Comments

John R. Logan: This study takes a substantial step toward including the full range of cities in research on urban China. This is a field where our knowledge is mainly limited to studies of major coastal cities such as Beijing and Shanghai, which have been the destination of the largest waves of migration and foreign and domestic investment in the last twenty years. Despite the dramatic size and expansion of such growth poles, China remains a relatively underurbanized country, and a large share of urban residents live in much more modest settlements. What are the trends in these places, and what are their prospects? Henderson shows that there are reasons to believe that medium-size cities may be even more productive than larger cities, and he suggests a shift in public policy to encourage their development.

I focus here on two questions. The first is how to study China, which requires sensitivity to the processes of population redistribution and the limitations of available data. The second is what can be learned more generally by studying Chinese urbanization.

How to Study China

As Henderson makes clear, "urban" is an administrative category in China. With regard to individual residents, the state distinguishes between people who have a rural household registration (hukou) and those with an urban registration. The majority of migrants are unable to change their registration status, and therefore find themselves living in cities or on the outskirts of cities without full citizenship rights enjoyed by other residents (including participation in the housing market and benefit of public services such as education and health care). With regard to places, the state has established a hierarchy in which those with higher status have greater authority and stronger claims on resources. The highest ranked are the four cities (Beijing, Shanghai, Tianjin, and Chongqing) that have the same political standing as provinces. Provincial capitals and prefecture-level cities are next in rank. The medium-size cities studied by Henderson are county-level cities, political entities that may be under the jurisdiction of larger prefecture-level cities, on the one hand, but control extensive rural territories within their boundaries, on the other. Towns and villages are smaller and have even less administrative autonomy.

The processes of urbanization involve redistribution among all of these components. The majority of migrants stay within their province of origin, and perhaps the largest number are moving just one or two levels in the urban hierarchy—from village to town, town to county-level city, and so on. At the same time, there is considerable urbanization of the countryside in two respects. First, one or more members of rural households may find a nonagricultural job in a nearby place, perhaps commuting a considerable distance to work. Without changing place of residence or hukou status, the person may be understood to have become urban in a significant sense. Second, the accumulation of small migrations over time causes places to move up the administrative hierarchy. The central government continually recategorizes some villages as towns, some towns as county-level cities, and so forth. In fact, considerable effort is required for social scientists to track urbanization in China, because national-level measures need to be recalibrated to adjust for these changes.

These conditions have implications for the interpretation of analyses of county-level cities. The universe of county-level cities is continually changing. This is why Henderson's OLS regression for county cities includes 361 cities, but the regressions with instrumental variables, which require data for the cities at an earlier time point, include only 183 cities. Half of his cases in 2001 simply did not exist as county-level cities in 1990. In addition, there are cities that were county level in 1990 but had advanced to prefecture status by 2001. Systematic bias could be introduced into the results if one limits attention to cities that did not change their administrative status in the previous interval. After all, these are the cases that did not grow as much or did...