وفور منابع و رشد اقتصادی در چین

This paper revisits the resource curse phenomenon in China and differs from the previous studies in four respects: (i) City-level data is used; (ii) A spatial variable is constructed to estimate the diffusion effect of natural resources among cities in the same province; (iii) The impact of resource abundance on economic development is investigated not only at the city level but also at the prefectural level in China; (iv) We use a functional coefficient regression model to deal with city-specific heterogeneity and, at the same time, analyze the transmission mechanism of the resource curse phenomenon. Our empirical results show that there is no significant evidence to support the existence of a resource curse phenomenon in China. On the other hand, we find that the degree of natural resource abundance in a city has a positive diffusion effect on the economic growth of neighboring cities within the same province at the city level, but not at prefectural levels. We attribute this to the urban bias policy.

Many studies attempt to determine whether or not natural resources serve as an important engine for economic growth. Their common finding is that the economic growth rates of natural resource-abundant countries are slower than those of natural resource-scarce economies (Leite and Weidmann, 1999, Papyrakis and Gerlagh, 2004, Rodriguez and Sachs, 1999, Sachs and Warner, 1995, Sachs and Warner, 1997 and Sachs and Warner, 1999, among many others). This widely accepted phenomenon is referred to as the “resource curse” in the literature. Moreover, some recent studies analyze which socio-economic variables yield a negative association between economic growth and natural resource abundance. For example, Gelb (1988) and Auty (1990) argue that resource rich countries are likely to pay more attention to rent-seeking behavior rather than other productive activities. Angrist and Kugler (2008) emphasize that abundant resources can be a source of civil conflict. Matsuyama (1992) and Sachs and Warner (2001) find that resource-abundant economies value their natural resource oriented goods higher than their manufactured goods, which could keep their economies at a low level of economic growth. Gylfason (2001) finds that the level of education is an important factor for determining the resource curse phenomenon. Kronenberg (2004) demonstrates that corruption is a major determinant for the appearance of the curse of natural resources. Papyrakis and Gerlagh (2007) extend this line of research from cross-country studies to one examining different regions within the same country. They investigate forty-nine U.S. states and find positive evidences of the existence of the resource curse. In particular, they find that resource abundance decreases investment, schooling, openness and R&D expenditure and while increasing corruption. However, a consensus among economists is far from reached about understanding the role natural resources play in economic growth. Habakkuk (1962) believes that resource abundance is one of the main reasons the U.S. economy surpassed the U.K. economy in the 19th century. Wright (1990) finds that the most significant feature of U.S. manufacturing exports during the early 20th century was an intensity in natural resources, and that abundant resources reflect advanced technology. Davis (1995) analyzes twenty-two mineral-based economies using different criteria and finds that the existence of resource curse is the exception rather than the rule. On the other hand, many economists find that the negative association between resource abundance and economic growth is not robust or insignificant when different measures of resource abundance and different frequencies of data are used. Sala-i Martin (1997) finds a negative association when the ratio of primary products to exports is used, a measurement of resource abundance advanced by Sachs and Warner (1995), while a positive association is obtained when the ratio of GDP to mining products is employed. Stijns (2005) shows that resource curse disappears when the resource abundance is measured in terms of energy and mineral reserves. Alexeev and Conrad (2009) argue that the current finding of the existence of resource curse, obtained by using an average growth rate starting from 1965, is possibly due to a dynamic pattern of refinement. In recent years, more and more economists have become interested in examining whether the resource curse exists in China. To the best of our knowledge, Zhang, Xing, Fan, and Luo (2008) is the first paper in the English literature to explore this important issue. Using provincial-level data from 1985 to 2005, they find that Chinese provinces with abundant resources perform worse than provinces with poor natural resources in terms of per capita consumption growth. However, when they use the subperiod sample of 1995 to 2005, this finding disappears. They attribute this change to the resource price liberalization launched in the mid-1990s. Xu and Wang (2006) employ provincial-level panel data from 1995 to 2003 and find evidence to support the existence of resource curse. Using panel data of eleven western provinces between 1991 to 2006, Shao and Qi (2009) verify the existence of resource curse in the western regions of China. Using the panel data of twenty-eight provinces, Ji, Magnus, and Wang (2010) find that although resource abundance has a positive impact on economic growth in China, resource dependence has a negative impact. Furthermore, using a varying-coefficient model, they find the effect of natural resource on economic growth varies with institutional qualities. Fang, Ji, and Zhao (2011) investigate the resource curse in China using city-level data. They argue that the controversial results of the existence of the resource curse partially result from using different resource abundance measures. In this paper, we revisit the curse of resources in China and analyze possible transmission mechanisms between resource abundance and economic development. Our paper contributes to the literature in four respects: (i). Instead of using province-level data, our analysis is based on all city-level data in China.1 Benefitting from the large number of prefectural-level observations, we adopt a cross-sectional econometric model rather than a panel data model.2 (ii). A diffusion variable based on the relative degree of resource abundance and the geographic distances between cities is constructed to capture the spill-over effect of resource abundance among cities within the same province. Hence, our study can distinguish two different effects. The effect of resource abundance represents whether the resource abundance of a city affects this city's long-run economic development, and the effect of the diffusion variable implies whether a city can gain a benefit from resource-rich cities within the same province. To our knowledge, this is a new contribution to the literature. (iii). We investigate the impact of resource abundance on economic development not only at the city level but also in the rural regions. We obtain different empirical outcomes for these cases. This difference may result from the urban bias policy. (iv). We analyze transmission mechanisms between resource abundance and economic development by employing a functional coefficient regression model. To analyze the transmission mechanisms, many studies adopt two least squares regressions separately (see Fang et al., 2011, Papyrakis and Gerlagh, 2004, Papyrakis and Gerlagh, 2007 and Shao and Qi, 2009; among others). Due to these two separated regressions, it is different to infer whether a particular transmission mechanism is important to the relationship between resource abundance and economic growth. Taking advantage of the functional coefficient regression, we can combine the two regressions into one and make a precise inference about the transmission mechanism. Moreover, this functional coefficient regression allows us to capture a nonlinear relationship between resource richness and economic development, providing some interesting economic stories that may be neglected in a simple linear model. Our results show that there is no support for the existence of resource curse phenomenon at the city level in China over the period 1997–2005. By applying the functional coefficient regression model, we find that the estimated effects of natural abundance on the economic growth of regional economies are significantly positive when the relative scale of the manufacturing industry, innovation (R&D) and openness are considered as transmission channels. In particular, we find a nonlinear relationship between natural resources and economic growth, which indicates that one unit of natural resources affects the growth of the regional economy differently depending on the levels of the transmission variables. We believe that this result is useful for Chinese policy makers to establish appropriate economic and political policies. Moreover, our empirical results for the diffusion effect show that an abundance of natural resources not only encourages local economic development but also boosts growth of the economy in other cities in the same province. This diffusion phenomenon is significant through the economic transmission channels such as manufacturing, innovation, human capital investment and openness at the city level but is not significant at the prefectural level. We attribute this difference to the urban bias policy. The next section discusses different measurements for resource abundance in China and their possible impacts on empirical results. Section 3 introduces the basic model and the construction of the diffusion variable. We also investigate the impact of resource abundance on economic development at the city level and in rural regions as well. Section 4 analyzes possible transmission mechanisms using the functional coefficient regression model, and the last section concludes.

We find that city-specific heterogeneity plays an important role in explaining the association between resource abundance and economic growth. This association is much more complicated than what is analyzed in a simple linear model. It may be a conceptual mistake misleading to summarize the complicated relationship into a binary situation: resource curse or resource blessing. Benefitting from the functional-coefficient setup and the nonparametric estimation, we identify nonlinearity in the impact of resource abundance on economic growth and also in the transmission channels. For example, we find that the impact of natural resources on economic growth depends on the level of industrialization of the area. At the city level, relatively under-industrialized cities can benefit more from resource abundance than those industrialized ones. Hence, policy makers should differentiate between areas when implementing policies. Contrary to the findings of many studies, we find empirical evidence against the hypothesis of the existence of the resource curse by using a specific resource measurement, the average fraction of workers in the mining industry compared to the total local population. We show that the previous measures of resource abundance possibly mistakenly categorize less-developed regions into the resource-abundant group. However, as we indicated, our resource measure might have its own limits. Our future research lies in developing new measures, particularly some direct measures of natural resources to test the robustness of our results. Identifying the association between resource abundance and economic development also crucially depends on how to economic development is measured. Although most studies employ GDP-based statistics to measure economic development, we realize that GDP is an imperfect indicator of people's welfare as it does not include the environmental pollution from exploring and developing natural resources. In particular, according to Chinese laws, any natural resources beneath the ground belong to the government. Zhang et al. (2008) find that resource-rich areas have higher ratio of state sector to private sector investments, on average, than in resource-poor areas. The state sector may obtain most benefit from the extraction of resources while the local community receives very little yet endure all the negative externalities. Due to a lack of data to measure people's welfare, our results are suggestive only. Further studies are called for once more data on either income or consumption at the city level becomes available. This argument also relates to the urban bias policy in China. Despite most natural resources being located in rural areas, we find that the diffusion effect is not robust for rural cases. However, a significantly positive diffusion effect exists among cities within the same province. We attribute this difference to the urban bias policy. Although the government has gradually liberalized the prices of most natural resources since 1994, state-owned firms still control the operation of the exploration and mining of large-scale reserves while the private sector can only mine small-scale reserves (Zhang, et al., 2008). The change of urban bias policy needs policy makers to set up new policies more aligned with the interests of local residents.