City centrality, population density and energy efficiency

We use continuous night light data and LandScan population data to construct a monocentric index to measure whether a city tends to be a monocentric or polycentric spatial structure, and estimate the impact of such structure on energy efficiency by using a two-way fixed effects model. In order to solve the endogeneity of urban structure, the historical census data and topographic relief data at the city level are used to construct the instrumental variables. We find that an increase in the urban monocentricity index would significantly reduce the level of the urban energy intensity. Meanwhile, the relationship between monocentricity and urban energy intensity is U-shaped. When the degree of monocentricity is at relatively low level, its impact on energy efficiency is positive, implying the dominance of agglomeration effect. When economic agglomeration is greater than a certain degree, the impact of monocentricity becomes negative which results from the dominance of congestion effect. Moreover, impact of urban structure on energy intensity varies with urban population density. An increase in population density would weaken the negative impact of monocentricity on energy intensity, that is, in sparsely populated cities, the monocentricity is more conducive to energy saving. In terms of city form, monocentric structure may improve energy efficiency in the cities with poor transportation infrastructure more than in the cities with good transportation infrastructure. Monocentricity facilitates energy efficiency primarily by increasing corporate innovation and saving commuting time and distance. We perform a series of robustness tests and the findings are basically consistent with the baseline result. Our findings suggest that the cities at this stage should continue to strengthen the development of monocentricity, pay attention to compactness of urban spatial structure, and develop the polycentric structure cautiously. Meanwhile, in the development process, the local population density and the carrying capacity of the transportation infrastructure should be taken into consideration to avoid the negative effects caused by the congestion in main center.