Feng, Yi. Democracy, Governance, and Economic Performance: Theory and Evidence
Gene Shackman
Feng, Yi. Democracy, Governance, and Economic Performance: Theory and Evidence. Cambridge, MA: MIT Press. xvii + 383 pp. Cloth, $40.00.
In Democracy, Governance, and Economic Performance, Yi Feng investigates the relationship between politics and economic growth. In particular, Feng “studies the direct and indirect effects of political institutions on economic growth” (p. 13). The institutions in question are political freedom, political stability, and policy certainty. Feng’s approach is to study “the general patterns of political regimes and economic growth in a cross-national setting” (p. 12).
Feng has several basic hypotheses. Policy uncertainty is the most important issue in economic growth. According to Feng, policy uncertainty causes problems by lowering investment. Where there is high wealth disparity (his proxy for policy uncertainty), there is less ability to develop policy consensus, which in turn leads to the inability of governments to adopt a consistent long-term growth policy. Under these conditions, people are less likely to invest in any growth-related capital. Second, political instability, or high probability of irregular government change (e.g., revolution), similarly slows economic growth. Where there is high probability of government change, people are less likely to invest in capital. Finally, political freedom has indirect effects on growth through political instability, inflation, investment, education, income distribution, property rights, and population growth. Countries with more freedom may, for example, develop public education systems or reduce income inequality, which will promote growth.
Feng wants to capture the effects of politics on long-term trends in economic growth, rather than short-run factors (e.g., transitional crises, external shocks). Thus, the outcome variable is the long-term average of economic growth, from 1961 to 1995. Feng also mentions that using long-term averages can control for within-country time-series inaccuracies. The predictor variables in his study are the three political variables, and also a variety of control variables such as initial level of GDP and primary school enrollment, investment share of GDP, inflation, and birth rates.
Feng uses cross-national analysis to determine how well these variables can predict long-term economic growth. He finds that free and stable political systems best promote a growth-oriented economic agenda, and that democracy is good for growth by promoting and supporting general education, facilitating demographic transitions, reducing income inequality, and supporting rules that protect private property rights. He also finds that political freedom “Granger causes” economic freedom, but not the reverse. One policy implication to which he points is that political reform may be the only way for improving economic performance for some countries, particularly for those in sub-Saharan Africa that lack other economic growth conditions, such as education, physical capital, income equality, and property rights. Feng also mentions that some countries (e.g., former Soviet Union states) may experience temporary difficulties during the transition to democracy, but that the transition will eventually lead to improved growth.
Unfortunately, there are a number of problems with Feng’s research, primarily methodological. For example, his model includes the proposition that “initial GDP per capita will have a negative effect on growth” (p. 71 ). Thus, the regression equations presented in his main analysis, in Chapter 4, include the variable GDP per capita in 1960. However, one of the variables used to calculate his estimate of irregular government change (political instability) is GDP per capita. That is, GDP per capita appears twice in the same equations.
Another problem with the main regression equations is multicollinearity. Feng presents a table (p. 75) showing the correlations among the variables in the growth equations. Several of the correlations are above 0.7, a typical indicator of multicollinearity. (1) Feng specifically writes that he will examine the intercorrelations for multicollinearity, but, aside from presenting the table of correlations, he does not discuss the problem nor does he indicate how he deals with it. The highly intercorrelated variables (birth per capita, GDP per capita, and elementary school enrollment) are all in most of his regression equations, so a discussion is important, but, as mentioned, missing.
Another problem centers around the time period Feng uses for economic growth (1961 to 1995). Feng writes, “The focus of this work is the secular trend of economic growth over a fairly long period of time, rather than dynamic change, transitional crises or external shocks” (p. 66). Yet, he also writes, “The included period is important, as it spans the 1960s and 1970s … and the late 1980s and early 1990s, which witnessed the dismantling of authoritarian regimes and their controlled economies” (p. 70). Surely the collapse of the Soviet system was one of the largest transitional crises in recent decades. That is, Feng specifically writes that he uses this time period because it includes a period of substantial and sudden economic transition, but he also writes that he does not want to study transitional crises. Similarly, the 1970s included a large external shock of the rapid oil price increases. Feng also mentions that he did not include the late 1990s because of the dot.com bubble and recession. Dynamic changes, crises, or shocks can be found in any period, so there seems to be no clear reason to avoid any particular time period, if the focus is on long-term growth.
There is also at least one problem with Feng’s interpretation of research. For example, Feng writes, “In their empirical study, Alesina, Ozler, Roubini, and Swagel (1996) unequivocally find that causality runs from political instability to economic performance” (p. 68). However, Alberto Alesina, Nathaniel Ropes Professor of Political Economy at Harvard University, writes that causality is not so clear, and that Feng’s statement is not quite right. (2) As a sociologist, this reviewer is well aware of the difficulties of establishing “unequivocal” results, and Feng should be as well.
Many of the problems in this work throw question onto some of his results–for example, that political freedom “Granger causes” economic freedom, but not the reverse. Despite the many problems, this book may be useful as supplemental reading in introducing students to some of the issues about economic growth.
(1) Albert Alesina, personal communication with reviewer, December 2003.
(2) Susan Carol Losh, “Guide 8: More technical aspects of regression.” Online course notes, Introductory Statistics: Description and Inference (2002), 01.sp02.fsu.edu/Guide8.html (accessed January 3, 2004); Alberto Alesina, Sule Ozler, Nouriel Roubini, and Philip Swagel, “Political Instability and Economic Growth,” Journal of Economic Growth 1 (June 1996):189-211.
Gene Shackman, Ph.D.
The Global Social Change Research Project
Albany, New York
COPYRIGHT 2004 Pi Gamma Mu
COPYRIGHT 2005 Gale Group