Does Air Pollution Affect Consumption Behavior? Evidence from Korean Retail Sales*

• Date: 01 September 2019
• Author: Hyunduk Suh,Jongmin Yu,Hyunju Kang
• Published date: 01 September 2019
• DOI: http://doi.org/10.1111/asej.12185

Does Air Pollution Affect Consumption

Behavior? Evidence from Korean Retail Sales*

Hyunju Kang, Hyunduk Suh and Jongmin Yu

Using monthly regional panel data on air quality and large retail store sales in

Korea, we empirically examine the effect of air pollution on retail sales. We

account for regional heterogeneity in air pollution and control for various macro-

economic and climatic factors that can affect retail sales. We also use the air qual-

ity indicator in the west coastal islands (affected by trans-border pollution but

uncorrelated with the economic activity in the mainland) as an instrumental vari-

able. The estimation results show that, in general, 1 additional day of PM

10

level

higher than 80 μg/m

3

reduces monthly retail sales by approximately 0.1 percent.

Nonetheless, an adaptive pattern emerges over time, particularly when the level of

air pollution in the previous month was severe.

Keywords: air pollution, PM

10

, consumption, large store retail sales, adaptation.

JEL classification codes: E21, E60, Q53.

doi: 10.1111/asej.12185

I. Introduction

Air pollution is hazardous to human health, and people’s awareness of this has

grown over time. Bad weather conditions, as well as bad air quality, measured

by concentrations of materials such as particulate matter (PM), SO

2

,NO

2

and

O

3

, can affect individual behavior. It is also plausible that severe air pollution

makes people averse to going outside to buy goods and services, therefore

affecting consumption behavior. The aim of the present study is to empirically

investigate whether high levels of PM have adverse effects on consumption. PM

is floating dust in the air, usually generated by burning fossil fuels such as coal

and oil, as well as being present in exhaust gases from factories and automo-

biles. PM stands out from other types of local environmental problems such as

*Kang: Korea Capital Market Institute, 143 Uisadangdaero, Yeongdengpo-gu, Seoul 07332,

Korea. Email: hjkang326@gmail.com; Suh (corresponding author): Department of Economics, Inha

University, 100 Inharo, Michuhol-gu, Incheon 22212, Korea. Email: hsuh@inha.ac.kr. Yu: Depart-

ment of Economics, Hongik University, 94 Wausan-ro, Mapo-gu, Seoul 04066, Korea.

Email: yucono@hongik.ac.kr.

We are grateful to Hyomi Jang for excellent research assistance. This work was supported by the

2019 Hongik University Research Fund.

© 2019 East Asian Economic Association and John Wiley & Sons Australia, Ltd

Asian Economic Journal 2019, Vol.33 No.3, 235–251 235

soil or water pollution, because it spreads out to a much wider area than the lat-

ter types of pollution, with consequently greater effect on consumption. More-

over, it is different from greenhouse gases that are known to cause climate

change-related disasters, because while PM can affect individual behavior even

in a short period of time, the latter do not. Compared to other types of air pollut-

ants such as SO

2

,NO

2

,O

3

,SO

x

and NO

x

, PM is more distinctive because it is

visible and individuals may be more sensitive in reacting to it.

The existing literature lacks research that directly relates air pollution to con-

sumption behavior. However, the effects of other atmospheric factors such as

temperature, snow and rainfall on specific consumer items have been studied

extensively, following the work by Steel (Steele, 1951), which initially suggested

that weather can affect store sales. Maunder (1973) shows that weekly national

sales are influenced by the national level weather index, and consumption is par-

ticularly affected by abnormal temperature and precipitation, whose effects vary

across seasons. Starr-McCluer (2000) points out that weather affects the national

level of retail sales. However, this effect is only significant with monthly fre-

quency data, while it is nonsignificant when longer data frequencies are consid-

ered. This is because people can catch up on delayed spending due to bad

weather within the next couple of months, once weather conditions improve.

Parsons (2001) uses daily sales data to measure the impact of precipitation fre-

quency and maximum temperature, highlighting that daily consumer decisions

can be influenced by climatic factors. Bahng and Kincade (2012) show that high

temperature variation influences seasonal purchases in the Korean women’s

clothing market. Bertrand et al. (2015) distinguish between seasonal and abnor-

mal weather factors, and use them to examine the impact of weather on

national-level apparel sales in France. Tran (Roth Tran, 2016) confirms the

impact of weather on outdoor or sporting goods brands, which are particularly

sensitive to climatic factors. As for economic activities other than consumption,

Bloesch and Gourio (2015) support the view that weather has a significant,

though short-lived, effect on economic activities such as production. Boldin and

Wright (2015) propose a methodology that controls for the effect of weather

such as temperature, snowfall and hurricanes on national nonfarm payroll

employment.

Regarding the effect of air pollution represented by PM on socioeconomic

activities, Ebenstein et al. (2016) find that a higher level of PM

2.5

exposure dur-

ing test exams can cause a significant decline in test scores. Heyes et al. (2016)

measure the effects of air pollution (PM

2.5

) in the Manhattan area on investment

returns in the New York Stock Exchange. Li et al. (2017) highlight the impact

of air pollution on labor outflow and labor migration in China. They find that

labor forces in educated, male and younger groups are more sensitive to air

pollution.

In this study, we aim to find empirical evidence of the channel through which

bad air quality weakens consumption activities. Air pollution is usually more

severe in urban than rural areas, and is directed by air flow. Therefore, it tends

ASIAN ECONOMIC JOURNAL 236