RDP 2005-04: Monetary Policy, Asset-Price Bubbles and the Zero Lower Bound 1. Introduction

In a low-inflation economy, the bursting of an asset-price bubble can have significant and long-lasting consequences, both for the economy and for the operation of monetary policy. In Japan, the collapse of a major bubble in property and stock prices in the early 1990s ushered in over a decade of weak growth and declining price pressures – culminating, by late 1998, in ongoing consumer price deflation. This, in turn, has seen the Bank of Japan constrained in its actions, for over five years, by the zero lower bound (ZLB) on nominal interest rates. Likewise, the tech stock crash in the United States in 2000 marked the start of an economic downturn which saw the year-ended growth rate of the core personal consumption expenditure price index decline briefly to below 1 per cent, and prompted concerns for a time that the US Federal funds rate might also reach the ZLB.

These examples suggest, at the very least, that the interaction between asset-price bubbles and monetary policy is an important one for policy-makers, especially if operating in a low-inflation environment. If asset-price bubble collapses represent a primary mechanism by which otherwise well-functioning economies may become seriously destabilised, even to the point where monetary policy becomes constrained by the ZLB, this raises key questions for policy-makers as to how they might be able to forestall, or at least reduce the fall-out from, such collapses. These questions relate not just to how policy-makers might wish to react pre-emptively as asset-price misalignments develop, but also to choices about the framework within which policy is set.

This paper uses a simple, stylised two-equation model, due originally to Ball (1999a) and Svensson (1997), to explore these questions. More precisely, it builds upon recent work by Gruen, Plumb and Stone (2003), in which the Ball-Svensson model was augmented by the inclusion of an asset-price bubble. Gruen et al then used this augmented model to investigate the implications of such bubbles for optimal policy settings, under a variety of assumptions both about the bubble's stochastic behaviour, and about the degree to which this behaviour can be influenced by the actions of policy-makers.

Gruen et al (2003) highlighted two conflicting influences on policy-makers attempting to handle a developing bubble. On the one hand, they are likely to become more confident as time passes that observed asset-price rises do indeed constitute a bubble, and so become more willing to respond actively to these rises. At the same time, however, they would become increasingly conscious of the negative effects on the economy from the bubble's eventual bursting – effects which they would be anxious not to compound, given the delay with which monetary policy changes flow through to real activity.

As a result of these competing influences Gruen et al found that, even with an excellent understanding both of the economy and of the parameters governing a bubble's stochastic behaviour, it may be unclear whether policy-makers would wish to tighten policy in the face of such a bubble, beyond the degree to which they would do so based on an efficient markets view of asset prices. Their results highlighted the stringent informational requirements therefore inherent in a pre-emptive policy approach to asset-price bubbles – and the need for delicate judgements, in pursuing such a strategy, about both the process driving the bubble and its likely sensitivity to monetary policy.

In this paper we extend the work of Gruen et al (2003) by removing one simplification built into their modelling approach. This was the assumption that, whenever the economy is struck by a large negative shock, policy-makers can set the real interest rate as far below neutral as desired, regardless of the current level of inflation. By contrast, in this paper we impose a ZLB on nominal interest rates as a constraint on the actions of policy-makers attempting to deal with a developing asset-price bubble. We then examine the implications this has for both: the behaviour of policy-makers who believe that they understand the stochastic properties of the bubble; and the policy framework within which they must make their decisions.

With regard to the former, policy-makers who wish to react pre-emptively to a growing bubble must now take into account whether their current actions might result in them being unable to set the real interest rate optimally in subsequent periods, whenever the bubble bursts. Moreover, in doing so, they must allow for the possibility not merely that their actions might become constrained in the period in which the bubble actually bursts, but also that this might occur with a lag (as the bubble's collapse flows through to lower inflation, so reducing the amount by which the real interest rate can be set below neutral).^[1]

As regards the latter issue, if inflationary expectations are at least partially backward-looking then the level of inflation immediately prior to an asset-price bubble collapse clearly becomes important, in the presence of a ZLB on nominal interest rates. Hence, this constraint may influence decisions about aspects of the policy framework itself, such as policy-makers' preferred choice of target inflation rate.

Footnote

Note that our focus in this paper is on the effect which these possibilities (that is, the presence of the ZLB) might have on the interest rate recommendations of policy-makers, over and above whatever direct impact the presence of the bubble itself might have on these recommendations in the absence of the ZLB. Note also that our focus on the impact of the ZLB on policy-makers' thinking while a bubble is still growing is in contrast to much of the recent research on the ZLB, which has focused on how policy-makers should react once the ZLB has been reached. A brief review of where this paper sits within the recent literature on both asset-price bubbles and the ZLB is provided in Appendix A. [1]