Institutions and the Political-Economy of Economic Growth: A Panel Data and Cross Section Application to Institutional Change and Economic Transition in the Post War Period

Publikation: Working paperForskning


This research revisits one of the oldest questions in the study of economic growth. How institutions (defined as rules designed by humans to govern behavior in their social space including the economic and political sphere) affect economic growth. This question has often been investigated in ad hoc ways by using a simple dummy approach. In the paper is explored the efficiency channel which has been ignored in past research both in the political and economic literature on institutions and growth. A very simple model is build for application purposes alone (low theory) where the objective is to show that a democracy is the only viable set-up to achieve long run economic growth.

In the empirical parts of the paper is combined a panel data and cross section approach to investigate the influence of the following institutional typologies under a general scenario of transition or institutional change: market-led autocracies, state-led autocracies, countries in transition, oligarchies and democracies. The results support somewhat in opposition to prior research using this particular typology of institutions that democracies and market-led autocracies are almost equally good. However, the low number of observations available for countries that are de facto market-led autocracies strongly reduces the validity of this result both for the panel data analysis (19 observations) and in particular for the cross section analysis (2 observations). Furthermore, the very worst dictatorships may not be in the sample because we cannot sample on them. Only democracy is viable from a long run perspective and in particular securing the motivation of human capital to invest effort and thereby bring about economic efficiency.

These results are confirmed by using the traditional cross section specification in Mankiw, Romer and Weil (1992) and it is shown that in this investigative framework the dummy approach is appropriate.

Therefore the main difference between the present results and past results in the study of democracy with the PolityIV dataset lies in the combined typology more than in the panel data application. From an overall methodological viewpoint it is concluded the results are extremely robust.
Antal sider44
StatusUdgivet - 23 nov. 2012
Udgivet eksterntJa

Citer dette