The Great Compression


Paul Krugman recently blogged:

The middle-class society I grew up in didn’t evolve gradually or automatically.  It was created,
in a remarkably short period of time, by FDR and the New Deal.  As the
chart shows, income inequality declined drastically from the late 1930s
to the mid 1940s, with the rich losing ground while working Americans
saw unprecedented gains.

As you can see, the share of the top ten percent (not counting capital gains) falls steeply at about 1937 and flattens out by about 1942-43, with a slight uptick just afterwards.  But I am puzzled by Krugman’s description of the process.  A few points:

1. 1937-38 were disastrous years for the American economy and also for the middle class, mostly because of bad contractionary monetary policy.  This can be considered a second Great Depression and it aborted a recovery in process.  Robert Higgs has shown convincingly that World War II was an economic disaster, look at the figures at consumption don’t be fooled by aggregate gdp which is inflated by production for the war.

2. Therefore I am not sure when the "unprecedented gains" came during this period.  Yes 1940 was a year of recovery (and part of 1939) but is the claim that the middle class was created in that year?  Surely not but then I am confused.

3. Krugman cites "strong unions, a high minimum wage, and a progressive tax system" as driving the Great Compression — did these factors change so notably in 1937-44?

4. In real terms, relative to the time, the new federal minimum wage of 1938 was not especially high, even compared to the minimum wage today.  And the data are for pre-tax incomes, which means that progressive taxation is unlikely the major part of the story about the distribution of pre-tax income (noting the Yglesias caveat that the rate of tax will influence pre-tax incomes through labor supply).

5. Here are some data on U.S. unionization.  The history does broadly track Krugman’s time frame, as long as one does not obsess over 1937-43 as being special.  Note that optimistic estimates of the union wage premium today run about 15 percent, so it is hard to see unionization as the dominant factor in the change in the distribution of income.

6. The wartime economy, and the scarcity of labor, bid up wages for black workers, women, and many remaining non-drafted male laborers (due to forced saving, however, consumption was low).

7. My explanation of the break in the chart emphasizes a combination of events: top incomes were crushed by the depression of 1937-8, the war economy put a further lid on top of high incomes (for reasons of law and norms; surely the phrase "wartime wage and price controls" deserves to be uttered at least once), and the war economy bid up wages for many people near the bottom and middle of the distribution.  The wealthy classes financed a disproportionate chunk of World War II, it seems.

8. Krugman mentions none of the factors listed in #7.  Admittedly I may be wrong, but are not those factors obvious candidates for an explanation?

9. It is a vitally interesting question why postwar America stayed at this new percentile distribution of income (more or less), even after recovery from the war.  It may be possible to defend a version of Krugman’s broader hypothesis — policy matters for income distribution — in this setting, but the story then puts greater stress on both the equalizing effects of catastrophes and also on path-dependence.  That story might also suggest that strongly negative real shocks would be needed once again to make income distribution much more egalitarian.

10. How about this for an alternative story: "Crush the incomes at the top and then make the fat cats pay much higher wages to protect the world and become a superpower.  Impose wage and price controls as well.  See how long it takes before these distributional effects — which don’t exactly match the distribution of economic talent– reverse themselves in the aggregate."  I’m not sure that’s right but at least it seems to match more of the history.

I am sensitive to the claim that many people misinterpret the words of Paul Krugman, so please do read his whole post.  If he addresses these matters in more detail in his forthcoming book, I will let you know.


I note that the incomes used in the graph excluded capital gains. In the presence of high marginal tax rates, larger incomes tend to be diverted into financial instruments. For example, take a look at the population of partnerships (Statistics of Income, IRS) and consider the rapid growth in the 70's and 80's. A lot of this growth was due to real estate partnerships used as tax shelters.

The increase in non-capital gains incomes would seem to track the marginal tax rate cuts rather well.

(PS: I used to work on this particular study as the lead survey statistician.)

I agree with Jack. Suppose all wealth was confiscated and destroyed, and all forms of employment is outlawed. Look at how equitable this society is, everyone has the same income ($0).

Paul's point is the main thing of relevance here, but lets generalize it a bit further. The private company car and the expense account are part of this story too, as are, in fact, all of the tax-burden driven transitions from direct expenditures to "benefits", "expenses" etc. Remember that top (federal only) tax brackets were 87%-90% of income for most of this time. People at the top simply were not and could not have been working for income under such circumstances. They were getting compensation in other ways.

I didn't see a link to the study shown in the graph, but a few questions need to be asked. Is that pre-tax or post-tax income? Does it count income from government transfers? In-kind government transfers? Cash income only or benefits? Not including capital gains is only the first of many ways a graph of this sort can be juggled. While policy might or might not effect inter-personal compensation levels, there is absolutely no doubt that tax policy greatly effects how rich individuals structure their compensation.

What has everyone got against vertical gridlines these days? Without them, you have to count data points from 1917 to figure out that the Great Compression-- all of it!-- occurred in 1941-43. Wartime inflation seems the likeliest cause; its persistence after the war may have something to do with the continued poor performance of bonds all the way up to 1981. Leaving capital gains out of a measure of income for the top 10% will tend to exaggerate the importance of the bond market.

3. Krugman cites "strong unions, a high minimum wage, and a progressive tax system" as driving the Great Compression -- did these factors change so notably in 1937-44?

Unions were at their peak after the imposition of NIRA (passed 1933). It is reasonable to imagine that it could have taken a few years to have its full effect. Union membership peaked at about 34%. The NRA also introduced a minimum wage, new income taxes and other taxes, strict price controls, public works programs and some very heavy handed regulation.

Between 1941 and 1945 more than 1/3 of the top 100 American corporations were seized either in whole or in part.

I absolutely believe that these things had a huge impact on the economy including compressing wages, dragging all incomes down, suppressing growth, scaring away investment and driving businesses into bankruptcy.

Its a wonder that we ever climbed out of the great depression at all, and with fits and starts (thank you Jimmy Carter) finally have managed to recreate much of that intense growth that we had pre-FDR.

I'd like to see how the top 1% and 5% and bottom 1% and 5% fared over the years. It might show something very different. I suspect that the rise of the top 10% is due mainly to the dramatic rise in income of formerly upper middle-class people, due to changes in society that reward talented and intelligent people (as opposed to the "middle class era," where rich people made their money by inheriting it). In other words, it's due to the rise in income of a substantial portion of the bottom group of the top 10%.

Neither Paul Krugman nor Tyler Cowen, nor for that matter any of the commenters so far, realy have it. The Great Depression had very little to do with this. It was almost entirely a matter of World War II and its very strong labor market, plus the pushing of the top marginal tax rate to 94% after the war started (it was 75% in 1939). Actually looking closely at the figure that Tyler presents shows it. The sudden drop in the Gini does not start until 1940 (OK, a year before the US got into the war, but not going back to 1937, much less 38 or 39).

Anyone wanting to see an authoritative source on this matter should look at the old, but not disproven, book by Morgan Reynolds and Eugene Smolensky from 1977, _Public Expenditures, Taxes, and the Distribution of Income: The United States, 1950, 1961, 1970_ Academic Press (yes, they do talk about what happened before 1950, despite the title).

To Pat Mathews:

Although there is data on income distribution going back into the 19th century (big surprise, the 1890s appear to be the peak of inequality), it gets progressively less reliable as one goes further back in time, already pretty shakey once we are back as far as the 1920s.

Imposing punitively high income tax rates like we had during "the great compression" and thereafter will do wonders for the tax avoidance industry wherever you are in the world.

Same goes for wage and price controls, people still find a way to get paid, such as employer paid health insurance that is wrongly not recorded as income. This occurred during "the great compression" as well and stayed.

Just because income isn't getting recorded by the IRS as individual income doesn't mean that there isn't income, and I'm not talking about tax evasion here, noting that tax avoidance is legal and tax evasion is illegal. Warren Buffett seems to have mastered the art of tax avoidance and has paid very little (in relative terms) in taxes over the decades even as his net worth soared. Nevertheless it seems pretty apparent that it is due to Buffett's mostly untaxed labor that his holding company conglomerate has soared in value.

Looking at the graph it becomes apparent that there was a huge surge in income inequality after the 1986 tax act. This tax act eliminated an awful lot of tax shelters, and also reduced the need for them by severely lowering the income tax rate. Did employers suddenly start paying their top employees more money at that time? I seriously doubt it. The more likely explanation is that during that two year jump in the graph what happened is that "the rich" simply restructured the vehicles their income was taxed under, and suddenly it wound up being recorded as individual income whereas before it was still income but not officially recorded as such. If such an obvious tax shift is being recorded without an asterisk, it seems logical to think that maybe the whole chart is an omitted asterisk.

Perhaps a more properly conceived chart, taking into account the unseen income as well as the seen income, would show a much more gradual rise in inequality from say 1942 to present, with an upward slope the whole way similar to the decade from the mid 70's to the mid 80's.

Paul Krugman may wish to think that incomes were more equal during that long middle period, but it is more likely a statistical illusion (there are three kinds of lies: lies, damned lies, and statistics), at least in huge part.

happyjuggler0 wrote, This tax act eliminated an awful lot of tax shelters, and also reduced the need for them by severely lowering the income tax rate.

So I take it you're on record in favor of removing the current favorable treatment of capital gains?

If such an obvious tax shift is being recorded without an asterisk, it seems logical to think that maybe the whole chart is an omitted asterisk.

The entire chart needs an asterisk, because citing marginal rates is a pretty meaningless exercise without information about the definition of taxable income and without the bracket structure.


I am most definitely not in favor of raising anyone's taxes, but I can be convinced to radically lower everyone's taxes and radically lower government spending. My post was merely meant to be an alternative explanation to Krugman's "we need more big government" agenda.

Don't forget the immigration restrictions that were passed in 1924 and took effect in the later 1920s and then were repealed in 1965, with the new law taking effect in 1968. It took awhile for immigration momentum to build up, but it was becoming significant by the later 1970s.

This is by no means the whole story, but economists shouldn't ignore this piece of the puzzle.

A very comprehensive comment by Happyjuggler0.


The US suffered less than Germany or Japan since its war production machinery was not obliterated. However, there was a long transition of refitting the capital stock to the production of consumer goods after the war. This was interrupted by the Korean conflict. This refitting involved remaking and scrapping/replacement of the war production capital base. However, the US sufferred mightily during the war as can be seen by any analysis of consumption during and immediately after WWII. This loss in consumption is the symptom of capital destruction/misallocation. That people don't understand this is why they always claim that WWII ended the Great Depression. The Great Depression did not end until after WWII.

Pardon me... Precisely what is the definition of "economic talent" if not the ability to thrive economically in whatever the given economy of the day happens to be? The ability to thrive economically in some sort of imaginary laissez fair system?


I am confused. What Happyjuggler wrote was that it was not evasion he was talking about, but rather the use of legal loopholes that you wrote of.

Also, very few immigrants today, or in the early 20th century, were "well-off entrepreneurs" when they arrived in this country. Many of them became such people within a generation's time, yes, but not on arrival.

"War does not increase economic wealth- it destroys it in the most efficient way possible short of boating capital out to the middle of the ocean and sinking it."

Actually, war DOES boat capital out to the middle of the ocean and sink it. WWII especially.

And the USA is lucky that it was one country in WWII that didn't change its statistics by reducing the denominator.

With all of the caterwauling that Krugman does on behalf of economic justice, why does he not simply support a move to a consumption tax, adjusted so as to first untax family spending to the poverty level (a/k/a FairTax).

The effective percentages, that different income groups would pay under a FairTax consumption tax, is calculated by crediting a monthly "prebate" (advance rebate of projected tax on necessities) against all likely spending by citizen families (1 member and greater per Dept. of HHS poverty-level data). (Under the plan, a single person would receive ~$200/mo; a family of four ~$500 - in addition to working members no longer having tax withholding confiscated from the fruits of their labor every two weeks.) Prof.'s Kotlikoff and Rapson (10/06) concluded,

"...the FairTax imposes much lower average taxes on working-age households than does the current system. The FairTax broadens the tax base from what is now primarily a system of labor income taxation to a system that taxes, albeit indirectly, both labor income and existing wealth. By including existing wealth in the effective tax base, much of which is owned by rich and middle-class elderly households, the FairTax is able to tax labor income at a lower effective rate and, thereby, lower the average lifetime tax rates facing working-age Americans.

"Consider, as an example, a single household age 30 earning $50,000. The household’s average tax rate under the current system is 21.1 percent. It’s 13.5 percent under the FairTax. Since the FairTax would preserve the purchasing power of Social Security benefits and also provide a tax rebate, older low-income workers who will live primarily or exclusively on Social Security would be better off. As an example, the average remaining lifetime tax rate for an age 60 married couple with $20,000 of earnings falls from its current value of 7.2 percent to -11.0 percent under the FairTax. As another example, compare the current 24.0 percent remaining lifetime average tax rate of a married age 45 couple with $100,000 in earnings to the 14.7 percent rate that arises under the FairTax."

Further, per Jokischa and Kotlikoff (circa 2006?) ...

"...once one moves to generations postdating the baby boomers there are positive welfare gains for all income groups in each cohort. Under a 23 percent FairTax policy, the poorest members of the generation born in 1990 enjoy a 13.5 percent welfare gain. Their middle-class and rich contemporaries experience 5 and 2 percent welfare gains, respectively. The welfare gains are largest for future generations. Take the cohort born in 2030. The poorest members of this cohort enjoy a huge 26 percent improvement in their well-being. For middle class members of this birth group, there's a 12 percent welfare gain. And for the richest members of the group, the gain is 5 percent."

The robust growth of consumption, since the Great Depression, has been relentless. As a tax base, it seems far superior to current wage (indiv) /current year (business) earnings. The added bonus is that it would be, essentially, self-regulating: the more one spends, the more tax is paid. And gone, for good, the the many tax shelters that are available to the rich, only.

To all those who say that WW2 was a economic disaster: under what criteria? Of course (private)consumption fell during that period, but war production is a form of public consumption, so overall consumption rose. Arguably, the investments in mass production and R&D during the war set up the period of high growth after the war.

To all those who say that WW2 was a economic disaster: under what criteria?

There are two ways an economy can "recover" from a depression: (1) the economic capital is reallocated and remade to produce things that consumers want, or (2) consumers are convinced to want what economic capital finds easy to produce.

War does the latter. Certainly, beating the daylights out of Germany and Japan is a second best solution to economic depression. Nonetheless, it is preferable to getting the daylights beat out of yourself. During World War II, consumers wanted war production and the placement of valuable labor in foxholes. Calling it a recovery, however, means paying too much attention to numbers and too little attention to what people actually want.

Can we agree that Krugman needs his PhD revoked? Or at least have an asterisk next to his name, saying "Not an actual economist."


I maintain that my hypothesis that the big jump in individual income after the 1986 tax act was in fact due to the 1986 tax act, and that that was because people were legally handling their income somewhere other than on their individual income tax forms.

What is your alternative hypothesis? That the reduction in federal headline taxes paid from 50% to 28% spurred people to work harder a la what Art Laffer and the WSJ predicted? In other words, that Reaganomics worked as advertised? If so, then I don't see how this increase in inequality is a bad thing, it is merely spurring our most productive people to work where before they chose to be idle. They aren't depriving anyone of anything by working harder, on the contrary they are trading something valuable with the rest of us that otherwise wouldn't have existed, and thus by revealed preference is a good thing for the other 99% of us.

Or are you going to simply say that that massive surge in inequality was a random event that just happened to coincide with a major tax act that had as a major part of its intent the notion that individual income would now be reported as individual income? Cough.

I can see why someone on the economic Left would want to try such a head in the sand approach, but it is a glaring flaw in Krugman's notion of increased inequality. However I don't think it is wise for us to allow him to simply finesse away this issue, not if we are going to radically readjust our country's economic policy towards redistributionism like he wants us to.

I think Steve Sailer was suggesting that immigration might have contributed to income inequality since 1968, but not so much prior to 1924. The difference in income between the average American and average immigrant before 1924 was not as large as it was after 1968.

Immigration increases the size of the total economy, but the biggest gains go to the immigrants, those who own capital, and star professions, like sports, finance, and entertainment. But they weren't the only new participants. Women entered the workforce in large numbers as well.

Rich people were sheltering their income post-New Deal while middle class Americans earned high wages because labor force participation was restricted. In the 1970's, immigration surged and women entered the workforce in large numbers. In the 1980's, the tax shelters disappeared.

It would do wonders here to do a simple google search. I found this paper

The structure of wages narrowed considerably during the 1940's, increased slightly during the 1950's and 1960's, and then expanded greatly after 1970. The era of wage stretching of the past two decades has been a current focus, but we return attention here to the decade that was witness to an extraordinary compression in the wage structure. Wages narrowed by education, job experience, region, and occupation, and compression occurred within these cells as well. For white men, the 90-10 differential in the log of wages was 1.414 in 1940 but 1.060 in 1950. By 1985 it has risen back to its 1940 level. Thus the recent widening of the wage structure has returned to it a dispersion characteristic of fifty years ago. We explore various explanations for the rapid compression in the wage structure during the 1940's and for its maintenance during the subsequent decade or more. We first assess the hypothesis that the Great Depression left the wage structure in 1939 more unequal than in the late 1920's, but we find evidence to the contrary. World War II and the National War Labor Board share some of the credit for the Great Compression. But much belongs to a rapid increase in the demand for unskilled labor at a time when educated labor was greatly increasing in number. These same factors caused the wage structure to remain compressed until its expansion during the past two decades.

I do not think this bodes well for Krugman's policy hypotheis. But I think K's hypothesis is more political than economic.

"Can we all agree that a strong and vibrant middle class is vital to the national good? And, can we all agree that the middle class has gotten smaller since, say, the beginning of the Viet-Nam war?"

No, and no.

"granted the serfs of 100 years from now will be much better off than they were in Tsarist Russia" ... "Maybe we need another cataclysmic war that will require the rich to pay for it and that will restrict the labor markets to a point that pay for workers will have to rise..."

You would like to see a massive war, with millions of deaths come about in order to deplete the resources of the well off, reduce supply of workers and compress incomes so that it appears more equal.

I would like to see the poor - who are not serfs at all, but are free - continually increase their purchasing power just as they have been over all these years, via markets and entrepreneurship, without war and death and regulation. Growth helps the poor-- technology helps the poor in real terms, cheap products help the poor and everyone else by increasing their standard of living. Whether it looks like the middle spectrum is broader or more condensed is really irrelevant to me. The absolute purchasing power of the worst off and of the median are all that matters-- and they continue to rise.

"10. How about this for an alternative story: "Crush the incomes at the top and then make the fat cats pay much higher wages to protect the world and become a superpower. Impose wage and price controls as well. See how long it takes before these distributional effects -- which don't exactly match the distribution of economic talent-- reverse themselves in the aggregate." I'm not sure that's right but at least it seems to match more of the history."

Thomas Piketty has analysed the french "great compression" in this book, which I think has not been translated into english :

He shows that the first part of the chart is the same in France as in the US. Mostly, french top incomes were reduced in the 30s and the 40s by the great depression, the war economy, and war destructions. But then, top incomes did not come back to their prewar level, and have roughly stayed this way until now : this is why french inequality is now lower than in the US, it has (mostly) not reverted.
Piketty explains this with changes in tax policy. After WW2, France adopted a high level of progressive income tax, and high inheritance taxes, compared to what prevailed before. For Piketty, these tax changes explain why inequality remained low in France, countering the process in which high incomes become family fortunes. Tax policy changed in the US during the early 80s, the progressivity of income taxes, especially, decreased. The same happened in the UK, with the same effect : a rise of the share of top incomes.
So following Piketty, you have it right for the first part of the story : "middle class america" was created by the war economy and the depression, not by new deal policies. But Krugman has it right on the second part : what he calls the "great divergence" came from politically-driven changes in the progressivity of income and inheritance tax, not from a reversion to a "normal" tendency.


No, I don't want to see another stupid war that might destroy the economy. Oops, about four years late for that.

I did not say the poor were serfs now. I said that, if something doesn't change, they may well be in 100 years.

Do you really believe that a vibrant middle class is not vital to the economy? And you think the upper and/or the lower classes have gotten smaller in the last 40 years (that would have to have happened if the middle class shrank).

Rising tides may lift all boats. But it helps to first have a boat.

As the only industrial game in town, US companies could afford to pay its workers more, as they had little to no competition......

Hi! 我是一個室內設計.師,工作很忙,但很愉快,看到顧客滿意的笑容,我心里像開了花一樣。

Comments for this post are closed