# RECORD 6a582c18a0b55c4939a0e4812ae293fd34b678de # TITLE: Capitalization of Public School Quality into Housing Prices: An Empirical Analysis Based on School District Housing for Public Primary Schools in Shanghai # YEAR: 2018.0 VENUE: nan DOI: 10.4236/chnstd.2018.74025 # AUTHORS: Tiancheng Zhou # PAGES: 42 CANDIDATE TABLE PAGES (1-based): [12, 13, 16, 17, 18] # ABSTRACT: This paper examines the capitalization of public school quality into housing prices in the specific context of China’s compulsory education system and “district correspondence enrollment” policy, which stipulates a strict correspondence between residential estates and enrollment into public primary schools. Taking 344 neighborhoods in 7 main districts of Shanghai, China as its sample, this paper carries out detailed descriptive analysis of the data. It also employs the traditional Hedonic Price Model and 2SLS regression method to quantitatively calculate the exact capitalization rates while isolating eight non-school attributes affecting housing prices. It yields the conclusion that for every one-rank improvement in the quality of the corresponding public primary school, the average housing price of a neighborhood is projected to increase by 3.1%, 2.8%, and 1.9%, on the citywide, urban, and suburban scale, respectively. The capitalization effect of public school quality in housing prices is statistically significant, so the status-quo of the distribution of public educational resources in Shanghai is still considerably unequal. In the final section of this paper, the significance of this research is discussed and comprehensive policy recommendations and action plan are given in an attempt to mitigate the school district housing fever and educational inequality. ===== PAGE 1 (score 3, numeric-paren 0, table-words 0, var-words 3) ===== Chinese Studies, 2018, 7, 286-327 http://www.scirp.org/journal/chnstd ISSN Online: 2168-541X ISSN Print: 2168-5428 DOI: 10.4236/chnstd.2018.74025 Nov. 13, 2018 286 Chinese Studies Capitalization of Public School Quality into Housing Prices: An Empirical Analysis Based on School District Housing for Public Primary Schools in Shanghai Tiancheng Zhou Shanghai Pinghe School, Shanghai, China Abstract This paper examines the capitalization of public school quality into housing prices in the specific context of China’s compulsory education system and “district correspondence enrollment” policy, which stipulates a strict corres- pondence between residential estates and enrollment into public primary schools. Taking 344 neighborhoods in 7 main districts of Shanghai, China as its sample, this paper carries out detailed descriptive analysis of the data. It also employs the traditional Hedonic Price Model and 2SLS regression me- thod to quantitatively calculate the exact capitalization rates while isolating eight non-school attributes affecting housing prices. It yields the conclusion that for every one-rank improvement in the quality of the corresponding public primary school, the average housing price of a neighborhood is pro- jected to increase by 3.1%, 2.8%, and 1.9%, on the citywide, urban, and sub- urban scale, respectively. The capitalization effect of public school quality in housing prices is statistically significant, so the status-quo of the distribution of public educational resources in Shanghai is still considerably unequal. In the final section of this paper, the significance of this research is discussed and comprehensive policy recommendations and action plan are given in an attempt to mitigate the school district housing fever and educational inequa- lity. Keywords Capitalization of Educational Quality, Housing Prices, School District Housing Fever, Shanghai, China, Hedonic Price Model, Two-Stage Least Squares (2SLS) Regression Method How to cite this paper: Zhou, T. C. (2018). Capitalization of Public School Quality into Housing Prices: An Empirical Analysis Based on School District Housing for Pub- lic Primary Schools in Shanghai. Chinese Studies, 7, 286-327. https://doi.org/10.4236/chnstd.2018.74025 Received: October 11, 2018 Accepted: November 10, 2018 Published: November 13, 2018 Copyright © 2018 by author and Scientific Research Publishing Inc. This work is licensed under the Creative Commons Attribution International License (CC BY 4.0). http://creativecommons.org/licenses/by/4.0/ Open Access ===== PAGE 12 (score 14, numeric-paren 0, table-words 3, var-words 4) ===== T. C. Zhou DOI: 10.4236/chnstd.2018.74025 297 Chinese Studies Table 2. Descriptive analysis of independent variables, citywide. Variable Unit Average Standard Deviation Maximum Value Minimum Value Quality of the Corresponding Public Primary School no unit 20.31 11.41 44 1 Housing Price RMB/m2 76,949.81 24,548.04 150,621 28,641 Distance to the Corresponding Public Primary School Meters 748.13 504.75 4600 10 Distance to the District’s CBD Meters 2141.80 1439.08 7700 70 Property Management Fee RMB/m2 per month 2 1.72 14.8 0.3 Green Coverage Ratio % 34.03 9.24 70 10 Elevator no unit 0.52 0.50 1 0 Residential Area m2 92.41 50.92 404 18 Floor Area Ratio no unit 2.21 1.09 8.9 0.36 Building Age Years 19.54 12.27 178 1 Table 3. Descriptive analysis of independent variables, urban. Variable Unit Average Standard Deviation Maximum Value Minimum Value Quality of the Corresponding Public Primary School no unit 14.53 7.58 29 1 Housing Price RMB/m2 88,311.57 18,676.47 150,621 50,933 Distance to the Corresponding Public Primary School Meters 650.94 383.03 1900 10 Distance to the District’s CBD Meters 2059.59 1376.79 7700 70 Property Management Fee RMB/m2 per month 2 1.96 14.8 0.13 Green Coverage Ratio % 32.72 9.17 70 10 Elevator no unit 0.62 0.49 1 0 Residential Area m2 92.10 53.37 404 18 Floor Area Ratio no unit 2.47 1.08 5.68 0.38 Building Age Years 20.80 13.83 178 1 environmental and economic situations in Shanghai, they prove the validity and real-life significance of the sample and data sources of this paper and lay the foundation for following technical analyses. The number of KPS in each district of Shanghai is listed below in Table 5 and illustrated in Figure 2. The four districts with the most KPS are Xuhui, Hua- ngpu, Pudong, and Yangpu in order. These districts are all in the urban area and share the characteristic of advanced socioeconomic activities and educational ===== PAGE 13 (score 10, numeric-paren 0, table-words 2, var-words 4) ===== T. C. Zhou DOI: 10.4236/chnstd.2018.74025 298 Chinese Studies Table 4. Descriptive analysis of independent variables, suburban. Variable Unit Average Standard Deviation Maximum Value Minimum Value Quality of the Corresponding Public Primary School no unit 34.23 5.98 44 22 Housing Price RMB/m2 49,993.49 13,077.58 80,060 28,461 Distance to the Corresponding Public Primary School Meters 978.72 662.43 4600 40 Distance to the District’s CBD Meters 2336.86 1566.97 7100 110 Property Management Fee RMB/m2 per month 1.07 0.71 3.5 0.2 Green Coverage Ratio % 37.16 8.72 70 17 Elevator no unit 0.30 0.46 1 0 Residential Area m2 93.13 44.82 303 29 Floor Area Ratio no unit 1.57 0.86 8.9 0.36 Building Age Years 16.55 6.45 38 2 Table 5. Number of key primary schools (kps) in shanghai, by district. District Number District Number District Number Xuhui 14 Pudong 14 Huangpu 13 Yangpu 10 Minhang 9 Jiading 9 Baoshan 8 Jing’an 7 Putuo 6 Jinshan 6 Songjiang 6 Hongkou 5 Changning 4 Fengxian 3 Qingpu 2 Figure 2. Number of KPS in Shanghai, by district. resources, and thus relatively higher housing prices. The suburban districts, by contrast, generally have less developed economies and lower-quality educational resources. 0 2 4 6 8 10 12 14 16 Nubmer of KPS Districts ===== PAGE 14 (score 6, numeric-paren 0, table-words 0, var-words 6) ===== T. C. Zhou DOI: 10.4236/chnstd.2018.74025 299 Chinese Studies Figures 3-4 offer a clearer observation of the more pronounced capitalization effect of public school quality into housing prices in the urban districts by visua- lizing the distribution of housing prices. The distribution of housing prices in Huangpu District is illustrated graphically below in Figure 3 where the peaks and troughs correspond to the different housing prices of neighborhoods in dif- ferent school districts and the trend of change between them. One school may correspond to several neighborhoods whose housing prices decrease as the dis- tance to the school increases. The conspicuous peaks are likely to be caused by sought-after schools whose housing prices are significantly higher than those of others. The varying heights of the peaks illustrate the different housing prices for different schools. Figure 4 and Figure 5 illustrate the distributions of housing prices in Xuhui and Changning Districts, respectively. Similar to Figure 3, the peaks and troughs correspond to neighborhoods in different school districts. But the unique cha- racteristic setting these two districts apart from Huangpu is the differences be- tween housing prices of different neighborhoods are obvious in Xuhui and Changning. In Huangpu, however, there isn’t such sharp difference. 3.3. Hedonic Price Model To calculate the exact capitalization rates, the traditional Hedonic Price Model is employed. The basic assumption of our econometric model assumes that hous- ing price reflect the market values of public school quality and other characteris- tics. After referring to existing researches on real estate valuation and considering the quantification of the variable studied here, the logarithm functional is used to establish the basic model. Specifically, positive and continuous independent variables (e.g., distance, building age, floor area ratio) are adopted in logarithmic form, while the variables quantified through scoring or dummy variables (namely quality of corresponding public primary school and elevator) are adopted Figure 3. Distribution of housing prices in Huangpu district. 0 20000 40000 60000 80000 100000 120000 140000 160000 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 Housing Prices ===== PAGE 16 (score 31, numeric-paren 0, table-words 11, var-words 4) ===== T. C. Zhou DOI: 10.4236/chnstd.2018.74025 301 Chinese Studies The results of regression analysis and variance analysis on the citywide scale are given in Table 6 and Table 7, respectively. The variance analysis confirms that the model fits very well with the experimental data and so the coefficients it yielded can be used to explain the sample. Every coefficient fits with its expected sign, and their P-values further verify that most coefficients are statistically sig- nificant. Specifically, the coefficient of the quality of the corresponding public primary school is −0.031, meaning that housing price increases by 3.1% for a one-rank improvement of its corresponding school (which is equivalent to its numerical rank decreasing by 1). The residual plot and line fit plot for the variable “Quality of the Correspond- ing Public Primary School” are cited below in Figure 6 and Figure 7 to illustrate its statistical significance. The results of regression analysis and variance analysis in the urban areas are given in Table 8 and Table 9, respectively. The variance analysis confirms that the model fits very well with the experimental data and so the coefficients it yielded can be used to explain the sample. The P-values of the coefficients fur- ther verify that most of them are statistically significant and fit with their ex- pected signs. Specifically, the coefficient of the quality of the corresponding pub- lic primary school is −0.028, meaning that housing price increases by 2.8% for a one-rank improvement of its corresponding school (which is equivalent to its numerical rank decreasing by 1). Table 6. Regression analysis, citywide. Variable Coefficient Standard Deviation t Stat P-value Intercept 12.156 0.199 61.061 2.862E−183 Quality of the Corresponding Public Primary School −0.031 0.001 −33.790 8.616E−110 Distance to the Corresponding Public Primary School −0.027 0.010 −2.657 0.008 Distance to the District’s CBD −0.007 0.011 −0.663 0.508 Property Management Fee 0.029 0.019 1.536 0.125 Green Coverage Ratio 0.002 0.032 0.072 0.942 Elevator 0.003 0.022 0.123 0.902 Residential Area −0.003 0.024 −0.106 0.916 Floor Area Ratio −0.016 0.020 −0.775 0.439 Building Age −0.034 0.023 −1.466 0.144 Table 7. Variance analysis, citywide. Name df SS MS F Significance F Regression Analysis 9 44.456 4.940 208.954 1.9458E−131 Residuals 334 7.896 0.024 Sum 343 52.352 ===== PAGE 17 (score 16, numeric-paren 0, table-words 4, var-words 4) ===== T. C. Zhou DOI: 10.4236/chnstd.2018.74025 302 Chinese Studies Table 8. Regression analysis, urban. Variable Coefficient Standard Deviation t Stat P-value Intercept 12.210 0.207 59.068 8.565E−142 Quality of the Corresponding Public Primary School −0.028 0.001 −22.069 6.56E−59 Distance to the Corresponding Public Primary School −0.012 0.010 −1.146 0.253 Distance to the District’s CBD −0.024 0.012 −1.948 0.053 Property Management Fee 0.016 0.020 0.823 0.411 Green Coverage Ratio 0.032 0.031 1.005 0.316 Elevator −0.010 0.024 −0.429 0.668 Residential Area −0.001 0.025 −0.041 0.968 Floor Area Ratio −0.054 0.023 −2.381 0.018 Building Age −0.071 0.023 −3.122 0.002 Table 9. Variance analysis, urban. Name df SS MS F Significance F Regression Analysis 9 11.335 1.259 68.296 3.085E−60 Residuals 232 4.278 0.018 Sum 241 15.613 Figure 6. Variable “quality of the corresponding public primary school” residual plot. Figure 7. Variable “quality of the corresponding public primary school” line fit plot. -1 -0.5 0 0.5 1 0 10 20 30 40 50 Resiuals Variable “Quality of the Corresponding Public Primary School” Residual Plot ===== PAGE 18 (score 21, numeric-paren 0, table-words 6, var-words 5) ===== T. C. Zhou DOI: 10.4236/chnstd.2018.74025 303 Chinese Studies The results of regression analysis and variance analysis in the suburban areas are given in Table 10 and Table 11, respectively. The variance analysis confirms that the model fits very well with the experimental data and so the coefficients it yielded can be used to explain the sample. The P-values of the coefficients fur- ther verify that most of them are statistically significant and fit with their ex- pected signs. Specifically, the coefficient of the quality of the corresponding pub- lic primary school is −0.019, meaning that housing price increases by 1.9% for a one-rank improvement of its corresponding school (which is equivalent to its numerical rank decreasing by 1). 4. Conclusion This paper establishes hedonic regression analysis to investigate the capitaliza- tion effect of public school quality into housing prices and quantitatively calcu- late the exact capitalization rates. Using the data of 344 residential neighbor- hoods over 7 main districts in Shanghai, it yielded the following empirical re- sults: 1) The quality of the corresponding public primary school has a significant positive effect on housing prices. The compulsory educational facilities are capi- talized into estate prices in Shanghai. Under the strict enforcement of the “dis- trict correspondence enrollment” policy, families are willing to pay higher prices for houses so that their children can enroll in better schools. 2) The capitalization rate of public school quality is calculated to be 3.1%, 2.8%, and 1.9% on the citywide, urban, and suburban scales, respectively. When the rank of the corresponding public primary school improves by one, the aver- age housing price in the neighborhood increases by the given percentage. 3) The capitalization effects of public school quality into housing prices exhi- bit a Matthew Effect, i.e. the amount of capitalization increases rapidly as the quality of the corresponding public primary school approaches the top citywide, Table 10. Regression analysis, suburban. Variable Coefficient Standard Deviation t Stat P-value Intercept 10.743 0.520 20.645 2.677E−36 Quality of the Corresponding Public Primary School −0.019 0.003 −6.246 1.289E−08 Distance to the Corresponding Public Primary School −0.042 0.021 −2.001 0.048 Distance to the District’s CBD 0.042 0.023 1.806 0.074 Property Management Fee 0.016 0.039 0.407 0.685 Green Coverage Ratio 0.069 0.095 0.725 0.470 Elevator 0.075 0.049 1.511 0.134 Residential Area 0.049 0.057 0.868 0.387 Floor Area Ratio −0.043 0.043 −0.995 0.322 Building Age 0.049 0.062 0.786 0.434 ===== PAGE 19 (score 10, numeric-paren 0, table-words 3, var-words 4) ===== T. C. Zhou DOI: 10.4236/chnstd.2018.74025 304 Chinese Studies Table 11. Variance Analysis, Suburban. Name df SS MS F Significance F Regression Analysis 9 1.850 0.206 7.884 1.5289E−08 Residuals 92 2.399 0.026 Sum 101 4.249 Note: When the sample is divided into urban and suburban areas for separate analyses, its size diminishes, so the coefficients of a few variables disagree with their expected signs or have P-value greater than 0.1. Al- so, due to the particularity of the designation of school districts and the real estates in Shanghai, many old, unmodern neighborhoods are in close proximity to high-ranking public primary schools near the city cen- ter while the neighborhood and structural attributes vary quite erratically especially in suburban areas. Yet the P-value, residual plot, and line fit plot for the school quality variable in all three dimensions are statisti- cally significant and can thus reliably reveal the causational effects and validate the conclusions of this study. while the effect is much milder for schools at a middle or middle-lower level. Since the increase in housing price is given in percentages, the higher the rank, the more the capitalization effect compounds. On the citywide scale where a larger disparity in school quality is present, the measured capitalization rate is higher than in separate samples. These patterns reflect the serious inequality of access to public educational resources behind the school district housing fever: The gap between low- and high-quality educational resources widened and be- came even more unaffordable for the low-income families. 5. Further Implications 5.1. Research Significance As one of the few existing studies conducted on Shanghai’s School District Housing Fever, especially after the strengthening of the “district correspondence enrollment” policy in 2014, this paper examines the status-quo of the local capi- talization effect. Most notably, it deals with the lack of sufficient data on public school quality by synthesizing multiple sources. It also samples housing infor- mation of 344 neighborhoods in Shanghai and considers seven control variables that account for every aspect in the estate valuation. Moreover, by using the 2SLS regression method, it avoids the endogeneity bi- as that has been an issue in most existing researches in the field so that it was able to yield statistically significant results which are also consistent with real-life situations. 5.2. Policy Recommendations and Action Plans to Mitigate the School District Housing Fever The conclusions of this study reveal that the status-quo of the distribution of public educational resources in Shanghai is still considerably unequal. The “dis- trict correspondence enrollment” policy itself is a demand-side policy as it regu- late the method for enrollment, and so are most of the policies and measures adopted in an attempt to improve China’s educational inequality, such as desig- nating common school districts for higher-ranking and lower-ranking schools