Long-term Growth
Testing the breaks

While standard neoclassical models of growth simply assume that aggregate output and output per worker grow at a constant rate and the more recent theoretical literature on `endogenous growth' considers the conditions under which growth rates may rise over time, the corresponding empirical evidence has been mixed. In Discussion Paper No. 965, Research Affiliate Dan Ben-David and David Papell apply recent developments in time-series analysis to investigate whether output grows at a constant rate over extended periods. of time. They use aggregate and per capita data on the annual output of 16 countries over up to 130 years to conduct two tests of the unit root hypothesis. The first allows no break in the trend, and its results indicate that the unit root hypothesis cannot be rejected but provide little evidence of constant growth since 1860. The second allows for a break in each country's trend function, associated with a sharp fall in GDP in most cases, and this modification enables the unit root hypothesis to be rejected for both aggregate and per capita real GDP in most cases. Ben-David and Papell then calculate the steady-state growth rates in each sub-period, which reveal that post-break growth rates were twice the pre-break rates on average for aggregate GDP and two-and-a-half times the pre-break rates for per capita GDP.

This is not sufficient to distinguish the neoclassical and endogenous growth models, which both predict this outcome during the return to the steady-state path. The neoclassical model also predicts that growth rates should revert to their pre-break values once the previous steady-state path is attained, but Ben-David and Papell show, however that faster growth usually continues even after that point has been reached and eventually surpassed. This suggests a possible bridge between Romer-type predictions of increasing growth and Olsonian accounts of the rols of major social upheavals in breaking up distributional coalitions that improves resource allocation and promotes faster growth. Their results also demonstrate the significant relationship between the extent of the fall in the level of GDP and the subsequent relative rise in its post-transition steady-state growth rate.

The Great Wars, the Great Crash, and the Unit Root Hypothesis: Some New Evidence About An Old Stylized Fact
Dan Ben-David and David H Papell

Discussion Paper No. 965, June 1994 (IT)