Transfer outlays are included in several tables of total government spending, but only at current prices. As is cutsomary in such inquiries, these are related to GDP although transfer payments, of any kind, are not part of gross product, inclusive (GNP) or exclusive (GDP) of net factor income from abroad. Even with its shortcomings, the OECD report is a valuable addition to the literature on public expenditures. The individual investigator, aware of the constraints inherent in institutional research, will simply have to generate his own series on real expenditure.

Deflation of transfer outlays

In the absence of official deflators for government transfer expenditure (G) and total spending (G), several writers have experimented with alternative means of deflating these outlays." Depending on the source material and the object of the investigation, these studies have relied either on a consumer price index or the "government purchases" deflator to produce a constant-price series for total government expenditure, including transfer outlays.

To this writer's knowledge, there is only one precedent for the separate deflation of transfer outlays a staff paper by the U.S. Office of Management and Budget (OMB) designed expressly for the needs of that agency and made available to other users of Federal budget data.18 This document will be discussed more fully in Part IV. Here we are interested only in its treatment of transfer outlays.

On the several categories into which the Federal budget was divided, two are relevant here: (1) payments to individuals, directly and via grants to sub-national governments; and (2) net interest on the Federal debt. The former was deflated by the consumer price index (CPI), and the latter by the implicit price deflator for total GNP.19

The decision to use the CPI, rather than the implicit price deflator for personal consumption expenditures (PCE), was made on the ground that "it is better known to the general public." 20 Given the OMB's needs, the choice is understandable; the comparative merits of Paasche and Laspeyres indexes, however, might have been taken into account.

The decision to deflate the Federal debt by the GNP deflators is not explained, either in the text or in the "notes on deflators." Presumably, the most general price index available was deemed superior to one more narrowly-based. Ideally, an index reflecting ownership of the debt, or distribution of interest among owners would have been preferable. Experienced users of price indexes, however, are aware that there is no perfect deflator. Improvisation is especially appropriate in the deflation of government expenditures.

The source material for this study-the national accounts statistics compiled by the UN-provides few options for the deflation of transfer outlays. Consumer price indexes are available in a supplementary source-the UN Statistical Yearbook-but these indexes employ base-year quantities as weights. The implicit price deflator for private consumption expenditure, on the other hand, uses current-year quantities and is therefore consistent with the deflators from GDP and government consumption. This deflator appears in column 2 of Table 6. Increase in transfer expenditure, 1950–77

The undeflated amount of transfer expenditure in 1977 is shown in Table 6, column 1, as an index based on 1950. In each country these outlays grew by a large, multiple over the 27-year period. The smallest increase-10.7 times-occurred in the United States, where transfer expenditure by government rose from $22.7 billion in 1950 to $265.9 billion in 1977. In Greece, where the largest increase took place, the 1977 outlay for transfer programs was 126.1 billion drachmas-an increase of 47.5 times over the period: Sweden, occupying the middle position in this array, expanded its outlays 30.7 times over the period.

17 See, for example, Beck (1976), Pluta (1979), Tussing and Henning (1974), and Wagner and Weber (1977) 18 OMB, The Budget in Constant Dollars (Technical Staff Paper Series BRD/FAB 75-1). The OMB de flators and constant-price budget data are now widely used in Congressional reports and budget documents published by OMB.

19 Deflation of net interest was accomplished in three steps: (1) conversion of debt held by the public into real debt; (2) expression of constant-price data on debt as an index based on fiscal year 1969; and (3) multiplication of base year net interest by the index value for each year. This method was defended on the ground that "a percentage change in real debt should be reflected by an equal percentage change in constant price net interest." (p. 8)

20 Ibid., p. 7.

TABLE 6.-GOVERNMENT TRANSFER OUTLAYS, 1977: INDEXES OF AMOUNT, PRICE, AND VOLUME

Source: January 1979, "Economic Report of the President," table B-77, p. 272.

In real terms the expansion of transfer outlays was more modest (column 3). These increases ranged from a low of 206 percent (UK) to a high of 10 times (Netherlands) over the 1950 volume. In the median country (France) the 1977 volume of transfer outlays was 7.4 times that of 1950.

The price indexes used to deflate transfer expenditure (column 2) ranged from a high of 585 (Finland). Sweden again occupies the middle ground, with a 1977 price index of 408. The reader may find it helpful to see these indexes converted into annual rates of increase and to compare them with their counterparts for government consumption expenditure (Gc):

In studies which use the G. deflator for transfer outlays, the expansion of real transfer outlays will appear to be more modest than in the present study.

Growth rates of transfer expenditure

The indexes of the preceding section demonstrate that each country expanded its transfer outlays by a large multiple over the study period. Interpretation of these results can be facilitated by converting the indexes into annual rates of increase, for the full period and several segments thereof. Table 7 displays the results for un-deflated data, and Table 8, for the deflated amounts of transfer expenditure.

TABLE 7.-ANNUAL GROWTH RATES OF GOVERNMENT TRANSFER OUTLAYS IN CURRENT PRICES

Source: January 1979, "Economic Report of the President," table B-77, p. 272.

TABLE 8.-ANNUAL GROWTH RATES OF GOVERNMENT TRANSFER OUTLAYS IN CONSTANT PRICES

Source: January 1979, "Economic Report of the President," table B-77, p. 272.

Between 1950 and 1977 the annual growth rate of transfer expenditure in current prices ranged from a low of 9.5 percent (United States) to a high of 15.5 percent (Greece). The median rate of increase was 13.7 percent (Sweden).

In several cases the growth rate dipped during the 1960's and rose-sometimes sharply in the Seventies (through 1977 or, in Denmark and Ireland, through 1976). The general pattern, however, is one of a rising growth rate over the suc cessive segments of the full period. From 10.4 percent in the 1950's, the median growth rate rose to 12.7 percent in the 1960's and accelerated to 17.9 percent in the Seventies.

As Table 6 indicated, much of the expansion since 1950 in government transfer outlays was a response to higher prices. Social security benefits, for example, are now indexed in many countries. Following several months of successive increases in the consumer price indexes, recipients become entitled to higher benefits. For other transfer programs, where increases are not automatic, the tendency of legislators is to appropriate larger amounts simply to compensate for past increase in the price level.

Table 8 shows the annual growth rate of transfer outlays after adjustment for inflation. Median rates no longer exhibit the previously noted pattern of an unbroken rise over successive periods. From an annual rate of 8.4 percent in the 1960's, the growth rate drops to 7.7 percent in the next period; and in 1977 the

median value of real transfer outlays was only 3.9 percent higher than in the preceding year. In one case-Finland-the 1977 outlay was unchanged from that of 1976; and in the UK, only 1.9 percent higher.

The behavior of real transfer outlays since 1970 suggests a slowdown in the growth of this component of government spending. Compared with real expenditure for collective consumption, however, these growth rates are relatively high. An assessment of the full impact of these changes is deferred until we have examined total government spending (Part IV).

We can, however, now summarize the growth of transfer expenditure (G) and compare it with the behavior of spending for collective consumption (G.). In real terms G, grew at twice the annual rate (7.7 percent) as G. (3.8 percent). As a proportion of real GDP, the median value of real transfer outlays rose from 9.3 percent in 1950 to 20.6 percent in 1977. The real share of GDP accounted for by government consumption expenditure, on the other hand, scarcely grew-from 12.1 percent in 1950 to 12.7 percent in 1977.

IV. SIZING UP THE PUBLIC SECTOR

How large is the total public sector? Has it grown in real, as well as nominal terms, since 1950? The disaggregated approach we have followed in this study should provide a better perspective on these questions than was heretofore possible. Building on the data of parts II and III, we are now in a position to estimate total government expenditure at constant prices, analyze the behavior of these aggregates, and use them in ratios representing real size of the public sector.

A by-product of this procedure is the deflator for total government expenditure, displayed in Table 9 (columns 5 and 6) and in appendix table A-3.21 A weighted average of the indexes used to deflate consumption and transfer expenditure, this deflator is the first of its kind. In the absence of a published deflator for total expenditure, earlier studies of public-sector size either ignored the problem or turned to deflators appropriate only for one of the components of total spending. Total government spending since 1950

In nominal terms the increases in government spending betwene 1950 and 1977 ranged from a low of 9.7 times (United States) to a high of 43.4 times (Greece). The median increase was 26.5 times (Sweden). Because 1977 totals were not available for Denmark and Ireland, index values are shown also for 1976 (column 1 of table 9). Low, median, and high positions for 1976 are held by the same countries as for 1977.

TABLE 9.-TOTAL GOVERNMENT EXPENDITURE AT CURRENT AND CONSTANT PRICES

1 Calculated from amounts in national currency (appendix tables A-1 and A-2). 2 Not available.

Source: January 1979, "Economic Report of the President," table B-77, p. 272.

21 Methodological issues in deflating total government expenditure are discussed in Beck (1979b), “Estimating Changes in Real Size of the Public Sector," Economics Letters, vol. II, no. 3.

The behavior of real expenditure is quite different. Not only are the increases more modest, but the extremes of the range are now occupied by a different set of countries. The smallest increase-136 percent between 1950 and 1977-took place in the UK, and the largest-564 percent occurred in the Netherlands. France with an increase of 409 percent-occupies the median position in this array (column 4). For 1976 (column 3) the low, median, and high positions are held by the same set of countries.

The data in columns 3 and 4 were obtained by combining the deflated values of consumption and transfer expenditure and expressing the result as an index of the 1950 total. Unlike the constant-price data of Tables 4 and 6, the volume indexes of Table 9 are not a product of the standard procedure for deflating a current-price series. That option was not available here. A deflator for total government spending was, instead, generated indirectly by finding the constantprice total and dividing it into the current-price total.

The deflators of Table 9 (columns 5 and 6) represent an improvement over what was previously available to investigators of public-sector size. It is clear from the procedure followed that the deflator is only as good as the method used to obtain total government spending in constant prices. In this study the total was arrived at by separately deflating the two main categories of government spending-consumption and transfer outlays. Further progress requires a more disaggregated approach, and depends on the ability of researchers to develop a set of deflators that fit the smaller components to total government spending.

Among the countries for whom data appear in column 6, Switzerland had the smallest increase 195 percent-between 1950 and 1977. The largest increase-687 percent occurred in Finland, and the median rise-375 percent-was that of Austria. For 1976 (column 5) Sweden occupies the mid-point of the range, while the low and high positions are held by the US and Finland, respectively.

The indexes of Table 9 are sweeping measures of the growth in total government spending and its deflator, over the full study period. Annual rates of increase, for segments as well as the full period, facilitate interpretation of the overall results. These growth rates are discussed in the next section.

Growth rates of government expenditure

The median annual increase in total spending, before deflation and for the full period, was 13.1 percent (Sweden). Increases ranged from 9.2 percent (US) to 15.1 percent (Greece).

In all cases the growth of government spending substantially exceeded that of GDP, a matter that is discussed in depth below.

Between the Fifties and Sixties the median growth rate of nominal expenditure advanced slightly-from 10.6 percent to 11.7 percent-and rose sharply in the Seventies to 16.3 percent. Annual growth rates of real expenditure, calculated from Appendix Table A-2, are displayed in Table 11.

TABLE 10.-ANNUAL GROWTH RATES OF TOTAL GOVERNMENT EXPENDITURE IN CURRENT PRICES

Source: January 1979, "Economic Report of the President," table B-77, p. 272.