It's the leverage, stupid!

In the 30 months following the 2000 stock market peak, the S&P 500 fell by about 45%. Yet the U.S. recession that followed was brief and shallow. In the 21 months following the 2007 stock market peak, the equity market fell by a comparable 52%. This time was different: the recession that began in December 2007 was the deepest and longest since the 1930s.

The contrast between these two episodes of bursting asset price bubbles ought to make you wonder. When should we really worry about asset price bubbles? In fact, the biggest concern is not bubbles per se; it is leverage. And, surprisingly, there remain serious holes in our knowledge about who is leveraged and who is not.

We’ve known for generations that the stock market can go up and down without upsetting the economy too much. As Paul Samuelson quipped in the 1960s, “Wall Street Indexes have predicted nine of the past five recessions.” The following chart of the four-quarter growth rate in the S&P 500 and U.S. real GDP makes the point. Note two things. First, there is a dramatic difference in the scales: the change in the stock price index ranges from -45% to +50% (with a standard deviation of 16 percentage points), while real growth ranges from -4% to +9% (with a standard deviation of around 2½ percentage points). And second, the stock market “overpredicts” business cycle downturns: over the past half century, there were 8 recessions, but the S&P 500 dropped below year-ago levels 14 times.

S&P 500 and U.S. Real GDP (Percent Change from Four Quarters Earlier)

Sources: FRED and Yahoo Finance.

Sources: FRED and Yahoo Finance.

So the stock market matters for growth, but not all that much. Rising stock prices do raise household wealth and lower the cost of new capital, but several factors mute the impact on the aggregate economy. First, the consumption of people who own equity is not very sensitive to stock price fluctuations because the owners are generally wealthy and can easily smooth their spending in the face of transitory disturbances. Second, the holders of U.S. equities typically do not borrow to finance these assets.

This highlights the key point: it is leverage that matters most. Several years prior to the crisis, one of us examined data on equity and housing booms and concluded that the latter worsen growth prospects and create outsized risks of very bad outcomes while the former did not. The recent crisis confirms this pattern, as the collapse of U.S. housing prices brought down the related House of Debt and the financial system with it (see this earlier post).

A notable exception proves the point. When the equity bubble in Japan burst in 1990, the largest Japanese banks were among the biggest equity holders. Their leveraged equity losses (combined with the leveraged losses on their real estate-related lending) turned a number of leading banks into financial zombies.

So when we see asset price booms, the real question is how leveraged are the owners, and how leveraged are the intermediaries. Looking at the specifics of the 2007-09 crisis, the issue is the sensitivity of the balance sheets of U.S. large banks to changes in real estate prices. To get some sense of this, we have plotted the ratio of real estate loans to equity for U.S. domestically-charted banks in the following chart. The blue line represents the 25 largest banks, while the red line is for all banks. [Because of regulatory changes in the early 1990s, we start in 1994.] In the years prior to the crisis, the ratios were below 3 for large banks and greater than 3 for all banks (implying that smaller banks had a ratio around 4).

U.S. Commercial Bank Real Estate Exposure (as a ratio of equity)

Source: Federal Reserve Board, H.8 release.

Source: Federal Reserve Board, H.8 release.

We need to emphasize that the ratios plotted in this chart understate the exposure – at least on the part of large banks, who were able to game the regulatory system by increasing their off-balance sheet exposure.

A classic example of such regulatory arbitrage was the expansion of special investment vehicles (SIVs) in the years before the crisis. To reduce their regulatory capital needs, bank holding companies moved mortgage-backed securities (MBS) off their balance sheets into SIVs that issued asset-backed commercial paper (ABCP) to finance these assets. Such aggressive maturity transformation – financing long-term assets with short-term liabilities – is inherently risky. When mortgage defaults began and the value of the MBS fell, SIVs could no longer roll over their ABCP. At this point, banks had to make good on the back-up lines of credit they had extended to their SIVS. These contingent credit lines were off-balance sheet exposures to real estate.

Derivative instruments provide an additional avenue by which a financial intermediary can create leveraged real estate exposure without a regulatory capital charge. The classic pre-crisis example was American International Group (AIG), originally an insurance company. Through its Financial Products Group, headquartered in London, AIG managed to amass $446 billion in notional exposure as a seller of credit risk protection via credit default swaps (CDS) on MBS. In essence, they were selling insurance on mortgages.

Another way to highlight the role of leverage is to ask how the relatively small asset class of subprime mortgages could have triggered a crisis like that of 2007-2009. In 2007, outstanding subprime debt was less than $2 trillion, or only about 3% of all financial liabilities. Even if all subprime mortgages had become worthless – they didn’t – the loss of wealth would have been equivalent to a stock market decline of less than 8%. The reason that subprime debt mattered so greatly is that the losses sat on the balance sheets of leveraged intermediaries (including AIG, which required a nearly $200-billion government rescue).

Post-crisis changes in regulation have made it more difficult for banks and other intermediaries to amass off-balance sheet exposures, so the recent numbers in the chart probably are less understated than before the crisis. Put differently, the decline in U.S. commercial bank real estate exposure likely is greater than we are able to measure.

Disturbingly, however, we know of no source that reports the aggregate off-balance sheet exposure of banks (or other financial intermediaries) to real estate. We can only hope that regulators and governments are acting to fill this enormous and serious gap in our knowledge.

All of this leads us to draw two simple conclusions. First, investors and regulators need to be on the lookout for leverage; that’s the biggest villain. In the United States and many other countries, mortgage borrowing has been at the heart of financial instability, and it may be so again in the future. But we should not be lulled into a sense of security just because banks’ real estate exposure has declined. If leverage starts rising in real estate or elsewhere – on or off balance sheet – then we should be paying attention.