values would, therefore, underestimate the real transfer. Since the capitalized value of the subsidy stream is being computed, attempts to correct for these effects would have to include inflation forecasts over the relevant time horizon.

Effective Rate of Interest

The most important operational issue in the foregoing analysis is selection of the correct opportunity cost interest rate, i*. In some cases, the correct choice of the discount rate may be so difficult that it cannot be done with any confidence. Or, it might be that all rates are floating and an annual type of analysis must be used. In these instances, an alternative approach that does not involve a discount rate, that is, the ERI, might be preferred. The ERI bears the same relationship to the former analysis as the calculation of the internal rate of return does to the net present value in project analysis.19 The analogy results because a loan is just like a “backward project” in the sense that the benefits of the “project” (loan disbursements) come initially, while the costs (debt service payments) are spread over the lifetime of the loan.

The ERI is defined as that rate of interest which, if used as a discount rate, would reduce the net present value of the loan to zero. Thus, the ERI is the same as the rate of interest on any loan on which interest must be paid on outstanding balances at all times (and on which there are no other costs payable).20 Therefore, there may be no need to calculate the ERI because it is the same as the quoted rate on the loan. However, when loans include special arrangements, such as grace periods, during which payments are reduced below interest at the quoted rate, then the ERI diverges from this quoted rate and thus must be calculated directly.

Calculation of the ERI involves solving the following equation for r:



article image

Estimation of credit subsidies using ERIs is done in the same way as was described above for annual subsidies. The ERIs in both government borrowing and lending are calculated, then, the cost of servicing the outstanding balance of loans is calculated for each. The difference between the borrowing and lending cost is the subsidy.

Rationale and Economic Effects

Official credit programs are usually claimed to be necessary to correct failures in private capital markets. Such failures have been assumed to exist wherever potential borrowers cannot acquire credit at a “reasonable” cost. This may be due to inadequate flows of information, making risk assessment difficult and inaccurate; it may be due to monopolistic elements in the intermediation process or to other factors that inhibit the mobility of capital;21 it may be due to inherent flaws in specific security instruments;22 or it may result from the complete absence of financial institutions, as is sometimes the case in rural areas.

Related rationales (which also could justify other budgetary expenditures) are those relating to the exploitation of economic externalities, pursuit of social goals (including the stabilization of the economy), and alteration of the market-determined distribution of income. The first type of rationale is particularly prevalent in developing countries because it arises naturally in the planning context. Frequently, for example, governments attempt to divert credit into sectors when they are believed to generate more backward and forward linkages in the economy, or where there may be other reasons for the divergence between private and social costs, such as that arising in the field of education which has prompted the establishment of official student loan programs.23

Although any or several of these reasons may legitimately give rise to a specific official credit program, many such schemes still exist long after the original conditions that motivated their adoption have changed. When such programs operate in areas where there are no longer significant market failures, the official assistance is best understood as a reallocation of credit, usually at subsidized interest rates, to specific activities or borrowers.24 A large part of these flows also may be pure income transfers because the loans would have been made anyway by the private market, but at significantly higher rates of interest. This process undoubtedly results in important sacrifices of economic efficiency, as it siphons credit away from other uses that have stricter risk or return criteria, while delivering it to users selected, at least partially, on the basis of noneconomic criteria.

One of the major recognized economic effects of many official lending programs is, therefore, some sacrifice in the rate of economic growth. As suggested above, however, this does not necessarily follow when social and private returns are not equal, because then a well-designed official credit program could encourage investments with high social returns, even though private returns were relatively low. But even if government officials were able to select projects with high social (but low private) returns, it can be argued that interest rate subsidies are inefficient instruments in this context because they also distort factor prices in the process.25 If government intervention in credit markets were substantial, these distortions would encourage inappropriate capital-intensive production techniques in existing production processes and movement into new products that were more capital intensive. Thus, the economy would start down the wrong technological path, making subsequent reversals of development strategy more difficult. The opposite side of the same issue is, of course, that such “successful” credit programs increase unemployment, frequently in economies with a surplus of labor.

Relatively cheap loans over significant periods of time can also profoundly affect the financial structure of private enterprises in the sectors involved, creating, for example, relatively high ratios of debt to equity. This configuration of the corporate sector’s balance sheet can then inhibit the free conduct of monetary policy, since an abrupt raising of interest rates could bankrupt the business sector if rates on outstanding debt were adjustable.26

While one might legitimately wonder why it is necessary to be concerned about these matters, given the growth performance of some countries that have high debt-equity ratios (e.g., Japan and Korea), the argument here relates to the riskiness or vulnerability of the resultant structure—that is, extremely high debt-equity ratios may not be a problem during rapid inflation and growth, but they may become a severe handicap if growth and inflation slow quickly, or if the cost of capital rises abruptly. Moreover, there is the question of what the growth rate might have been without the distortions.

These results are somewhat impressionistic because surprisingly little empirical work has been done on the real economic effects of direct government lending programs. Much of what is known is based on the experience of the United States and is presented in an excellent survey article by Aragon.27

The available evidence does not permit firm conclusions, even for the United States, and, of course, what does seem evident may not be readily transferable to widely different economies that are characterized by significantly smaller and less unfettered capital markets. The major uncertainties concerning the economic impact of official credit programs are summarized by Aragon into two basic questions: (i) Do official credit programs produce lasting changes in the composition and volume of credit? (ii) Do such changes produce predictable shifts in the allocation of economic resources?

Earlier studies, covering the period from the mid-1950s to the mid-1960s concluded that the answer to both questions was yes; however, these studies analyzed only the first-round effects of the lending programs and not the various financing and portfolio reactions of private agents that tended to offset the original flow of credit. Studies adding these effects into the analysis (carried out during the mid-1970s) produced conclusions that contradicted the earlier work with regard to most aggregate, long-run effects, although the short-run effects, which were significant, were similar. Other studies found that, even when changes were produced in the overall composition of credit, subsequent transformations of real assets frequently did not occur. For example, one study found that mortgage loans financed a variety of financial assets and other nonhousing real assets.

With regard to stabilization issues and the contribution toward full employment, Aragon concludes that the effect of credit programs seems to vary according to the specific goals and assumptions of particular schemes and, for all purposes, the stance of monetary policy. Specialized program objectives, such as income redistribution, often conflict with maximization of economic efficiency and growth, as already suggested. Programs aimed at improvement of capital market efficiency and the provision of high-risk capital, however, have sometimes promoted innovation, investment, and growth. Increased spending was found most likely to occur when credit was extended to marginal or needy borrowers, especially in market-perfecting programs directed at small businesses. These, however, were only the initial results. The final impact on overall spending depended crucially on the level of accommodating monetary expansion.

When the money supply remained relatively fixed, federal credit activities in the United States simply resulted in displacement of private lenders and borrowers—that is, “crowding out” occurred.28 This tended to offset any expansionary impact.29 Additional national income tended to result only when the money supply was expanded. Moreover, there usually has been an important asymmetry between credit-program crowding out and budgetary deficit-financing crowding out. Budget financing needs have generally been highest in periods of deep recession, while those credit programs have been greatest during periods of expansion. Therefore, the probability of crowding out effects would seem to be greater for credit than for budget financing, although this would not necessarily be the case if large fiscal deficits were accompanied by tight monetary policy.

Whether increased spending translated into real output, or simply nominal income, growth tended to depend on the relative elasticities of supply in various sectors and the sectoral distribution of the credit. When resource utilization was high in favored sectors, expanded official credit resulted only in price increases and almost no change in real output, even without concurrent monetary expansion. This resulted from the fact that the shift in expenditure composition raised prices in the stimulated sector, but the high prices were not offset by price deflation in the sectors that were crowded out. For example, increases in mortgage credit frequently resulted in decreased business credit. The increased demand for housing in periods of tightness in the housing market was simply inflationary, as there was no offset in business sector prices. If accompanied by a supportive monetary policy, the official credit activity in the mortgage field tended to be even more inflationary.

Tempering all these results, however, are a number of particular problems that make empirical work on the aggregate economic effects of credit programs extremely difficult and definitive conclusions virtually impossible. First, generic differences in purposes among programs result in different impacts on real and financial variables—for example, market-perfecting credit programs should have different impacts from income-redistributing credit programs. Second, credit programs have different growth stages that alter the degree and diffusion of their impact—therefore, the effects may depend partly on how long the program has existed. Third, both the financial and real effects of specific credit market interventions have complex, lagged patterns, which can only be evaluated within the context of comprehensive econometric models. Since the level of aggregation must be high, only the very largest of such programs can realistically be evaluated at all. Finally, overall financial and economic conditions, as well as the mix of fiscal and monetary policy, largely determine the effects of credit programs.

Despite all the difficulties, however, there seems to be a consensus that several undesirable consequences can be associated directly with official credit programs, such as displacement of private borrowers and lenders; encouragement of foreign borrowing; insertion of a policy “wedge” into market decisions; creation and maintenance of large, inefficient bureaucracies; complications in the coordination of stabilization policies; and, sacrifice of economic efficiency without corresponding increases in the total supply of investible funds, so that the overall rate of economic growth suffers.

Fiscal and Budgetary Implications

The foregoing sections have shown that it is the time dimension of lending that distinguishes it from other budgetary expenditure items. Since subsidized credit schemes are common in a wide variety of countries (both developing and industrial), currently observed fiscal deficits may simply reflect, to a surprisingly large degree, past credit subsidy commitments. Using the estimation methods described above on Korean data from the 1970s, explicit interest subsidies in direct government lending were found to have been a major determinant of the Government’s fiscal position, equal to at least half of the central government’s deficit, on average.

Moreover, because the lending process contributes to the creation of long-term structural deficits, the short-term flexibility of fiscal policy may also become severely restricted. Chart 4 depicts an eight-year government program that annually lends an amount designated as 100 at 5 percent a year interest. This scheme is financed by government borrowing at 10 percent a year. It is assumed that both the lending and borrowing involve five-year bonds (the principal due in one lump sum at maturity, while annual payments of interest only are due prior to maturity); lending occurs on January 1 and repayment on December 31. As the figure shows, the annual subsidies grow to an equilibrium level, which remains even after no new net lending occurs and current new loan commitments are being met fully with repayments from past loans (as in years 6–8). The only way these annual subsidies can be stopped is by terminating the program and retiring the debt (years 9–12).

Chart 4.
Chart 4.

The Structural Fiscal Burden Built into Official Subsidized Lending Programs

Note also that in each year in which government intermediation is undertaken at these relative terms (years 1–8), the government commits itself to an explicit interest subsidy of 18.95, which is the present value of the annual subsidy stream discounted at 10 percent a year, or the government’s opportunity cost of funds. Alternatively, the total subsidy being given at the outset of the eight-year program is 111.23.

The concealed nature of such credit subsidies does not make them any less subsidies, and, like any government outlay, they are always financed by taxes, borrowing, or monetary growth at higher levels than would otherwise have occurred, or by other expenditures being lower than they would otherwise have been. Their concealed nature may mean, however, that they are more difficult to control than other expenditures because they may escape scrutiny when austerity forces budget cuts to be made. This is not to argue that all government lending programs should be ended. As explained earlier, there are circumstances in which credit subsidies are appropriate and justifiable. Still, they should be required to vie for the limited public resources on an equal basis with other competing claims. This suggests that their true magnitude—that is, the present discounted value of the subsidy stream—be entered explicitly in the budget in the originating year as an expenditure item.30


  • Aragon, George, “Federal Credit Programs,” in Federal Finance: The Pursuit of American Goals, Vol. 6 of Special Study on Economic Change, 96th Congress, 2nd Session (Washington: Government Printing Office, December 1980), pp. 35899.

    • Search Google Scholar
    • Export Citation
  • Bennett, James T., and Thomas J. Di Lorenzo, Underground Government: The Off-Budget Public Sector (Washington: The Cato Institute, 1983).

    • Search Google Scholar
    • Export Citation
  • Blejer, Mario I., “Interest Rate Differentials and Exchange Risk: Recent Argentine Experience,” Staff Papers, International Monetary Fund (Washington), Vol. 29 (June 1982), pp. 27079.

    • Search Google Scholar
    • Export Citation
  • Break, George F., Federal Lending and Economic Stability (Washington: The Brookings Institution, 1965).

  • Break, George F., “Issues in Measuring the Level of Government Economic Activity,” American Economic Review (Nashville, Tennessee), Vol. 72 (May 1982), pp. 28895.

    • Search Google Scholar
    • Export Citation
  • Brunner, Karl, ed., Government Credit Allocation: Where Do We Go From Here? (San Francisco: Institute for Contemporary Studies, 1975).

  • Carron, Andrew S., “Fiscal Activities Outside the Budget,” in Setting National Priorities, The 1982 Budget, ed. by Joseph A. Pechman (Washington: The Brookings Institution, 1981), pp. 26169.

    • Search Google Scholar
    • Export Citation
  • Fry, Maxwell J., “Interest Rates in Asia” (unpublished, Washington: International Monetary Fund, June 25, 1981).

  • Harvey, Charles, Analysis of Project Finance in Developing Countries (London: Heinemann, 1983).

  • Healy, James P., “The Institutional Framework for Financial Policy in Korea” (unpublished, Washington: International Monetary Fund, August 1981).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, Government Finance Statistics Yearbook, Vol. 6 (Washington, 1982).

  • Johnson, Omotunde E.G., “Direct Credit Controls in a Development Context—The Case of African Countries,” in Government Credit Allocation: Where Do We Go From Here?, ed. by Karl Brunner (San Francisco: Institute for Contemporary Studies, 1975), pp. 15180.

    • Search Google Scholar
    • Export Citation
  • Jones, E.P., and S.P. Mason, “Valuation of Loan Guarantees,” in U.S. Congressional Budget Office, Conference on the Economics of Federal Credit Activity, Part II: Papers (Washington: Government Printing Office, September 1981), pp. 34977.

    • Search Google Scholar
    • Export Citation
  • Larkins, Dan, $300 Billion in Loans: An Introduction to Federal Credit Programs, Domestic Affairs Study 6 (Washington: American Enterprise Institute for Public Policy Research, September 1972).

    • Search Google Scholar
    • Export Citation
  • McKinnon, Ronald I., and Donald J. Mathieson, How to Manage a Repressed Economy, Essays in International Finance, No. 145(Princeton, New Jersey: Princeton University Press, December 1981).

    • Search Google Scholar
    • Export Citation
  • McMurray, David W., “Evaluating Alternative Financing Packages,” World Bank External Debt Division Working Paper, No. 1982–3 (Washington: World Bank June 1982).

    • Search Google Scholar
    • Export Citation
  • Nowzad, Bahram, R.C. Williams, and others, External Indebtedness of Developing Countries, IMF Occasional Paper, No. 3 (Washington: International Monetary Fund, May 1981).

    • Search Google Scholar
    • Export Citation
  • Organization for Economic Cooperation and Development (OECD), Development Assistance Committee, Development Cooperation (Paris: OECD November 1980).

    • Search Google Scholar
    • Export Citation
  • Plantes, M.K., and D. Small, “Macroeconomic Consequences of Federal Credit Activity,” in U.S. Congressional Budget Office, Conference on the Economics of Federal Credit Activity, Part II—Papers (Washington: Government Printing Office, September 1981), pp. 165.

    • Search Google Scholar
    • Export Citation
  • Premchand, A., “Study Examines Government Lending Programs in OECD and Selected Developing Countries,” IMF Survey (Washington), Vol. 11 (August 2, 1982), pp. 23437.

    • Search Google Scholar
    • Export Citation
  • United States, “Feasibility of Explicit Recognition of the Interest Subsidy in Connection with Separate Budgeting of Loans,” President’s Commission on Budget Concepts, Staff Papers (Washington: Government Printing Office, 1967), pp. 33545.

    • Search Google Scholar
    • Export Citation
  • United States, Office of Management and Budget, Office of the President of the United States, Special Analyses: Budget of the United States Government, Fiscal Year 1978 (Washington: Government Printing Office, 1977).

    • Search Google Scholar
    • Export Citation
  • United States, Congress, Congressional Budget Office, Loan Guarantees: Current Concerns and Alternatives for Control (Washington: Government Printing Office, August 1978).

    • Search Google Scholar
    • Export Citation
  • United States, Congress, Congressional Budget Office, Loan Guarantees: Current Concerns and Alternatives for Control: A Compilation of Staff Working Papers (Washington: Government Printing Office, 1979).

    • Search Google Scholar
    • Export Citation
  • United States, Congress, Congressional Budget Office, Federal Credit Activities: An Analysis of the President’s Credit Budget for 1981 (Washington: Government Printing Office, February 1980).

    • Search Google Scholar
    • Export Citation
  • United States, (1981a), Congress, Congressional Budget Office, Federal Credit Activities: An Analysis of the President’s Credit Budget for 1982 (Washington: Government Printing Office, April 1981).

    • Search Google Scholar
    • Export Citation
  • United States, (1981b), Congress, Congressional Budget Office, Small Issue Industrial Revenue Bonds (Washington: Government Printing Office, April 1981).

    • Search Google Scholar
    • Export Citation
  • United States, (1981c), Congress, Congressional Budget Office, “The Benefits and Costs of the Export-Import Bank Loan Subsidy Program” (mimeographed, Washington: Government Printing Office, March 1981).

    • Search Google Scholar
    • Export Citation
  • United States, (1981d), Congress, Congressional Budget Office, Conference on the Economics of Federal Credit Activity, Part II: Papers (Washington: Government Printing Office, September 1981).

    • Search Google Scholar
    • Export Citation
  • United States, (1981e), Executive Office of the President, Office of Management and Budget, Special Analyses: Budget of the United States Government, Fiscal Year 1982 (Washington: Government Printing Office, 1981).

    • Search Google Scholar
    • Export Citation
  • United States, (1982a), Congress, Congressional Budget Office, The Federal Financing Bank and the Budgetary Treatment of Federal Credit Activities (Washington: Government Printing Office, January 1982).

    • Search Google Scholar
    • Export Citation
  • United States, (1982b), Congress, Congressional Budget Office, Federal Credit Activities: An Analysis of the President’s Credit Budget for 1983 (Washington: Government Printing Office, March 1982).

    • Search Google Scholar
    • Export Citation
  • United States, (1982c), Congress, Congressional Budget Office, Reducing the Federal Deficit: Strategies and Options (Washington: Government Printing Office, 1982).

    • Search Google Scholar
    • Export Citation
  • United States, Executive Office of the President, Office of Management and Budget, Budget of the United States Government, Fiscal Year 1989 (Washington: Government Printing Office, February 1988).

    • Search Google Scholar
    • Export Citation
  • van Wijnbergen, Sweder, “Short-Run Macro-Economic Adjustment Policies in South Korea: A Quantitative Analysis,” World Bank Staff Working Paper, No. 510 (Washington: World Bank, November 1981).

    • Search Google Scholar
    • Export Citation
  • von Furstenberg, George M., “Distribution Effects of GNMA Home Mortgage Purchases and Commitments Under the Tandem Plans,” Journal of Money, Credit, and Banking (Columbus), Vol. 8 (August 1976), pp. 37389.

    • Search Google Scholar
    • Export Citation
  • Wattleworth, Michael, “Credit Subsidies in Budgetary Lending,” (unpublished, Washington: International Monetary Fund, 1983).

  • Weidenbaum, Murrey L., “An Economic Analysis of the Federal Government’s Credit Programs,” in U.S. Congress, House Committee on Banking, Currency and Housing, Hearings on Loan Guarantees and Off-Budget Financing, 94th Congress, 2nd Session, November 10, 1976 (Washington: Government Printing Office, 1976), pp. 15375.

    • Search Google Scholar
    • Export Citation
  • Weidenbaum, Murrey L., Subsidies in Federal Credit Programs (Washington: American Enterprise Institute, Reprint No. 4, September 1972).

  • Williams, Richard C., with G.G. Johnson, and others, International Capital Markets: Developments and Prospects, 1982, IMF Occasional Paper, No. 14 (Washington: International Monetary Fund, July 1982).

    • Search Google Scholar
    • Export Citation
  • World Bank, World Debt Tables, 1981, External Debt Division, EC-167/81 (Washington: World Bank, December 1981).


The author wishes to acknowledge Etienne Gilard, who as a Summer Intern at the Fund in 1982 contributed substantially to the background work for this paper, and to thank his colleagues in the Fund, who provided many valuable comments and suggestions.


The methodology has been applied to Korean data with satisfactory results. The results of the application are summarized in a longer paper by the author (see Wattleworth (1983)).


This is the size of the direct loan portolio; another $507 billion in guaranteed loans was outstanding, while government-sponsored enterprises had an additional $581 billion of loans outstanding. Thus, directly and indirectly, the Government had influenced the allocation, on subsidized terms, of $1,322 billion of outstanding credit, equivalent to 29 percent of GNP in 1987. See United States (1988).


The theoretical case for this argument is developed in McKinnon and Mathieson (1981).


For an introduction to loan guarantee analysis, see United States (1978 and 1979). Methods developed to measure the value of loan guarantees are largely based on the contingent claims method; see, for example, Jones and Mason (1981). Other papers in this volume are also of interest. More recent work by the Office of Management and Budget of the U.S. Federal Government simply estimates the subsidy inherent in guaranteeing loans as the difference between the amount the borrower pays and the amount required to reinsure the loan.


This point is made in various places. See, for example, either United States (1977), p. 87, or Aragon (1980), p. 374. Also, for an example of how this mechanism works through industrial development bonds or industrial revenue bonds in the United States, see United States (1981e), pp. 184–86, and United States (1981b).


This distinction is also made in Aragon (1980), p. 373.


The subsidy (or grant) element—a concept introduced later—is defined as the subsidy value stated as a percentage of (the present value of) the loan’s face value—here equal to 23.7 percent (see Organization for Economic Cooperation and Development (1980, p. 241).


This approach to the problem was inspired by an excellent book by Harvey (1983).


It is worth noting here that an alternative methodology (the effective rate of interest (ERI)) introduced later would give roughly the same ordinal ranking of the alternatives. Thus, the ERI on the last loan is 6.05 percent a year, on the third option 9.09 percent a year, and on the first two it equals the quoted loan rate of 10 percent a year.


Of course, the formulas work when i* < ERI on the loan. In applications, all such loans should be included in the analysis, so the government’s position on balance can be derived. In an earlier application to Korean data (see Wattleworth (1983)) many loans analyzed involved negative subsidies at the i* rates used.


Break (1982). The terms “sub-” and “infra-” marginal require some explanation. If i* were the market-clearing interest rate with a fixed supply of credit, then submarginal borrowers would be those along the demand schedule below i*; inframarginal borrowers would be those along the demand schedule above i*. The upshot of this argument is that a meticulous study will probably have to be program-specific. For an excellent example of such an analysis, see von Furstenberg (1976).


One should not infer from the whole line of reasoning in this section that a government should simply charge its own borrowing rate on its lending. Clearly, a misallocation of resources in the economy would still result, and the implicit credit subsidies would not be removed. As Break (1965), pp. 36–39, has shown, if this were done, the resultant government lending programs would be overexpanded and social welfare would be reduced.


This statement is based on the presumption that for instruments that are identical except for their currency of denomination, the only reason for an interest rate differential between them should be expected changes in the exchange rate between the two. This is known as the Fisher hypothesis. While there may be other reasons for departures from Fisher parity (e.g., transactions costs, differential taxation, political risk), a major determinant is assumed to be exchange risk (see Blejer (1982, p. 271)).

The use of government bond yields as estimates of i* is only a proxy solution. The ideal measure of the government’s margin of cost of the use of resources would be the social rate of discount. But there is no consensus on how this should be measured. See Larkins (1972, p. 34).


The share of adjustable rate notes in total bonds floated has been small historically, although maturities have shortened somewhat (see Williams, Johnson, and others (1982, pp. 49 and 55)).


For an application of this approach, see United States (1981c).


See Appendix II, “Inflation and Debt Service,” pp. 42–45 in Nowzad, Williams, and others (1981).


See Harvey (1983, p. 12).


See Harvey (1983, p. 7).


For example, in the United States, state chartering of savings and loan associations and banks historically prevented excess loanable funds in surplus regions from flowing to areas of excess demand; see Plantes and Small (1981, pp. 14–15).


Such as originally existed in U.S. residential mortgage instruments, which lacked liquidity and carried onerous terms for borrowers and associated high risks for lenders before the Government created the Federal Housing Administration and Federal National Mortgage Association (a secondary market); see Aragon (1980, p. 359).


For a somewhat different, but related, discussion of the reasons frequently given by the local authorities for direct credit market intervention in developing countries, see Johnson (1975).


Fry (1981, p. 38).


This process has been amply demonstrated in international capital markets and sovereign external debt since the onset of the debt crisis in 1982.


Unless otherwise indicated, the remainder of this section comes directly from Aragon’s (1980) article; the interested reader should refer to it directly for greater depth and thoroughness.


As Weidenbaum (1976, p. 162), explains: “This … occurs for a variety of reasons. The total supply of funds is broadly determined by household and business saving and the ability of banks to increase the money supply …. The normal response of financial markets to an increase in the demand for funds by a borrower, such as is represented by a federal credit program, is an increase in interest rates so as to balance out the demand for funds with the supply of saving. But the federal government’s demand for funds is ‘interest-inelastic’ … and the interest-elasticity of saving is relatively modest. Thus, weak and marginal borrowers will be ‘rationed’ out of financial markets in the process, while the Treasury and other borrowers pay higher rates of interest.”


Estimates of the extent of crowding out are very difficult to make and somewhat uncertain. One such estimate, however, found that for every $1 billion in loan guarantees extended by the federal government, between $736 million and $1.3 billion in private investment was crowded out in 1980. See Bennett and Di Lorenzo (1983).


The U.S. Administration has recently proposed that the U.S. Congress adopt this procedure. See the United States (1988, pp. 6b-8).

Methodological Issues: Occa Paper 59