A History of America's Disappearing Middle Class
By Paul Krugman
Friday 09 March 2007
Economist and New York Times columnist Paul Krugman explains in simple terms how the American economy went from having the world's most dynamic middle class to being on the verge of a rich-poor state in only 30 years.
The following is excerpted from the keynote speech delivered by Paul Krugman at the Economic Policy Institute's recent conference on The Agenda for Shared Prosperity.
... One thing I've been noticing on multiple debates in public policies - climate change is another one - is there seems to be an almost seamless transition from denial to fatalism. That for 15 or 20 years the people would say, "No, what you're saying is not happening." And then almost immediately they'll turn around and say, "Well, yeah, sure it's happening, but there's nothing that can be done about it."
And that's kind of the way a lot of the discussion now goes on inequality. That there is really nothing you can do to arrest this. That it's all the invisible hand driving this growth in inequality, and there's nothing you can do to really change it - well, maybe better education. But while education is very much a good thing, it's the all-American way of dodging problems. Since everybody approves of it, you say we should have better education but wave away the pretty strong evidence that while it's a good thing, it won't make very much difference. So there's this general sense that you can't do anything.
And I don't think that that's what the historical record suggests. That in fact when we look at it, there appears to be quite a lot that the political process can do about inequality. Just to say, there's the obvious. Obviously, even if you look at the United States right now, the tax and social insurance system makes an enormous difference.
But the amount of inequality in the United States is substantially less than it would be if we did not have still at least somewhat progressive taxation, and still a pretty extensive, though not nearly extensive enough, system of social insurance. And that makes a big difference. Certainly if you're looking at say the United States versus Canada, a lot of the difference between the two countries is just that Canada has more of a better safety net financed by somewhat higher taxation.
And if you're looking for a progressive agenda, certainly from my point of view, a large part of that ought to be straightforward orthodox stuff, which is still very hard to do politically. It would be essentially restoring progressivity of the tax system, and using the revenue to improve social insurance and, above all, health care.
So, if you say what would I really like if I went into a Rip Van Winkle sleep and woke up ten years from now, I'd like to wake up and discover that we have a national health care in some version with the necessary funding supplied in part by higher taxes on me, or actually, the top two percent of the income distribution. But people a lot richer than me, of course. But it's not the whole story that the only thing you can do is taxes and social insurance. And the arc of history for the United States suggests that there's actually a lot more that can happen.
If you look back across the past 80 years or so of the United States, what you see is that in the 1920s, we were for practical purposes still in the gilded age. That may not be the way the historians cut it, but in terms of the actual distribution of income, so far as we can measure it in terms of the role of status and general feel of the society, we were still an extremely unequal royalist society.
By the time World War II was over, we had become the middle-class society that the baby boomers in this audience grew up in. We had become a much more equal society. That high degree of equality began to go away - depending on exactly which numbers you look at - during the late 70's, maybe a little earlier than that. And at this point we're basically back to pre-tax and transfer to the levels of inequality that we had in 1929.
So there is this great arc to the middle class, away from gilded age to middle-class society and then back to the new gilded age, which is now what we're living in. And there are really two puzzles about that. One of them is a political puzzle, which is why instead of leaning against these trends, politics has actually reinforced them. Why it is that U.S. politics moved left in the age of a relatively middle-class society, and moved right as society got more unequal?
A na ve view of politics would say that, "Gee, when a few people are winning a lot and most people are lagging behind, people ought to be voting for more social insurance and more progressive taxation, not less." And we have some understanding of why that doesn't happen. It has to do with the role of money, organization and all of these other things that affect politics. That story also helps us understand why politics gets so nasty.
If you actually look at some of the measures - I'm really into quantitative political science these days - of political positions that political scientists calculate, it does look as if what the main thing that moves actually over time is in fact the Republican party. The Democratic Party has not - at least with northern Democrats - gotten significantly more liberal over the past. They haven't moved much at all over the past 30 years.
But the Republican Party, which had largely converged on the Democrats in the age of Eisenhower, has moved sharply to the right. And so that one party, in effect, moves with the income of the top 5 percent or 1 percent of the population. So that seems to be the story. I mean, we can think about reasons why that might be true. But the other puzzle, and this comes to the question of the conference, is what drove these changes? How did we become largely middle class?
Why have we become a much more unequal society once again? And the standard, what economists like to say, is "Well, it's all invisible hand. It's all market forces." The history doesn't seem to look like that, if we ask how did the society we had in 1947, which is when a lot of our data start, come into existence.
Was it a gradual process whereas the economy developed and we got out of the early days of the American industrial revolution, we gradually moved towards middle classness? Well, no, historically it happened in an eye blink. In this Claudia Golden and Bob Margot classic paper, they call it the great compression. As late as the late 30s, the income distribution appears to be highly unequal.
By the time you wake up in 1946 or so, it's highly equal. And how did that happen? A lot of it was more or less deliberate compression of wage differentials during World War II. But if you were or had standards, supply and demand for different types of labor, you'd say that should last only as long as the wage controls lasted. It should have sprung back to where it was, but it didn't. It actually stayed quite equal for another 30 years at least. You ask, what buttressed that? Well, partly it's the rise of a powerful union movement, which is at least in large part a change in the political climate, but then remained in place for several decades more.
Other things we're not sure. But it looks more or less as a leveling of the income distribution. Obviously we want to be careful about the words. No one presumably in this room, and certainly not me, is advocating Cuba. We're not calling for a flat income distribution. But the relative equalization that seems to have taken place was engineered by a combination of top-down politics and grassroots organization that made people want a more equal society in the 30s and the 40s, and they got it.
And it remained for quite a long time. Now, that started to come apart roughly 30 years ago, and there's been a large increase in inequality since then. As people probably know, I've written about the part that is sort of polite to talk about, which is the rising premium for highly educated workers. But that's only part of it. Even more spectacular is the increase in inequality of the far-right tail of the income distribution.
The CEOs and high school teachers who got roughly the same number of years of formal education haven't exactly had the same growth in income over the past 30 years. So, there's this vast increase in inequality at the top. What do we think caused that? I actually just had to do a class on that. It was in my international trade class, but we were doing the trade and inequality stuff.
And the question is what do we think is underlying the rise in inequality in the United States? And searching for metaphor, I actually ended up with the "Murder on the Orient Express." Not for what actually happened but for the way we described it. In "Murder on the Orient Express," somebody is killed and there are 12 suspects. The question is which of them did it and the answer actually is all of them. The official economic story about rising inequality is one in which we have a whole bunch of villains, which all seem to be playing a role.
So we've got skill bias and technological change, which is shifting demand towards highly educated workers. We've got growing international trade with increased imports of labor-intensive products further reducing demand for less educated workers. We have immigration, possibly similar in its effect to trade. We have the falling real value of the minimum wage contributing at the bottom end. We have some affected unionization driving the change in income distribution.
Finally, in terms of at least the after-tax distribution, we have changes in taxes which have, in general, reinforced rising inequality. It could be true, but it's kind of funny that all of these different things should be working in the same direction. In "Murder on the Orient Express," there is an elaborate conspiracy that means that all 12 of the potential suspects were actually in collusion. It's a little hard to see how all of these factors and economics are in collusion.
Okay, I think that what we can say is that the political climate matters more for the distribution of income than the economic models that we know how to work with and would seem to suggest more than our models capture. If you ask me practically what I want done now, I think that the most important agenda thing right now is, in fact, to work on the taxes and social insurance side, because that is concrete and you can get stuff.
But there is a lot of reason to believe that a change in the political climate in various ways can do a lot more than you would think just from looking at the taxes and social insurance. Let me give you two pieces of evidence that I looked at. One is that there is some really interesting, though intellectually disturbing, work by my colleague, Larry Bartell who is in the Princeton Politics Department and has just looked at what happens to income growth at different points in the income distribution under administrations of the two parties.
Now there shouldn't be a big difference really because at any given historical period, the visible policies are not all that different. Certainly there is a pretty significant shift from Clinton to Bush and there was, in fact, a pretty significant shift from Bush to Clinton previously. But it's in taxes and it really shouldn't be very obvious at pre-tax distribution of income. And yet what Bartell finds is actually there is a really striking difference. Inequality on average rises under Republicans. At least in the bottom 80 percent of the income distribution, it's stable or falling under Democrats. The top 1 percent just kept on rising right through, but there is at least a surprising, fairly robust correlation.
The other thing I would say is timing. There's a very clear co-movement over time between income inequality and both the political polarization and the rightward tilt of our politics. It's pretty clear that the rising inequality over the past 30 years has been associated with a rightward shift of the political center of gravity, mainly because of the Republican Party shifting to the right.
You might say that's the causation running from income distribution to politics. But if you actually then just start to look at it through history, the timing actually seems to be reversed. The rise of an aggressive or rightwing movement and the rise of a really major assault on the New Deal great society legacy both come before the big shift in income distribution takes place.
The emergence of the modern right is something that obviously dates back to Goldwater, but really becomes a political force in the '70s. You don't really see the big changes in income distribution until the '80s. So it looks almost as if, just in this crude sense, politics is leading the economic changes. How could that operate? I just want to talk about two things. I suspect that there are quite a few channels that we don't really perceive, but there are two that are fairly clear. One of them is unionization.
Obviously, private sector unions were very important in the U.S. 30 years ago and have very nearly - not completely, but very nearly - collapsed, and they are down to eight percent of private employment. Why did that happen? You will often see people saying - well, that's because of de-industrialization, and because of the decline of manufacturing. But that is actually not right. It's not right in two ways.
First of all, arithmetically, most of the decline in unionization is a result not of the decline in manufacturing share, but of the decline of the unionization of manufacturing itself. So the big thing that happens is that there is a collapse of unionization within the manufacturing sector and then of course also a smaller share of manufacturing in the economy, but it's much more dramatic on the collapse within the sector.
The other is that there is no law that says that unionization should be a manufacturing phenomenon. What it really is, to the extent that there is a story, is that large enterprises are more likely to be to be unionized. The reason why the high tend of unionization was also a period when manufacturing was the core of the union movement, is that at that time, large enterprises were largely a manufacturing phenomenon.
Now we have a service economy in which there are a lot of large service sector enterprises. Not to put too fine a point on it, but why exactly couldn't Wal-Mart be unionized? It doesn't face international competition. There is no obvious reason why it wouldn't be possible to have a strong union in Wal-Mart and in the big box sector and other parts of the economy. And just think of how different the whole political economy would look if the service sector enterprises were unionized.
Not necessarily all the effects would be positive, but it would certainly be very, very different. What happened? Why did manufacturing unionization collapse? Why didn't the emerging service sector get unionized? And the answer is actually pretty straightforward and pretty brutal. It's politics and aggressive employer behavior enabled by politics.
I have seen estimates of a fraction of workers who voted for a union and who were fired in the early '80s. They range from a low of one in 20 to a high of one in eight. There is no question that aggressive, often illegal, union busting is the reason the union movement declined. And the change in the political climate that began in the '70s clearly played a role in making that possible.
Now how important is all of that? You may have seen that there have been a number of estimates of the effects of unions on income distribution. It's funny. People will often say that those estimates are small and actually they typically are roughly comparable in size to typical estimates to the effect of international trade on income distribution. So these are both in secondary and the standard accounting to technological change, but both fairly significant.
What is more, there are a lot of reasons to think that those estimates are not capturing a lot of the story. As the people who do them will concede, what they basically do is say: What if the workers were paid, unionized and non-unionized workers were paid the same as they are now, and just do a sort of shift share analysis. What that doesn't capture - and they know it, but there is just no way to do it better - is the effect of a strong union movement on the bargaining position of workers who are not unionized, but might be.
It doesn't capture the effect of a strong union movement and possibly disciplining insider behavior by executives and so on down the line. So it is likely that that is a much more important story than we typically give it credit for being. Let me just give you my other piece of the story, executive compensation. There's a raging debate now of how much of the soaring executive compensation is insider self-dealing, and how much of it is just market forces.
I went back and was looking at what people said about executive compensation when it was low, just 40 or 50 times the average worker salary. [Here are] some quotes: "Managerial labor contracts are not, in fact, a private matter between employers and employees." "Parties such as employees' labor unions, consumer groups, Congress and the media create forces in the political media that constrain the types of contracts." And so on down the line.
A lot of discussion was of the role of the political climate that was basically hostile to outrageous paychecks and constrained it. Where are these quotes from? They are actually from [economists] Michael Jensen and Kevin Murphy writing, saying people have complained that there are not enough incentives in executive pay. They are saying that what we really need is to have executives get more stock options and stake in the firm - in other words, all of the stuff that has happened since then.
So back when executive pay was low, 40 or 50 times average pay, it was actually the defenders of higher executive pay that complained that it was actually non-market forces that were constraining executive pay. Now of course that disclosing of pay has happened, the same side of the debate says it's ridiculous to claim that social norms and political forces have any role in this. But I think it's actually quite clear that it did. We can argue about which is the natural market outcome. But the point is, in fact, that we had a society 25 years ago in which there were some constraints imposed by public opinion, by strong unions, by a general sense that there were things that you don't do.
And maybe that led firms to make a decision to think of there being a sort of tradeoff between a "let's have a happy high morale" workforce, or let's have a super star CEO and squeeze the workers for all we can. There were some things that tilted the balance in that decision.
Okay, are we going to do another great compression? Hopefully not. The reason I say hopefully not is even FDR needed World War II to be able to carry out that sort of wholesale social engineering that took place. I am not looking forward to having a replay of that. I think that if we get serious, as some of us hope we do, and we actually do have a swing back in the political pendulum that - in addition to the direct stuff - we can do a lot to change the climate in the many small ways that add up to a lot of impact on the bargaining power of workers.
The ability of the bottom 80 percent of the population to get a bigger share of the total pie - I think that if we get there, we may be surprised at just how successful we are at moving ourselves back, at least part of the way, to the kind of middle-class society that people like me grew up in.