An interesting new paper from Hess Chung, Jean-Philippe Laforte, and David Reifschneider of the Board of Governors, and John Williams of the SF Fed:
Have We Underestimated the Likelihood and Severity of Zero Lower Bound Events?, by Hess Chung, Jean-Philippe Laforte, David Reifschneider, and John C. Williams, January 7, 2011: Abstract Before the recent recession, the consensus among researchers was that the zero lower bound (ZLB) probably would not pose a significant problem for monetary policy as long as a central bank aimed for an inflation rate of about 2 percent; some have even argued that an appreciably lower target inflation rate would pose no problems. This paper reexamines this consensus in the wake of the financial crisis, which has seen policy rates at their effective lower bound for more than two years in the United States and Japan and near zero in many other countries. We conduct our analysis using a set of structural and time series statistical models. We find that the decline in economic activity and interest rates in the United States has generally been well outside forecast confidence bands of many empirical macroeconomic models. In contrast, the decline in inflation has been less surprising. We identify a number of factors that help to account for the degree to which models were surprised by recent events. First, uncertainty about model parameters and latent variables, which were typically ignored in past research, significantly increases the probability of hitting the ZLB. Second, models that are based primarily on the Great Moderation period severely understate the incidence and severity of ZLB events. Third, the propagation mechanisms and shocks embedded in standard DSGE models appear to be insufficient to generate sustained periods of policy being stuck at the ZLB, such as we now observe. We conclude that past estimates of the incidence and effects of the ZLB were too low and suggest a need for a general reexamination of the empirical adequacy of standard models. In addition to this statistical analysis, we show that the ZLB probably had a first-order impact on macroeconomic outcomes in the United States. Finally, we analyze the use of asset purchases as an alternative monetary policy tool when short-term interest rates are constrained by the ZLB, and find that the Federal Reserve’s asset purchases have been effective at mitigating the economic costs of the ZLB. In particular, model simulations indicate that the past and projected expansion of the Federal Reserve's securities holdings since late 2008 will lower the unemployment rate, relative to what it would have been absent the purchases, by 1½ percentage points by 2012. In addition, we find that the asset purchases have probably prevented the U.S. economy from falling into deflation.
And, from the conclusions:
VII. Conclusions and further research The zero lower bound has been an important constraint on monetary policy in many countries over the past several years. The fact that many central banks have encountered the ZLB should not have come as a surprise—previous research using empirical models that were not based primarily on the Great Moderation period did predict that the ZLB would be a relatively frequent constraint on monetary policy in a low inflation environment. What has been a surprise is the magnitude and duration of the constraint imposed by the ZLB in the United States and in some other countries.
Our analysis has identified a number of factors that may have led to an underestimation of the extent to which the ZLB may affect macroeconomic outcomes. Going forward, researchers who seek to assess the probability and effects of hitting the ZLB will need to confront the issues raised by these factors. First, relying on model stochastic simulations that assume constant parameters and variances, and so abstract from data and parameter uncertainty, contributes to an underestimate of the probability of encountering the ZLB. Our results indicate that time-varying parameters, measurement error, and parameter uncertainty can noticeably raise the estimated probability of hitting the zero lower bound, indicating that future research should incorporate these factors in the analysis. Second, researchers need to find ways to ensure that model-generated probability distributions adequately account for relatively rare tail events, even if the data in the model’s estimation sample does not include any such events. This adjustment can be accomplished by using long samples in estimating the shock variances, or by using methods that incorporate a priori on tail events and making the distribution of these events less sensitive to recent data. Finally, our analysis shows that one can obtain quite different answers depending on the model used in the analysis. For example, we find that an estimated DSGE model estimated over the Great Moderation period predicts that it is extremely unlikely that the Federal Reserve could get stuck at the ZLB for a year or longer, while other models that feature stronger intrinsic persistence view such outcomes as much more likely. This range of results indicates that research on the ZLB should explicitly integrate a range of models, including models that allow for structural change.
Based on counterfactual model simulations run using a variety of monetary policy rules, we find that the ZLB has importantly constrained the ability of conventional monetary policy to limit the depth and duration of the current slump. However, our results also suggest that alternative monetary policy instruments, such as asset purchases, have been effective at mitigating the adverse macroeconomic effects of the ZLB. In particular, we find that the asset purchases undertaken by the Federal Reserve over the past two years, plus those currently underway, are roughly equivalent to a 300 basis point reduction in short-term interest rate. Model simulations suggest that the additional stimulus provided by these purchases is keeping the deterioration in labor market conditions from being noticeably worse than it otherwise would be; the asset purchase program may also be keeping the economy from falling into deflation. Future analysis of the effects of the ZLB will need to take account of the potential use of such alternative tools.
Finally, we should note that we have ignored a number of factors in our analysis. First, the effects of the ZLB can depend on the assumption regarding expectations formation. Roberts and Reifschneider (2006) and Williams (2006) find that deviations from rational expectations magnify the effects of the ZLB. Second, in most cases we abstracted from the problems of real-time measurement of data and natural rates that add additional uncertainty to the setting of policy and macroeconomic outcomes. Third, our analysis of ZLB risks in the United States has not taken account of the experience of other countries; future studies should put more weight on the cross-country evidence, which shows a frequent pattern of financial crises followed by deep recessions. And fourth, we did not take into account fiscal policy actions that may step in when the economy is constrained by the ZLB.
The fourth point about fiscal policy points to an extension of the analysis that needs to be conducted. With success defined as reducing the unemployment rate by 1.5% -- far short of what is needed to return unemployment to reasonable levels -- fiscal policy measures have a key role to play when the economy is stuck at the ZLB. Thus, we need to know more about how monetary and fiscal policy can interact and produce the best possible outcome.