Why Small Booms Cause Big Busts

BERKELEY – As bubbles go, it was not a very big one. From 2002 to 2006, the share of the American economy devoted to residential construction rose by 1.2 percentage points of GDP above its previous trend value, before plunging as the United States entered the greatest economic crisis in nearly a century. According to my rough calculations, the excess investment in the housing sector during this period totaled some $500 billion – by any measure a tiny fraction of the world economy at the time of the crash.

The resulting damage, however, has been enormous. The economies of Europe and North America are roughly 6% smaller than we would have expected them to be had there been no crisis. In other words, a relatively small amount of overinvestment is responsible for some $1.8 trillion in lost production every year. Given that the gap shows no signs of closing, and accounting for expected growth rates and equity returns, I estimate that the total loss to production will eventually reach nearly $3 quadrillion. For each dollar of overinvestment in the housing market, the world economy will have suffered $6,000 in losses. How can this be?

It is important to note that not all recessions cause so much pain. Financial blows in 1987, 1991, 1997, 1998, and 2001 (when some $4 trillion of excess investment was lost when the dot-com bubble burst) had little impact on the broader real economy. The reason why things were different this time can be found in a recently published paper by Òscar Jordà, Moritz Schularick, and Alan M. Taylor. Large credit booms, the authors show, can greatly worsen the damage caused by the collapse of an asset bubble.

Historically, when a recession is caused by the collapse of an asset bubble that was not fueled by a credit boom, the economy is roughly 1-1.5% below what it otherwise would have been five years after the start of the downturn. When a credit boom is involved, however, the damage is significantly greater. When the bubble is in equity prices, the economy performs 4% below par, on average, after five years – and as much as 9% below par when the bubble is in the housing market. Given these findings, it is clear that the distress experienced since the beginning of the economic crisis is not far out of line with historical experience.