By Simon Sturn and Gerald Epstein
A large body of cross-country time-series literature shows that financial development—predominantly measured by private credit as a percent of GDP—fuels growth. But, in light of the many recent episodes of finance driven crisis, these results seem curious. Haven’t we seen that periods of rapid credit expansion are also often periods of economic crisis?
The answer to this puzzle might have to do with the time horizon under consideration. A broad theoretical literature argues that credit demand and supply are correlated with growth in the short-run. Credit demand is “pro-cyclical”—firms are reluctant to borrow and invest during business-cycle slumps, periods of low demand and high uncertainty, while the opposite is true for business-cycle booms. Credit supply is also pro-cyclical, as banks are less willing to lend during recessions, when banks have less capital and borrowers have lower net worth, than during upturns.
Finance and growth, then, are correlated in the short run, but this does not imply that finance also causes long-run growth. Therefore, it is crucial to address the short-run pro-cyclical fluctuations of credit in empirical studies on the impact of finance on growth. Otherwise, the true long-run growth effect of financial development will be overstated.
How long is the “long-run” in the empirical literature?
The standard literature on finance and (long-run) growth does not entirely ignore this possible problem, but as we show in our paper these approaches are not up to the task.
The “by-now-standard approach” to deal with short-run fluctuations in the empirical cross-country time-series literature is to transform annual data into five-year non-overlapping periods, which allows one “to focus on long-run economic growth.”
But it is doubtful that five year averaging should purge short-run fluctuations from the data. In fact, according to NBER’s Business Cycle Dating Committee, the average business cycle in the United States from 1960 to 2009 lasted about 6½ years, with the shortest cycle about 2 years and the longest around 11 years. The Euro Area Business Cycle Dating Committee finds an average length of the business cycle of 9½ years since the mid-1970s, with a minimum of nearly 4 years and a maximum of more than 1½ decades. Also, the output gap measures constructed by the OECD and IMF for several rich countries show business cycles between 2 and 15 years of length. Averaging the data over five-year periods is therefore unlikely to smooth out cyclical variations in growth.
Purging business-cycle effects
We address this methodological shortcoming and re-estimate the impact of finance on growth in a dynamic panel data regression set-up for 130 countries over the period 1965 to 2009. We explicitly purge business cycles by including different measures of the output gap, which is the difference between current output and output growing at its trend growth rate.
This is what we find: First, once short-run fluctuations are explicitly purged, the impact of finance on growth is consistently and significantly smaller. This suggests that much of the previous literature following the “by-now-standard approach” overstates the true effect of finance on long-run growth. Their results are contaminated by short-run fluctuations of growth and credit over the business cycle.
Second, we confirm the finding of previous literature that the effects of finance on economic growth become much lower once observations for more recent years are included in the sample. For example, we find that finance boosts economic growth significantly if the sample is estimated until 1989, but not when estimated until 1999 or 2009. This suggests that something in the finance-growth nexus has changed. Perhaps this is the creation of more destructive and predatory forms of finance in recent decades?
Third, our results are consistent with those of several recent papers which argue that financial development measured as private credit increases growth initially, but its effect diminishes and even turns negative at a level of about 90 percent of GDP, which is reached by many developed countries.
In sum, the conventional wisdom among economists that finance boosts growth may mainly reflect results from the pre-financial-markets-liberalization area, combined with a mistaken statistical approach. In fact, our results confirm that the positive finance-growth relationship started to vanish decades ago, not only with the recent crisis.
Triple Crisis welcomes your comments. Please share your thoughts below.