The Value of Air Quality in Chinese Cities: Evidence from Labor and Property Market Outcomes

Abstract

Using a dual-market sorting model of workers’ location decisions, this paper studies the capitalization of air pollution in wages and property prices across Chinese cities. To account for endogeneity of air pollution in the determination of wages and property prices, we exploit quasi-experimental variation in air quality induced by a policy subsidizing coal-based winter heating in northern China, and document a discontinuity in average air quality for cities located north and south of the policy boundary. Using data for all 288 Chinese cities in 2011, we estimate an equilibrium relationship between wages and house prices for the entire system of Chinese cities, and specify a regression discontinuity design to quantify how variation in air quality induced by the policy affects this relationship locally. Our preferred estimates of the elasticity of wages and house prices with respect to $\text {PM}_{10}$ concentration are 0.53 and ${-}$ 0.71 respectively. At the average of our sample, the willingness to pay for a unit reduction in $\text {PM}_{10}$ concentration is CNY 261.28 ($\simeq$ USD 40.50), with a significant share reflected in labor market outcomes.

Appendices

Appendix A: Data Sources and Summary Statistics

Variable	Definition	Mean	SD	p10	p25	p50	p75	p90	Source
Wages	Average yearly wages (1000 CNY/Year)	35.96	8.24	27.02	30.48	34.62	39.81	47.05	NBSC (2012a)
Male	Share of population: male (%)	52.77	8.80	42.48	50.20	52.43	57.93	62.51	NBSC (2012b)
College	Share of population: college graduates (%)	8.10	4.85	3.93	5.02	6.73	8.98	15.07	NBSC (2010)
Unemployment	Unemployment rate (%)	3.30	0.74	2.20	2.80	3.50	3.90	4.20	NBSC (2012a)
Migrants	Share of population: immigration (%)	21.64	32.78	6.40	8.62	13.24	22.39	42.46	NBSC (2010)
House prices	Average house prices (1000 $\hbox {CNY}/\hbox {m}^2$)	4.12	2.47	2.40	2.73	3.38	4.37	6.68	NBSC (2012a)

Variable	Definition	Mean	SD	p10	p25	p50	p75	p90	Source
Piped water	Share of houses with piped water (%)	63.25	21.46	35.00	45.00	61.50	83.00	93.00	NBSC (2010)
Fitted toilets	Share of houses with toilet (%)	70.68	16.50	49.00	58.00	73.00	84.00	90.00	NBSC (2010)
Fitted shower	Share of houses with shower (%)	51.19	22.79	18.00	32.50	52.00	67.50	81.00	NBSC (2010)
Density	Population density (1000 persons /$\hbox {km}^2$)	0.47	0.54	0.08	0.18	0.34	0.62	0.87	NBSC (2012a)
Land supply	Traded land area (100 $\hbox {km}^2$)	0.13	0.42	0.02	0.04	0.07	0.12	0.22	CLRA (2012)
Coastal	Indicator equal to 1 if city is on the coast	0.09	0.28	0.00	0.00	0.00	0.00	0.00	NGCC (2012)
Universities	Number of universities per 10,000 residents	0.01	0.01	0.00	0.01	0.01	0.01	0.02	NBSC (2012a)
$\text {PM}_{10}$	Average $\text {PM}_{10}$ concentration ($\upmu \hbox {g}/\hbox {m}^3$)	77.44	19.11	53.00	64.00	78.00	91.00	100.00	MEP (2012)
$\hbox {Polluted days}^{\mathrm{a}}$	Number of days with API $\ge$ 100	21.94	21.59	0.00	4.50	17.00	34.00	48.00	NBSC (2012a)
Huai policy	Indicator equal to 1 if city is included in the River Huai policy	0.43	0.50	0.00	0.00	0.00	1.00	1.00	NGCC (2012)
Latitude	Distance to latitude $33^{\circ }$ (degree $^{\circ }$)	$-$ 0.06	6.67	$-$ 9.27	$-$ 5.17	$-$ 0.56	4.71	8.80	NGCC (2012)
Temp mean summer	Mean temperature in summer ($^{\circ }\mathrm {C}$)	26.10	2.87	22.24	23.66	26.85	28.32	29.37	CMA (2012)
Temp mean winter	Mean temperature in winter ($^{\circ }\mathrm {C}$)	1.91	8.34	$-$ 9.62	$-$ 3.04	3.27	6.83	13.16	CMA (2012)

1. 288 observations
2. $^{\mathrm{a}}$Definition of API is in Appendix B. Exchange rate for 2011: CNY 1 $\simeq$ USD 0.155

Appendix B: Definition of the Air Pollution Index (API)

The conversion method from pollutant concentrations to API values is available from the website of China’s Ministry of Environmental Protection. First, average concentration of air pollutants are reported from the monitoring stations in each city (cities with a population of over 3 million people are required to use at least eight monitoring stations to measure urban air quality). Specifically, concentrations of $\text {PM}_{10}$, $\hbox {SO}_{2}$ and $\hbox {NO}_{2}$ are measured with daily averages.

The API is then constructed from piecewise linear indexes computed for each pollutant. Conversion of concentrations into pollutant-specific indexes is as follows:

$$\begin{aligned} I_k= \frac{I_{{ high}} - I_{{ low}}}{C_{{ high}} - C_{{ low}} }\left( C_k -C_{{ low}}\right) + I_{{ low}} \end{aligned}$$

(B1)

where $I_k$ is the pollution index of pollutant k, $C_k$ is pollutant k’s concentration, $C_{{ low}}$ is the concentration threshold below $C_k$ (defined in Table 4), $C_{{ high}}$ is the concentration threshold above $C_k$, $I_{{ low}}$ is the index threshold corresponding to $C_{{ low}}$, and $I_{{ high}}$ is the index threshold corresponding to $C_{{ high}}$. The API then represents the highest value of those indexes:

$$\begin{aligned} \hbox {API} = \hbox {max} \left( I_{PM_{10}},I_{SO_2} ,I_{NO_2}\right) \end{aligned}$$

(B2)

Note that the scale for each pollutant is non-linear, as is the final API value. An implication is that an API of 100 is not equivalent to a doubling of air pollution associated with an API of 50.

Table 4 Air pollution index (API).

As shown in Table 4, an API of 100 is regarded as the threshold of air pollution that can lead to non-negligible health consequences. Thus days with an API under 100 are defined as a “Blue Sky” days (i.e. days in which air pollution poses little or no risk to human health). When the API is between 101 and 200, air pollution may generate slight irritations to breathing and lead to heart diseases. When API is between 201 and 300, air pollution will aggravate the symptoms of patients with cardiac and lung diseases, even healthy people will be noticeably affected. When API is above 300 air is severely polluted and may cause irritations and other symptoms. People are advised to avoid outdoor activities.

Appendix C: First Stage Regression Results

Table 5 First stage results for pollution in the wages equation (Table 2)

Table 6 First stage results for house prices in the wages equation (Table 2)

Table 7 First stage results for pollution in the house prices equation (Table 3)

Table 8 First stage results for wages in the house prices equation (Table 3)