Income and social communication: The demographics of stock market participation

• Author: Ming Gao,Longkai Zhao,Juanjuan Meng
• Published date: 01 July 2019
• DOI: http://doi.org/10.1111/twec.12777
• Date: 01 July 2019

ORIGINAL ARTICLE

Income and social communication: The

demographics of stock market participation

Ming Gao

1

|

Juanjuan Meng

2

|

Longkai Zhao

2

1

School of Economics, Peking University, Beijing, China

2

Guanghua School of Management, Peking University, Beijing, China

KEYWORDS

disposable income, social communication, stock market participation

1

|

INTRODUCTION

China's stock market has experienced tremendous growth over the past 20 years. It is now the lar-

gest market among developing countries and has significant impact on international finan cial mar-

kets. Compared to a more mature market, the nascent development stage of the Chinese market

provides a unique opportunity to analyse people's participation decisions when a society as a whole

encounters the market.

This paper systematically evaluates the determinants of the new investors' participation deci-

sions using the officially compiled account opening data at a monthly frequency and on the pro-

vince level. Different from the literature that often focuses on one or a few dimensions, we

evaluate the relative importance of as many dimensions as possible: disposable income, demo-

graphic variables, macroeconomic factors, stock market conditions, and social communication

measures on both the level and the change of the participation decision. In addition to the stan-

dard variables, we introduce various social communication measures including the number of text

messages and catering enterprises in a given province and stock market information gained

through social communication. These measures are important because the Chinese stock market is

dominated by retail investors,

1

whose information sources are very limited, and investing in the

stock market is likely to be heavily influenced by social interactions in developing countries such

as China, in which the operation of the society hinges on social relationships. As opposed to a

more mature market, the rapid development of an economy and the stock market in developing

countries attracts investors who possess little prior knowledge of the stock market (Bekaert, Har-

vey, & Lundblad, 2001), and social communication is an important channel for information

exchange.

We present several key findings: first, the level of participation rate is predominately deter-

mined by the income factor, followed by various measures of social communication. Second,

1

According to the China Capital Market Development Report by the China Securities Regulatory Commission, the propor-

tion of retail investors whose total value of stock holdings and cash balances were below 100,000 RMB in their security

account was 81% (90%) in the Shanghai (Shenzhen) Stock Exchange in 2007, respectively.

Received: 28 March 2017

|

Revised: 10 November 2018

|

Accepted: 15 December 2018

DOI: 10.1111/twec.12777

2244

|

© 2019 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/twec World Econ. 2019;42:2244–2277.

social communication plays the most important role in the growth of the participation rate, acting

as a multiplier to stimulate stock market participation. These effects are more pronounced in high‐

income, high‐education, high‐population‐density groups and during the bull market period. Third,

to understand how omitted variable bias can possibly affect the results, we compare the regressions

of a single dimension to that of all combined dimensions. We find that most of the estimates con-

cerning the proportion of college graduates become insignificant in the joint regressions. Market

conditions such as past returns and turnover still keep the results significant, but they become

unstable. Among macro factors, interest rate shows consistent and significant negative results in

the change in the participation rate. These results suggest that in analysing participation decisions,

omitted variable biases can be a severe concern.

This study has several differences from the literature. First, the literature often discusses demo-

graphic factors, social communication factors and other factors separately. In this study, we com-

bine these channels to evaluate the relative importance of each factor. By comparing regressions

that use only one dimension and that combine all dimensions, we can understand how possible

omitted variable bias affects the results. Second, the broad existing literature on household partici-

pation behaviours primarily relies on individual‐level survey data from the United States and

Europe (Campbell, 2006; Cole & Shastry, 2009; Guiso, Haliassos, & Jappelli, 2003; Vissing‐

Jorgensen, 2003), which are often on a small scale and are potentially subject to sample‐selection

issues and response inaccuracies. Instead, our paper relies on officially compiled participation data

from China aggregated at the province level. The data cover the whole country and accurately cor-

respond to the size of the population. By being aware of both the advantages and disadvantages of

the aggregate data, we provide complementary evidence to the literature by showing that income

and social communication factors that prove significant at the individual level also survive the

aggregation and have robust macro impact. We believe this complementary evidence has indepen-

dent value. Third, compared to the previous studies that focus on developed markets, the analysis

in this paper provides evidence from the developing markets context so that we are able to under-

stand households' participation decisions when the entire society encounters a stock market on a

large scale. In this case, the finding of social communication contributes new knowledge to the

literature.

Our paper is related to the literature that investigates stock market participation. Household

wealth has been the focus of many recent papers, yet the findings have not been conclusive. Wach-

ter and Yogo (2010) develop a model with the prediction that the portfolio share in risky assets

rises with wealth. Carroll (2002), and Calvet and Sodini (2014) show that higher‐income house-

holds are associated with less risk aversion and accordingly allocate a higher share of their wealth

to stocks. Lynch and Tan (2011) predict that higher labour income will result in lower stock mar-

ket participation because labour income risk is negatively correlated with stock returns. On the

contrary, Cocco, Gomes, and Maenhout (2005) suggest that labour income is similar to holding

safe assets, and therefore, households with higher labour income will invest more in risky assets.

Some papers also explain why participation increases with wealth from the perspective of fixed

participation cost (Haliassos & Bertaut, 1995; Vissing‐Jorgensen, 2003). Empirically, using an

international market survey, Guiso, Haliassos, and Jappelli (2000) find that stock market participa-

tion is correlated with wealth in some markets, yet in Italy and Germany, about half of wealthy

households have no direct or indirect stock holdings.

A growing body of studies also focuses on the relation between social communication and

stock market participation. For instance, Hong, Kubik, and Stein (2004) show that households are

more likely to participate in the market when they are more active in their community. Similarly,

Li (2014) finds that investment knowledge and experience are shared among extended family

GAO ET AL.

|

2245

members. Gao and Fok (2015) show that households with strong family and social interaction are

more likely to save, invest in risky assets and borrow. Liang and Guo (2015) find that Internet

access mitigates the influence of social interaction on stock market participation . Meanwhile, Kaus-

tia and Knüpfer (2012) show that recent stock returns of local peers affect stock market participa-

tion in Finland. Our results show that the social communication effect is preserved even when

individual decisions are aggregated at the province level.

The paper is organised as follows. Section 2 discusses the institutional background. Section 3

introduces the data and presents the summary statistics. Section 4 provides the empirical results

and robustness checks. Section 5 concludes.

2

|

INSTITUTIONAL BACKGROUND

The Shanghai Stock Exchange and the Shenzhen Stock Exchange, the two major stock exchanges

in China, were both established in December 1990. Since then, the market has experienced tremen-

dous growth and development. According to the WIND database, the number of listed companies

reached 2,827 at the end of 2015—up from only 10 companies in the early 1990s—with a total

market capitalization of US$ 8.16 trillion. One important difference between the Chinese stock

market and the US stock market is that retailed investors dominate the Chinese market, whereas

the major players in the US market are institutions.

However, China's economy has large regional differences. For instance, coastal province s attract

far more foreign investments than inland provinces because of China's export‐oriented development

strategy. China is also facing a widening income gap. According to the National Bureau of Statis-

tics, the Gini coefficient of 2013 reached 0.473, which suggests relatively large income inequality.

Different from developed economies, the great heterogeneity of the Chinese economy and demo-

graphics across provinces provide an excellent opportunity for us to investigate the determinants of

households' stock market participation decisions.

It is important to note that the 2005–07 split‐share structure reforms occurred during our sample

period of 2005–10. Before the reforms, approximately two‐thirds of domestically listed A‐shares

were government‐owned and not tradable; after the reforms, these shares were publicly tradable.

The reforms and the resulting market expansion are closely related to China's 2008 market bubble.

Many investors during this period were attracted to the market and the resultin g rapid growth of

participation provides valuable variation for analysing participation decisions. In heterogeneous

analysis, we also distinguish between the reform period and post‐reform period via bull and bear

market types to understand whether our results are robust to this event.

3

|

DATA DESCRIPTION AND SUMMARY

3.1

|

Data description

The empirical analysis relies on monthly panel data from January 2005 to December 2010, aggre-

gated at the province level.

2

Our data consist of three parts: account opening data, province‐level

characteristics and stock market data.

2

The province‐level participation data after 2010 are adjusted for which provinces the brokerage houses belonged to based

on the new changes. As a result, the data before and after 2010 are inconsistent in terms of the number of total participants

in each province. We therefore use the data before 2010 for the sake of consistency.

2246

|

GAO ET AL.