RDP 2011-04: Assessing Some Models of the Impact of Financial Stress upon Business Cycles 8. Conclusion

Models that incorporate financial features pertaining to credit and debt are increasingly appearing in the macroeconomic literature. To date much of the assessment of the success of this augmentation of traditional models has involved recording the magnitude of impulse responses to various financial shocks and whether the shocks can give a better account of the variation in output over time. Our paper provides a complementary approach, namely whether the augmented models provide a better explanation of the business or growth cycles and whether they can replicate some stylised facts about the relationship between recessions and credit. To demonstrate this approach we took two models representative of common ways of introducing financial factors into macroeconomic models – those of Gilchrist et al (2009) and Iacoviello (2005). While financial factors can play a role in particular cycles, generally it seems that the average cycle characteristics of these models are not affected much by their introduction. The Gilchrist et al model managed to replicate some of the stylised facts but failed to do so for others, for example, that credit crises produce long duration recessions. This points to the need to either add extra features or perhaps combine existing ones. Finally, successful prediction of recessions ultimately involves an ability to predict the signs of future output growth rates. To gauge whether the models examined here can predict recessions, we ask how important to current output growth rates are the component of current shocks that are unpredictable using past information. We find they are very important. Consequently, the models imply that future growth rates in output are largely dependent on future shocks. Since these are unpredictable using current information this severely limits the predictive ability of the two models.