THE IMPACT of governmental fiscal activity on a country’s employment, income, and prices is of great economic importance. Not only is a significant share of national income disposed of through the government budget; more important is the tendency in many countries to manipulate the budget with a view toward regulating employment, income, and prices. The intelligent use of the budget for this purpose—or even the restriction of the budget to a neutral role—requires accurate knowledge of the budget’s effects on the economy. The size of the budget surplus or deficit is customarily accepted as an indicator of these effects. In most cases, however, surpluses and deficits as usually computed do not accurately measure the budget’s true deflationary or inflationary effect, and further adjustments are needed before this can be ascertained.1 An examination of the questions which arise in this connection is of special interest at the present time when, in forecasting, the prospect of budgetary deficits has again to be given serious attention. The present study suggests the corrections which should be applied to published budget statements to convert them into an economically meaningful form, as well as the further adjustments which have to be made to determine the inflationary or deflationary effects of the budget.


THE IMPACT of governmental fiscal activity on a country’s employment, income, and prices is of great economic importance. Not only is a significant share of national income disposed of through the government budget; more important is the tendency in many countries to manipulate the budget with a view toward regulating employment, income, and prices. The intelligent use of the budget for this purpose—or even the restriction of the budget to a neutral role—requires accurate knowledge of the budget’s effects on the economy. The size of the budget surplus or deficit is customarily accepted as an indicator of these effects. In most cases, however, surpluses and deficits as usually computed do not accurately measure the budget’s true deflationary or inflationary effect, and further adjustments are needed before this can be ascertained.1 An examination of the questions which arise in this connection is of special interest at the present time when, in forecasting, the prospect of budgetary deficits has again to be given serious attention. The present study suggests the corrections which should be applied to published budget statements to convert them into an economically meaningful form, as well as the further adjustments which have to be made to determine the inflationary or deflationary effects of the budget.

THE IMPACT of governmental fiscal activity on a country’s employment, income, and prices is of great economic importance. Not only is a significant share of national income disposed of through the government budget; more important is the tendency in many countries to manipulate the budget with a view toward regulating employment, income, and prices. The intelligent use of the budget for this purpose—or even the restriction of the budget to a neutral role—requires accurate knowledge of the budget’s effects on the economy. The size of the budget surplus or deficit is customarily accepted as an indicator of these effects. In most cases, however, surpluses and deficits as usually computed do not accurately measure the budget’s true deflationary or inflationary effect, and further adjustments are needed before this can be ascertained.1 An examination of the questions which arise in this connection is of special interest at the present time when, in forecasting, the prospect of budgetary deficits has again to be given serious attention. The present study suggests the corrections which should be applied to published budget statements to convert them into an economically meaningful form, as well as the further adjustments which have to be made to determine the inflationary or deflationary effects of the budget.

Accounting Corrections of Budget Figures

The terms “inflationary” and “deflationary” are used in this study to indicate additions to or subtractions from the amount of money expenditure, i.e., the flow of purchasing power. Under conditions of full employment, changes of this kind are accompanied by corresponding price changes; under other conditions, the accompanying changes in production are likely to be more important.

The inflationary or deflationary effects of government income and expenditure thus defined can be approximately measured by ascertaining the net budget deficit or surplus. The accurate calculation of the net deficit or surplus is seldom a simple task, however, and published figures usually have to be qualified by corrections.

Before a budget can be satisfactorily analyzed in order to determine the corrections needed, its legalistic, conventional form may have to be changed to one with economic significance. The economic significance is to be found in the impact of total government receipts and expenditures upon the economy during the period for which the budget is computed. To obtain such a measure, data on various extra-budgetary governmental activities, e.g., receipts and disbursements by government trust funds (for example, social insurance funds) and profits and losses of government commercial enterprises, have to be combined with the budgetary data. In this connection, it may be desirable to consolidate the accounts of other government units with those of the national government. Expenditures in any given year which have been charged to a previous budget rather than to the current budget must also be added, e.g., payments of accrued interest obligations on maturity of government bonds. On the other hand, items for which budgetary outlays are recorded but which have not involved a cash payment should be deducted from total expenditure, e.g., charges for accruing interest obligations on government debt where interest is not paid currently.

In some countries a sizable volume of purchases is both budgeted and paid for after acquisition. Such expenditures should be transferred to the budget of the period in which acquisition actually occurs. An adjustment of a similar kind is needed for long-term government construction projects carried out by private contractors even though they may be both budgeted and paid for at the time of completion; these should be treated as government investments (financed in effect by borrowed funds) equal in value to the work done each year, for the inflationary impact must occur at the time when the work is done.

If the ratio of government expenditure to national income is desired, intra-governmental transactions which appear more than once in the budget have to be eliminated. On the other hand, in measuring merely the budgetary surplus or deficit, any double counting of this kind will cancel out. There is, however, one exception to this automatic canceling. If purchases of government bonds by a governmental agency are recorded as an expenditure in the budget, they must be eliminated, since the offsetting transactions—sales of government securities—will not be included in the budget as an offsetting “revenue” item.

The balance remaining after these adjustments corresponds approximately to the change in government debt and cash holdings during the fiscal year. Further adjustments may be required because of various special transactions. Devaluation profits (or revaluation losses) and write-downs of the value of the government’s capital assets must be eliminated, for they are purely bookkeeping items not involving money transactions. Since purchases of gold or silver from domestic sources place new money in circulation, they are inflationary like any other purchases of goods and services and therefore should be included among expenditures when the cash deficit is computed. This is true whether the government makes payment from its existing cash and deposits or makes a costless payment by creating new money. Seigniorage profits should not be included in revenues, and expenditures made with these profits should therefore be counted as deficit spending. In the same way, new currency issues used for financing expenditure should be counted as deficit spending even though they do not constitute a drain on government cash and deposits.

To the extent that governmental expenditures are made abroad, the governmental budget deficit, normally an inflationary force, does not affect the domestic income stream. These expenditures should therefore be removed from the budget—unless the balance of payments is simultaneously examined for an offsetting, deflationary item (imports for government use)—since otherwise there would be a deficit, indicating an inflationary impact which in fact does not exist. Allowance must be made here, however, for the influence of import restrictions. Shortages of foreign exchange may necessitate variations in the amount of private imports equal and opposite to variations in government imports; in this case government expenditure on imports should generally be considered inflationary.

For reasons similar to those outlined above, government revenues from abroad—interest receipts or payments for services—should be excluded from budgetary revenues. Local currency proceeds from such sources as the sale of ECA imports should also be excluded if the view is taken that these imports involve essentially a private (international) rather than a governmental operation; their deflationary effect will then be attributed to (private) international trade rather than to a budgetary surplus. Likewise, government expenditures for financing an export surplus (e.g., ECA grants by the United States) should be deducted from budgetary outlays if their inflationary effect is attributable to private (international) activity rather than to any government deficit resulting from the expenditure, i.e., if the government expenditure constitutes merely compensatory official financing of a balance of payments surplus which can be considered as privately generated rather than constituting in effect a purchase of goods and services by the government itself. Similarly, the payment of members’ quotas to the International Monetary Fund has an inflationary effect on the economy of the paying country, if at all, only when it is drawn upon to finance a balance of payments surplus, and at that point the payment should be considered as compensatory official financing of a payments surplus and therefore eliminated from budget expenditure. This procedure is in conformity with the customary procedure when balance of payments deficits or surpluses are financed by government exchange stabilization funds: the government’s gain or loss of local currency is excluded from the budget revenue or expenditure.

The Budget’s Inflationary Impact

The corrections so far listed may for the most part be regarded as accounting corrections for which, in general, sufficient information can be found in published government accounts. The cash budget deficit or surplus, thus corrected, gives a first approximation to the direct effect of fiscal activities on income (the “multiplicand” of the so-called multiplier analysis), i.e., the additions to or subtractions from the flow of purchasing power arising directly out of budget transactions. This will subsequently be referred to in this paper as the “income effect” of the budget. If this cash budget is in balance, government adds to the flow of purchasing power no more than it takes from it. To measure this effect more accurately, however, further refinements are required, for which estimates, rather than accounting corrections, will have to be made. Certain revenue collections do not necessarily constitute a withdrawal from the flow of purchasing power, and certain payments do not necessarily constitute an addition to it. For example, since revenues collected from idle hoards are nondeflationary, a balanced budget based on such receipts would be inflationary. Likewise, certain expenditures going directly into recipients’ idle hoards are not inflationary, and a deficit due to such outlays would not be inflationary. All adjustments of this kind which are discussed below might in some circumstances be applied more easily to changes from the preceding year in the various budget items than to the totals of these items. A recasting of the analysis for this purpose would, however, involve no significant change in its substance; therefore, all references to the inflationary significance of the total values of any item may be understood as applying equally to changes in the item.

Assumption of elastic credit supply

Revenues paid out of taxpayers’ savings or capital may be treated provisionally as nondeflationary, and expenditures going directly into the recipients’ capital, or, in some cases, into current savings, as non-inflationary. This involves the assumption of complete elasticity of the credit supply and complete mobility of funds throughout the capital market. (Alternatively, the assumption might be made of a credit supply with some mobility and elasticity and a demand for credit which is independent of the interest rate within the relevant range.) All studies which measure the inflationary effect of the budget by reference to the budget deficit in fact implicitly assume either mobility and elasticity in the credit supply or deficit financing by a central bank which is not obliged to vary credits to others as a result of its government credit operations.

This assumption can be expressed in another way by postulating that all revenue collections out of savings or capital are paid exclusively—ultimately or directly—out of idle holdings of cash or bank deposits or idle funds in the sense of bank loan-making capacity which would not otherwise have been utilized; and, similarly, that all payments going directly into capital or current savings add to these balances. Unless this postulate is valid, i.e., unless credit is elastic and mobile, a government deficit not financed by the central bank cannot be wholly inflationary. For, unless commercial banks have cash holdings substantially in excess of their normal minima, a government’s use of loanable funds must, if the central bank creates no new liquidity, deprive some private investors of funds needed for their projects; hence such a government deficit would not create an equivalent net addition to the flow of purchasing power. Likewise, insofar as a budget surplus used to repay the public debt held outside the central bank results in an expansion of private investment, the surplus is not deflationary.

Adjustment of revenue figures

Income and capital taxes. Granted the above assumptions about credit, it follows that capital levies, estate taxes, taxes on capital transfers (e.g., stock exchange taxes), the part of income taxes paid out of current savings, and revenues derived from the sale of assets not newly created (land, securities, and possibly buildings) are in general not deflationary forms of government revenue. Such revenue should be added to the cash deficit (or deducted from the cash surplus) in order to measure the true inflationary impact of the budget. This procedure will require qualification in the case of capital taxes which are in fact paid out of income and lead to a reduction of consumption.

In practice, of course, the assumption of credit elasticity does not hold fully. Nevertheless, a sound working hypothesis is to eliminate taxes on capital and to adjust, suitably, for taxes on income. It may be impossible to determine within reasonable limits of accuracy what part of income tax collections is actually paid out of savings, for the marginal propensity to save (the proportion of any change in net income which is embodied in a change in saving) may vary over time and/or among income levels and be very difficult to measure. The use of the marginal propensity to save in conjunction with the large deductions which taxes make from income is justified only if it remains approximately constant as income changes. If the propensity is not approximately constant, an average value for the range of income involved must be used.

The attainment of a high degree of precision here may not in itself be a matter of any great importance, however. Since the important indirect effects of the budget on private activity necessarily have to be disregarded, it is scarcely worth while to insist on great precision in the allocation of taxes. But where, as in the United States, income taxes are a very large part of total revenue, some adjustment must be made; the greater degree of precision thereby attained would, however imperfect, still have some real significance. For a budget to which items of this kind make a substantial contribution, the significance of these adjustments will always be considerable, because their importance has to be assessed in relation to the cash deficit and not to the budget as a whole. At the very least, a “balanced” budget of this sort should be characterized as inflationary.

Business income taxes. The effects of business income taxes are especially difficult to measure. In part they represent profits which otherwise would have been paid out to individuals as income, and to this extent they could be treated as deflationary in the same way as personal income taxes. In part, business income taxes represent funds which would have been invested within the business taxed; but insofar as the assumption that loanable funds are easily obtained is tenable, the blocking of self-financing should not be considered a deterrent to investment. On the other hand, the fact that taxes must be paid reduces the desirability of investment; insofar as the investment discouraged would have been made out of purely domestic resources, the tax is on this count deflationary; insofar as the investment would have been made out of imported resources, it is not.

Income taxes and imports. Even where income taxes are paid at the expense of consumption rather than of savings, their effect is not deflationary to the extent that the reduction in consumption is concentrated upon imports. The effect of the budget on the volume of imports might, of course, be disregarded on the ground that from the point of view of the monetary authorities the balance of payments is as important as inflation or deflation, and that it is the impact of the budget on purchasing power plus balance of payments which is significant. This interpretation is not always valid, however; the effect of the budget on the total flow of purchasing power within the country is of at least equally great interest. Granted the objective of measuring changes in purchasing power, it might still be argued that the budget’s effect on the balance of payments is so indirect that account should be taken of it only in the balance of payments schedule. But just as it has been agreed to make an adjustment in the government’s accounts for the deflationary effect of taxes paid out of savings when actually that correction appears in the savings-consumption-disposable-income schedule, so the effect of taxes on the balance of payments should be applied as an adjustment to tax revenues in the budget.

Therefore, allowance should be made for the marginal propensity to import. This is not the propensity estimated for the country as a whole, but only that which relates consumer imports to disposable personal income. It may vary widely according to the level of individual income, and its measurement may be very difficult. However, this does not justify neglect of the correction, for in many countries the marginal propensity to import for consumption may be as high as 40 or 50 per cent. If income-tax payers have a negligible marginal propensity to save but a marginal propensity to import of the order of 40 or 50 per cent, personal income tax revenues have to be reduced by 40 or 50 per cent for an estimation of the budget’s income effect. Thus neglect of this consideration may produce serious errors; even a rough, approximate adjustment would be better than none.

Such an adjustment cannot be made without reference to import regulations. Where import restrictions exist, it is necessary to consider separately the effects of additions to taxation, which may lower imports, and reductions of taxation, which may not be permitted to cause increases in imports. An increase of taxation may in some countries merely result in reducing the amount of the unsatisfied demand for imports, without changing the effective demand. If it goes further and curtails the effective demand for imports by those on whom the tax falls, the reduction of these imports may or may not be offset by relaxing restrictions on imports desired by other classes in the community. Therefore, in calculating the deflationary effects of tax increases or the inflationary effects of tax reductions, the adjustment made for the marginal propensity to import must be limited to the amount of the actual change in imports which the import restrictions permit.

Excise taxes. Excise taxes should be considered when the budget’s income effect is measured. Insofar as these taxes are paid by the producer (and his employees and suppliers), they constitute a drain on his money income, rather than on consumers’ real income, and should be treated in the same way as income taxes. Excise taxes paid by consumers require a different treatment. They do not reduce disposable money income and therefore need not affect money saving. However, if—as is likely—the average propensity to consume or to save is in part a function of real as well as of money income, excise taxes may also to some extent be paid out of savings. For, since they increase the cost of living, they have much the same effect as income taxes in reducing the real disposable income corresponding to any money income, and therefore they divert part of any given money income from savings to consumption.

The case for adjustment is weaker here than with respect to income taxes, however; in general, excise taxes tend to affect individuals with low incomes and therefore low average and, probably, low marginal propensities to save, while income taxes are paid primarily by those with high incomes and high marginal propensities to save. Moreover, even if the propensity to consume or to save may be assumed to be entirely a function of real income, the reduction in savings attributable to excise taxation will be less than is indicated by the marginal propensity to save. Given the reduction in real income caused by the cost of living increase which is due to the collection of an excise tax, real saving will be reduced by that amount multiplied by the marginal propensity to save. But by virtue of the increased cost of living, even an unchanged quantity of money saving will have a diminished value in real terms. Consequently, part of the needed reduction in real saving takes place even if money saving does not decline at all, and a reduction in money saving smaller than that indicated by the marginal propensity to save will suffice to produce the required reduction in real saving. The decline in money saving could even be zero. This will be true when the marginal propensity to save is equal at all relevant income levels to the average propensity to save, for then real saving is always the same proportion of real income. Hence money saving from any given money income is constant whatever the price level, so that excise taxes which are felt as price level rises do not cause savings to diminish; that being the case, the taxes are paid entirely at the expense of consumption. Excise taxation is thus more deflationary than income taxation, as far as its direct effects on the flow of purchasing power are concerned. The excess of the actual tax revenue over that part of it which is truly deflationary may be only (say) half as great for excise taxes as for income taxes.

However small the effect of an excise tax may be on savings, a low marginal propensity to save need not be associated with a low marginal propensity to import and—as with income taxes—the deflationary effect of excise tax revenues may have to be reduced to the extent to which they are paid at the expense of imports (provided that imports are free to vary). Where untaxed imports are substitutable for taxed domestic goods, care should be taken in applying the marginal propensity to import as a correction. For any reduction in imports due to the effect of excise taxes on real income would be counteracted to some extent by an increase in imports due to the relative cheapening of imports in comparison with the newly taxed goods. Hence the excise tax would diminish imports to a smaller extent than is indicated by the marginal propensity to import, which means that its deflationary effect would be proportionately strengthened.

Import taxes. Import taxes (tariffs, unfavorable import exchange rates in a multiple rate system, etc.) comprise a very important part of budgetary revenue in many countries. Therefore, adjustments applied to this category of revenue would be highly significant. Provided that import controls do not impose a prior limitation, import taxes will reduce the quantity of imports by raising their cost. Consequently they will produce two effects: increased government revenue (deflationary) and a larger export surplus (inflationary). Since the latter at least partly offsets the former, the total volume of import tax revenues overstates their deflationary effect. The taxes are collected in part or in full from the foreign exporter, whose sales volume is diminished, rather than from the domestic consumer, and thus may have their deflationary effect in the foreign rather than in the home economy.

Where the elasticity of demand for taxed imports is unity in the relevant range, total expenditure on these imports is unchanged; the same amount of purchasing power will then go out of circulation, whatever the rate of import taxes. The more going to the government, i.e., the higher the import taxes, the less will “leak away” in the form of payments to foreigners for imports. Thus complete national accounts would show two equal (nominally) inflationary and deflationary components: improved balance of trade, increased government revenue. In a study confined to the income effect of the budget, the two should be consolidated, i.e., canceled out against each other.

This conclusion does not hold, however, if a tax is considered to have an “inflationary effect”—irrespective of its effect on the level of money demand—if it raises prices. Import taxes on goods with unit demand elasticity would then have to be counted as inflationary, for, although total money expenditure is unaffected by them, the quantity of imports is reduced by the taxes and therefore the total quantity of goods on which the given amount of money is spent is diminished. In consequence the price level, or, more precisely, the prices of imports, must rise. This kind of inflationary effect is less important, however, than the one usually considered; price increases caused by the imposition of import taxes are not accompanied by any direct pressure on the supply or the prices of domestic resources.

Where the elasticity of demand for taxed imports exceeds unity, the effect of import taxes is inflationary on either definition of inflation, for the increase in the cost of imports reduces the total outlay on imports (including their import tax component); purchasing power will therefore be diverted to domestic uses.2 Where the elasticity of demand is less than unity, import taxes have some deflationary effect since, as the cost of imports increases, total outlay on them also rises. Hence part of import tax revenues is collected at the expense of purchasing power which would otherwise have been available for domestic uses, rather than at the expense of foreign exporters’ purchasing power. To calculate the deflationary effect of such taxes, an approximate correction on this account may be made by omitting a proportion of tax revenues equal to the (properly weighted) elasticity of demand for taxed imports; e.g., if elasticity is one, import taxes should be omitted from the budget revenues; if elasticity is ¼, they should be reduced by 25 per cent.

These corrections are valid only under the assumption of an infinitely elastic supply of imports, and unless the import supply elasticity is very high or the demand elasticity is close to 1 a more general correction formula will be needed. This formula can be derived easily:


where Ng and Nd are the supply and demand elasticities, respectively, for taxed imports. As Ng becomes large the formula approaches import tax revenue multiplied by Nd, the correction given in the last paragraph. (Note that where tax rates are high—say 50 per cent of the ex-tax price—the formula becomes unreliable.)

The deflationary effects of import taxes will also be modified by their effects on real income and thereby on savings. It was suggested above that the effect of excise taxes on savings would be relatively small since they are generally paid by the nonsaving poor. Import taxes, however, in many countries are levied to a large extent on the luxury imports of the rich. Hence their effect on savings should not be disregarded. The reduction in real saving which may result from such taxes equals the product of the marginal propensity to save and the fall in real income caused by the taxes, but the pertinent factor—the reduction in money saving—will be a somewhat smaller amount (see the section on excise taxes, above). The fall in real income caused by the taxes is (approximately) the entire tax revenue and not merely the deflationary portion of the taxes. This can be illustrated by the unit-demand-elasticity case: although the withdrawal from domestic money expenditure caused by the taxes is nil, the quantity of goods purchased with that expenditure must be reduced, since part of the expenditure is diverted to the payment of taxes. The amount of taxes paid approximates the sacrifice of goods and, hence, the reduction in real income. (If foreign supply is not completely elastic, the reduction in real income will be less than this amount, since part of the import tax is covered by a reduction in the price which the foreign exporter charges.) After this savings effect is taken into account, the net inflationary value of an import tax is found to equal the revenue from the tax minus the part of the tax paid at the expense of foreign exporters and minus the associated reduction in savings.

Special treatment is required for a country whose import tax revenues are derived in significant amount from taxes levied for the purpose of equalizing the tax burden on imported goods with that on competing goods produced domestically. The effect on the value of imports of the collection of equal taxes on foreign and domestic supplies of a given product would differ from the effect of the collection of a tax on foreign supplies only; in the former case, the demand elasticity for the foreign product as indicated by the change in the quantity of imports accompanying a given change in tariff would appear lower. Under the assumption that foreign and domestic supplies would maintain their relative shares of the market, this adjustment can be handled by attributing to the demand for imports of the product the same elasticity as that for the product as a whole. On this basis, the correction proposed above can be utilized.

Adjustment of expenditure figures

Noninflationary expenditures. Among budget expenditures, war damage compensation payments (in many cases) and outlays for the purchase of real property not newly created are noninflationary payments into capital, and therefore should be deducted from the cash deficit or added to the cash surplus in determining the budget’s income effect. Purchases of “old” business securities also would fall in this category. Strictly speaking, loans to private business, if included among budget expenditures, should be eliminated, since they too are payments into capital. In many countries, however, there will be in practice few such loans except where risk or the scarcity of credit rules out the alternative of bank or saving financing. In that event, the loans will make inflationary expenditures possible; whether these are attributed to the budget or to private investment will depend on the scope of the investigation being made.

Transfer payments. A distinction should be made between payments for the purchase of goods and services and transfer payments, such as pensions, bonuses, interest on debt, consumer subsidies, etc. A transfer payment merely restores to the taxpayers funds collected from them. The government, in making these expenditures and collecting the taxes needed for their financing, plays no role therefore, as far as the expenditure of purchasing power on goods and services is concerned. This statement must be qualified to some extent, however, since the taxpayer and the transfer-payment recipient may have different marginal propensities to consume or to import with respect to disposable income; insofar as transfer payments and the taxes which finance them together involve a redistribution of income to the lower income groups, they may be somewhat inflationary. On the other hand, during a year in which their value is increased, such tax-subsidy operations may also produce a counteracting deflationary effect, for they involve declines in the income velocity of circulation. This is because they involve two extra (though “nominal”) turnovers of the active money supply—the payment of taxes and of transfers—between the receipt of income by the taxpayers and its expenditure by the transfer recipients.

Where subsidies applied to specific consumer expenditure items are important in the budget, it will be worth while to attempt to measure their inflationary significance by the method used in finding the deflationary significance of excise taxation, rather than by the simpler method applicable to income taxation. This follows from the fact that subsidies, like excises, have a direct effect on the price level rather than on the amount of disposable income. Whichever procedure is adopted, the import subsidies paid in a multiple exchange rate system by means of an overvalued import exchange rate should be combined with the more obvious forms of subsidy.

Expenditures on goods and services for the government’s own use constitute a withdrawal of goods and services from the private sector of the economy, and hence constitute inflationary pressure. But transfer payments, having something of the character of gifts, are not a claim on goods and services or an addition to national income. Thus, as they are spent by the government, they have no inflationary impact. It is only as they are re-spent by transfer payment recipients that there will be such an impact. Hence, in finding the contribution to the flow of domestic money demand created by such budget expenditures, the portions of the total which the transfer recipients save or spend on imports must be disregarded, and only that which they spend on domestic resources must be counted.

Under conditions of full employment, the distinction between these two types of expenditure is of special importance when the budget is studied for its effect on the price level or on the creation of forced or undesired savings. Under such conditions, the diversion of productive resources or their products to the government, due to a deficit expenditure by the government, reduces the supplies available for the rest of the economy and therefore adds to the inflationary pressures; transfer payments, however—which do not constitute any such claim on goods and services—do not cause such pressures at the first “turnover” of the funds.

On the other hand, where there is unemployment and underproduction, government expenditures on goods and services will not divert resources from the use of the private sector, for the government actually “creates” the productivity of the resources which it wants for its own use. Whether the government is adding to disposable incomes through transfer payments or through purchases of goods and services is accordingly of little consequence. For example, employees who are kept on the government payroll in times of depression may still be regarded as productive labor; but if they were released by the government, they would have to be supported instead by relief (a transfer payment). The inclusion of the entire amount of the payments to these individuals in the measurement of the budget’s inflationary impact merely means that these individuals are considered as productive of national income (government services) and hence as employed, rather than as unemployed recipients of transfer payments. It does not mean that the whole of the payment to them involves a corresponding pressure on the prices of productive resources.

However, even in conditions of underemployment, the budget’s effect on income as well as on prices or forced savings is of interest. Hence, to measure this effect, the proper procedure is to reduce transfer-payment expenditures alone by the amount they increase the recipients’ savings and imports (as contrasted with their consumption of domestic products), and to accept without such adjustment any domestic goods-and-services expenditures, even though their effect on the price level may be the same as that of transfer expenditures.3

The decision whether public debt interest payments are to be defined as involving goods-and-services expenditures or transfer expenditures is important, both because debt interest recipients have higher marginal propensities to save than other transfer payment recipients and because debt interest is such an important part of total expenditures, e.g., over 15 per cent of the total in Canada and Switzerland and over 10 per cent in the Netherlands, Turkey, and the United Kingdom. In the definition of interest payments, the practice adopted is not uniform. If the debt is considered as incurred for productive investment, interest on it is customarily treated as a goods-and-services expenditure; otherwise, the interest is generally treated merely as a transfer payment.

Effects on investment and exports

No consideration has been given thus far to the marginal propensity to invest and the marginal propensity to export. Likewise, no mention has been made of the “acceleration principle”—the principle whereby inflations and deflations once started can feed upon themselves and become cumulative. Disregard of these factors is justified for two reasons.

First, their usefulness is less certain and their statistical measurement more difficult than for either the marginal propensity to save or the marginal propensity to import. Second, the acceleration principle applies to the working-out of the consequences of any inflationary impact and therefore plays no part in determining the size of that impact. (This must be qualified in dealing with the problem of induced tax revenues discussed below, for the amount of revenue induced does depend on the working-out of the consequences of an inflationary impact.) Similarly, the marginal propensities to invest and to export can be regarded as significant factors only in the working-out of the consequences of an inflationary impact. These propensities express the relationships of changes in investment and in exports to changes in national income—not to changes in the distribution of net income or of expenditure, at a given level of national income, between the government and the private sectors. But only relationships of the latter type can justify an adjustment of the cash budget deficit in order to ascertain the true inflationary impact of the budget, for they are the only relationships which cause the net addition to the flow of purchasing power associated with government revenue-expenditure activities to differ from the conventional cash deficit.

Exports and investments do, of course, have such relationships. Assume, for example, an increase in government expenditure which is matched by an increase in revenue whose deflationary effect (measured as described above) is equal to the expenditure’s inflationary effect. In this case no change in income should be produced. However, the reduction in private expenditure on domestic resources caused by the taxation increase may affect exportable products (and resources which produce them) to a greater extent than does the equal increase in government expenditures on domestic resources. Hence the supply, and therefore the value of exports, would tend to change, and this change would itself constitute an inflationary or deflationary impact. Similarly, if private investment is dependent not only on the level of national income but also on the level of private income net of income taxes, the otherwise neutral budgetary changes described above must cause a reduction in private investment and will thus be deflationary.

Although it is possible to find examples of export and investment relationships of a sort which make adjustments of the budget necessary, the more important relationships are probably those involving total income and not requiring adjustments of the budget. Here the inflationary impact of the deficit may evoke changes in investment and exports, but such changes occur only because the deficit is inflationary; they do not contribute to the determination of whether the deficit will be inflationary or not.


The preceding discussion is summarized in Chart 1. The budget’s inflationary impact is equal to the difference between expenditures creating an addition to the flow of domestic money expenditures (lower section of the first column) and revenues creating a withdrawal from the flow of domestic money expenditures (lower section of the second column). All other budgetary expenditures and revenues are without effect on the domestic money demand for currently produced domestic goods and services.

Chart 1.
Chart 1.

Inflationary Impact of the Budget

Citation: IMF Staff Papers 1951, 001; 10.5089/9781451949322.024.A003

The relative sizes of the items in the chart are merely illustrative. It would be possible for the budget to have an inflationary effect even though there were a cash surplus (which would appear in the first column of the chart). Similarly, a cash deficit might be accompanied by a deflationary impact for the budget as a whole; the lower part of the second column would then be larger than the lower part of the first column.

Implications of the Budget’s Income Effect

Independence of income effect from actual income changes

Since the intention of this study is to measure the budget’s contribution to the stream of purchasing power, it is of no consequence that deflationary forces from other sectors of the economy may offset the budget’s income effect so that prices and incomes are unchanged. A budget deficit used to moderate the decline in incomes and prices when private investment falls is as inflationary, in the sense in which the term is used in this paper, as an equal deficit which produces increases in income and prices when private investment is stable.

Distribution of inflationary effects over time

During a process of continuous inflation, it is likely that the deficit and the associated inflationary impact will tend to be concentrated at the end of the year rather than at the beginning. In these circumstances, the addition to the total of income which the deficit produces will not occur entirely within the same year, since the working out of the multiplier takes time. Even with an equal distribution of the income effect during the year, part of the multiplied effects of that portion of the total income effect which occurs in the last part of the year will affect the total income of following years. Hence the deficit of the year studied cannot be considered as entirely inflationary during that year. However, if a deficit of a similar rate existed during the latter part of the preceding year, no important error is introduced if all of the excess in the current year’s income above the level associated with a zero income effect is attributed to the current deficit.

In computing the time distribution of the income effect, it may be desirable to allocate the deflationary effect of tax collections to the periods during which the taxpayers—especially business taxpayers—accumulate reserves for meeting expected tax obligations, rather than to the points of time when tax collections are made. Thus, even though tax receipts may be concentrated in particular months of the year, the conclusion that the budget had a deflationary effect in those months would not be justified. Similarly, in calculating the inflationary effect of any annual deficit, it may be desirable to “date back” the part of tax revenues received which was actually accumulated by the taxpayer in the preceding year; in the year under consideration they constitute payments out of “capital” and hence are nondeflationary. This adjustment would be important, however, only for a year in which tax rates were changed; otherwise, current tax payments out of past accumulations of reserves for taxes would tend to be equal to current additions to reserves for future taxes.

A study which disregards the lag in the time of expenditure of transfer payments on goods and services relative to the direct disbursement by the government of other outlays on goods and services may slightly overstate the value of (a change in) the income effect. Transfer payments spent by the recipients on goods and services are counted as inflationary during the period in which they are received but not necessarily spent. Hence the positive income effect of a budget which includes transfer payments is overstated. However, unless there is a marked change in the rate of transfer payments during the period for which the budget is computed, the errors will not be important. Similar reasoning applies to income taxes (which, in effect, are negative transfer payments), since taxpayers may not cut their expenditures at the same time as their net incomes are reduced by increases in taxation.

Consistency of inflationary and deflationary income effect with income stability

Increases in savings and in imports, induced after the beginning of the budget period by the prior increases in income which result from the part of the deficit previously realized, are among the deflationary forces from other sectors of the economy which are disregarded when the budget’s inflationary impact is measured. By perhaps the middle of the year for which the budget is computed, the portion of the year’s deficit already realized may have produced such an increase in the level of savings and of imports (via the increase in income for which it is also responsible) that during the remainder of the year the increased savings and imports completely offset the inflationary pressure of the budget deficit. This situation will be indicated by a cessation in the rise of the income level.

Therefore, “inflationary significance” must be understood to mean the effect of the budget in increasing the total of income earned during the budget period beyond what it would have been in the absence of the so-called “inflationary” income effect. The inflationary significance does not necessarily indicate a force which is continuously inflationary in the sense of producing continuous increases in the level of income; a part of the effect will involve such increases but the rest merely sustain an increased level of income already established. This distinction is illustrated most clearly by a deficit which has been maintained at a constant rate during two or three successive years. During the second (or third) year there may be no change in incomes and prices, ceteris paribus, for during the first (or second) year there will have been induced increases in the levels of savings and imports, which will continue to counteract the inflationary effects of the budget deficit during the second (or third) year. The deficit of the latter year is therefore inflationary, not in the sense of causing further increases in the level (rate) of incomes and prices, but rather in the sense of sustaining the existing higher level. Without the inflationary impact of that deficit, incomes and prices would have to fall.

This example is not to be taken as implying necessarily that continuous budget deficits of constant size will cease after a time to raise the level of prices and income. There is a tendency for such a stabilization to be produced; and if the deficit is initiated in conditions of underemployment, or if imports can be permitted to rise relative to exports, stability will probably be reached. The level of savings and of imports can be expected to rise sufficiently to offset the deficit’s inflationary effect. But if the deficit is initiated during a period of full employment, the inflationary effect will take the form of price level rises and these may discourage adequate increases in savings. With that brake on inflation removed, there is theoretically no limit to the amount of price level inflation which can occur.

The inflationary significance of a budget deficit also needs careful definition if the deficit is unevenly distributed during the budget period. Assume as an illustration a country having some unemployment or some means of securing an (increase in the) import surplus so that an expansion of real income or at least of real consumption (accompanied by some increases in prices) is possible. If a deficit is initiated in one year and is entirely concentrated in the first half of that year, the first half year would show rising incomes (inflation) but the second half would probably, ceteris paribus, show falling incomes (deflation), since the deflationary impact of the higher rate of savings and imports associated with the higher income (or consumption) level would not continue to be counteracted by an inflationary impact from the budget. As a result, prices and incomes by the end of the year might even be at the level of the beginning of the year—the same level that would have existed with a zero income effect. Thus again the budget’s income effect must be treated as inflationary only in the sense of creating an increase in the total amount of income earned during the year, and not necessarily in the sense of producing increases in the rate of income and employment or prices at the end of the year compared with the beginning of the year.

Real or relative value versus money value of income effect

Two possible purposes for measuring the income effect of the budget should be distinguished. One purpose may be to express the net addition caused by the government to total claims on goods and services as a percentage of existing claims. In that case, the value of the income effect for each month of the year should be deflated by an index of the current income level. (In the absence of monthly data, the ratio of the total income effect for the whole year to total income for the year may be used, provided that the monthly incomes or the ratios of the monthly income effects to monthly incomes do not vary too widely.) The significance of measuring the income effect in this way arises from the fact that the pressure of the deficit on prices and employment varies with the size of the income effect relative to current income. This is particularly important under conditions of full employment, since income variation then implies more-or-less proportional price variations. The absolute values of portions of the income effect realized in different parts of the year are then not always comparable, since they may correspond to a variety of purchasing powers. With a constant rate of effect in money terms, the real value of successive portions of the effect and of the net real claim on goods and services will be constantly diminishing. Only when (and if) income and prices reach a plateau will this real value become stable. Hence deflation by current money income is necessary to produce meaningful figures.

The second purpose is to measure the budget’s effect for the multiplier; in this case, the income effect should not be deflated by current income. The fact that deficits realized during earlier parts of any year will raise income and will therefore cause voluntary savings to increase does not mean that later components of the deficit should be considered less inflationary (in the sense of income-creating) than earlier components. But deflation of an unchanging level of the deficit by rising levels of income would lead to such a conclusion. Hence for multiplier purposes the income effect should be expressed in absolute terms, or, if international comparability is sought, as a fraction of the income of some period in which the budget had either a “normal” or a zero income effect.

Induced tax revenues and inflationary significance of deficit

Theoretically, the budgetary injection of purchasing power should be measured without regard for the counteracting leakages which are the indirect result of that injection (induced savings and imports). Unfortunately, a deduction for some of these leakages is automatically made when the budget deficit is computed, and this deduction cannot be eliminated easily. The difficulty arises because funds leak out of the flow of purchasing power into tax payments as well as into savings and imports. There is (conceptually) a marginal propensity to pay taxes just as there is (conceptually) a marginal propensity to save or to import. The rise in income and prices engendered by the deficit increases imports and hence import tax payments, and also domestic consumption, and hence consumption tax and, of course, income tax payments.

The situation can be illustrated most clearly in terms of the leakage through imports and on the assumption that the price elasticity of demand for imports is unity. Assume an excess of government expenditures over government revenue at a time when the marginal propensity to save (minus any marginal propensity to invest) is zero (Case one). This will cause incomes to rise until the rate of imports has increased enough to offset the inflationary impact of the fixed rate of government deficit spending. But now suppose that in an otherwise similar situation some part of the given amount of revenue is derived from a 100 per cent ad valorem tax on imports (Case two). Again incomes will rise, because of the inflationary impact of the budget deficit, until the public’s expenditure on imports (which, however, now includes import taxes) is once more equal to the amount of the impact. Clearly the injection of purchasing power created by the budget is equal to the inflationary impact (or, in other terms, the multiplicand is equal to the impact). However, instead of being offset by the leakage from an equal (induced) increase in imports, the initial rate of income effect is now offset by a leakage in the form of an increase in imports—as seen from the standpoint of the foreign exporter or of balance of payments accounting—of half the amount plus a like leakage in the form of an increase in import tax payments. Since the leakage into revenue will be concealed in the form of a reduction of the deficit, the visible inflationary impact will be (something more than) one half as great as in Case one. However, the effective inflationary impact of the budget is clearly the same in both cases and—as argued above—the first gives the proper measure of that impact.

The same conclusion would be drawn from the two cases only if the induced tax revenues in the second case were eliminated. But for that to be done information would be necessary which in practice is not available in sufficiently accurate form, viz., the various marginal propensities, including the propensities to pay taxes (needed both for eliminating nondeflationary and noninflationary revenues and expenditures from the budget and for finding the total, multiplied effect on income and thereby the value of the part of that total effect which goes into induced tax payments), and the distribution of the budget’s income effect during the budget year and the speed with which the multiplier works itself out, i.e., the speed with which the inflationary impact contributed by the government brings about ramified increases in the incomes of other sectors of the economy (needed for finding what part of the total amount of induced tax revenue increases which must occur will occur during the year for which the budget is being investigated and will influence the size of that budget’s deficit).

Theoretically the problem could also be avoided by correcting the coefficient by which the initial impact is multiplied to express the total effect. In “multiplier” terminology, the case of the import tax involves, in comparison with the no-tax case, a decrease in the marginal propensity to import offset by the creation of a marginal propensity to pay import taxes. It is permissible in the multiplier technique to consider as multiplicand the actual (adjusted) budget deficit (gross of induced tax revenues) and to eliminate from parts of the multiplier the marginal propensity to pay taxes; the products of the two alternative multiplicands and their respective multipliers will be the same.

In practice, however, this way of avoiding the difficulty would also require the use of some of the information which is not obtainable. Moreover, the value for the budget’s inflationary impact presented in budget deficit analyses will normally be given without any qualifications concerning the size of the appropriate multiplier. The data which are of the widest interest and which will actually be presented (annual budgetary income effects in money or deflated terms) do not contain the adjustment; comparison of the income effects of different years or among different countries will therefore not be entirely meaningful. (This is, of course, also true insofar as the marginal propensities to save and import differ among years and among countries.)

Even if the appropriate multiplier is presented in conjunction with the effect of the budget on income, inclusion of induced revenue in the revenue component of the effect diminishes its practical value. The budget’s inflationary impact ought to show the size of the anti-inflationary measures (e.g., reduction in goods and services expenditure) which the government should take or should have taken. But where the size of the inflationary impact computed is influenced by tax revenues which result from that impact, the impact will be an underestimate of the anti-inflationary measures needed; insofar as inflation is avoided the induced revenue will not appear, and hence when the avoidance of inflation is considered such sources of revenue must be ignored.

The induced taxes problem might not seem to rise in instances where a budget deficit merely prevents income from falling (rather than actually raising it), for here there is no induced change in tax revenue. But if even in these circumstances the budget deficit must be considered inflationary because it causes a deviation from the existing income norm (see section above, Independence of income effect from actual income changes), it must be considered as having also an expansionist effect on those tax revenues which are dependent on the level of income. Hence consistency requires that only part of the observed tax revenues be counted and therefore that the income effect be defined as a value larger than indicated by the deficit actually observed.

The reasonableness of this procedure can be illustrated by comparing the stable-income situation with one in which the deflationary impact of a fall in private demand is not immediately offset by an increase in government expenditures. In that case, when the expenditure increase is finally made, tax revenues will at first be accumulating at a rate consistent with the lower income level reached; as a result, the rate of deficit then observed will be larger than that which occurs when the fall in private demand is immediately offset. But as the income effect of this larger deficit causes income to rise toward its original, predepression level, it causes tax revenues to rise also and thus causes the observable level of income effect to decline. Since the induced tax revenues should be disregarded in this case, consistency requires that they also be disregarded in the case where the expenditure increase occurs rapidly enough to prevent the (temporary) fall in income.

When no correction can be made for induced taxes, the income effect found will usually be of some value, even though its measurement is defective. And for the purpose of measuring the budget’s net contribution at any point of time to the domestic flow of money demand (see section above, Real or relative value versus money value of income effect), the income effect must in any case be measured without such an adjustment.

If import taxes, poll taxes, and license fees are the chief sources of revenue, a situation may arise in which no correction is required. Where imports are controlled so that they can vary with income only in the downward direction, the marginal propensity to pay import taxes must be taken into account when a budget surplus has to be considered; when, however, there is a budget deficit, the wish to import more will be frustrated by the controls, so that the change in import taxes will be zero. Similarly, where the number of licenses for which fees are paid is limited, as in the case of monopolies or public utilities, income and price variations can have no effect upon government income from the fees. In these cases, autonomous budgetary income effects will not produce offsetting induced changes in themselves, for budget revenue will not be sensitive to income changes. It is questionable, however, whether many instances of sufficiently close approximation to this requirement will be found. (At any rate, even though government revenues may not be dependent on the income level, government expenditure may be so dependent—e.g., variations in the amount of unemployment relief or in the cost of normal government activities—and the problem of deriving the significant inflationary effect from the observed effect would still exist.)

The one important instance in which the omission of an adjustment for induced tax revenues does not seriously impair the accuracy of the measure of the inflationary impact is where the inflationary impact is repressed. Here there are no income changes or associated tax changes to be estimated. The income effect does not increase the real or money value of output, and hence does not induce tax payments, since it merely displaces other (“repressed”) demand for a portion of the fixed amount of production. This treatment of the income effect may appear to conflict with that in earlier parts of the paper. However, since the reduction in other expenditures is in a sense the result of the budget itself, the argument above that an induced tax revenue correction is necessary, even though income does not change, does not apply here. And the reasoning, presented in the early part of this paper, that a government deficit which discourages an equal amount of private expenditure (on investment) cannot be considered as having an inflationary impact, is also irrevelant here, since “repressed inflation” is a concept intended to show the amount of available private purchasing power the use of which is (temporarily) prevented by the government deficit expenditure.

If the purpose of the measurement is to determine current additions to latent inflation, i.e., the accumulation of the pent-up money demand which may be released in the future, no correction for induced changes in tax revenues need be made, for taxes do constitute a withdrawal from pent-up money demand. This problem involves primarily the budget’s liquidity effect rather than directly its income or “inflationary” effect, and is therefore outside the scope of the present paper.


Mr. White, economist in the Financial Problems and Policies Division, is a graduate of Harvard University.


Additional studies of this subject are as follows:

Villard, Henry H. Deficit Spending and the National Income (New York, Farrar and Rinehart, 1941).

Wallich, Henry C. “Income-Generating Effects of a Balanced Budget,” Quarterly Journal of Economics, LIX, pp. 78–91 (November 1944).

Economic Survey of Denmark: National Budget for 1949 (Copenhagen, 1949).

Council of Economic Advisers. The Economic Report of the President Transmitted to the Congress (Washington, U.S. Government Printing Office, January 1949 and January 1950).

Brown, E. Cary. “Analysis of Consumption Taxes in Terms of the Theory of Income Determination,” American Economic Review, XL, pp. 74–89 (March 1950).

Econometrica, Vol. 13 (1945) and Vol. 14 (1946):

  • “Multiplier Effects of a Balanced Budget,” by Trygve Haavelmo, October 1945, pp. 311–18.

  • “.... Some Monetary Implications of Mr. Haavelmo’s Paper,” by G. Haberler, April 1946, pp. 148–49.

  • “.... The Implication of a Lag for Mr. Haavelmo’s Analysis,” by R. M. Goodwin, April 1946, pp. 150–51.

  • “.... Further Analysis,” by Everett E. Hagen, April 1946, pp. 152–56.

  • “.... Reply,” by Trygve Haavelmo, April 1946, pp. 156–58.


This discussion is entirely in terms of a price elasticity of demand for imports, defined with reference to changes in real income, and therefore in savings and consumption, which occur in response to changes in import prices at a given level of money income. No attention need be given at this point to the marginal propensity to import.


Expenditures on goods and services should not be reduced by the amount by which they increase the savings and imports of the provider of those goods and services: the total expenditures constitute an addition to money demand. (The savings resulting from the increased income which this demand produces cannot be considered as a deduction from that additional money demand itself. If they were so considered, consistency would require that the consumption expenditures resulting from the increased income be added to that money demand; such treatment would amount to computing the total “multiplied” impact of government expenditure—a concept very different from the customary measure of inflationary forces, and one which would combine the causal inflationary factors with the inflation they actually produce.)

IMF Staff Papers: Volume 1, No. 3
Author: International Monetary Fund. Research Dept.