A lot has been written on the so-called lean vs. clean debate in monetary policy that tries to resolve whether central banks should actively try to counter asset price bubbles during their build or just make sure they do everything they can in order to clean up in the aftermath of these bubbles bursting. To recap, there’s decent arguments for both sides of the debate: the clean camp, applying the “Greenspan principle”, generally cites the difficulties in spotting bubbles, often due to efficient market hypothesis concerns, as the main reason for not using the particularly blunt instrument of interest rate policy to deflate potential bubbles – after all, can we ever really know it is one until it bursts? The clean camp, while generally acknowledging the problems involved in bubble-spotting, tend to stress the damage to the real economy that financial asset bubbles can cause, highlight the potential inability of central banks to “clean up” under certain circumstances, such as when the zero lower bound on nominal interest rates is reached, as well as the inconsistency of asymmetrically responding to asset prices – not at all during the build-up of a bubble, yet strongly to its bursting. There are models that supposedly show that both the clean as well as the lean approach is the optimal one (e.g. Bernanke and Gertler on the clean side and Filardo on the lean side, just to name two – both .pdfs).
Yet in general most of the literature seems to, mostly implicitly but often also explicitly, assume that leaning against asset price bubbles would, at least over the short run, generally hurt the economy, thus potentially making it less attractive. Raising interest rates to combat a bubble in e.g. housing prices would obviously have the effect of hitting the rest of the economy as well. Yet this would imply an odd construction of the target of the central bank: again an asymmetric one, yet this time biased to the upside. Just as both inflation as well as real economic activity targets of central banks in classic Taylor rules are defined symmetrically, there is no strong enough reason an additional asset price target should not be as well. Staying with housing prices: For what it’s worth, it is probably impossible to reasonably expect a central bank to find the “correct number” a given housing price index should have (without even going into the different issues with the indexes themselves) at any given moment in time. However, and while still imperfect, it would seem reasonable to assume that the growth rate of housing prices should not stray too much from its historic average growth rate (1 year? 5 years? 10 years? Linear or smoothed trend? Different question…), thus providing a decent benchmark by which to judge current developments in the housing market.
This approach has several nice properties. First, the trend used to define the “desired” growth rate would be adjusted each period as new data comes in and the oldest data point is thrown out. If a long sample period is picked, this adjustment would be slow – essentially a sign of skepticism regarding potential structural shifts that could, in fact, raise the growth rate of housing prices and one that would interpret any deviation from trend growth, unless it holds up for a fairly long time, as “bubbly”. If the sample period to construct the trend is short, the target value obtained from it adjusts fairly quickly, ensuring that structural shifts in the housing market and the economy are accounted for fairly swiftly, while at the same time running the risk of neglecting bubble-like behavior. Evidently a middle ground would need to be found that is in tune with the central bank’s preferences, yet in either case both structural shifts as well as potential bubbles would be accounted for to a certain degree.
But the potentially more important feature that stems from the very way a “housing price growth target” (as a proxy for the financial market and for purposes of illustration, not necessarily “the” asset price that matters the most) might be constructed is that, during any given period, there will be quarters where housing prices grow faster than the target and those where they grow slower than the target (unless the growth of housing prices is either significantly and consistently upward or downward biased, i.e. non-stationary, which would seem unlikely). In fact, if housing price growth rates are, over time, stationary, then on average a central bank targeting housing price growth behaves in a strictly neutral fashion – there will be an equal amount of leaning against periods of higher growth as there will be of stimulus during periods of weaker growth. The rest of the economy should therefore, on average, not be much affected. So leaning against asset price bubbles does indeed hurt the rest economy – except during the times when it actually stimulates it.