Bubbles, Credit, and Their Consequences
Òscar Jordà, Moritz Schularick, and Alan M. Taylor
FRBSF Economic Letter 2016-27 | September 12, 2016
The collapse of an asset price bubble usually creates a great deal of economic disruption. But bubbles are hard to anticipate and costly to deflate. As a result, policymakers struggle to determine how they should respond, if at all. Evaluating the economic costs of past equity and real estate bubbles—with particular attention to how much credit grew during boom phases—can provide valuable insights for this debate. A recent study finds that equity bubbles are relatively benign. More danger comes from housing bubbles in which credit grows rapidly.
Asset price booms pose a challenge for monetary policy. It is difficult to separate optimism justified by future prospects about underlying fundamentals from optimism fueled by speculation about future prices alone. Matters are made worse when borrowed money is involved, as Fisher (1933, p. 341) reminds us. Sometimes the economic consequences of the boom-bust cycle typical of a bubble are well contained. A good example is the dot-com boom and bust of information technology stocks in the early 2000s. Other times the consequences are dire, as the collapse of the housing market and the Great Recession taught us.
How should a central bank respond to an asset price boom? Rudebusch (2005) provides a detailed road map. It all depends on how early one can tell that there is a bubble, and whether the costs of different policy interventions outweigh the benefits. Central banks monitor asset markets closely since they provide timely information on broader economic conditions. However, it is often difficult to tell the signal from the noise. Fluctuations in asset prices reflect a varying mix of fundamentals and speculation. Uncertainty about these mixed signals has often stayed the hand of policymakers, much like a monetary version of the Hippocratic admonition, “First, do no harm.”
There are good reasons to be cautious. When it comes to interest rate policy, the evidence seems to suggest that it is often best to “clean” rather than “lean.” Using estimates available from a wide variety of sources, Williams (2015) calculates that the interest rate hikes needed to prevent the housing boom of the mid-2000s would have inflicted a deeper economic slowdown than the Great Recession. However, these estimates are inevitably imprecise. On the opposite side of the ledger, Borio and Lowe (2002) argue for more forceful interest rate policy in low inflation environments while recognizing that it is difficult to identify financial imbalances in advance.
Our recent research (Jordà, Schularick and Taylor 2015) examines the connection between asset price bubbles, credit growth, and macroeconomic outcomes. This Economic Letterpresents our estimates of the economic costs of bubbles in equities and in real estate under a variety of credit scenarios. Bubbles happen somewhat infrequently, and seldom in housing markets. To maximize the sample of such episodes, we reach back in history to 1870 and look beyond the United States to 16 additional advanced economies (the complete list is available in Jordà et al. 2015).
Credit booms and asset price bubbles
A defining feature of advanced economies in the post-World War II era is the rise of credit documented in Jordà, Schularick, and Taylor (2016). This is visible in Figure 1, which displays the cross-country average ratio to GDP of unsecured and mortgage lending since 1870. Following a period of relative stability, both lending ratios grew rapidly after the war, with mortgages taking off in the mid-1980s. As a result, households have become more leveraged than ever. This is seen in Figure 2, which displays the ratio of the total value of mortgage lending to the value of the stock of U.S. housing. Before World War I that ratio stood at around 0.15, but since then the ratio has more than tripled.
Average ratio to GDP of unsecured and mortgage bank credit
Source: Jordà, Schularick, and Taylor (2016).
Ratio of total mortgages to total value of U.S. housing stock
Source: Jordà, Schularick, and Taylor (2016).
Most buyers use mortgages to buy homes, but few savers use borrowed funds to invest in the stock market. Thus, one might expect equity price busts to be less dangerous than collapses in house prices: A crash in the price of assets financed with external (rather than internal) funds is likely to have deeper effects on the economy. As collateral values evaporate, some agents will delever to reduce their debt burden, in turn causing a further collapse in asset prices and in aggregate demand. The more widespread this type of leverage is, the more extensive the damage to the economy. Integrating the role of credit into the analysis of asset price bubbles is therefore critical.
What is a bubble and how would you measure it?
There is no widely accepted definition of what an asset price bubble is. Intuitively, it is a situation where asset prices drift up from their fundamental value, that is the current valuation of the expected stream of payoffs, and then fall abruptly. And therein lies the rub. The fundamental value is neither directly observable nor easy to determine. Borio and Lowe (2002), Bordo and Jeanne (2002), Detken and Smets (2004), and Goodhart and Hofmann (2008) all use, one way or another, large deviations of asset prices from some reference level or booms and subsequent busts to identify bubble episodes.
We follow a two-pronged approach. First, we focus on spells of sizable asset price run-up—an event we expect to see less than one-third of the time in the data—and a subsequent collapse in prices of 15% or more. This simple rule of thumb would identify events such as the U.S. Great Depression, the mid-1980s Japanese real estate collapse, the 2000s dot-com boom and bust, and the recent collapse of housing markets in many countries. This is, of course, a backward-looking measure meant to sort the historical record for analysis. It is not designed to be a bubble-forecasting tool.