Monetary policy, bubbles, and the knowledge problem.

• Author: Posen, Adam S.
• Position: Report

The role of monetary policy in creating, not merely responding to, asset price bubbles is a provocative topic of relevance to central banking requiring empirical analysis. The tenor of the times following the global financial crisis is to take that extreme premise as a given, and to advocate a policy response of tightening monetary policy preemptively to prevent or pop bubbles--that is, to lean against the wind. Leading expressions of this view, set out before the current consensus emerged, include Bordo and Jeanne (2002); Borio and Lowe (2002); Borio and White (2003); Cecchetti, Genberg, and Wadhwani (2002); Roubini (2006); and White (2006, 2009), As argued in Posen (2009), however, the success of such a policy depends upon three empirically testable assumptions: first, that we can discern bubbles in real time from among the ongoing fluctuations in asset prices before it is too late; second, that the monetary instruments available to central banks do affect asset prices in a dependable fashion; and third, that it is worth it on net to preempt bubbles, despite the potential costs from lost output and increased volatility of doing so.

Amidst the understandable public outrage in the United States and elsewhere regarding the aftermath of the 2008-09 crash, it is easy to rush to judgment on monetary policy and to forget that all three of these assumptions remain far from established. Previously, several researchers have taken on the second assumption, and offered cross-national evidence that monetary policy instruments do not predictably or dependably influence asset prices. (1) Analyses of the bubble of the 1980s in Japan, often held up as the paradigmatic example of a missed opportunity to preempt an asset price boom with monetary tightening (or even of a bubble caused by monetary laxity), show that the case does not fit the paradigm--not least because real estate prices there rose by 50 percent in two years prior to monetary loosening, and continued rising after monetary tightening began. (2) One implication is that if small open economies facing apparent bubbles at present driven by capital inflows raise interest rates by anything short of hundreds of basis points, they will only attract more inflows and exacerbate their problems (Posen 2010).

This article challenges the validity of the first listed assumption necessary for leaning against the wind to succeed--namely, the assumption that monetary policymakers can correctly identify asset price bubbles in time to respond preemptively (or at least usefully). This is something where many policymakers even previously skeptical now feel they can be like Supreme Court Justice Potter Stewart and recognize obscenity in asset prices when they see it. (3) Some patterns do emerge if we look more carefully at the historical record of asset price booms and busts (Hellebrandt, Meads, Posen 2011), but, in light of those patterns, the prospect of getting the call right becomes very daunting. This is not because one should have huge faith that financial markets are always correct in their pricing of risk and reward, or that central bank decisionmakers can avoid making judgment calls as a principle. The difficulty arises because of the complex nature of asset price booms and busts, a complexity that seems to be overlooked in the advocacy of leaning against the wind.

Some Booms Are Different

Nowadays, all empirical macroeconomists are rifling on Reinhart and Rogoff's (2009) already classic book, This Time Is Different. In that spirit, consider a few facts about the diversity of types of bubbles, or at least of asset price booms and busts. It turns out that some things taken for granted about booms and busts are not true, and their impact and attributes are quite varied.

Even if one is to use judgment on asset prices from a forward-looking perspective, it certainly helps to have an objective baseline definition for what constitutes a boom or bust looking retrospectively for research. Hellebrandt, Meads, and Posen (2011) create lists of asset price booms and busts, both for residential real estate prices and for equity prices, for 17 developed economies since the 1970s. (4) Their lists were generated by two distinct methods for identifying booms and busts: one which looks at sustained periods of price growth more than 1.3 standard deviations above/below the four-quarter moving average of the growth rate for the series (by country and asset type), following Bordo and Jeanne (2002); the other which uses Hodrik-Prescott filter methods to identify a time-varying trend growth rate for the given series, and then looks for sustained large deviations or gaps from trend, following Goodhart and Hoffmann (2008) and Hume and Sentence (2009). To a perhaps surprising degree, the results of the analyses are robust across the two differently generated (and genuinely differing) lists.

Examining simple descriptive statistics on the duration and timing of booms and busts yields already ample evidence that...