Household debt, demographics, defaults, and spending

27 Jun 2010 by Jim Fickett.

This post is a broad overview of debt in the household sector: (1) In aggregate, household sector debt has been on an increasing trend for 6 decades (the boom only emphasized the trend). (2) Debt is not spread evenly (if it were it would not be a problem); young professionals and minorities have the most debt; the poor have the hardest time coping with their debt. (3) So far, the effects of debt reduction are most prominently seen in default, but a further hit to spending is likely.

It is a commonplace to speak of the “debt-burdened consumer”, who borrowed too much during the housing boom, and to say that households need to “rebuild their tattered balance sheets”. But what do such statements really mean?

Often analysts and reporters consider the situation for the country or the household sector in aggregate. Here is a graph of the assets, liabilities, and net worth of the household sector since 1952, the beginning of the current series (click for larger image; Internet Explorer users may want to turn off “Automatic image resizing” for a clearer image; data are from the Flow of Funds, table B.100.)

It is somewhat surprising, given current rhetoric, that liabilities have increased very gradually and rather evenly over the last 58 years. The liabilities curve never deviates far from the dotted line that runs from its start to its end. Yes, the accumulation stopped at times, was faster than average during the boom, and has pulled back since the bust. But overall, there is a clear trend running over decades.

How much debt is too much? At a sector level, aggregate debt is about 1.25 times aggregate disposable annual income, and also equal to about one fifth of aggregate assets. If this situation were spread evenly, it would probably not be a problem. 73% of the debt is mortgage debt, so think of a prototypical person with a mortgage balance of 11 months' salary, a car loan worth about 4 months' salary, and the value of house plus savings equal to about 72 months' salary. This hardly seems debt-burdened and tattered. And when you think about it, it doesn't seem too alarming that the older generation preferred to pay off the mortgage, but the younger generation is happy to pay rent to the bank and take the mortgage interest deduction.

The real issue, of course, is that the debt is not spread evenly. And the real question is how many people are vulnerable to temporary loss of income or changing interest rates. The aggregate level of debt is, to some extent, an issue, but not because some particular level is too high in and of itself. Rather, a higher aggregate level of debt may be a problem if it indicates the existence of more unfortunate individuals who have gotten in over their heads and are carrying far more than their share of the debt. There is certainly a correlation between aggregate debt and the number of stressed individuals:

The McKinsey study Debt and Deleveraging that I wrote about in January says,

our analysis also shows that total debt is rarely spread evenly within sectors, and that average levels of sector leverage mask pockets of very highly leveraged borrowers. It was these borrowers in each sector that got into trouble and caused most of the credit losses in the crisis. This suggests a need for far more granular tracking of debt and leverage within the economy.

We can get some idea of where those “pockets of very highly leveraged borrowers” lie from the 2007 Survey of Consumer Finances.

The first table below shows the leverage, meaning the ratio of liabilities to assets, in various subsets of the population. It turns out leverage is highest

  • in the middle and upper-middle class – the 40th to 90th percentile of income,
  • among those under 45 years of age, and
  • among minority ethnic groups

(The last point in the table, that leverage is high when net worth is low, doesn't tell us anything new – when debt is high and assets low, we get both leverage=debt/assets higher and net worth=assets-debt lower.)

The second table (click for larger image) shows the ratio of debt payments to overall income. The main thing this adds to the picture is that, although leverage is highest in the middle class, the percentage of families with really burdensome debt (over 40% of income going to debt payments), or with some payment more than 60 days past due, is highest in the lower class – income percentile under 20%.

So overall, young professionals and minorities have the most debt; the poor have the hardest time coping with their debt. Of course, even at this level of aggregation the generalities hide many specifics for individuals.

The demographic viewpoint can help an investor understand the trends. McKinsey suggest that since most of the problem debt is held by the middle class, a strong impact on spending is likely:

deleveraging by the middle class is likely to take a very different path than deleveraging by the poorest segments of socienty. Lower-income household have little or no savings, so deleveraging of these households is most likely to occur through default … Middle-income households have much lower default rates and instead deleverage by saving more and consuming less

Which brings us to the effects of deleveraging. There are two ways debt can be reduced – either the debt can be discharged through default, or it can be payed down by higher saving and lower spending. So the two possible effects of deleveraging are higher default rates or lower spending.

Despite the point just quoted from McKinsey, it seems that defaults are high among the better off, this time around. For example, from CoreLogic, via Reuters, we have this point today:

First American CoreLogic, which tracks U.S. real estate and mortgages, says the percentage of $1 million-plus loans more than 90 days delinquent rose to 13.3 percent in February, half again as high as the 8.6 percent overall delinquency rate.

The million-dollar delinquency rate has exceeded the overall delinquency rate since April 2008.

Although it is useful to keep an eye on the aggregate level of debt in the household sector, and to understand the demographics, I think the most useful information for investors is in the end effects: defaults and lower spending. So far, we can see

  • Overall deleveraging has been rather light (from the balance sheet graph above)
  • The most noticeable effect so far seems to be in defaults (see the bankruptcy graph in yesterday's post and the record and rising foreclosure rate)
  • There was a one-time effect on spending, but not as much of an ongoing effect as one might think (see graph here), given all the talk about excessive debt; this remains a surprise, and on the whole I think it is likely to change