All Commentary
Tuesday, March 1, 1977

Inflation versus Employment

For many years, it has been popularly assumed that inflation increases employment. This belief has rested both on naive and on more sophisticated grounds. 

The naive belief goes like this: When more money is printed, people have more “purchasing power”; they buy more goods, and employers take on more workers to make more goods.

The more sophisticated view was expounded by Irving Fisher in 1926:

“When the dollar is losing value, or in other words when the price level is rising, a businessman finds his receipts rising as fast, on the average, as this general rise of prices, but not his expenses, because his expenses consist, to a large extent, of things which are contractually fixed . . . . Employment is then stimulated—for a time at least. “1

This view contained a kernel of truth. But 32 years later, in 1958, the British economist A.W. Phillips published an article2 which seemed to show that over the preceding century, when money-wage-rates rose, employment rose, and vice versa.

This, too, seemed a plausible relationship. Given a period for the most part noninflationary, but in which capital investment and invention were raising the unit-productivity of labor, profit margins on employment would be rising, in some years much more than in others; and in these years the demand for labor would increase, and employers would bid up wage rates. The increased demand for labor would lead both to higher wages and to increased employment. Phillips may have seen what he thought he saw.

But Keynesian economists, struck by the Phillips thesis, and seeing in it a confirmation of their previous belief, carried it much further. They began to construct Phillips Curves of their own, based not on a comparison of wage rates and employment, but of general prices and employment. And they announced they had found there is a trade-off between unemployment and prices. Price stability and reasonably full employment, they asserted, just cannot exist at the same time. The more we get of the one the less we can have of the other. We must make a choice. If we choose a low level of inflation, or none at all, we have to reconcile ourselves to a high level of unemployment. If we choose a low level of unemployment, we must reconcile ourselves to a high rate of inflation.

This alleged dilemma has served as a rationalization for continued inflation in many countries when every other excuse has run out.

The Phillips Curve is a myth, and in the last few years it has been increasingly recognized as a myth. Here is a table comparing the percent changes in the Consumer Price Index, for the 28 years from 1948 to 1975 inclusive, with the per cent rate of unemployment in the same years.       


Change CPI Unemployment






















































































Economic Report of the President,

January, 1976; pp. 224 and 199.


I leave it to the Phillipists to make what they can of this table. The average annual price rise in the 28 years was 3.2 per cent, and the average unemployment rate 4.9 per cent. If the alleged Phillips relationship held dependably, then in any year in which the price rise (or “inflation” rate) went above 3.2 per cent, the unemployment rate would fall below 4.9 per cent. Conversely, in any year in which the “inflation” rate fell below 3.2 per cent, the unemployment rate would rise above 4.9 per cent. This relationship would hold for all of the 28 years. If, on the other hand, the Phillips Curve were inoperative or nonexistent, the probabilities are that the relationship would hold only about half the time. This is exactly what we find. The Phillips relation occurred in 15 of the 28 years but was falsified in the other 13.

Alternative Views

More detailed analysis of the table hardly helps. An economist who saw what happened only in the years 1948 through 1964 might have been excused for being impressed by the Phillips Curve, for its posited relationship held in 13 of those 17 years. But an economist who saw only what happened in the last 11 of those 28 years—from 1965 through 1975—might have been equally excused for suspecting that the real relationship was the exact opposite of what the Phillips Curve assumed, for in that period it was borne out in only two years and falsified in nine. And even the economist who seriously studied only what happened in the 1948-1964 period would have noted some strange anomalies. In 1951, when the CPI rose 7.9 per cent, unemployment was 3.3 per cent; in 1952, when prices rose only 2.2 per cent, unemployment fell to 3.0; and in 1953, when prices rose only 8/10 of 1 per cent, unemployment fell further to 2.9—the lowest for any year on the table.

Phillips statisticians can play with these figures in various ways, to see whether they can extract any more convincing correlation. They can try, for example, to find whether the Phillips relationship held any better if the CPI rise is measured from December to December, or if the calculations are remade to allow for a lag of three months, or six months, or a year, between the “inflation” rate and the unemployment rate. But I do not think they will have any better luck. If the reader will make the count allowing for one year’s lag between the price rise and the unemployment figure, for example, he will find the Phillips Curve contention borne out in only 10 and contradicted in the other 18 years. (I have referred to the rate of the consumer-price rise as the “inflation” rate because that is unfortunately the way the term is applied by the majority of journalists and even economists. Strictly, the term “inflation” should refer only to an increase in the stock of money. A rise of prices is a usual consequence of that increase, though the price rise may be lower or higher than the money increase. Insistence on the distinction between these two terms is not merely pedantic. When the chief consequence of an inflation is itself called the inflation, the real relation of cause and effect is obscured or reversed.)

A clearer picture of the relationship (or nonrelationship) of price rises and unemployment emerges if we take only the last 15 years of the 28 and make our comparisons for the average of five-year periods:

CPI rise Unemployment rate  rate

(per year)               (per year)











This table was suggested by one which appeared in Milton Friedman’s column in Newsweek of December 6, 1976. There are one or two minor changes.

In sum, the highest rate of “inflation” was accompanied by the highest rate of unemployment.


The experience in other nations has been even more striking. In August 1975 The Conference Board published a study comparing the percentages of the work forces employed with consumer price indices in seven industrial nations over the preceding fifteen years. By this measurement, in the United States, Canada, and Sweden, the relationship did not noticeably belie the Phillips Curve. (In our 28-year U.S. table, however, we saw that when the price-increase figure shot up in 1974 to 11 per cent from a rate of 6.2 per cent in 1973, unemployment also rose. If we look at 1975—not shown in the Conference Board study—we find that unemployment soared to 8.5 per cent though there was a similar high price rise-9.1 per cent—in 1975. Similarly, if we take what happened in 1975 in Canada, we find that though consumer prices rose in that year by the unusually high rate of 10.7 per cent, the index of manufacturing employment in Canada fell from 108.9 in 1974 to 102.8 in 1975.)

 In the four other countries in the Conference Board study, the relationship of employment and inflation was emphatically the opposite of that assumed by the Phillips Curve. The steady price rise in Germany from 1967 to 1973 was accompanied by an equally steady fall in employment. In Japan a rise of 19 per cent in consumer prices in 1973 and of 21 per cent in 1974 was accompanied by a fall in employment. In Italy, though consumer prices began to soar in 1968, reaching a 25 per cent annual rate in 1974, employment declined during the period. In some ways the record of Great Britain, where the Phillips Curve was invented, was the worst of all. Though consumer prices soared 18 per cent in 1974 from a rate of 4 per cent a decade earlier, employment turned downward. Not shown in the Conference Board compilation was the record of 1975 itself, when the British CPI soared 24 per cent—and employment fell further.

But informed economists, with memories, did not need to wait for the experience of the seventies to distrust the relationship posited by the Phillips Curve. In the last and worst months of the great German hyperinflation of 1920-1923, unemployment in the trade unions, which had been 6.3 per cent in August, 1923, soared to 9.9 per cent in September, 19.1 per cent in October, 23.4 per cent in November, and 28.2 per cent in December.

A Nest of Fallacies

There is a whole nest of fallacies wrapped in the Phillips Curve, and one of them is the implication that the absence of inflation is the sole or at least the chief cause of unemployment. There can be scores of causes for unemployment. One is tempted to say that there can be as many distinguishable causes for unemployment as there are unemployed. But even if we look only at the unemployment brought about by governmental policies, we can find at least a dozen different types of measures that achieve this—minimum-wage laws, laws granting special privileges and immunities to labor unions and imposing special compulsions on employers to make concessions (in the U.S., the Norris-LaGuardia Act, Wagner-Taft-Hartley Act, and so forth), unemployment insurance, direct relief, Social Security payments, food stamps, and so on. Whenever unions are given the power to enforce their demands by strike threats and intimidation or by compulsory “collective bargaining” legally imposed on employers, the unions almost invariably extort above-market wage rates that bring about unemployment. Unemployment insurance becomes increasingly generous year by year, and is today paid in some States for as long as 65 weeks. A study prepared for the U.S. Department of Labor in February 1975 finally conceded that “the more liberal the unemployment insurance benefits, the higher the unemployment rate will be.”

As long ago as 1934, when the New Deal was being enacted, the economist Benjamin M. Anderson remarked to me in conversation:

 “We can have just as much unemployment as we want to pay for.” The government is today buying a huge amount of it. Yet when the monthly unemployment figures are published, the overwhelming majority of commentators and politicians forget all about this, and attribute the high unemployment figure to insufficient Federal spending, insufficient deficits, insufficient inflation.

Another thing wrong with the Phillips Curve is the blind trust its compilers place in the official unemployment statistics. I am not speaking here merely about the amount of guesswork and sampling errors embodied in such statistics, but about the vagueness in the very concept of “full employment.” Full employment never means that “everybody has a job” but merely that everybody in the “labor force” has a job. And an immense amount of guesswork goes into estimating the “labor force.” Out of a total population estimated in 1975 at 213,631,000, only 92,613,000—or some 43 per cent—were estimated as being in the “civilian labor force.” These were part of the “noninstitutional” population 16 years of age and over, with certain deductions. As only 84,783,000 persons were estimated as being employed in 1975, this left an average of 7,830,000 “unemployed.”

Imprecise Measures

But none of these figures involved exact counts. They were all estimates—subject to various degrees of error. In any case the “unemployed” can never be exactly counted because of the subjective element. As the economist A.C. Pigou put it some forty years ago: “A man is only unemployed when he is both not employed and also desires to be employed.”

It is this second requirement that we can never measure. The U.S. Department of Labor Statistics counts a man as unemployed if he is out of a job and “looking for work.” But it is very difficult to determine whether a man is actually looking for a job or how much effort he is making. And when men and women are being paid enough unemployment insurance or relief or food stamps to feel no great urgency to take a job, the raw government statistics can give a very misleading impression of the hardships of all “unemployment.”

“Full employment,” as bureaucratically defined, is a completely unrealistic goal. It has never been realized in the official figures. Even if there were no governmental policies that created unemployment, it is hardly possible to imagine a situation in which, on the very day any person was laid off, he found a new job with wages and other conditions to his liking. People who give up jobs, and even those who are dropped from them, commonly give themselves an intentional vacation. There is always a certain amount of “frictional,” “normal,” or “natural” unemployment—averaging in this country, as officially measured, about 5 per cent—and government interventions that try persistently to force the figure below this average tend to create inflation and other distortions much worse than the alleged evil they are trying to cure.

To set up “full employment at whatever cost” as the sole or even chief economic goal, results in a distortion and perversion of all values.³

The Impact of Inflation

When we put aside all questions of exact quantitative determination and alleged Phillips curves, it is nonetheless clear that inflation does affect employment in numerous ways. It is true that, at its beginning, inflation can tend to create more employment, for the reason that Irving Fisher gave long ago: It tends to increase sales and selling prices faster than it increases costs. But this effect is only temporary, and occurs only to the extent that the inflation is unexpected. For in a short time costs catch up with retail selling prices. To prevent this the inflation must be continued. But as soon as people expect the inflation to be continued, they all make compensating adjustments and demands. Unions ask for higher wage rates and “escalating” clauses, lenders demand higher interest rates, including “price premiums,” and so on. To keep stimulating employment, it is not enough for the government to continue inflating at the old rate, however high; it must accelerate the inflation. But as soon as people expect even the acceleration, this too becomes futile for providing more employment.

Meanwhile, even if the inflation is relatively mild and proceeds at a fairly even rate, it begins to create distortions in the economy. It is amazing how systematically this is overlooked. For most journalists and even most economists make the tacit assumption that an inflation increases prices uniformly—that if the wholesale or consumers price index has gone up about 10 per cent in the last year, all prices have gone up about 10 per cent. This assumption is seldom made consciously and explicitly; if it were it would be more often detected and refuted.

The assumption is never correct. For (even apart from the wide differences in the elasticity of demand for different commodities) the new money that the government prints and pays out in an inflation does not go proportionately or simultaneously to everybody. It goes, say, to government contractors and their employees, and these first receivers spend it on the particular goods and services they want. The producers of these goods, and their employees, in turn spend the money for still other goods and services. And so on. The first groups spend the money when prices have still gone up least; the final groups when prices have gone up most. In addition, the growing realization that inflation will continue, itself changes the direction of demand—away from thrift and toward luxury spending, for example.

Misallocation and Waste of Scarce Resources

Thus, while inflation is going on it always brings about a misdirection of production and employment. It leads to a condition of temporary demand for various products, a malproduction and a malemployment, a misallocation of resources, that neither can nor should be continued once the inflation is brought to a halt. Thus, at the end of every inflation there is certain to be what is called a “stabilization crisis.” 

But even the distorted and misdirected employment cannot be indefinitely maintained by continuing or accelerating the inflation. For the inflation, as it goes on, more and more distorts relative prices and relative wages, and destroys workable relations between particular prices and particular wage rates. While some producers confront swollen and unmeetable demand, others are being driven out of business by wages and other costs rising far faster than their own selling prices. And as inflation accelerates it becomes impossible for individual producers to make any dependable estimate of the wage rates and other costs they will have to meet in the next few months, or their own future selling prices, or the margin between the two. The result is not only increasing malemployment but increasing unemployment. This was tragically illustrated, for example, in the last months of the German hyperinflation.

 Nor can the government mitigate the situation by any such further intervention as “indexing.” If it tries to insure, for example, that all workers are paid the average increase that has occurred in wages or prices, it will not only increase wages over the previous average but put out of business even sooner the producers who have not been able, because of lack of demand, to raise their selling prices as much as the average. Every attempt to correct previous distortions and inequities by government ukase will only create worse distortions and inequities. There is no cure but to halt the inflation. This is itself an operation not without its cost; but that cost is infinitely less than that of continuing the inflation—or even of trying to slow it down “gradually.”

In sum, an inflation can increase employment only temporarily, only to the extent that it is unexpected, and only when it is comparatively mild and in its early stages. Its long-run effect is to misdirect employment and finally to destroy it. The belief that inflation increases employment is perhaps the most costly myth of the present age.

1″A Statistical- Relation between Unemployment and Price Changes.” International Labor Review, June 1926, pp. 785-792. Milton Friedman has recently called attention to the article.

2″The Relation between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861-1957,”Economica, November, 1958, pp. 283-299.

3The present writer has discussed this question more fully in Ch. XXVI: ” ‘Full Employment’ as the Goal,” The Failure of the “New Economics,”1959.

  • Henry Hazlitt (1894-1993) was the great economic journalist of the 20th century. He is the author of Economics in One Lesson among 20 other books. See his complete bibliography. He was chief editorial writer for the New York Times, and wrote weekly for Newsweek. He served in an editorial capacity at The Freeman and was a board member of the Foundation for Economic Education.