Public Principles of Public Debt: A Defense and Restatement
By James M. Buchanan
- Ch. 1, The Economists and Vulgar Opinion
- Ch. 2, The New Orthodoxy
- Ch. 3, The Methodology of Debt Theory
- Ch. 4, Concerning Future Generations
- Ch. 5, The Analogy: True or False
- Ch. 6, Internal and External Public Loans
- Ch. 7, Consumption Spending, the Rate of Interest, Relative and Absolute Prices
- Ch. 8, A Review of Pre-Keynesian Debt Theory
- Ch. 9, Public Debt and Depression
- Ch. 10, War Borrowing
- Ch. 11, Public Debt and Inflation
- Ch. 12, When Should Government Borrow
- Ch. 13, Should Public Debt Be Retired
- Ch. 14, Debt Retirement and Economic Stabilization
- Appendix, A Suggested Conceptual Revaluation of the National Debt
A Suggested Conceptual Revaluation of the National Debt
This Appendix represents an attempt to apply the theory of public debt contained in the earlier chapters to the general problem of measuring the magnitude or size of the national debt. It is an exploratory effort designed to raise and to isolate the relevant issues rather than to resolve all of the complexities which may appear. With this quite limited objective in view, I have not tried to provide definitive solutions to the measurement problem. Insofar as the argument requires, I have used actual data on the national debt for illustrative purposes. But I should emphasize the illustrative usage as opposed to any presumed factual presentation or rearrangement of data.
II. The General Measurement Problem as Applied to Private Debt
Public debt has traditionally been measured in terms of the principal or maturity value, that is, the amount which must be repaid at the maturity date, and, except for securities issued at a discount, the amount of funds transferred to the government when the debt is created.*103 This apparently simple measurement has its origin in the treatment of private debt. Since the principal represents the payment necessary to discharge fully the obligation of the debt, it seems appropriate that this be used in measuring debt size. The implications of this measurement procedure do not seem to have been thoroughly examined.
Let us initially suppose that a riskless private loan is contracted and that the structure of interest rates remains stable through time. The interest rate paid on this loan will be a “pure” rate unalloyed by any risk premium. Competition will insure that this rate approximates that paid on other riskless loans in the economy. Under these circumstances there can be no question but that the size of the debt is best shown by the maturity value. This indicates the value of an alternative capital asset of identical risk characteristics which would be required to provide a yield sufficient to cover fully the debt obligation. In a slightly different sense, the maturity value also represents the capitalized value of the future payments stream if this stream is capitalized at the pure rate of yield on investment. In this particular example, the stream of payments is capitalized at the internal rate, obviously yielding the maturity value as a capitalized sum.
Let us now introduce a second model in which a private loan of some riskiness is contracted. We shall continue to assume that the structure of interest rates remains stable over time. In this case, the interest rate paid must include some risk premium; it must exceed that paid on the no-risk investment. The maturity value of the debt will not be equal to the capitalized value of the stream of payments, capitalized at the pure rate of yield. For example, suppose that we are considering a loan of $100 (which we may convert to a loan in perpetuity by the assumption of refunding at the same rate) at an interest rate of 10 per cent while the pure rate of yield is 5 per cent. The capitalization of a $10-payments stream in perpetuity yields a total value of $200. The debt claim will, however, be worth only $100 in the market. In other words, $100 will be sufficient to purchase an asset of equivalent risk characteristics. The $100 purchase will not, of course, provide a certain yield which will guarantee that the initial debt can be fully serviced. If the borrower desires a perfect hedge against the debt, he must purchase that asset which will yield the $10 with certainty. This asset will command a market price of $200. But this $200 asset will more than remove the debt obligation represented by the original loan. It will also remove from either borrower or lender the risk which the lender assumed in the original transaction. The purchase of the $100 asset is sufficient to place both borrower and lender in a position identical to that which they enjoyed prior to the debt transaction.
The measure of the size of private debt in terms of maturity values is less appropriate when the assumption of stability in the structure of interest rates is dropped. If the pure rate of yield on private investment changes subsequent to the contraction of a private loan, the capitalization process suggested above will yield a different value for debt than the principal or maturity value. And, for many purposes, this capital value is more useful in indicating the real weight of the debt. Suppose, to return to our first model, that a riskless loan of $100 is contracted at the pure rate of 5 per cent. Subsequently, the rate of yield on marginal investments in the economy falls to 4 per cent. The capitalization process now yields $125 as a measure of the debt rather than $100. This indicates that the full discharge of the debt obligation is now equivalent to the sacrifice of an alternative earning asset commanding a price of $125. The larger figure is obviously more appropriate as a measure if the debt instrument is marketable; but, when considered correctly, it is equally appropriate when debt instruments are not marketable.
III. Measuring Local Government Debt
The maturity value measure is applicable only in a more restricted sense to local government debt. Initially, let us assume once again that the structure of interest rates does not change over time. In this case, as with private debt, the maturity value of the debt will represent the market value of resources which must be sacrificed in order to finance the purchase of a capital asset of characteristics identical to the debt obligation. But the special feature of local government debt is that income from local government securities is tax exempt. This means that such securities may be marketed at a rate which is lower than the pure rate of yield on marginal investment in the private sector. The effects of this feature on the measurement problem must be examined.
Again it will be helpful to consider a simplified numerical example. Suppose that some local government issues a bond for $100 at a rate of 3 per cent while the pure rate of yield on private investment is 4 per cent. Again we shall assume that the maturing issues are continually refunded, allowing the fixed maturity security to be converted conceptually into a security of no maturity date. The $3 annual interest payment discounts to a value of only $75 when this payments stream is capitalized at the pure yield rate. But the market will operate so as to insure that an equivalent asset commands a market price of $100. The local unit of government may, if it desires, purchase a no-risk asset for $75 which will yield an income sufficient to service the debt. This suggests that the $75 provides a more useful measure for the size of the debt than the $100 maturity value or, in the case of the no-maturity security, the principal.
Actually, however, both the $75 and the $100 measure must be used. It is true that a tax payment of only $75 would be required to offset the debt. But the discharge of the debt will also eliminate from the economy a tax exemption which has a capital value of $25 under our assumptions. Therefore, the market value of resources which must be given up, in present or future periods, by individuals in order to discharge fully the obligation represented by the debt is $100. The maturity value measure is the more useful one when the problem is considered in this light. However, when it is recognized that the local taxpaying group may be quite different from the bond purchasing group, the measure becomes less useful. The bonds issued by single local units of government are normally marketed nationally. The advantages of the tax exemption feature are secured by federal taxpayers scattered throughout the economy. The capital value of the tax exemption is held, not by local taxpayers of the borrowing jurisdiction, but by bond purchasers from the entire nation. Therefore, it is not proper to attribute to local taxpayers the supplementary capital value of the tax exemption. The $75 figure more correctly measures the size of local government debt when the single local unit of government is considered in isolation.*104
On the other hand, when aggregate local debt is considered, the use of maturity values seems necessary. Those individuals holding the bonds of any one jurisdiction must be taxpayers in some local unit, and, therefore, the capital value of the tax exemption must be added in when all local debts are taken into account.
Just as in the case of private debt, when the assumption of stability in the interest rate structure is dropped, the use of principal to measure debt size is not acceptable, even in the measurement of aggregate local government debt. If interest rates increase after the sale of local securities, the borrowing jurisdiction may, if it chooses, repurchase its own securities for a price below that indicated by the principal of the loan. The size of the aggregate local government debt at any moment in time seems to be best measured by the amount of current tax collections which would be required to finance the purchase of an earning asset which will yield an income sufficient to service all outstanding debt plus the capital value of the tax exemption feature. To return to the numerical example, let us suppose that the pure rate of yield on marginal investment increases from 4 per cent to 5 per cent subsequent to the debt issue. The capitalization process yields a sum of $60 instead of $75 as the amount of taxes required to purchase an earning asset yielding a sum sufficient to service the debt. If the local unit is considered in isolation, this becomes the measure of the debt. If, on the other hand, the aggregate local debt is to be measured, the capital value of the tax exemption feature must also be calculated. The differential yield of $1 capitalized at 5 per cent rather than 4 per cent gives a total capital value of $20 rather than $25. When aggregate local debt is measured, the appropriate magnitude now becomes $80 instead of $100.
IV. The National Debt
The two preceding sections have shown that the normally accepted measurement procedures may not provide meaningful totals for private debt and for local government debt under certain conditions. The use of principal or maturity value to measure the magnitude of the debt obligation is even less applicable for the national debt. The fundamental reason for the difference lies, of course, in the possession of money-creating powers by the central or national government. As earlier chapters have shown, the existence of this power, along with that of pure debt creation, has led to some confusion. In the discussion which follows I shall propose an alternative way of evaluating the debt, at least conceptually, which should be of some assistance in clarifying the distinction between real or pure debt and monetized debt.
How large is the national debt? As of July 31, 1957, official records indicate that the national debt of the United States amounted to $272.5 billion. What does this figure tell us? This is somewhat more difficult to answer. It may provide some information concerning the amount of purchasing power which was transferred to government at some time in the past by individuals and institutions. There is, however, no way of knowing whether or not this purchasing power was actually transferred away from the private economy. All, none, or any portion of this purchasing power may have been created in the process of debt issue. The $272.5 billion figure tells us nothing about the real debt which was created, nor does it tell us anything about the amount of real resources which must currently be given up if we choose to discharge fully the debt obligation.
The confusion generated by the use of this measure may be readily illustrated by frequently encountered popular statements which claim that each man, woman, and child in the United States owes a national debt of some $1,600. This is computed by dividing the debt total of $272 billion by some current estimate for population. From this the inference is often drawn, implicitly or explicitly, that the full discharge of the debt obligation would require that additional current taxes in the amount of $1,600 be levied on each individual (on the average).
If we neglect for the time being the fact that interest rates have risen sharply since the issue of large portions of the national debt, the inference would seem valid in terms of the analogy with private debt. When the adjustment to present market values is made (an adjustment which applies equally to private and public debt), such a per capita computation should indicate the market value of real resources which each individual would have to sacrifice to discharge fully his per capita share of the national debt obligation.
It is obvious, however, that the inference is almost wholly incorrect when applied to national debt. Here the analogy between public and private debt appears to be false. The explanation is not difficult to find. A large part of the national debt does not represent pure debt at all. This part is essentially “money” both when issued and as held by individuals and institutions. This being the case, full tax financing of debt retirement would act to destroy “money” in the system, thereby generating serious destabilizing effects. The analogy breaks down here only because of this mixture of pure debt and “money” in what we normally refer to as national debt. As the analysis of this book has demonstrated, the analogy fully holds when pure or real debt is considered. Quite obviously money is not debt.
Any meaningful measure of the national debt should reflect the same information as that which is provided by the accepted measure of private debt. That is, this measure should indicate the capital value of resources which must be given up or sacrificed in order to discharge fully the debt obligation. It should indicate the total tax collection, in real terms, which is required to retire all national debt without, at the same time, exerting significant over-all effects on the absolute price level. In other words, the conceptual retirement operation should be neutral in its effects on the level of economic activity. If the use of maturity values as adjusted to take account of changes in interest rates fails this test, how may such a meaningful measure be constructed?
The solution is to be found in the capitalization process discussed above for private debt and local government debt. If the interest payments stream is capitalized at a rate indicating the pure rate of yield on marginal investments in the private sector, the resultant capital value will provide an accurate measure of the national debt in some meaningful sense. This approach was shown to be faulty in application to private debt and to local government debt in the aggregative sense. The capital value, calculated in this manner, was shown to diverge from the principal of the debt in the one case because of the failure to include a differential risk premium and in the other because of its omission of a differential tax exemption feature. But the market appropriately places some values, negative or positive, on these features. And the private individual is forced to abide by market evaluations in purchasing equivalent assets or in repurchasing debt instruments.
In the case of the national debt, however, the measure yielded by the capitalization process suggested is much more useful. Quite clearly, as Section V will demonstrate, the current interest charges capitalized at an estimated rate for the net or pure yield on capital investment will provide a figure far below either the maturity value of the national debt or for this latter value adjusted downward for the recent increases in the level of interest rates. It is equally clear that this difference is primarily due to the fact that the national debt instruments possess many characteristics of money. This being true, the adjusted maturity value does not reflect the value of real resources which would have to be given up to discharge the debt.
Money may be issued at zero cost. Therefore, that portion of the national debt which does represent “money” in its relevance to human behavior can be replaced with actual money, currency, without private people being forced to sacrifice real goods and services. The share of total debt, as measured, which represents genuine or pure debt can best be determined from the capitalization process suggested.
In the following section I shall attempt to apply this proposed measurement process to the national debt of the United States.
V. How Large Is the National Debt?
As of July 31, 1957, the national debt, as measured, amounted to $272.5 billion. This may be called, for our purposes, the principal or the maturity value measure. Since interest rates have risen since much of the debt has been issued, the first step in any evaluation is that of adjusting this value downward to reflect the reduction in capital value which has taken place. The government debt may be conceptually repurchased for less than the principal sum outstanding.
An extremely rough calculation suggests that the adjusted market value of national debt as of July 31, 1957, was $257 billion.*105 This figure represents the cash outlay which will enable the government conceptually to repurchase all outstanding debt, either directly through established markets or indirectly through “hedging” sale and repurchase of additional debt sufficient to offset nonmarketable issues.*106
There are two means through which the necessary cash outlay of $257 billion may be secured. Money may be printed directly, or taxes may be levied. The $257 billion figure tells us nothing concerning the breakdown between these two sources. To secure such a breakdown, the capitalization process suggested above must be introduced.
The annual interest charge on the national debt as of July, 1957, is estimated at $7.4 billion. This amounts to 2.7 per cent of the maturity value of $272.5 billion, and almost 2.9 per cent of the adjusted market value of $257 billion. Quite clearly neither of these represents an appropriate capitalization rate in the 1950’s. This rate should be representative of the pure rate of yield on capital investment at the margin of use. Without making any detailed attempt to determine this rate accurately, I shall make the assumption that this rate is 4 per cent.*107 If the $7.4 billion (assumed to be the value of the interest payments stream in perpetuity) is capitalized at a 4 per cent rate, we get a capital value of $185 billion, not $257 billion. This figure comes much closer to providing a measure of national debt in some “pure” sense. By the current sacrifice of $185 billion in privately owned earning assets or in consumption goods, the national debt can be fully retired, provided we can neglect the possible secondary effects of the retirement process itself on the structure of interest rates.*108 A more direct statement can be made in a slightly different manner. The net yield from $185 billion of earning assets in the private economy is obligated to the service of the national debt.
The remaining $72 billion represent that portion of the national debt which is, for the most part, “money” in its relevance to human behavior.*109 This suggests that, secondary effects aside, the national debt could be wiped off the books with a capital levy of $185 billion and a direct currency creation of $72 billion. The additional currency would be needed to offset the deflationary impact of the debt retirement, and to keep the whole operation “neutral” in its stabilization effects. Having removed the debt instruments, possessing much “moneyness,” there would have to be more nominal units of money introduced in order to prevent serious deflationary consequences. But this additional money needed may be created without cost, and, therefore, it should not be included in any estimate of “pure” debt.
It is to be emphasized that I am proposing a conceptual revaluation of the debt, not any actual attempt at retirement. Specifically, what is suggested is that the manner of measuring the debt be modified and that an additional and supplementary evaluation be made. The total process of debt measurement should look as follows:
|National Debt as of July, 1957
|Billions of Dollars|
|1. National debt, maturity value||$272|
|2. National debt, present market value||257|
|3. National debt, “pure”||$185|
|National debt, “monetized”||72|
This account would be useful in many respects. First of all, it would indicate more accurately the real burden of debt which is being shifted forward to future generations of taxpayers. This is the $185 billion, not the $257 or the $275 billion. The burden of the $72 billion, if it existed, was not a burden of pure debt but of inflation, and, as such, has already been shouldered. Future generations will be little affected by this portion of the nominal debt. Similar conclusions follow for the $18 billion difference between the maturity value and the present market value. This no longer exists as a debt obligation; this portion has been “retired” through the levy of a “tax” on the holders of government securities.
VI. Treasury Refunding Operations
One of the most important uses of this supplementary account would be that of providing an accurate check on the effects of Treasury refunding operations. Suppose that the Treasury succeeds in refunding a portion of the debt at a lower rate of interest. This is accomplished, assuming that the general pattern of rates is not changing, by replacing debt instruments possessing less “moneyness” with debt instruments possessing more “moneyness.” Let us assume that a particular operation of this sort reduces the annual interest charge from $7.4 to $7 billion. We shall continue to assume that the pure rate of yield is 4 per cent. This operation reduces the value of pure debt from $185 to $175 billion, while the value of the monetized debt is increased by $10 billion. In this way it becomes obvious that the refunding operation is equivalent to the retirement of pure debt. Future generations of taxpayers are relieved of an annual interest charge of $.4 billion, and individuals living currently are subjected to a possible burden of an additional $10 billion. In the full-employment setting, a refunding of this sort will be inflationary, and the $10 billion may be considered a tax on the holders of cash balances and government securities. The refunding will have shifted a real burden of debt from future taxpayers to these groups. If unemployment should be present, the inflationary consequences need not occur. And here the burden may be removed from future taxpayers without placing substantial real cost on individuals currently living. Under these conditions, this type of refunding is, of course, to be recommended.
The opposing case may now be considered. We assume that the Treasury succeeds in increasing the total interest charge on the debt. This is accomplished by replacing debt instruments possessing considerable “moneyness” with others which more closely resemble pure debt. Again for purposes of illustration, suppose that a particular operation increases the annual interest charge from $7.4 to $7.8 billion. This will increase the value of pure debt from $185 billion to $195 billion, assuming the same capitalization rate of 4 per cent. The operation is equivalent to the issue of $10 billion additional pure debt. The monetized debt is reduced by $10 billion.
This operation will be deflationary, at least relative to what would have taken place in its absence. But the deflation itself must relieve present taxpayers at the expense of future taxpayers in this case. The government secures no greater share of resources than before the refunding; but it agrees to pay more future income than before. Individuals, after the operation, hold more claims to future income. The net value of claims to current income (“money”) and to future income (pure debt) has not been modified. But claims to current income have been reduced and claims to future income increased. Those who give up the claims to current incomes in exchange for greater claims on future incomes are not harmed by the operation. They are purchasing pure debt instruments. On the other hand, those who are unaffected directly by the operation gain by the deflation imposed, assuming that we may neglect distributional consequences.
This analysis may be clarified somewhat by a more specific example. Suppose that Individual A holds, prior to the refunding, a security with a maturity value of $100 yielding only 3 per cent interest because of specific redemption features which allow this security to fill a near-money role in his portfolio. The Treasury offers him in exchange a $100 security which yields 4 per cent but which does not carry with it these “moneyness” features. The individual accepts the offer and the exchange is made. Clearly, future taxpayers are charged with the additional $1 of interest. Individual A will find it necessary to reduce his rate of spending on current real goods and services sufficiently to restore his liquidity position. He will find it necessary to withdraw approximately $25 from circulation. The goods and services so released will become available to the whole social group. Other individuals will be benefited by the refunding while Individual A will have undergone merely a transformation of his assets. In this simple model, the net effect on the current generation must be beneficial.
The conclusions reached on such a simple model must be modified if we introduce leverage effects stemming from fractional reserve banking. For example, if the refunding operation should take the form of retiring debt held by the central bank and replacing it with debt held by individuals, the reduction in liquidity occasioned by the retirement may be some multiple of the actual maturity value of the debt involved. In this case the refunding operation has the effect of removing “powerful” money from the system and replacing it with “weak” money. Whether or not this is desirable will depend on the stage of the cycle in which the refunding operation takes place. Refunding at higher yields is equivalent to borrowing solely to prevent inflation. This case was analyzed somewhat more fully in Chapter 11.
The analysis to this point has assumed that the level or pattern of interest rates does not change with any Treasury refunding. But clearly interest rates do change, and this must now be taken into account. Let us return to the first example in which the total interest charge is reduced from $7.4 to $7 billion. In saying that this operation reduces pure debt from $185 to $175 billion, the old rate of 4 per cent for the net yield on private investment was employed. But a refunding of this magnitude would tend to reduce the level of interest rates, and thus the appropriate capitalization rate. And if the annual interest payments stream is reduced, but at the same time the capitalization rate is reduced, will the amount of pure debt, as calculated in the manner proposed, necessarily be changed? It may readily be demonstrated that the amount of pure debt must also be reduced under these conditions, although by less than in the previous example. It is true that the operation may reduce both the interest payment and the appropriate rate of capitalization. But it is impossible for the Treasury operation alone to reduce the rate of capitalization, defined as the estimated net yield on zero-risk investment in the whole economy, proportionately with the interest payment. The interest charge is calculated for the national debt alone; the rate of capitalization is taken from the whole economy. The refunding operation must, therefore, effect an increase in monetized debt and a decrease in real debt.
Similar conclusions follow when the opposite sort of refunding is considered. Here the general level of interest rates will tend to increase, and thus the appropriate discount rate. But this rate cannot increase proportionately with the interest charge. Pure debt must increase and monetized debt decrease.
The necessity for taking the change in the rate of capitalization into account along with the change in the payments stream need not make the conceptual revaluation much more difficult. Since the purpose is that of allowing us to define somewhat more specifically the effects of a refunding operation, the revaluation can be conducted in an ex poste sense, that is, after both the payments stream and the capitalization rate have maintained their newer levels.
VII. Relation with Simons’ Proposal
Henry Simons, in his famous paper on debt policy,*110 suggested that all of the national debt should be refunded into consols or transformed into currency. The revaluation proposed here is a means of accomplishing the purpose desired by Simons without necessarily undertaking the drastic steps which he suggested. The revaluation proposal is based on an acceptance of the fact that the public debt instruments, as issued, will continue to fall anywhere along the spectrum between currency and consols, or more properly put, between currency and pure debt instruments.*111 The revaluation serves to separate these two aspects of the national debt. In a sense it represents a conceptual refunding along the lines which Simons suggested. By revaluing debt through the capitalization of the annual interest charge on the basis of the net yield on no-risk investment, we are essentially isolating that portion of the debt which can be refunded as consols, that form of debt as far removed from currency as is possible. The remainder (subtracting the first item from present market rather than maturity value) can be considered as being refunded into currency. This step alone will clarify discussion of the debt, and it would seem to be relatively less important whether or not an actual refunding along these lines takes place. The confusion which has been based on the fact that actual debt instruments possess features of both currency and pure debt would be substantially eliminated by reference to the account proposed.
In such a situation, the local government is merely taking advantage of an opportunity to make a profit through arbitrage. The differential between the rate of yield on municipals and on private bonds is, of course, exaggerated in the example. But so long as any differential at all exists, the operation outlined here is possible.
Local units of government are normally prevented by charter from investing in private securities. They may be allowed, however, to invest in federal government securities. And here, too, if a differential in rate should be present, local taxpayers can be relieved of a large portion of their normal public expenditure burden through a similar operation.
The implication of this appears to be that, if federal income tax rates should remain high, the stiffening of debt limit laws restricting local government borrowing or the removal of all investing opportunities may prove desirable.
All data employed in making this calculation were taken from Treasury Bulletin for September, 1957.