GS: What is the "Right" Level of Household Debt?

US Daily 

 

·         Households owed $13.5 trillion to their lenders as of the end of the first quarter, or 120% of after-tax income.  With the housing bust and credit crunch, this stock is gradually declining.  But what is the “right” level of debt and how long will it take to get there?

·         The mortgage market could ultimately shrink from over $10 trillion (trn) to perhaps $8trn.  Prime mortgage lending clearly serves an important economic purpose and should remain broadly stable, though some decline is possible given lower prices and possibly lower initial loan-to-value ratios.  Nonprime and home equity lending suffered from much higher default rates in the crisis; originations have collapsed and so the stock of debt should (and will) stabilize at much lower levels.  Consumer debt, currently $2.4 trn, would have to see a peak-to-trough decline of as much as 20% to bring it back in line with debt ratios that were sustained without problems in the past. Of course, debt levels are just one metric of the deleveraging process, and others--such as the debt service calculations described in our latest US Economics Analyst--are also relevant.

·         If the recent pace of deleveraging continues, and household income grows modestly, it could take the better part of a decade to reach this new equilibrium.  However, since it is the change—not the level—of saving that matters for growth, debt reduction would not necessarily prevent GDP growth from recovering to above-trend levels well before the end of the deleveraging process. 

Households owed $13.5 trillion to their lenders as of the end of the first quarter, or 120% of after-tax income.  (Note: figures discussed below for “household” debt include nonprofit organizations, although available data suggest that the vast majority—more than 90%—are in fact direct obligations of households. The numbers referenced in the text are also organized in the table below.)   The peak household debt-to-disposable income ratio of 130% was set in the second half of 2007, as the credit bubble was just beginning to burst.  This was more than double the ratio a quarter-century before, itself a record at the time.  In the aftermath of the housing bust and credit crunch, many market participants wonder just how far down debt needs to go.  Put another way, what is the “right” level of household debt?

Exhibit 1: What a Stable Ratio of Household Debt Might Look Like

The bulk of household debt – $10.2 trillion or three-quarters of the total – is mortgage-related.  Of this, $1 trillion are home equity loans, roughly $3 trillion are nonprime mortgages (subprime or “Alt-A”), and the remainder are prime (both conforming and jumbo).  The total and home equity figures come from the Fed’s latest Flow of Funds report, while the nonprime mortgage stock is an approximation that starts with the estimate from “Home Prices and Credit Losses: Projections and Policy Options” (GS Global Paper 177, January 13, 2009) and adjusts it downward based on the decline in the number of subprime mortgages outstanding over the past two years.  Household mortgage debt peaked at $10.6 trillion in early 2008, and we believe the total stock could ultimately decline by as much as $3 trillion, with the contraction concentrated in nonprime mortgages and home equity loans:

1. The prime mortgage market clearly serves an important economic and social purpose—allowing households to become homeowners during their early working years, rather than having to wait until near retirement to own property—and thus should remain the biggest share of household debt for years to come. However, the stock of prime mortgages could nonetheless decline slightly over the next few years for two reasons.  First, home prices have fallen approximately 30% from their peak and are now back to late-2003 levels on a Case-Shiller basis.  With lower prices, purchasers need smaller mortgages.  (Of course, only a fraction of currently outstanding loans were made near peak prices, so the implied decline in the mortgage stock from lower prices is much smaller than 30%.)  Second, initial loan-to-value ratios (LTVs) could become slightly more conservative.  Against these reasons for a declining prime mortgage stock, some current nonprime mortgage borrowers may be able to refinance into prime loans.

2. The future of nonprime mortgages is much dimmer, for obvious reasons.  Nonprime originations have collapsed following a surge in delinquencies and defaults and the evaporation of demand for securitizations.  The “right” level of nonprime mortgage debt thus appears to be well below current levels. If the stock of nonprime mortgages ultimately falls by half to two-thirds—a decline which would still leave nonprime debt outstanding somewhat above levels at the beginning of the decade—this would represent the disappearance of $1.5-$2trn in mortgage debt.  This precipitous decline represents a combination of fewer nonprime loans, lower home prices, and much lower initial LTVs (though the Federal Housing Administration currently finances loans up to 96.5% LTV, we presume that figure will eventually decline).  The number of subprime mortgages has already fallen by about one-third from peak levels, according to the Mortgage Bankers’ Association.    

3. Home equity loans seem likely to contract significantly as well. As we have showed in past research, mortgage equity withdrawal was highly correlated to the rate of home price appreciation; a mere leveling out of home prices, let alone the destruction of equity experienced in recent years, should have reduced home equity originations considerably.  As home equity loans are a relatively minor portion of the total mortgage stock, we simply assume the “right” stock of home equity loans is half the current level, although one could easily argue for a bigger decline.

Beyond mortgages, households owe another $2.4trn in consumer debt.  This is slightly more than 21% of disposable income, down from a peak of nearly 25% in the 2001-2005 period.  Deleveraging in this sector seems likely to continue, at least in the near term.  Consumer debt hovered between 15-20% of household disposable income in every year from 1960-1994 without a major financial crisis, and so in our view a reasonable first-pass approximation would be a return to the lower end of that range.  This would imply a decline to a stock of $2trn of consumer debt given current incomes, a little higher in future years assuming incomes continue to grow modestly. 

Our guesstimate of the equilibrium household debt stock, summarized in the table, is therefore in the $11 trillion range, or about $2½trn below current levels. This includes $8trn of mortgage debt. (One way of cross-checking the “appropriate” level of the mortgage stock is simply to roll it back to 2003, when housing prices and homeownership were in the neighborhood of current levels and the nonprime boom was still getting started.  At that point, the entire US residential mortgage stock was still less than $7 trillion.  An important difference is the lower overall level of home equity now and for the foreseeable future.)  It also includes $2.2trn for consumer debt and $0.8trn of other debt not classified as mortgages or consumer credit.

How long would it take to get back to this level?  The debt stock can decline either because of net paydowns or because of defaults.  Over the past year, the decline has been only about $250 billion or roughly 2% of the total.  At this pace, cutting the debt by $2½trn would take a decade.  In the first quarter of 2010, the annual rate was about $400bn or about 3%; at this pace, the “equilibrium” stock would be reached in just over six years.  Taking the simple average of these two estimates suggests a deleveraging that could last around eight years.  Given that ongoing increases in nominal income would lift the equilibrium debt stock accordingly—the figures above were all calculated based on current incomes, home prices, and so on—the risks may be tilted to a slightly shorter time.

This simple back-of-the-envelope calculation seems broadly plausible given historical precedent.  A McKinsey Global Institute study of deleveraging episodes (“Debt and deleveraging: the global credit bubble and its economic consequences,” January 2010; deleveraging is defined here as declines in the debt-to-GDP ratio) suggested that typical “belt-tightening” episodes lasted six to seven years.  Our previous analysis of the “Big Five” crises identified by Carmen Reinhart and Kenneth Rogoff (see “After the Private Sector Retrenchment,” US Economics Analyst 10/23, June 14) suggests that private sector saving remains high for many years after the crisis, with little appreciable decline until eight years after the crisis (which would correspond to six years from now).   

Suppose household debt did stabilize at about $11trn eight years from now.  What debt-to-disposable income ratio (DDIR) would this imply? If nominal after-tax income grew 3% per year, just over half its pre-crisis pace, this would imply a DDIR of about 78%, down from 120% currently.  We last saw this ratio in 1988, with interest rates at significantly higher levels, the technology boom still in its infancy, and “subprime” two decades from being a household word, and in fact the household DDIR stayed between 58% and 85% from the early 1960s through the mid-1990s.  This seems like a reasonable guess for the “right” level of the household DDIR, and consequently a plausible outcome as well, though it’s certainly possible to imagine milder results, particularly while significant government support and low interest rates continue.  More severe outcomes would probably require a sharp rise in interest rates or an extended period of even slower disposable income growth. Of course, debt levels are just one metric of the deleveraging process, and others--such as the debt service calculations described in our latest US Economics Analyst--are also relevant.

We emphasize that forecasting a long period of deleveraging is definitely not the same thing as saying the United States will have a “lost decade.”  Deleveraging does not automatically mean low growth beyond the initial period when saving rates adjust, which is when headwinds are most severe.  It’s the change in the saving rate, not the level, that matters for growth.  While saving is rising in the early phase of deleveraging—as it did in 2008 and 2009—growth suffers.  Once saving is at desired levels, it’s the pace of income growth that matters.  High nominal growth—whether from real activity or inflation—actually would facilitate deleveraging, since it’s much easier to pay down debt when your income is rising.   So even though saving is likely to remain high for years, this does not automatically mean growth must remain low. 

Andrew Tilton