America’s household debt binge *was* about income inequality

Since 2006, a yawning gap has opened up between the average level of real household spending and the remarkably stable long-term trend:

The standard explanation is that the trend’s stability was misleading.

For one thing, aggregate consumption spending grew much faster than disposable incomes thanks to the steady decline in the average savings rate since the early 1980s. Not surprisingly, America’s household savings rate hit bottom just before consumption spending began falling away from its long-term trend.

Even more important than this aggregate story, however, was the changing distribution of income. While the soaring fortunes of those at the top is well-known, the less-appreciated corrollary is that those in the bottom half experienced barely any growth in their real market incomes:

Adding in the impact of taxes and cash transfers doesn’t change much, especially if you focus on the period before the crisis:

(Data come from the Thomas Piketty / Emmanuel Saez / Gabriel Zucman Distributional National Accounts.)

This matters because people who have a lot of money generally don’t spend that much if they get any more. Their material needs are already mostly satisfied, so any extra income is mostly appreciated for the status it accords and used to purchase assets. In contrast to the vast majority of the population that saves basically nothing, the rich save almost half their earnings:

Given all this, one might have reasonably expected America’s skewed income growth since the 1970s to have depressed consumption growth relative to total income growth while boosting the aggregate savings rate. Yet that’s the opposite of what actually happened.

The apparent contradiction between the income and consumption data can be reconciled by looking at changes in borrowing and saving. Many Americans saved progressively less and less while boosting their borrowing to support consumption growth far in excess of income growth. At least, they did until 2006/7.

As Saez and Zucman show, the household savings rate for Americans in the bottom 90 per cent of the wealth distribution collapsed from about 7 per cent in the mid-1980s to 3 per cent in the mid-1990s to negative 3 per cent by the late 1990s to negative 7 per cent by the mid-2000s:

Many Americans also exploited looser credit conditions and rising house prices to temporarily boost consumption spending above what would have been sustainable based on underlying income trends. During the peak of the bubble, this home equity extraction boosted household purchasing power by about 8 per cent, compared to an average of about 1-2 per cent during the mid-1990s.

Why are we bringing all this up now? Because JW Mason has released a draft of an intriguing new paper saying this standard story is wrong.

According to him, the boom in debt can be explained by the failure of interest rates to fall in line with the slowdown in nominal income growth. Meanwhile, consumption growth has been overstated because of bad methodology. Measured properly, he argues, it hasn’t grown much and is probably being driven by high-earners.

The first part of Mason’s argument is the most interesting. As he explained in more detail in an earlier paper with Arjun Jayadev, more than all of the reported increase in American household indebtedness from the mid-1980s through the late 1990s can be explained by high real interest rates. Households’ “primary deficit”, defined as the change in the stock of debt minus interest payments and plus defaults, was consistently negative throughout this period:

Had interest rates stayed steady at their 1946-1983 average and everything else played out as it actually did, American household indebtedness wouldn’t have changed at all between the early 1980s and the late 1990s (finely-dashed line):

None of this, however, explains what happened in the 2000s, a borrowing binge that looks even more extreme once the relatively high real interest rates of the 1980s and 1990s are accounted for.

The next part of Mason’s argument is that little of the money raised from this borrowing binge was actually spent on consumption, a claim that partly rests on an unusual distinction between consumption and investment. Noting that the vast majority of the household debt stock consists of mortgages, followed by auto loans and student loans, Mason says “most borrowing does not finance consumption” but is instead undertaken to buy assets.

But this is misleading.

For starters, much of the growth in mortgage borrowing, particularly in the 2000s, was about extracting equity from housing people already owned, rather than buying additional units. Just because this borrowing is collateralised by a fixed asset doesn’t mean it should be considered “investment”.

Atif Mian and Amir Sufi have extensively documented how the combination of rising house prices and collapsing lending standards boosted consumption during the boom. Once the process reversed, it was the places with the least growth in mortgage debt, such as Texas, that fared best. By contrast, the places with the biggest growth in mortgage debt — and consumption — during the boom have had depressed consumption ever since.

At least money spent to acquire housing can be considered an investment in the national accounts. The same can’t be said for spending on motor vehicles or schooling.

One way to distinguish between spending normally thought of as consumption and spending typically defined as investment — and avoid falling into a definitional rabbit hole — is to consider how easily the thing being bought can be resold in the future, and at what price.

Over time, stocks and bonds tend to generate significant returns above inflation. Houses, generally speaking, tend to maintain their nominal value over time, and often their real value, too. New cars and trucks, by contrast, lose about a fifth of their value in their first year and continue to depreciate after that, even if they’re well-maintained.

Education is even less of an investment, since it’s literally impossible to recoup the money spent on schooling on short notice by selling your degree or certification. There is no secondary market. So in at least one important respect, spending on clothes, furniture, and video games is more of an “investment” than spending on education.

The best you can hope for is that you learned something useful, or made the right connections, so that your earnings are higher after being in school than they otherwise would be. Even when borrowing to get a degree is a sound decision — and it often isn’t — there’s a good reason the money spent on schooling is classified as consumption in the national accounts.

To be fair, there is another way to distinguish between investment spending and consumption spending: does the thing you’re spending on produce long-term benefits?

If you buy something that boosts your long-term earning power, or can be used for a long time, thereby limiting your future expenses, it could reasonably be considered as an investment rather than a consumption expense. This is often how businesses treat their spending, amortising out the cost of big-ticket items and taking depreciation charges along the way.

Boffins don’t use this approach for household spending because it is extremely difficult to apply it consistently. It’s not just new cars and trucks that have long lives, but everything from picture frames to socks. (More on that here.)

In fact, strictly defining consumption as spending on goods and services that can’t be reused would more or less reduce the category to groceries and energy. Money spent on checkups at the doctor’s can (potentially) generate much greater long-term benefits than spending on a new car.

Even spending on live entertainment and restaurants can be thought of as investment because the experience can be enjoyed again and again as a memory. The memory might even provide greater benefits as time passes — a much better investment than spending money on a depreciating car. From this perspective, it’s difficult to explain why spending on schooling should count as “investment”, but not spending on concert tickets or holiday cruises.

This is not how Mason chooses to proceed. Instead, he ignores his earlier discussion about the difference between investment and consumption and constructs a measure of spending that only tracks household cash outflows.

This is reasonable. “Owners’ equivalent rent” isn’t actually a drain on anyone’s income, while health spending on behalf of households paid by the state is already accounted for in taxes, inflation, and interest rates. Mason also rightly excluded spending by nonprofit institutions on behalf of households.

The result is this chart, showing how household consumption is supposedly lower now, as a share of total output, than it was in the 1960s, and has been stable since the early 1970s:

The discrepancy between the two lines is determined entirely by Mason’s decision to exclude health spending. (It’s relatively straightforward to replicate the chart by using tables 2.4.5 and 7.12 of the National Income and Product Accounts, although for the sake of consistency you should use different denominators for the two ratios.)

None of this matters that much, though. From 1990 through 2006, the difference between the growth in consumption spending as commonly measured and Mason’s narrower definition was less than 5 per cent:

Even by his own methodology, it looks as if consumption growth hasn’t been particularly overstated. By contrast, the 10 percentage point collapse in the aggregate household savings rate — excluding all imputed forms of income and expenditure — suggests Americans really did impair their balance sheets for the sake of keeping up consumption:

Mason has presented some interesting work, particularly on what he calls the “debt disinflation” of the 1980s. But the overall thrust of his argument is not convincing.

Related links:
The New York Fed thinks it’s time to make cashout refis great again — FT Alphaville
Stop pretending America’s housing boom had nothing to do with lending standards — FT Alphaville
Yes, looser credit — and fraud — drove the housing bubble — FT Alphaville