Chapter XXVII “THE NATIONAL INCOME APPROACH”

No analysis of Keynesian economics would be complete without at least some discussion of what is variously called “aggregrative” economics, “macro-economics,” and “the national income approach.”

Many of his disciples are under the impression that it was Keynes who created “the national income concept.” This is pure fantasy. Efforts to calculate the national income have a long history. Though Keynes does have a great deal to say about “aggregative” economics (which we have already analyzed), his discussion of the national income in the General Theory is confined, in fact, to two or three pages, which mainly refer to earlier studies by A. C. Pigou, Colin Clark, and Simon Kuznets.

Yet “the national income approach” owes at least part of its present vogue to Keynesian ways of thinking, and therefore a few words may properly be said about it here. A thorough discussion would call for a volume in itself, but I shall attempt no more than a few sketchy comments.

1. Is National Income Determinate?

The first thing to be emphasized about the national income is that it is an arbitrary, and from the standpoint of scientific precision an indeterminate, figure. The ablest students of the subject have recognized this. I need merely refer to the fine pioneering study of Simon Kuznets.¹

Kuznets devotes his entire first chapter, of fifty-seven pages, to a discussion of the problems embedded in the very concept of “national income.” He begins:

The statistician who supposes that he can make a purely objective estimate of national income, not influenced by preconceptions concerning the ‘facts,’ is deluding himself; for whenever he includes one item or excludes another he is implicitly accepting some standard of judgment, his own or that of the compiler of his data. There is no escaping this subjective element.

Kuznets goes on to show that estimates of the national income necessarily involve legal and moral considerations. Should we include “the compensation of robbers, murderers, drug peddlers, and smugglers”? And how shall we “draw a line between economic activity and economic goods on the one hand and active life in general and its stream of satisfactions on the other”? Should “washing, shaving, and playing for amusement on the piano” be treated as economic activity? “When judged by the attributes of satisfaction-yielding, scarcity, and disposability, they do not differ from the same activities carried on for money as services to other people (nursing, barbering, and giving concerts).”

And yet Kuznets decides to include only items that “are dealt in on the market.” This of course excludes all do-it-yourself activities (which in total are probably enormous). It excludes all the products of the family economy, including all the activities of housewives. So we get to such paradoxes as these: When a man marries his cook, the value of her work disappears from the national income accounts. When an opera singer sings professionally, she is considered as adding the equivalent of her salary to the national income. When she sings for charity or for friends, it doesn’t count.

How are we to prevent double counting at a hundred points? If we count the income of doctors and dentists, should we, or should we not, deduct it from the income of patients?

What is it that we are trying to measure, anyway? What is the difference between “economic activity” and “active life in general”? How, except by arbitrary “value judgments,” do we distinguish between “productive” and “unproductive” activities? Are we trying to measure “national income produced,” “national income paid out,” “national income spent,” or “national income consumed”?

No doubt today most laymen (and a large number of statisticians and economists!) assume that all these problems must have been satisfactorily solved, because they read daily in their newspapers official figures showing exactly what the national income, “personal income,” “disposable personal income,” and above all the “gross national product” or “GNP,” were not only in past periods, but at what annual rate they are currently running. And these figures are presented with great precision, with decimal points. Few laymen are aware that these figures are made up not of definite items which can be lined up and counted, but in large part of estimates subject to error.

Let us take a few quite recent illustrations. The President’s annual Economic Report of January, 1958, boasted in its opening paragraph that the nation’s GNP, or output of goods and services, in 1957 totaled $434 billion, “5 per cent larger than in the preceding year.” Only later in the report were we explicitly told that “four-fifths of this increase was accounted for by rising prices,” and that therefore “in physical terms, the increase was only about 1 per cent.” In July of 1958, however, the national income estimates received one of their periodic revisions, and the Department of Commerce statisticians decided that our GNP in 1957 was not $434 billion but $440 billion, and that our 1956 GNP was not $415 billion but $419 billion. Yet in “1957 prices,” we were informed, our 1956 GNP was $435 billion.

This brings us to one of the great problems in estimating national income. It is measured in a dollar which has itself no fixed value. In a period of inflation, all values are falsified. Today the most frequently cited over-all figure is not that of national income but of gross national product, or GNP. I shall therefore use this for purposes of illustration. For 1939 the GNP was estimated at $91.1 billion; for 1957 it was estimated at $440.3 billion. Here is an apparent quadrupling, or better, of the GNP. But when the government statisticians restate the figures in “constant dollars” (specifically in “1954 dollars”), they find that the GNP in 1939 has to be raised to $189.3 billion and that the 1957 GNP has to be lowered to $407 billion. In other words, “real” GNP did not quadruple but only about doubled in the 18-year period.

The government statisticians get this result by dividing actual dollar totals by an index number of prices for each year. They print, in fact, a separate table of “implicit price deflators” for the gross national product figures for each year based on an index number of 100 for 1954. The price deflator for 1939, on this basis, is 48.1, and for 1957 is 108.2. If we take the GNP in 1939 at the prices that prevailed in that year it comes, as we have seen, to $91.1 billion. But if we translate 1957 national income into 1939 prices, we get, instead of $440.3 billion, only $195.7 billion for 1957. This does not look nearly as impressive. If, again, we divide these figures by the population, we find a much lower rate of per capita growth than we are at all likely to gather from the crude over-all figures.

But though inflation and the changing value of the dollar make comparative over-all national income figures quite misleading, is it, in fact, possible to correct the comparison by applying “implicit price deflators”? Only approximately; never accurately. As Kuznets and every other serious student of index numbers has pointed out, goods never remain the same for two years in succession, either in relative quantities or in comparative quality, and no index number can be completely “scientific.”

There is one further factor that distorts and falsifies comparative national income figures. It is a factor I do not recall ever having seen discussed in connection with these figures, yet it goes to the heart of the whole problem of measurability.

Larger crops often have a smaller total dollar value than smaller crops. (Hence crop-restriction schemes.) But this merely illustrates a wider principle. Economists have pointed out since the time of Adam Smith that it is not “value-in-use,” but scarcity, that determines “value-in-exchange,” or money price. Water is an indispensable commodity that ordinarily commands no price at all. If more and more things became plentiful (except dollars), the national income, as measured in dollars, might begin to fall. If we could imagine a situation in which everything we could wish for was in as adequate supply as air and water, we might have no (monetary) national income at all!

When one seeks to be clear about basic principles, it is never a bad idea (in spite of the ridicule that has been heaped upon it since the days of Karl Marx) to go back to “Crusoe economics.” Suppose, then, we begin with a community of just two persons, one of whom raises beans (say 1,000 pounds) and the other of whom raises potatoes (also 1,000 pounds). This is their total wealth. The total wealth (or, if we wish, income) of the community is thus 1,000 pounds of beans plus 1,000 pounds of potatoes. But, someone may wish to know, which is the wealthier—Ben, who raises beans, or Peter, who raises potatoes? And what is the total wealth (or annual income) of the community expressed in terms of some common measure?

Suppose Ben and Peter exchange their beans and potatoes at a ratio of a pound for a pound (to such an extent as to bring the relative marginal utilities of each to both of them into equilibrium). And suppose we elect to regard the potatoes as the “medium of exchange” and the “money of account.” Then the total income of the community is obviously 2,000 “pounds-of-potatoes,” made up of 1,000 pounds of potatoes and 1,000 pounds of beans a year.

But now certain paradoxical results appear. Suppose Peter doubles the amount of potatoes he grows, while Ben raises only the same amount of beans. Then the income of the community has risen, in real terms, to 2,000 pounds of potatoes plus 1,000 pounds of beans. We might be tempted to conclude that, in terms of the common “standard-of-value,” the income of the community was now 3,000 “pounds-of-potatoes.” But because potatoes were now twice as plentiful, and beans were unchanged in supply, Ben might demand, and Peter be willing to pay, two pounds of potatoes for every pound of beans. But this would mean that the supply of beans was twice as valuable as before. Therefore the total income of the community, as expressed in potatoes, would not be 3,000 “pounds-of-potatoes,” but 4,000.

Suppose, on the other hand, it was the supply of beans that had doubled, and Peter was able to demand and get two pounds of beans for every pound of potatoes. Then the income of the community, measured in “pounds-of-potatoes,” would not be 3,000 “pounds” but only 2,000.

So our “national income” figure, expressed in a common medium of exchange or “money-of-account,” does not express any absolute total at all, but merely an internal relationship of marginal values (times quantities). We could go on to illustrate this by a more complex “model,” assuming, say, a hundred different commodities, one of which would be gold, and assuming that a certain weight of gold, a “dollar” (or one-thirty-fifth of an ounce), was the medium of exchange and the “money-of-account.” It would then be easy to show that an increase in the other ninety-nine commodities would by no means mean a proportionate increase in the national income as measured in “dollars,” and yet that a doubling of the amount of “dollars” alone might double the national income as expressed in dollars.

Nor would it be possible to “correct” for these paradoxical results, except in an inaccurate and untrustworthy fashion, by using “implicit price deflators” or inflators. And if the problem of translating money-value income into “real,” or heterogeneous physical, income is insoluble, still more so is the problem of translating either into “psychic” or “enjoyment” income. Hence the impossibility of a “scientific” comparison of the income of “Russia” and the “United States.”

In brief, national income estimates have a very limited value, a far more modest value than is now commonly supposed. They might have some value in comparing the national incomes of two different countries—if the figures in both countries were compiled by the same methods and (largely arbitrary or conventional) standards, if both countries had the same monetary standard (say gold), and if complete freedom of currency convertibility and of trade prevailed. Such comparisons have little value when currency ratios are fixed by government ukase or exchange control rather than by free markets or free convertibility into a common commodity.

2. Its Dangers for Policy

It is impossible, in sum, to arrive at a precise, scientific, objective, or absolute measurement of the national income in terms of dollars. But the assumption that we can do so has led to dangerous policies, and threatens to lead to even more dangerous policies.

Policy implications, in fact, are already found in the “national income approach.” For this embodies an attempt to deal with economic problems starting from an arbitrarily constructed “whole,” from a “collective,” and not from acting individuals. This “macro-economic” as differentiated from the “micro-economic” approach raises first of all the question: Why is the “nation” considered the collective to be chosen and not the state (State of New York), the municipality (City of New York), the borough (Manhattan) or, on the other side, the continent (America)—or the whole world? The chief answer to this question is that the choice of the collective is determined mainly by political considerations. Many of our American “progressives” aim at an equalization of incomes within the United States, but not at a world equalization. This political tendency explains, also, why these people are always talking about the “distribution” of the national income and not about the contribution of the various individuals and groups of individuals to its coming into existence. Logically, the contribution problem ought to be considered first. Much of the “national income” discussion is dominated by the Marxian thesis according to which goods are “socially” produced and afterwards individually appropriated.

I have said that though the government compiles quarterly estimates both of gross national product and of national income, it is the former figure that is much more frequently cited. This is partly because it appears earlier (as a private firm knows its gross income before it knows its net income), and partly because it is the larger figure. National planners love big figures. We are constantly being told that “we” (the government) can easily afford to spend or give away (say, to foreign governments) this or that huge sum because it is after all only such-and-such a percentage of our gross national product. No one would dream of considering such reasoning valid as applied to a private firm. The average industrial company’s net profit, for example, amounts (1956-7) to only 5 or 6 cents on every dollar of sales.

There are great deductions to be made from gross national product before we can estimate national income. For example, in 1957, gross national product was estimated at $440.3 billion, whereas national income was estimated at only $364 billion. In arriving at the latter figure some $34 billion was deducted for depreciation charges and some $38 billion for indirect business taxes. But depreciation charges are the result of estimates. The “right” amount of depreciation is never precisely known. Contrary to the belief of laymen (and even of many accountants) a depreciation charge is not so much an estimate of past deterioration as a forecast of future probabilities. It is never known, for example, when an old machine is going to be made obsolete by a new invention. And particularly in a period of monetary inflation, such as we have been undergoing for the last generation, depreciation charges are systematically underestimated, because they fail to allow for ever mounting replacement costs.

Another bad practice to which a too literal reliance on national income figures has led is that of insisting on the urgency of a certain “rate of growth” of the national income, no matter what level it has already reached. Thus a report of the Rockefeller Brothers Fund in 1958, looking ten years ahead, came up with the remarkable discovery that an economic “growth rate” of 5 per cent a year would lead to a bigger growth in ten years than a 3 per cent rate or even a 4 per cent rate.

This insistence on achieving or maintaining a certain “rate of growth” is the result of several misconceptions. Professor G. Warren Nutter has pointed out that there is “a long-run tendency... for the industrial growth rates to slow down, or retard, as the level of production gets higher.” There are several basic explanations of this. One has to do with a trick of percentage figures. Another has to do with a physical satiety point in human needs. If only one family in a country has a bathtub, and the next year 50 families get one, the rate of growth is 5,000 per cent. But once everybody has a bathtub net growth stops. This principle applies to houses, automobiles, radios, television sets, etc.

In addition, as we have just noticed a little while back, as more and more things become plentiful (except dollars) there might even be a tendency for the national income figures to reflect this by falling, because prices might fall faster than output rose.

Still another practical danger of the religious use of national income figures is that it can lead to a confusion or reversal of economic cause and effect. The national income of a given year is the total result of all the production and transactions during that year. In this respect the national income figures are similar to the account books of a private firm. But more and more, in current discussion, one finds the national income figure treated as a cause of production. The national income is thought of as the purchasing power that automatically creates and buys the production. The truth is that the national income is the production itself, looked at from another side. Broadly speaking, national income does not cause national production, but national production causes national income. Insofar as the causation is the other way ‘round, it is because of the truth in that very Say’s Law that the Keynesians and national income addicts tell us has been discredited.

The national income figures seem to have given birth to all sorts of cause-and-effect fallacies. For example, if we look at the composition of the national income figures for, say, 1957, we find that part of the GNP total of $440.3 billion is arrived at by including $87.1 billion for “government purchases of goods and services.” When the national income figures of $364 billion for that year are broken down into specific industries, we find that nearly $43 billion is accounted for by “government and government enterprises.” It is easy to jump to the conclusion, which Keynesians do, that if it were not for these $87 billion of government purchases or these $43 billion of government payrolls and enterprises, the national income would be just that much less. People with a less favorable opinion of the role of government would point out that whatever the government spends it takes away from somebody in taxes. (This applies also to the hidden tax involved in monetary inflation.) Undoubtedly, such government employees as policemen, firemen, judges, and road-builders do increase (by an unascertainable amount) real national income. But it may be questioned whether such agencies as price controllers, rent boards, the Tariff Commission, the crop restriction agents of the Department of Agriculture, or the National Labor Relations Board do not bring about a net reduction of the real national income, in spite of the fact that they increase it according to the government figures.

If we think of the national income as a mere lump over-all sum in dollars, and it falls short of some “goal” by x billion dollars, it is a tempting step for economic planners to assume that the x billion dollars could be easily supplied by that much deficit spending, or even by printing that much money. This leads indirectly to inflation. For we can raise our national income to any figure we want simply by depreciating the dollar enough to raise prices to reach that income.

In Germany, in 1923, the national income (in marks) actually rose to hundreds of billions of times higher than its previous level, because the paper mark was depreciated to one-trillionth of its former purchasing power.

To be sure, when explicitly taxed with the point, economic planners will say that their goal is a national income of x billions “in dollars of present purchasing power.” But they forget this qualification in actual practice. They are always citing the latest national income figures in terms of the latest and most inflated dollar. They do not stop to remind us, or even themselves, how much the national income would have to be written down to reflect the price level of, say, twenty years ago.

“The national income approach” has become one of the inportant incitements to inflation. For the easiest and surest way to get constantly bigger national income figures is not by increasing output and consumer satisfactions, but by constantly shrinking the measuring rod, by constantly depreciating the dollar.

It remains to be pointed out, finally, that economic forecasting based on “aggregative economics” or “the national income approach” has been a failure. David McCord Wright, who declares that: “In practical experience, the Keynesian forecasters have quite a poor record,” cites in evidence “the egregious failure of most Keynesian forecasts after World War II,” which was “very largely due to an unexpected upward jump of the consumption level.” Similarly, he adds, “in 1953 and again in 1958 the Keynesian models of mechanical interrelationships between investment and consumption did not work out.” ²

This judgment corroborates that of John H. Williams: “The consumption function in particular has given the mathematicians... an ideal concept for building models of national income and making forecasts. Thus far, the forecasts have been almost uniformly bad.” ³

¹National Income and Its Composition, 1919-1938. (New York: National Bureau of Economic Research, 1941), 2 vols.

²Science, Nov. 21, 1958, pp. 1261-1262.

³American Economic Review, May, 1948, p. 284.