[Paper Review] International evidence on business cycle magnitude dependence

Corrado Di Guilmi, Edoardo Gaffeo+2 more|ArXiv.org|Jan 26, 2004

Monetary Policy and Economic Impact15 references3 citations

TL;DR

This paper investigates whether business cycle turning points—expansions and recessions—are more likely to end as their magnitude increases, using parametric hazard models on 16 countries from 1881 to 2000. It finds positive magnitude dependence in recessions (longer recessions are more likely to end), while expansions show no magnitude dependence overall, though post-WWII expansions exhibit negative magnitude dependence, indicating they become less likely to end as they grow larger.

ABSTRACT

Are expansions and recessions more likely to end as their magnitude increases? In this paper we apply parametric hazard models to investigate this issue in a sample of 16 countries from 1881 to 2000. For the total sample we find evidence of positive magnitude dependence for recessions, while for expansions we are not able to reject the null of magnitude independence. This last result is likely due to a structural change in the mechanism guiding expansions before and after the second World War. In particular, upturns show negative magnitude dependence in the post-World War II sub-sample, meaning that in this period expansions become less likely to end as their magnitude increases.

Motivation & Objective

• To assess whether the likelihood of business cycle turning points increases with cycle magnitude.
• To investigate whether expansions and recessions exhibit magnitude dependence in their duration-to-end dynamics.
• To identify structural shifts in business cycle behavior, particularly around World War II.
• To examine whether the mechanism governing cycle duration differs before and after WWII.
• To provide international evidence on the dependence of cycle duration on magnitude using a consistent econometric framework.

Proposed method

• Parametric hazard models are applied to estimate the hazard rate of cycle turning points as a function of cycle duration.
• The hazard rate is modeled as a function of cumulative duration, allowing for time-varying failure intensity.
• The model distinguishes between expansions and recessions, analyzing each separately.
• The sample is split into pre- and post-World War II sub-samples to detect structural changes in cycle dynamics.
• The analysis uses annual data on real GDP for 16 countries spanning 1881 to 2000.
• Statistical inference is conducted using likelihood ratio tests and confidence intervals to assess magnitude dependence.

Experimental results

Research questions

• RQ1 Do longer recessions have a higher probability of ending compared to shorter ones?
• RQ2 Is there evidence of magnitude dependence in expansions, such that longer expansions are more likely to end?
• RQ3 Does the relationship between cycle duration and turning point likelihood change after World War II?
• RQ4 Are expansions in the post-WWII period characterized by decreasing hazard rates as they grow longer?
• RQ5 What structural shifts in business cycle dynamics can be identified around the mid-20th century?

Key findings

• Recessions exhibit positive magnitude dependence: the longer a recession lasts, the more likely it is to end.
• Overall expansions show no significant magnitude dependence, failing to reject the null of magnitude independence.
• Post-World War II expansions display negative magnitude dependence, indicating that longer expansions are less likely to end.
• The structural break around WWII explains the apparent absence of magnitude dependence in the full sample.
• The hazard rate for expansions decreases with duration in the post-WWII period, suggesting increasing resilience or stabilization mechanisms.
• The results are robust across different model specifications and country samples, supporting the existence of time-varying dynamics in business cycle behavior.