Re-estimating growth and convergence for developed economies: 1870--1990

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)

First Committee Member

John Devereux - Committee Chair


This study is an assessment of the accuracy of long span comparative estimates of per capita Gross Domestic Product (GDP). These estimates are based on backward projections from a single GDP benchmark using domestic growth rates. An alternative is to construct multiple GDP benchmarks based on direct price comparisons. This study presents new benchmark estimates of relative GDP per capita for eleven developed economies for the period 1870 to 1990. A comparison of the benchmark GDP estimates with the long span GDP estimates within each benchmark year serves as a test of the reliability of the long span estimates.There are significant discrepancies between the long span and benchmark GDP estimates. Reliance on the long span GDP estimates is found to result in the mis-measurement of the forces of catch up and convergence across economies. For example, the long span estimates find convergence in the pre world war one era but no evidence of catch up. In contrast, the benchmark GDP estimates show significant convergence and catch up during this period.The long span GDP estimates also lead to the mis-measurement of growth for individual economies. The long span and benchmark GDP estimates show radically different income paths for the UK relative to the US over the period 1870 to 1990. The long span estimates describe a gradual, eighty-year British relative decline, from 1870 to 1950. The benchmark GDP estimates, however, show that the UK kept pace with the US from 1870 to 1930. Moreover, the UK relative decline is concentrated in just twenty years, from 1930 to 1950. Finally, the post-war resurgence of the UK economy is stronger than is currently estimated.The long span GDP estimates therefore provide flawed measures of relative GDP per capita. Yet they are widely used in empirical macroeconomics, economic growth and economic history. The findings in this study suggest that this data should be used with caution. In particular, the data may be unable to support the sophisticated techniques to which it is currently subjected.


Economics, General; Economics, History

Link to Full Text


Link to Full Text