How much should we trust five-year averaging to purge business cycle effects? A reassessment of the finance-growth and capital accumulation-unemployment nexus

Averaging data over five years is the canonical approach to purge business cycle effects in cross-country/time-series studies. This neglects that business cycles have been found to last 7–10 years on average, with substantial heterogeneity within and between countries. We assess if five-year averaging leads to insufficiently purged cyclical effects and biased estimates by replicating studies that, based on five-year averaging, find growth-enhancing effects of finance for a panel of 130 countries, and unemployment-reducing effects of capital accumulation for a panel of 20 OECD countries. To test if five–year averaged data is purged from cyclical effects we add output gap measures as controls, which are highly significant and strongly narrow the effects of the key explaining variables, often resulting in insignificant estimates. Too much finance is robustly found to harm growth. Easily implementable alternatives to five–year averaging are discussed.