House Prices Are Important: A Simple Explanation for Why the U.S. Middle Class Has Fared Poorly

The New York Times has launched the Upshot, its new vertical, with an article by David Leonhardt and Kevin Quealy on the fact that low- and middle-income households in the United States are no longer the most affluent in the world. If you read the Agenda regularly, you’re already familiar with some of the data Leonhardt and Quealy use in their report, but they do an excellent job of covering the waterfront. The trouble with Leonhardt and Quealy’s analysis is that they ignore one extremely important factor that helps explain lagging U.S. performance.

First, let’s summarize Leonhardt and Quealy’s central finding. Median income in Canada tied with median United States income in 2010 (at $18,700, or $75,000 for a family of four after taxes), and there is reason to believe that it has since surpassed the U.S. median. The U.S. median was unchanged from 2000 to 2010 while the Canadian median had increased by 20 percent over the same interval after adjusting for inflation. And the gap between the U.S. median income and median incomes in a number of Western European countries, including Britain, the Netherlands, and Sweden, is smaller than it was in 2000. Households at the 20th percentile of the income distribution in the U.S. have significantly less disposable income than their counterparts in Canada, Sweden, Norway, Finland or the Netherlands, and the U.S. incomes were higher at this level 35 years ago (when the composition of the bottom fifth of the U.S. income distribution was quite different, due in part to an ongoing influx of less-skilled immigrants and the impact of mass incarceration on earnings and wealth accumulation, among other factors).

So what accounts for America’s deteriorating edge and Canada’s strong performance? Leonhardt and Quealy identity three central factors: (1) educational attainment in the U.S. is rising more slowly than in other market democracies (an issue we often discuss in this space); (2) U.S. business enterprises offer higher compensation to top-end employees, like CEOs, and less to other employees; and (3) governments in Canada and Western Europe do more to redistribute income.

What I found extremely surprising is that Leonhardt and Quealy ignore the role of housing wealth. Consider the following chart from the Organization of Economic Cooperation and Development:

This chart is from May of 2013, but the broad pattern hasn’t changed too dramatically much since. Houses appear to be correctly valued in the United States, where prices are rising after a substantial correction (perhaps you remember it). Houses are undervalued in countries that are still reeling from the after-effects of the crisis (Greece, Ireland, Portugal, Slovenia, Slovakia, and the Czech Republic), where prices are still falling; and in Germany and Switzerland, where growth in household disposable income and favorable financing conditions are now leading to rising prices. Houses appear to be overvalued and prices are falling in several European countries, including Spain, the United Kingdom, Belgium, Denmark, Finland, the Netherlands and one non-European country, Australia. This matters because house price declines will undermine the financial health of households.

And finally we have a special category of countries where houses appear to be overvalued yet prices have continued to rise, including Canada, Norway, New Zealand, and Sweden, albeit to a more modest degree. The OECD warns that countries in this category are the most vulnerable to a price correction.

Why would house prices matter for Leonhardt and Quealy’s analysis? Couldn’t it just be a coincidence that Canada has fared well and that it has yet to experience a housing price correction — while the U.S. housing boom came to a miserable close in 2006–7, the Canadian housing boom has continued more or less without interruption? Should we ignore the fact that house prices are still overvalued in Britain, the Netherlands, and Sweden, the countries where, according to Leonhardt and Quealy, the gap in median incomes with the United States is closing?

Housing matters for a number of reasons. When house prices are high, construction employment tends to be high as well, and construction employment is particularly beneficial for less-skilled workers. As of last fall, construction employment accounted for 7.6 percent of all employment in Canada and wages have been growing at a fast clip. Kerwin Kofi Charles of the University of Chicago’s Harris School of Public Policy and Erik Hurst and Matthew Notowidigo, both of the Booth School, have found that rising construction employment helped absorb workers displaced by declining manufacturing employment during the 2000s. When the housing boom was followed by the housing bust, this shift came to a close, and the result was particularly painful for low- and middle-income households. In 2013, Stephen Bronars observed that the absence of a recovery in construction employment in the United States does much to explain why the labor market looks so grim. There has been a modest improvement in construction employment since then, and one could argue, as I have, that we are on the cusp of a structural shift in construction employment. But there is no question that the Canadians are doing much better on this front. (One important caveat is that homeownership rates varied across the market democracies in the pre-bust era, as demonstrated by a look at homeownership rates in 2004. In some countries, like Austria, Denmark, and the Netherlands, homeownership rates were in the 50–55 percent range. In others, like Australia, Belgium, Canada, the U.S., and the U.K., they were in the 68–72 percent range. Germany, at 41 percent, and Spain, at 83 percent, were outliers. The wealth destruction caused by house-price declines appears to have a sharper effect on low- and middle-income consumption in societies where the middle class owned their homes, or rather where they owned some share of their homes, than in societies where the middle class was more likely to rent — a fact that should inform policymakers going forward.)

The more important reason housing matters is that house prices have an impact on consumption levels. Atif Mian of the University of California, Berkeley and Amir Sufi of the University of Chicago’s Booth School of Business have argued that the central driver of the decline in employment levels between 2007 and 2009  was the drop in demand caused by shocks to household balance sheets. Mian and Sufi tested this proposition by comparing employment declines in high-leverage counties, in which households had higher ratios of debt to disposable income, and in low-leverage counties, in which households had lower ratios. And they found that high-leverage counties saw a steeper drop in “non-tradable employment” than low-leverage counties, which led them to conclude that consumer-demand shocks were indeed key. In House of Debt, their brilliant new book, which I’ve promised not to write about at length until publication, Mian and Sufi detail the ways in which the housing bust damaged the economic well-being low- and middle-income households across the country.

Essentially, inflexible debt contracts led to massive wealth destruction for low- and middle-income borrowers while leaving wealthy savers largely unscathed. Because low- and middle-income borrowers have a higher marginal propensity to consume, the shift of wealth from borrowers to creditors sapped demand, thus leaving us with a weak economy. Mian and Sufi go into detail concerning how we might have avoided this fate, and though their line of thinking has tended to be embraced by the political Left, many of their conclusions are, or should be, congenial to pro-market conservatives, an idea that I intend to address at greater length. But for now it is enough to say that to ignore the role of housing wealth in understanding the trajectories of the U.S. and Canadian economies is to ignore a central, if not the central, part of the picture.

Apart from this elephant in the room, Leonhardt and Quealy’s analysis is unobjectionable. It is true that the rate of improvement in U.S. labor-force quality has slowed down, as Robert Gordon and others have observed, and as Tino Sanandaji and I have discussed. Other market democracies engage in more income redistribution. It’s not clear to me that the amount of redistribution the U.S. engages in is the problem; rather, I would argue that we ought to engage in less redistribution via tax expenditures and more, perhaps, via in-work benefits. That said, it seems like perfectly reasonable to pay attention to how taxes and transfers impact disposable income. Their claims regarding compensation within U.S. business enterprises don’t strike me as particularly compelling, not least because employment levels in the U.S. are so much lower than they are in Sweden, which seems like a good place to start. Yet it’s at least possible that internal firm dynamics play some role. I do think that it is a mistake to overlook the contrast between selective, employment-oriented immigration policies in countries like Canada and Australia and U.S. immigration policy, but this is a complicated issue and I can see why Leonhardt and Quealy chose not to focus on it.

But here’s the thing: If the U.S. had done more to address the wealth destruction that followed from the housing bust, it is hard to deny that middle- and low-income households would be in a much better position. Indeed, the really scary thing is that, as Mian and Sufi have argued, it’s not clear that low- and middle-income Americans are in a less vulnerable position now than they were before the bust. And if Canada ever does see a house-price correction, as seems at least plausible, it is not clear that the contrast between Canada and the United States will look quite so favorable a few years down the road.

And finally, I’ll make a brief political point. Many conservatives believe that to win Latino voters, they need to take a particular position on immigration reform. They might instead consider paying more attention to the fact that Hispanic household wealth fell by 66 percent from 2005 to 2009, and that many Latino families, and indeed many middle-income families of all backgrounds, are still reeling from the wealth destruction of that era.

P.S. Derek Thompson offers a nice variation on this theme, with additional charts.