The Spatial Distribution of Population in 48 World Cities: Implications for Economies in Transition By Alain Bertaud∗ and Stephen Malpezzi† December 17, 2003 Comments Welcome *166 Forest Road Glen Rock, NJ 07452-3017 USA [email protected] http://alain-bertaud.com †The Center for Urban Land Economics Research The University of Wisconsin 975 University Avenue Madison, WI 53706-1323 USA [email protected] http://www.bus.wisc.edu/realestate 2 Alain Bertaud, until recently Principal Urban Planner at the World Bank, is an independent consultant. Stephen Malpezzi is Professor, and Wangard Faculty Scholar, at the Department of Real Estate and Urban Land Economics, and an associate member of the Department of Urban and Regional Planning, of the University of Wisconsin-Madison. Comments and criticisms are particularly welcome. We have received useful comments from numerous colleagues, including Robert Buckley, Paul Chesire, Peter Colwell, Peter Englund, Roy Gilbert, Daniela Gressani, Gregory Ingram, Christine Kessides, Kyung-Hwan Kim, Somik Lall, Bertrand Renaud, Margret Thalwitz, Bengt Turner, Lee Travers, Anthony Yezer, and participants in a seminar presented to the University of Uppsala’s Center for Housing Research in Gävle, Sweden. Yongping Liang drafted the second appendix. These colleagues bear no responsibility for remaining shortcomings. This paper has been supported by the World Bank. Malpezzi's work on this paper has also been supported by the UW's World Affairs and the Global Economy (WAGE) Initiative, and the Center for Urban Land Economics Research. Opinions in this paper are those of the authors, and do not reflect the views of any institution. Keywords: URBAN DECENTRALIZATION, POPULATION DENSITY, SPRAWL, LAND USE, HOUSING, TRANSITION ECONOMIES 3 4 Executive Summary This paper is one of the first attempts to examine urban form systematically around the world. While we have drawn a wide sample, comprising cities from all regions and all levels of development, in this paper we make special reference cities in “transition economies” (mainly in the World Bank’s ECA Region, comprising Central and Eastern Europe, and some of the countries of the Former Soviet Union). By “systematically,” we mean this is the first paper that examines a range of measures of urban form constructed in a largely consistent manner from largely consistent data. Urban economists have studied the spatial distribution of population intensively for the past 40 years, drawing on a longer literature in geography and other social sciences. We present data on selected measures of urban form for about 50, mostly large, metropolitan areas. While we discuss policy issues in light of the special concern of the ECA region, our cities are drawn from all major regions, from rich countries and poor, from market economies and economies in transition to market. Empirical Findings In this paper we have calculated, on a consistent basis, population density gradients for almost 50 large cities in some twenty countries. We have also constructed an alternative measure of city population dispersion. In a variety of second-stage analyses, we have examined several potential determinants of urban form. Our focus has been on income, population, and the nature of the regulatory regime. To a lesser extent, we have examined the role of natural constraint (physical geography), and transport mode. The first important finding is that in many cities – perhaps a surprisingly large number, to some – the negative exponential density gradient implied by the standard urban model fits the data quite well. On the other hand, in a number of cities, population density departs a lot from the standard model. A few cities that depart substantially are cities, like Seoul, that are usually characterized as market economies but have extraordinary regulatory environments for land use; some are centrally planned cities, like Moscow or Brasilia; and others are cities that developed under apartheid, like Capetown. In a second stage model we find that density gradients flatten with income, with city population, and with falling transportation costs, as the standard urban model predicts. We also find that cities with extremely repressive urban regulations, as in South Africa, Korea and Russia, have flatter (sometimes inverted) population density gradients. However, improving our measurement of the urban policy environment remains an important subject for future research. In several respects, our results confirm the findings from the two-decade earlier study of Mills and Tan (1980). The negative exponential function is a useful framework for studying urban form (although we have also analyzed a competing measure of dispersion). Like Mills and Tan, but based on more data, we find persistent and strong patterns of decentralization as incomes rise and cities grow in population. However, we have evidence that many other things affect density, as Mills and Tan suggest; in this paper, we have shown that the effect of the 5 regulatory regime of the city is profound. We have also demonstrated that natural constraint matters, but so far to a lesser extent than man-made constraints. Socialist (transition) cities are more likely to be stringently regulated, or most often to be in the ‘mixed case’ where a stringently planned, non-market urban form has overlaid an existing market-driven city. However, once mode of regulation is controlled for, there is no identifiable residual effect of socialism. Regulation is the transmission mechanism. Changing regulations will, over a very long period, change the form of the city. The regulatory environment, and the existence of natural constraint, are the most consistent systematic determinant of our dispersion index. Income and population play a role, but much less than in density or in the gradient. Policy Implications What are the implications of these preliminary results for those concerned with urban form, or more directly, with city efficiency in transition economies? We first document that socialist cities are more likely to have stringent regulations, or “mixed cases,” where a period of heavily regulated socialist development overlay an earlier period of market-oriented development. Moscow is the notable exception to this pattern, partly because of its longer history under socialism. Moscow is important on its own terms, but it is not necessarily an accurate representation of a “typical” socialist city. Furthermore, some of the most distorted urban forms are not socialist cities, but rather cities that developed under apartheid or that were built de novo from idealized plans, often as capitals, such as Brasilia. (Some Soviet cities developed for special purposes, such as Magnitogorsk or Novosibirsk were built de novo, but we have not as yet had the opportunity to analyze them.) However, while it is good news that Moscow is probably an extreme case, not the prototype, of the socialist city, we can infer that the distortions in urban form caused by the mixed regulatory environment are still significant. Based on our own research, and on broadly related research such as that discussed in World Bank (1993 c), Renaud (1995 a, b) Buckley and Mini (2000) and Buckley, Ellis and Hamilton (2001), we discuss some of the more specific operational issues that arise from these distortions: -- The need to consider location in rent reforms. In the past, rents for socialist units were rarely related to location. In many policy discussions of rent reform, the focus is often on raising rents, without necessarily focusing on locational differentials. With respect to countries still working through a stock of social housing, given the excess supply of units on the fringe of the city, it may not be easy or even fair to substantially raise rents for poorly located units, such as Moscow’s large peripheral estates. -- Conversion of industrial land to other uses. Another policy implication is that in cities like Moscow and Cracow, significant industrial parcels will have to be recycled for other uses, primarily residential. Some countries, including Russia, 6 have mooted differential property tax rates for industrial uses. This would be a mistake, as it would retard this necessary redevelopment. However, redevelopment of large contiguous parcels of industrial land, as often found in ECA cities, can be difficult and expensive. Former industrial land is often heavily polluted and requires a cleaning of topsoil and other remediation before conversion. In addition, even in centrally located areas additional primary infrastructure has to be built to allow residential and commercial development. -- The rehabilitation or progressive elimination of poorly located “panel” housing. First, in many socialist cities, though most spectacularly in Moscow, a great deal of past investment has been in housing that is worth very little, not only because of poor design and maintenance and incentives, but location. In many cities of Central and Eastern Europe market rents in “panel housing” is below maintenance cost. This creates a real conundrum for city managers. On the one hand, investing large sums of money to upgrade this existing stock would be risky, at best, given their location. On the other hand, at least in some cities these units are such a large proportion of the stock that it will take some years for markets to build sufficient replacement stock to house the population. The best policy is probably to undertake some “satisficing” maintenance, while accelerating the development of real estate markets that can fill the gap. This would include, but not be limited to, regulatory and planning environments conducive to market-driven infill as well as some greenfield development. -- Interaction with demographic trends. Most ECA cities face falling populations over the next decades. In spite of the falling population there is still pressure on housing because household size is also falling, and many apartments in downtown areas have been transformed into offices or even torn down (as in Warsaw) to make room for office and commercial space. As incomes increase we can predict that the desire to consume more floor space per person will also fuel demand for more housing. Many small apartments (outside panel housing) are being consolidated into larger ones, thus diminishing the number of units. We have therefore a paradox of a concurrent falling population and a housing shortage. -- Using regulations to “reshape” socialist cities. Transition economy planners should strive to avoid the twin problems discussed by Mills (2002 a, b): regulating to increase density, sometimes at the same time regulating to reduce density, without a clear idea of the costs and benefits of such regulation. Rather than micromanage density, whether by regulations preventing the development of single family housing in peripheral areas, or minimum plot size and unrealistic floor area ratios, planners should strive to broadly follow the market (Wallace 1988), regulating with a light touch when needed to deal with concrete and identifiable external costs. -- The need to monitor urban spatial structures, and the regulatory environment. Mapping land uses, corresponding densities, housing prices and changes in prices 7 are among the ways planners should monitor their cities. With this baseline data, planners would move on to studying whether permitted land uses, floor area ratios and the like are broadly appropriate. Too many planning exercises, like the development of many master plans, are static one-off exercises. Planners need to view plans and regulations as dynamic, just as the city is dynamic. In many ECA cities, planners consider that the monitoring of land prices and rents is not relevant to their job. In fact, because ECA cities lack interest group feed backs as mentioned above, monitoring the evolution of real estate prices and rents is the most useful tool to guide future land legislation and primary infrastructure investment. In some ways, the most important prescription for planners in transition economies is, in a version of the Hippocratic oath, “to do no harm.” As planning and regulatory systems continue to develop along with the market, policymakers in transition economies need to let the substantive requirements of their cities guide their design of regulations, rather than blindly copy forms and institutions from other countries. Otherwise-different experiences in cities as diverse as Brasilia and Seoul demonstrate that inappropriate planning and regulation can exist in otherwise market-oriented economies, but at great cost in suboptimal land use, increased commuting costs, and a weakened public fisc. As the Korea case points out (Kim 2001), it can be very difficult to unwind a distorted planning system once in place. This paper highlights the centrality of regulation in determining city form and economic efficiency. However while the paper – and associated literature – has made some progress, further improvement is necessary. There is still some work to be done on measuring form, but it is even more apparent we still lack a completely satisfactory metric for municipal regulation. A completely satisfactory metric would have to be (1) conceptually well-founded, (2) empirically implementable at reasonable cost, and (3) eventually widely calculated for a range of cities to facilitate useful comparisons. Such a measure would provide a quick and cheap tool to assess when perverse regulatory behavior is damaging urban development. The World Bank Group is well placed to make such an effort, given the focus on institutions and “good governance” within the Bank. Of course, the regulatory environment is a broad concept. One initial focus on measuring the urban policy environment should surely be land use regulations. Land use regulations can be a major problem in ECA cities, because they often contribute to the weakening of old urban centers and push “modern” development across municipal boundaries. The costs of “auditing” these regulations is low relative to potential benefits, especially now that most regulatory maps are in digital form. Such an analysis can be included as part of a City Development Strategy exercise. A debate on the issue of “urban shape” is often well-received in ECA countries, especially when clients are offered illuminating comparisons to other European cities. Future Research We have found that our most basic results are very robust to changes in model and sample; the numbers may change slightly, but qualitative results will almost surely remain robust. Still, we 8 have some thoughts on a longer run research agenda. There are of course potential gains simply from increasing the size of the database and the range of cities we study. Other large gains could come from continued efforts to systematize and coordinate data collection on cities, including but not limited to the kinds of data we examine here. The Housing and Urban Development Indicators Project, initiated at the Bank and UNCHS (Habitat), and now mainly carried out under UN auspices, could be one possible avenue for such an effort. Many individual efforts, like the Indicators Project, Bank-sponsored data collection on urban form by Newman and colleagues, and our own effort, could make a greater contribution with stronger design, including but not limited to (a) more overlap among sampled cities, (b) wider geographic and country profile scope (including more ECA countries), and (c) less of a laundry-list approach and more focus on improving and collecting key indicators of economic and environmental outputs, and measures of the policy environment. 9 Introduction This paper is the first attempt to examine urban form systematically around the world, with special reference to “transition economies” (mainly in Central and Eastern Europe, and some of the countries of the Former Soviet Union). By “systematically,” we mean that, while we draw on a large literature outlined below, this is the first paper that examines a range of measures of urban form constructed in a largely consistent manner from largely consistent data. Urban economists have studied the spatial distribution of population intensively since the pioneering work of Alonso (1964), Muth (1969) and Mills (1972). Of course this work has a longer history, traceable at least back to von Thunen (1826), including studies by other social scientists such as Burgess (1925), Hoyt (1959) and Clark (1951). In addition to population distribution, there is a related empirical literature on the distribution of real estate prices (e.g. Follain and Malpezzi 1981) and the distribution of wages and incomes over space (Eberts 1981, Madden 1985). Much of this literature is ably surveyed by McDonald (1989). Broad reviews of the theoretical models behind this empirical work can be found in, for example, Wheaton (1979) Straszheim (1987) and Arnott et al. (1998). Beginning with Colin Clark, there have been many hundreds of published studies of population density patterns for cities in developed countries. Despite the size of this literature, only a relative handful of studies have been published for developing and/or transition economies. Notable exceptions include the World Bank’s City Study of Bogota and Cali (Ingram and Carroll 1981; Mohan 1994; Ingram 1998); Asabere’s studies of Accra; several studies of Korea, such as Follain, Renaud and Lim (1979), Mills and Song (1979) and Son and Kim (1998); the path-breaking comparative study of Mills and Tan; and a series of studies by Alain Bertaud, mostly unpublished 1