Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
China Economic Review 22 (2011) 172–181 Contents lists available at ScienceDirect China Economic Review The incentive role of creating “cities” in China Lixing LI ⁎ China Center for Economic Research, National School of Development, Peking University, Beijing 100871, China a r t i c l e i n f o Article history: Received 12 April 2010 Received in revised form 5 December 2010 Accepted 5 December 2010 Available online 24 December 2010 JEL classification: H11 R11 P26 H77 O40 a b s t r a c t China operated an urbanization policy by which counties could be given city status between 1983 and 1997. The policy had substantial impacts on the new administrative entities, including more discretionary power and fiscal independence. Such “county-to-city upgrading” provided the central government with an instrument to reward localities. Using a large dataset covering all counties during 1993–1997, I show that upgrading is not an automatic procedure that endorses the high urbanization levels in existing counties. Although official guidelines for upgrading counties to cities were published, these requirements were largely ignored in practice. Instead, economic growth rate was the key factor in determining which counties obtained city status. This paper interprets the creation of county-level cities through upgrading as part of the incentive structure of Chinese local officials. The importance of both fiscal and political incentives facing the local government in promoting economic growth is highlighted. © 2010 Elsevier Inc. All rights reserved. Keywords: Incentive mechanism Economic growth Regional competition Urbanization China 1. Introduction In developing countries, local governments play a key role in economic growth. For example, in China, the local government not only provides public goods and services to its residents but is also deeply involved in economic activities. In most regions, building infrastructures, offering inexpensive land and preferential tax policies are the main jobs of the local officials, whose goal is to attract investments. Although not allowed to directly engage in borrowing, local governments have established companies to borrow from banks and invest in urban infrastructures. The effort of the local government and its attitudes toward the market can easily become the binding constraint for regional development. Thus, a major concern of the central government in developing countries is to give local governments the right incentive to spur market development and economic growth (World Bank, 2008).1 Under a hierarchical political system in which local officials do not face election pressure from citizens, the central government must rely on its administrative authority to create effective incentive mechanisms (Zhuravskaya, 2007). For example, fiscal decentralization and the promotion of local officials are two important institutions that provide market incentives to local governments in China (Huang, 2002; Jin, Qian, & Weingast, 2005; Li & Zhou, 2005; Xu, in press). ⁎ Tel.: + 86 10 62759485; fax: + 86 10 62751474. E-mail address: [email protected]. 1 The Growth Report (page 4) by the World Bank states that “the administration must also attract and retain talented people, by offering better pay, promotions, and recognition to officials.” 1043-951X/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.chieco.2010.12.003 L. Li / China Economic Review 22 (2011) 172–181 173 The current paper examines a specific incentive mechanism that is related to urbanization policies in developing countries. In China, awarding “city” status to existing counties is the dominant way to create new urban administrative units in the reform era. The so-called county-to-city upgrading is not only an endorsement of high urbanization levels in existing counties but also provides local governments many political and fiscal benefits, thus providing the center with an instrument to reward localities. Furthermore, economic growth rate, which was not written in the formal requirements, plays an important role in determining which counties can be upgraded to cities. Thus, county-to-city upgrading forms part of the incentive structure of Chinese local officials in promoting economic growth. Using data for all counties in China during 1993–1997, I show that the official minimum requirements for county-to-city upgrading are not enforced in practice. Instead, economic performance is the key to obtaining city status. I provide evidence on the relationship between a county's economic growth rate and its probability of obtaining city status after controlling for urbanization levels and other requirements, such as industrial output and fiscal strength. This reflects a principal–agent relationship between the center and the local governments in which the center uses incentives to induce high growth rate and the localities compete to obtain rewards, including city status. The current paper draws extensively on the growing literature on the relationship between urbanization and economic development. In China, urbanization is featured with the emergence of a large number of new cities, which stands in contrast to the typical urban growth process of other countries, for which the main channel is the expansion of existing cities (Anderson & Ge, 2005). Studies on China's urbanization and city creation are not rare (e.g., Au & Henderson, 2006a,b; Chung, 1999; Deng, Huang, Rozelle, & Uchida, 2008), but the political economic driving forces have largely been ignored.2 The present paper makes the first attempt to interpret city creation as an incentive mechanism to promote economic growth. This paper also highlights the coexistence of fiscal decentralization and political centralization in China. This unique institutional setting is attracting growing interest (Blanchard & Shleifer, 2001; Zhang, 2006; Zhuravskaya, 2007). While the competition and commitment role of fiscal decentralization has been discussed extensively (e.g., Jin et al., 2005; Qian & Weingast, 1997), some recent studies argue that the centralized power structure is also crucial in creating yardstick competition among local officials and constraining the practice of local protectionism (Bai, Tao, & Tong, 2008; Bo, 2002; Li & Zhou, 2005). In this paper, upgrading is the result of centralized decision-making and, at the same time, gives local officials more discretion over revenue collection. The current work provides a chance to learn political centralization and fiscal decentralization in the same setting. The remainder of the paper is organized as follows. Section 2 introduces the background of county-to-city upgrading. Section 3 describes the data and presents evidence on the non-enforcement of formal upgrading requirements. Section 4 presents empirical evidence on the incentive role of awarding city status. Section 5 concludes. 2. County-to-city upgrading in China There are two main sub-provincial administrative levels in China, prefecture and county. County-level jurisdictions have approximately three thousand individuals and include counties, county-level cities and urban districts.3 At the end of 2008, there were approximately 1700 counties and 368 county-level cities. In the present paper, “upgrading” refers to the reclassification from a county into a county-level city. During upgrading, the entire county is labeled a “city.” Thus, a city includes not only the relatively urbanized section, but also large expanses of rural areas. The administrative units created through upgrading are different from the typical understanding of the term “city” (Chan, 1997). The creation of cities through upgrading stands in contrast to the experience of most countries, where “city” and “county” fall into different administrative categories. Cities typically only govern a small area where urban population is concentrated. Rather than reclassifying an entire county, a new city is normally created inside a county. Furthermore, the decision is often made through a vote. For example, in the U.S., a new city could be created by adopting a home rule charter (Hennessey, 2008), while in Brazil, new municipalities are established through local voting (Naritomi, Soares, & Assuncao, 2007). At the beginning of China's economic reform in the early 1980s, restricting the size of large cities and promoting small cities and towns was adopted as the major strategy for achieving urbanization (Saich, 2008). Because setting up a new city government incurs great administrative costs, upgrading existing counties into cities became the major form of city creation in China (Liu & Wang, 2000), constituting more than 90% of county-level cities established after 1982. By 1997, nearly 15% of counties had obtained city status. The total number of cities grew quickly from less than 250 in 1982 to more than 650 in 1997 (See Fig. 1). The official rule for regulating county-to-city upgrading first appeared in 1983, when cities had shown the advantage in attracting investment and the demand for city status increased in the coastal provinces. Under some rough requirements proposed by the Ministry of Personnel and Ministry of Civil Affairs, nearly 100 counties obtained city status during 1983–1986. The number of cities continued to rise, and the central government raised the minimum requirements for city status in 1986 and 1993. The 1993 requirements set different standards for counties with different population densities. The three major requirements include the industrialization level, urbanization level, and fiscal strength (see Table 1). Growth rate is not formally written into the documents as a requirement. It is apparent from Table 1 that the standards vary by population density. The entry barriers for counties with lower population density were set lower than those with higher population density. This seems to imply that, according to the central government, setting up cities does not follow the typical notion of urban agglomeration. 2 3 For a general discussion on the political economy of city formation, see Henderson and Becker (2000). Autonomous counties in the minority area are counted as counties. For a detailed description of the Chinese administrative system, see Chung (1999). 174 L. Li / China Economic Review 22 (2011) 172–181 800 700 number 600 500 400 300 200 100 19 1981 1982 1983 1984 19 85 1986 1987 1988 1989 1990 1991 1992 19 93 1994 1995 1996 19 97 1998 2099 2000 2001 20 02 2003 2004 05 0 year county-level cities prefecture-level cities all cities Fig. 1. Number of cities in China. Table 1 Minimum requirements for county-to-city upgrading. Source: “The Report on Adjusting the Criteria for the Designation of New Cities” by the Ministry of Civil Affairs in 1993 (Available in Zhang & Zhao, 1998). Population density (person/km2) N 400 100–400 b100 Percentage of counties in this category 25% 45% 30% Industrialization level 1.5 billion 80% 150 k 30% 60 million 100 1.2 billion 70% 120 k 25% 50 million 80 0.8 billion 60% 100 k 20% 40 million 60 Urbanization level Fiscal strength Industrial output value (yuan) Share of industrial output value in gross output value Urban population (engaged in non-agricultural production) Share of urban population Fiscal revenue (yuan) Per capita fiscal revenue (yuan) As will be shown later in this paper, these formal minimum requirements for upgrading are not enforced in practice. Many counties that do not meet these requirements gained city status nonetheless. These widespread upgrading cases have masked China's true urbanization level (Zhang & Zhao, 1998; Zhou & Ma, 2003). In large part because of these problems, the central government eventually stopped the policy of county-to-city upgrading in 1997, freezing all applications in hand. Since then, no new cities have been established at the county level. Rather, the total number of cities has remained stable.4 County-to-city upgrading has not been resumed until now. To discuss the incentive role of this upgrading policy, this paper will focus on the period before 1997. Although obtaining city status does not change a county's rank in the administrative hierarchy, its government gains much more political power, which is the reason for the term “upgrading.” For example, the party secretaries of many county-level cities are able to enter the standing committee of the prefecture-level party committee and enjoy a deputy-prefecture political rank.5 The administrative authority is also expanded (e.g., in terms of governing local finance, trade, and transportation activities; see Table 2 for details). In Hubei province, there were experiments to place the budget of county-level cities under the direct supervision of the provincial government, thus bypassing the prefecture level.6 Table 2 also lists other benefits associated with city status, among which the most prominent is that cities are granted more quotas to convert farm land into non-farm use. Through land conversion, local governments can obtain a substantial amount of revenue (Lichtenberg & Ding, 2009).7 Moreover, these benefits are not only for the local officials, such as larger government size and higher official rank, but for the entire county as well.8 For example, additional revenues and favorable policies typically result in more urban infrastructures. In sum, localities were eager to obtain city status in the 1990s (Ren & Wang, 1999). At the same time, upgrading is costly to the central government. For example, the expanded government size increases salary expenses that are paid out of the central budget (Wong & Bird, 2008). In many cases, the central government must provide 4 After 1997, some county-level cities were further upgraded into prefecture-level cities, and some counties were transformed into urban districts of prefecture-level cities. The total number of cities did not change (Chan, Henderson, & Tsui, 2008). 5 The political privilege also brings the local government closer to higher-level governments, thus facilitating the supervision from above. If this is counted as the cost of upgrading, such a small cost is overshadowed by the political and fiscal benefits. In fact, we have observed great enthusiasm for city status among counties in the mid 1990s. 6 This is called "line item under province", or Shengji Jihua Danlie. 7 Land revenue belongs exclusively to the local government. In 2009, the total land revenue was 1424 billion RMB yuan, an amount that is equivalent to 1/5 of the budgetary fiscal revenue. 8 For simplicity, this paper ignores the divergence of objectives among different local officials in a county and assumes that they have the common interest of obtaining city status. L. Li / China Economic Review 22 (2011) 172–181 175 Table 2 Benefits of being a city: an incomplete list. Category Benefits Source Tax and fee Cities enjoy a higher urban construction tax (7% compared to 5% for counties); could collect the surcharges levied on the issuing of motorcycle registration. In Liaoning province, cities could get 1 to 2 million additional subsidies each year after upgrading. Cities get more quotas to convert farm-land to non-farm use After achieving the status of “line item under province” (Shengji Jihua Danlie), cities could report directly to the provincial administration to ask for investment project. Cities have more authority on foreign trade and exchange management; gains authority over police recruitment and vehicle administration; could establish the branch of custom and large State-Owned banks; could approve projects with higher cap of investment. Cities could establish more branches of government and have a larger size of government employees. At times, the bureaucratic rank and salary of officials are raised after upgrading. Cities generally carry greater prestige and are more attractive to outside investors. Chung and Lam (2004); Zhang and Zhao (1998) Land-related Favorable policy Administrative power Government size Rank and salary Reputation Zhang (2006); Ping (2006) Liu (2004); Su (2000); Zhang and Zhao (1998) Chung and Lam (2004); Du (1993) Ren and Wang (1999) Liu (2005) Gu (1997); Chung and Lam (2004); Wang, Ji, and Lin (1998) Note: Given the volatility of Chinese policies, the benefits are continuously changing overtime; benefits listed are not necessarily effective during the same period. subsidies to new cities for their infrastructure construction. In addition to these fiscal burdens, inequality could also rise as a result of the upgrading policy. Because counties that are upgraded are generally rich ones, the favorable policies and subsidies for new cities are likely to enlarge regional disparity. Thus, as a way of rewarding local governments, the number of city titles is limited, and the central government faces constraint when making upgrading decisions. 3. Data and descriptive evidence 3.1. Data Systematic data on Chinese counties is rare, especially for the period before 1990. To the best of my knowledge, only one paper has used county-level data to study the incentives of local governments (Guo, 2007). The data used in the current paper is from the yearly series of Sub-Provincial Public Finance Statistics published by the Ministry of Finance. It covers all county-level jurisdictions and includes detailed public finance and basic socio-economic information. I use upgrading to capture the political rewards associated with county-to-city upgrading, which differ from the typical measure of promotion. The specific benefits associated with city status make upgrading a reliable measure of rewards. In contrast, the existing literature that examines political promotion often faces the problem of identifying when the mobility of an official is a true promotion and when it is not.9 Compared to promotion, upgrading also suffers less from the omitted-variable problem caused by unobserved personal relationships. Because the promotion decision of local officials is made by the government at only one level higher, personal relationships surely play an important role. In contrast, because upgrading is costly, the decision is made in a deliberate manner (Wu, 2000). In fact, applications for upgrading are submitted to the Ministry of Civil Affairs through each provincial government, and the State Council is responsible for the approval decision. This strict process is normally beyond the influence of county leaders. Data used in the current analysis range from 1993, the first year of the published data, to 1997, when upgrading stopped. Data of Xizang (Tibet) were dropped because many variables had missing values. A total of 99 upgrading cases were identified through official records from the Ministry of Civil Affairs. Table 3 shows the summary statistics for upgrading and non-upgrading cases. Because GDP data are not available at the county level, Gross Value of Industrial and Agricultural Output (GVIAO) is the best consistent measure of a county's total economic activity. The official requirements on upgrading include industrialization level, urbanization level, and fiscal strength. Measures for these requirements include industrial output value, share of industrial output value in GVIAO, urban population, share of urban population in total population, and per capita fiscal revenue. All of the output and revenue measures were adjusted to 1993 constant prices using the yearly GDP deflator. Except for growth rate, all variables are averaged from 1993 to 1997. As summarized in Rawski and Xiao (2001), there are concerns about the accuracy of China's economic statistics in the literature. Here I address the issue of data quality from three perspectives. First, price reforms in 1993–1994 led to evaluations of economic statistics based on market prices and eliminated the double counting problem. As noted by Au and Henderson (2006b), data at sub-provincial levels “at least up to 1997, are viewed as being extremely high quality” and “are less likely to be manipulated, compared to manipulation at the level of the provinces and centre.” The data used in the current paper are at the county level and mainly cover the period of 1993–1997; thus, data manipulation should be less of a problem. Second, the public 9 Part of the reason is that the actual power and political benefits enjoyed by an official may be hard to discern using only his or her job title under the Chinese political system. For example, when a provincial governor is assigned to be the director of the provincial People's Congress, this seemingly lateral move typically means a loss of power and marks the end of his or her political career. While Li and Zhou (2005) categorize this type of mobility into the same category as demotion, Bo (2002) simply drops these observations from his study. 176 L. Li / China Economic Review 22 (2011) 172–181 Table 3 Summary statistics (by upgrading status, 1993–1997). Non-upgrading Gross value of industrial and agricultural output (GVIAO,10 K yuan) Per capita GVIAO (yuan) Growth rate of GVIAO (1994–1997) Industrial output value Share of industrial output value in GVIAO Population (10 K) Urban population (10 K) Share of urban population Budgetary fiscal revenue (10 K yuan) Fiscal revenue per capita (yuan) Upgrading Obs Mean Obs Mean 8,048 8,048 6,349 8,048 8,046 8,053 8,046 8,046 8,053 8,051 145,563 3,220 0.164 97,858 0.527 41.7 4.93 0.138 4255 92.0 99 99 99 99 99 99 99 99 99 99 651,154 8,896 0.246 558,708 0.732 65.2 11.04 0.198 6,980 122 Note: All of the output and revenue measures have been adjusted to 1993 constant prices using the yearly GDP deflator. Growth rate is calculated using adjusted values. Table 4 Number of counties that satisfied each upgrading requirement in 1993. Requirement Criterion for this requirement Number of counties (total is 1,678) Met one criterion Industrialization level Urbanization level Fiscal strength Industrial output Share of industrial output Urban population Share of urban population Fiscal revenue Per capita revenue 329 290 212 212 444 976 (19.6%) (17.3%) (12.6%) (12.6%) (26.5%) (58.2%) Met both 158 (9.4%) 67 (4.0%) 352 (21.0%) finance dataset used in this paper is compiled by the Ministry of Finance. It is difficult to manipulate statistics on local public finance activities, such as fiscal revenue, because every statistic reflects actual money flow. In fact, Tsui (2005) and Zhang (2006) used this dataset and did not find any significant problems. Third, if the local officials in a county attempted to inflate statistics to obtain city status, they would first inflate industrial output, urban population and fiscal revenue to meet the official requirements, which are explicitly announced. Thus, checking whether these requirements were satisfied could provide useful information on local officials' behavior. If the majority of counties that received an upgrade did not meet these requirements, it would be hard to believe that false reporting is widespread. 3.2. Non-enforcement of upgrading requirements Because the central government created rules to regulate upgrading, a natural hypothesis is that the center aims to enforce these rules. However, many scholars (e.g., Chung & Lam, 2004; Liu & Wang, 2000) have observed that the official minimum requirements for upgrading are not enforced in practice. While eligible counties compete for upgrades, some ineligible counties received city status due to special treatment. Thus, the conjecture is that the formal minimum requirements for upgrading were not enforced in practice. The data allow me to directly determine whether counties that received an upgrade met those requirements. The three major requirements include industrialization level, urbanization level, and fiscal strength. Each requirement has an absolute criterion and a relative (or per capita) criterion. For counties with different population densities, these upgrading requirements differ. Based on the 1993 data, I summarized the number and percentage of counties that met each criterion as well as those that met both criteria of a specific requirement (Table 4).10 This table shows that the requirement on the urbanization level is quite stringent, while the fiscal strength requirement is less binding. I then assigned all county-year observations of 1994–1997 into different groups according to whether the three requirements were satisfied. For example, for a county with a population density above 400 people per square kilometers to satisfy the fiscal requirement in a particular year, its fiscal revenue must have exceeded 60 million RMB yuan, and the per capita revenue must have been higher than 100 RMB yuan. Table 5 lists the number of observations in each of these groups according to whether upgrading 10 In the official requirements shown in Table 1, “urban population” refers to those who are engaged in non-agricultural production. There is no direct report of this variable in the current dataset. Instead, I have data on the population with urban registration status for each county. For each population density group, I also know the average ratio of the population engaged in non-agricultural production to the population with urban registration status. I multiplied these two numbers to obtain an estimate of the population engaged in non-agricultural production for each county and used this number to determine whether a county met the requirement on the urbanization level. L. Li / China Economic Review 22 (2011) 172–181 177 Table 5 Number of county-year observations by upgrading status and specific requirements satisfied (1994–1997). Urbanization requirement Not satisfied Industrialization requirement Non-upgrading cases Upgrading cases Fiscal Fiscal Fiscal Fiscal requirement requirement requirement requirement not satisfied satisfied not satisfied satisfied Satisfied Not satisfied Satisfied Not satisfied Satisfied 4583 1005 24 14 212 375 14 31 96 83 2 8 5 36 0 6 Table 6 Number of county-year observations by upgrading status and number of requirements satisfied (1994–1997). Number of requirements satisfied 0 1 2 3 Total Non-upgrading cases Upgrading cases 4583 24 1313 30 463 39 36 6 6395 99 occurred. To ease the reading of the results, I further combined cells with an equal number of satisfied requirements and summarized them in Table 6. Among the 99 cases of upgrading, only 6 met all three requirements, 39 met two, 30 met only one, and 24 met none. On the other hand, there were 36 cases for which all three requirements were met but the counties were not upgraded. This table clearly demonstrates that the official requirements were not enforced in practice. 4. Importance of growth rate in obtaining city status Because the formal rules were not enforced in practice, upgrading should not be an instrument for urbanization. The central government must have had other considerations when making decisions. That is, informal rules existed that were not explicitly written in the official documents but nonetheless played an important role in the upgrading decision. Although unusual, this has been a normal practice in the Chinese political system. In this section, I investigate the incentive role of county-to-city upgrading in terms of promoting economic growth under the principal–agent relationship between the center and localities. 4.1. Empirical strategy Awarding city status is a way to promote urbanization. More importantly, because city status provides many benefits to local governments, it could be used to provide incentives to local governments to achieve certain goals. Growth has been widely regarded as a critical component in the central government's preference set. As stated in Section 2, upgrading is costly to the central government. Thus, the central government faces a problem of promoting growth with limited resources; therefore, it must select appropriate counties to receive city status. With knowledge of the center's true preference, local governments compete to obtain the lucrative benefits associated with city status. In the following, I directly examine how the upgrading prospects related to each county's growth rate. To show the importance of growth rate, Fig. 2 displays the number of upgrading cases from 1981 to 1997 as well as the corresponding yearly GDP growth rate. We can clearly observe several waves of upgrading around 1983, 1988, and 1993, which roughly correspond to the periods when China experienced high economic growth. This figure suggests that high growth may generate high demand for upgrading at the national level. To test the positive effect of growth rate on upgrading, I conducted the analysis at the county level and controlled for other confounding factors that affect upgrading. It is evident that industrialization level, urbanization level and fiscal strength are important factors because they were listed in the official documents. Thus, the main hypothesis of this paper is that, conditional on industrialization level, urbanization level, and fiscal strength, the higher the growth rate, the more likely upgrading occurred. I modeled upgrading as a discrete choice problem faced by the central government.11 For county i at year t, the upgrading decision Yit depends on the county's growth rate Git and a vector of other county-specific control variables Zit, such as industrial output, urban population, and fiscal revenue. I used a logit model to estimate the impact of these variables on the probability of upgrading: PrðYit = 1Þ = Φðβ0 + Git β1 + Zit β2 Þ = expðβ0 + Git β1 + Zit β2 Þ = ð1 + expðβ0 + Git β1 + Zit β2 ÞÞ: 11 If we assume that 1) all counties started to have a chance to upgrade in 1983, the beginning year of the policy, and 2) upgrading to city status is an ultimate outcome for most counties with sufficient time, then upgrading could be alternatively treated as a duration problem. Obtaining city status means the ending of spells, and the probability of upgrading is the corresponding hazard. However, because I can only follow the subjects for 5 periods, it is difficult to extract any meaningful information on the shape of the hazard of ending the spell. Thus, a duration model is not adopted in this paper. In fact, a linear model could be used to approximate a duration model. Currie and Neidell (2005) adopted a similar method to estimate the determinants of infant death. 178 L. Li / China Economic Review 22 (2011) 172–181 60 18% 16% 50 14% 12% 10% 30 % number 40 8% 6% 20 4% 10 2% 0% 19 84 19 85 19 86 19 87 19 88 19 89 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 82 83 19 19 19 8 1 0 year number of upgrading (left axis) growth rate of GDP (right axis) Data source: GDP is from the 2006 China Statistical Year book published by the National Bureau of Statistics; number of upgrading is from Liu and Wang (2000). Fig. 2. Number of county-to-city upgrading cases and yearly GDP growth rate. To estimate this model using data from different years, I treated a county that was upgraded in year t as if it contributed t-1993 county-year observations to the sample. The dependent variable was coded as 1 in the year that the county was upgraded and 0 in all other years. All county-year observations after upgrading were dropped.12 This sample construction approach is similar to that used by Currie and Neidell (2005). 4.2. Baseline results I closely followed the official requirements for upgrading and controlled for industrial output value, share of industrial output in gross output, urban population, share of urban population in total population, and per capita fiscal revenue. This specification allows me to estimate the effect of growth rate conditional on industrialization level, urbanization level, and fiscal strength. All of these control variables were lagged by one year to reflect the timing of upgrading decision, which should be based on realized local performance. Because counties vary greatly in size, I applied a log transformation to those level variables to reduce the effect of outliers. Table 7 Column 1 reports the estimation results, listing coefficients and robust standard errors clustered at the prefecture level. The key independent variable, growth rate of GVIAO, had a positive and significant coefficient. This finding is consistent with the hypothesis that, conditional on industrialization level, urbanization level and fiscal strength, a higher growth rate increased the probability of achieving city status. Among the five control variables that measure the official requirements, urban population and fiscal revenue per capita had positive and significant coefficients. While industrial output and its share in total output had positive and insignificant coefficients, share of urban population had a negative and insignificant coefficient. To further control for the variation of upgrading policy across years, which is common for all counties, I added year dummies (Column 2). The only change to the estimates was that the coefficient on share of urban population became positive but remained insignificant. In Column 3, I added province dummies. This is equivalent to asking the following question: what makes a county stand out from its peers in the same province in competing for city status? The results were similar to Column 1, suggesting that growth was important in the within-province competition. Then I controlled for both year and province fixed effects (Column 4). There was little change to the estimates of control variables, except that industrial output became significant. In all of these models, growth rate remained positive and significant. Because several provinces do not govern any county that was upgraded during the sample period, adding province dummies resulted in dropping counties in these provinces from the estimation. To compare the coefficient size between the baseline model and the model with year and province dummies, Column 5 reports the estimation results using the same sample as in Column 4 without controlling for year and province dummies. Overall, adding province and year dummies generated a larger coefficient on growth rate, which suggests that they may capture the effect of omitted variables. For example, if the upgrading policy is in favor of provinces with low growth potential, such a bias in policy would have a negative correlation with growth and a positive correlation with upgrading. Therefore, including province dummies eliminated the effect of this omitted variable and increased the coefficient on growth rate. In the following alternative specifications, I will report results with province and year dummies. 12 Because no city was downgraded to a county, we do not include cities into this discrete choice problem. L. Li / China Economic Review 22 (2011) 172–181 179 Table 7 Positive effect of growth rate on upgrading—logit model. Growth rate of GVIAO Log(industrial output) Share of industrial output Log(urban population) Share of urban population Log(fiscal revenue per capita) Year Dummies Province Dummies Sample size (1) (2) (3) (4) (5) 0.473** (0.211) 0.033 (0.257) 1.934 (1.267) 1.949*** (0.369) − 1.304 (1.864) 1.097*** (0.272) 0.519** (0.261) 0.367 (0.301) 1.322 (1.375) 1.584*** (0.396) 0.976 (2.002) 0.650* (0.352) Yes 0.546** (0.224) 0.456 (0.282) 2.028 (1.345) 2.189*** (0.404) − 1.374 (2.334) 1.390*** (0.276) 0.426* (0.219) 0.103 (0.267) 1.903 (1.788) 1.978*** (0.367) − 1.341 (1.317) 1.360*** (0.244) 6,436 6,436 0.749*** (0.275) 1.130*** (0.323) 0.950 (1.454) 1.600*** (0.395) 2.251 (2.313) 0.684** (0.332) Yes Yes 5,649 Yes 5,649 5,649 Note: Coefficients and robust standard errors clustered at the prefecture level are listed. Significance levels of 10%, 5% and 1% are represented by *, ** and ***, respectively. Except for growth rate and year dummies, other control variables are lagged by one year. To assess the economic magnitude of the effect of growth rate, I also calculated the marginal effect. In the estimation of Column 4, the mean marginal effect of growth rate on upgrading probability equals 0.01. This implies that when the annual growth rate increased by ten percentage points (e.g., from the average growth rate of 16% to 26%), the probability of upgrading increased by 0.001, which is approximately 7% of the average upgrading probability of 0.015.13 The change in probability of upgrading if industrial output moved from the 25th percentile to the 75th percentile was 0.022, which is approximately 1.4 times the average upgrading probability. For urban population and fiscal revenue per capita, this value was 0.026 and 0.007, respectively. These values are in reasonable ranges. 4.3. Robustness checks To check the sensitivity of the baseline results, I adopted alternative definitions of growth rate as well as various specifications and techniques to estimate the model. The results are reported in Table 8. Because upgrading decisions are made throughout a year, it is likely that the center refers to the previous year's performance when making decisions. To determine whether the results are sensitive to different timing, I lagged growth rate by one year and lagged all other independent variables by two years. The use of lagged values could also partially deal with the potential reverse causality problem. The results (Column 1) are generally similar to the baseline results, with an even larger coefficient on growth rate, suggesting that the results are robust to the different timing of variables. By including all counties in the estimation, I implicitly assume that every county has a chance of upgrading and thus faces incentives. In fact, some counties have little chance of obtaining city status; therefore, the upgrading policy does not provide any incentive for them to increase growth rates.14 One group of these counties is the nationally designated poor counties, which receive special subsidies from the center every year. The official document in 1993 (which also set the requirements on upgrading) explicitly states that these counties generally should not be considered for an upgrade. Thus, I dropped these counties and reestimated the baseline model (Column 2). The results were similar. Furthermore, I adopted alternative measures for growth rate. Chen, Li, and Zhou (2005) showed that relative performance matters more than absolute performance in evaluating provincial officials. Following their method, I subtracted the average growth rate of counties in the same prefecture from each county's growth rate and generated an alternative growth measure − relative growth rate (Column 3). The relative growth rate had a positive and significant coefficient, and the size was even larger than the previous measure of growth rate. This suggests that relative growth rate is also used as a reference in the evaluation of counties' economic performance. Following Li and Zhou (2005), I used average growth rate as another alternative growth measure, where the averaging is from the beginning of the sample to the current year. This measure corresponds to an evaluation process that is based on both the current and past growth rates. The results (Column 4) show that, although not statistically significant, the coefficient on average growth rate had a magnitude similar to the baseline results. These exercises show that the baseline results were not driven by a particular definition of growth rate. The current sample covers several years, and the upgrading policy may vary across time. Additionally, growth rates in different years may not be comparable to each other. For example, a growth rate of 14% was below average in 1994, but was above average 13 This magnitude is comparable to that estimated in Li and Zhou (2005). They showed that if the GDP growth rate of a province increased by the same amount (10 percentage points from mean), the probability of provincial leaders obtaining a promotion increased by an amount that is 24% of the average probability. 14 Ideally, I should only look at those counties that joined the competition for city status. Unfortunately, the list of counties that submitted an application or did any preparation is not available. 180 L. Li / China Economic Review 22 (2011) 172–181 Table 8 Positive effect of growth rate on upgrading—robustness checks. Growth rate of GVIAO Log(industrial output) Share of industrial output Log(urban population) Share of urban population Log(fiscal revenue per capita) Year dummies Province dummies Sample size Lagged timing Exclude poor counties Relative growth rate Average growth rate Split sample: 1994–1995 Split sample: 1996–1997 One crosssection (1) (2) (3) (4) (5) (6) (7) 1.107** (0.470) 1.191** (0.482) 1.514 (2.158) 1.679** (0.664) − 0.468 (2.741) 1.574*** (0.444) Yes Yes 3,904 0.739*** (0.282) 1.011*** (0.337) 1.169 (1.488) 1.618*** (0.400) 1.579 (2.390) 0.625* (0.335) Yes Yes 3885 0.814** (0.392) 1.06*** (0.315) 1.903 (2.259) 1.66*** (0.395) 1.121 (1.429) 0.732** (0.331) Yes Yes 5,649 0.663 (0.438) 0.98*** (0.317) 1.617 (1.421) 1.73*** (0.400) 1.332 (2.314) 0.752** (0.335) Yes Yes 5,654 0.592* (0.345) 1.128*** (0.382) 1.451 (1.770) 1.648*** (0.481) 1.518 (3.209) 0.340 (0.444) Yes Yes 2,699 0.686* (0.401) 1.272** (0.591) − 0.704 (2.524) 1.739** (0.722) 6.517* (3.470) 1.348** (0.575) Yes Yes 2,511 0.967 (0.884) − 0.276 (0.474) 2.50 (2.03) 3.58*** (0.598) − 8.01*** (3.57) 3.65*** (0.487) Yes 1,486 Note: Coefficients and robust standard errors clustered at the prefecture level are listed. Significance levels of 10%, 5% and 1% are represented by *, ** and ***, respectively. In column 1, growth rate is lagged by one year; other control variables are lagged by two years. in 1997. Controlling for year fixed effects could partially lessen this problem by allowing different intercepts for different years, but the marginal effect of growth rate may also change. Ideally, I should estimate the model year-by-year to check the robustness of the effect of the growth rate in different years. Due to the limited sample size and few upgraded cases in some years, I split the sample into two periods: 1994–1995 and 1996–1997. The estimation of these two sub-samples (Columns 5 and 6) was consistent with the baseline model. In addition, I estimated the model in a cross-section setting where each county served as one observation. In this specification, the dependent variable equals one if the county obtained an upgrade during the sampling years, and the independent variables are the average values across these years. Although the small sample size generates larger standard errors, the results provide evidence on the positive effect of growth rate (Column 7). These tests demonstrate that, even though the upgrading policy may vary across time, growth rate plays a crucial role. In sum, the regression results show that the effect of growth rate on the chance of obtaining city status is robust. Whether in an absolute or relative sense, growth rate plays the key role in upgrading. While the requirements included in the official guidelines are not enforced in practice, the central government tends to upgrade counties with faster growth rates into cities. 5. Conclusion Based on the example of creating cities in China, this paper relates urbanization policy with a fundamental question faced by the central government in developing countries of how to provide incentives to local governments. The current work found that economic growth plays an important role in determining whether a county obtains city status. The policy of “county-to-city upgrading” serves as an incentive mechanism to promote economic growth, forming part of the incentive structure of motivating local officials, Chinese style. However, it should be noted that upgrading only provides a one-time incentive to localities. This incentive mechanism is backward-looking in the sense that the rewards are for past economic performances only and may lack long-term effects. It has been argued in the literature that China is underurbanized, both from the perspective of international comparison and judged by the efficiency standard.15 Au and Henderson (2006a,b) proposed that various explicit and implicit restrictions on migration are the main reason for China's underurbanization. Lin, Cai, & Li (1996) regarded these restrictions as being endogenous to the development strategy that favors capital-intensive heavy industry over light industry and agriculture. The current study provides another possible reason. Instead of being the natural outcome of agglomeration, city creation has been used as an incentive instrument to serve political purposes, but likely at the cost of promoting real urbanization. Providing incentives through upgrading has not only affected the urbanization level in China but has also fundamentally changed China's administrative and governance structure. Such changes were made by the central government, likely under pressure from localities, and are different from the political coalition or government decentralization in developed countries (Alesina, Baqir, & Hoxby, 2004; Hennessey, 2008; Weese, 2009). Many county-level cities are now governed by prefecture-level cities. The Constitution of the People's Republic of China does not provide for the administration of one city (county-level) by another (prefecture-level) (Chan, 1997). Recently, promoting urbanization has been listed high on the policy agenda. The 15 By international standards, China's urban population ratio seems to be significantly lower than many other countries at the same development level (World Bank, 2008; Zhou & Ma, 2003). From the efficiency point of view, Au and Henderson (2006a,b) shows that Chinese cities are far from optimal, suggesting that further city expansion will bring about income growth. It is also commonly believed that China's urbanization lags behind its industrialization process. Sridhar and Wan (2007) find that China's urbanization rate is 10 percentage points lower than its industrialization level. The presence of a large number of dependents not being able to unite with the migrant workers in cities indicates that urbanization lags behind industrialization in a real sense (Lu, Huang, Lu, & Zhou, 2007). L. Li / China Economic Review 22 (2011) 172–181 181 urbanization policy has been shifted toward a balanced development of large, medium, and small cities, as well as towns. Understanding the political economy of city formation in the 1990s would provide an important reference for future city development. Acknowledgements I want to thank Peter Murrell, Roger Betancourt, Alan Drazen, Shenggen Fan, Athar Hussain, Ginger Jin, Melissa Kearney, James Kung, Yi-Min Lin, Wallace Oates, Xinqiao Ping, John Wallis, Chenggang Xu, Li-An Zhou, two anonymous referees, and seminar participants at Fudan, HKUST, IFPRI, University of Maryland, SHUFE, SWUFE, Zhejiang University, CES 2006 Shanghai meeting, AEA 2007 Chicago meeting, and the 2009 Seoul National University WCU-Hong Kong University Conference for helpful comments. References Alesina, Alberto, Baqir, Reza, & Hoxby, Caroline (2004). Political Jurisdictions in Heterogeneous Communities. Journal of Political Economy, 112, 348−396. Anderson, Gordon, & Ge, Ying (2005). The Size Distribution of Chinese Cities. Regional Science and Urban Economics, 35, 756−776. Au, Chun-Chung, & Henderson, Vernon (2006a). How Migration Restrictions Limit Agglomeration and Productivity in China. Journal of Development Economics, 80(2), 350−388. Au, Chun-Chung, & Henderson, Vernon (2006b). Are Chinese Cities Too Small. Review of Economic Studies, 73(3), 549−576. Bai, Chong-En, Tao, Zhigang, & Tong, Sarah Y. (2008). Bureaucratic Integration and Regional Specialization in China. China Economic Review, 19(2), 308−319. Blanchard, Olivier, & Shleifer, Andrei (2001). Federalism with and without Political Centralization: China versus Russia. IMF Staff Paper, 48, special issue (pp. 171−179). Bo, Zhiyue (2002). Chinese Provincial Leaders: Economic Performance and Political Mobility since 1949. Armonk, N.Y.: M.E. Sharpe. Chan, Kam Wing (1997). Urbanization and Urban Infrastructure Services in the PRC. In Christine Wong (Ed.), Financing Local Government in the People's Republic of China. Hong Kong: University of Oxford Press. Chan, Kam Wing, Henderson, Vernon, & Tsui, Kai Yuen (2008). Special Dimensions of Chinese Economic Development. In Loren Brandt, & Thomas Rawski (Eds.), China's Great Economic Transformation. New York: Cambridge University Press. Chen, Ye., Li, Hongbin, & Zhou, Li-An. (2005). Relative Performance Evaluation and the Turnover of Provincial Leaders in China. Economics Letters, 88(3), 421−425. Chung, Jae Ho. (1999). Cities in China: Recipes for Economic Development in the Reform Era. London and New York: Routledge. Chung, Jae Ho., & Lam, Tao-chiu (2004). China's 'City System' In Flux: Explaining Post-Mao Administrative Changes. China Quarterly, 180, 945−964. Currie, Janet, & Neidell, Matthew (2005). Air Pollution and Infant Health: What Can We Learn From California's Recent Experience? Quarterly Journal of Economics, 120(3), 1003−1030. Deng, Xiangzheng, Huang, Jikun, Rozelle, Scott, & Uchida, Emi (2008). Growth, Population and Industrialization, and Urban Land Expansion of China. Journal of Urban Economics, 63(1), 96−115. Du, Qiming (1993). City: Urban Designation vs. Administrative Division. Zhongguo Fangyu (Chinese Territory Administrative Divisions and Place Names)(5). Gu, Jie (1997). County-to-City Upgrading and from County-Executive to City-Mayor. Zhongguo Xingzheng Guanli (China Administrative Management)(7). Guo, Gang (2007). Retrospective Economic Accountability under Authoritarianism: Evidence from China. Political Research Quarterly, 60(3), 378−390. Henderson, Vernon, & Becker, Randy (2000). Political Economy of City Sizes and Formation. Journal of Urban Economics, 48(3), 453−484. Hennessey, Jessica (2008). The Adoption of Constitutional Home Rule: A Test of Endogenous Policy Decentralization.: University of Maryland Manuscript. Huang, Yasheng (2002). Managing Chinese Bureaucrats: An Institutional Economics Perspective. Political Studies, 50, 61−79. Jin, Hehui, Qian, Yingyi, & Weingast, Barry (2005). Regional Decentralization and Fiscal Incentives: Federalism, Chinese Style. Journal of Public Economics, 89(9–10), 1719−1742. Li, Hongbin, & Zhou, Li-An. (2005). Political Turnover and Economic Performance: The Incentive Role of Personal Control in China. Journal of Public Economics, 89 (9–10), 1743−1762. Lichtenberg, Erik, & Ding, Chengri (2009). Local Officials as Land Developers: Urban Spatial Expansion in China. Journal of Urban Economics, 66(1), 57−64. Lin, Justin, Yifu, Cai, Fang, & Li, Zhou (1996). China’s Miracle: Development Strategy and Economic Reform. Hong Kong: Chinese University of Hong Kong Press. Liu, Bangwei (2005). Urbanization: From "County-to-City Upgrading" Back to a Normal City System. Tianzhong Xuekan (Journal of Tianzhong School), 20(4). Liu, Chunbin (2004). City Govern County or County Govern City? Zhongguo Shuiwu (The Journal of China Tax Issues), 8, 60−61. Liu, Junde, & Wang, Yuming (2000). Zhidu yu Chuangxin - Zhongguo Chengshi Zhidu de Fazhan yu Gaige Xinlun (On the Development and Reform of China's Urban System). Nanjing: Dongnan University Press. Lu, Zheng, Huang, Qunhui, Lu, Tie, & Zhou, Weifu (2007). The Process and Problems of Industrialization and Urbanization in China. The Chinese Economy, 40(1), 6−30. Naritomi, Joana, Soares, Rodrigo, & Assuncao, Juliano (2007). Rent Seeking and the Unveiling of ‘De Facto’ Institutions: Development and Colonial Heritage within Brazil. NBER Working Paper No. 13545. Ping, Xinqiao (2006). The Performance Evaluation of Local Budget System and Index Design. CCER working paper No.C22006018, Peking University. Qian, Yingyi, & Weingast, Barry (1997). Federalism as a Commitment to Preserving Market Incentives. Journal of Economic Perspectives, 11(4), 83−92. Rawski, Thomas, & Xiao, Wei (2001). Roundtable on Chinese Economic Statistics Introduction. China Economic Review, 12(4), 298−302. Ren, Mingda, & Wang, Xiaomin (1999). Why There Is a Passion on County-to-City Transformation? Shehui (The Society)(7). Saich, Tony (2008). The Changing Role of Urban Government. In Shahid Yusuf, & Tony Saich (Eds.), China Urbanizes. : The World Bank. Sridhar, Kala Seetharam, & Wan, Guanghua (2007). Firm Location Choice in Cities: Evidence from China, India, and Brazil. UNC-Wider research paper No. 2007/56. Su, Donghai (2000). West Development and Turning Guyuan Prefecture into a City. Shichang Jingji Yanjiu (Research on Market Economy)(2). Tsui, Kai-yuen (2005). Local Tax System, Intergovernmental Transfers and China' S Local Fiscal Disparities. Journal of Comparative Economics, 33, 173−196. Wang, Yuxi, Ji, Lijia, & Lin, Yang (1998). Some Investigation and Thoughts on Issues about County-to-City Upgrading in Liaoning Province. Zhongguo Fangyu (Chinese Territory Administrative Divisions and Place Names)(1). Weese, Eric (2009). Political Mergers as Coalition Formation: Evidence from Japanese Municipal Amalgamations.: Department of Economics, MIT Manuscript. Wong, Christine, & Bird, Richard (2008). China's Fiscal System: A Work in Progress. In Loren Brandt, & Thomas Rawski (Eds.), China's Great Economic Transformation. New York: Cambridge University Press. World Bank (2008). The Growth Report: Strategies for Sustained Growth and Inclusive Development. The World Bank: Washington D.C. Wu, Bowei (2000). The Impact of Administrative Division on Socio-economic Development. Zhongguo Fangyu (Chinese Territory Administrative Divisions and Place Names)(3). Xu, Chenggang, in press. “The Fundamental Institutions of China's Reforms and Development”. Journal of Economic Literature. Zhang, Li, & Zhao, Simon (1998). Re-Examining China's "Urban" Concept and the Level of Urbanization. China Quarterly, 154, 330−381. Zhang, Xiaobo (2006). Fiscal Decentralization and Political Centralization in China: Implications for Growth and Inequality. Journal of Comparative Economics, 34 (4), 713−726. Zhou, Yixing, & Ma, Laurence (2003). China's Urbanization Levels: Reconstructing a Baseline from the Fifth Population Census. China Quarterly, 173, 176−196. Zhuravskaya, Ekaterina (2007). Whither Russia? A Review of Andrei Shleifer's A Normal Country. Journal of Economic Literature, 45(1), 127−146.