Are prior restrictions on factor shares appropriate in growth accounting estimations?

Several studies make different prior assumptions on the magnitude of factor shares and scale of production when accounting for economic growth. The initial Solow estimations for instance assumed a capital share of 0.3 and constant returns to scale. Most authors have subsequently used the same restrictions just because they were used in previous studies even when production in the countries under study may not necessarily be taking place under constant returns to scale and capital share may be a value not any close to 0.3. This is likely to distort growth accounting estimation results. This study investigates whether these prior restrictions on factor shares and scale of production as commonly used in the literature are appropriate and whether the Solow assumptions are likely biased even in developed countries. Using Kenyan data and structural vector autoregressions, the main findings are; first, in all cases of the unrestricted estimations, the share of physical capital is less than 0.16. An estimation which imposes 0.3 as the share of physical capital in this case would therefore not be in line with the data generating process leading to biased results. Secondly, in all cases, the explanatory power of the model decreases when the restrictions are imposed implying that the restrictions are not appropriate in growth accounting exercises. The paper further disputes the “styled fact” in growth studies that the Solow assumptions may be relevant for developed countries but not developing countries and concludes that the Solow assumptions may be biased even in developed countries.