Contrary to the common approach in the literature, the economic and other forces that push countries toward democratization are continuous rather than discrete. This paper argues that failure to account for the latent variable of "incipient democracy" can bias estimates of democracy's determinants. The paper presents a new avenue by which economic integration can foster democracy, one that focuses on the means for democratization rather than the motive. This strengthening of civil society is identified as a necessary component of economic integration with modern distributed production, though we would not expect to see it in autocracies dependent on natural resource trade. The arguments are applied to the case of China.
The question of how and when countries democratize has been a vital one for both academics and policymakers. Different academic disciplines have taken their own distinctive approaches, but a leading one in economics and political science has been to compare countries at different stages of democratization and search for the factors that correlate with progress. This type of approach rests on the assumption that progress toward democratization is readily observable and comparable across countries.
This paper focuses on the earliest stages of a move to democracy and argues that progress may not be observable. Instead, progress toward democracy operates as a latent variable. This approach can serve to organize thinking about democratic progress, highlight the effects of economic integration on democratic development. These benefits of integration are distinct from the hypotheses prevalent in the literature, since they focus more on the means by which pressures for change are translated into actual change, rather than just the existence of the pressures themselves.
If incipient democracy builds as an unobserved, continuous, latent variable, this casts doubt on the econometric validity of the pervasive cross-country regressions that have been used to identify causal forces in democratization. By failing to account for the continuous nature of the underlying variable, conventional regression estimators are biased. This paper thus adds to the growing list of econometric critiques of the crosssectional regression approach to explaining democracy. . . .
Click here to view the full document as an Adobe Acrobat PDF.
Philip I. Levy is a resident scholar at AEI.