A Study on the Factors Influencing Household Consumption from a Money Demand Perspective: Evidence from Chinese Urban Residents

Abstract

Based on the classic Keynesian theory of money demand and city panel data, in this study, we investigate the impacts of different money demand motives on urban household consumption in China, and provide ideas for promoting sustainable growth in household consumption in China. The results of this study show the following: First, in general, the theory of money demand motivation can adequately explain household consumption in large and medium-sized cities in China. Second, the CPI time series has a significant adverse effect on the real money demand of most households. Third, residents significantly reduce food consumption to satisfy speculative money demand for financial instruments, and the lower the income level, the keener households are to invest in high-risk stocks. Fourth, even for high-income households, the precautionary money demand generated via the purchase of commercial insurance still has a significant crowding-out effect on their total consumption. Social security spending, which reduces the precautionary money demand of households, has a much more positive impact on high-income households.

1. Introduction

According to the composition of GDP, the growth of a country’s economy can be driven by exports, investment, and consumption. China focused heavily on the role of exports as a driver of the economy until 2008. After that, given the serious impact of the sub-prime crisis on China’s export demand, the focus of China’s economic growth strategy shifted to be driven by consumption. In 2021, official World Bank figures showed that, among the world’s major economies in the same period, China’s final consumption as a share of GDP was 54.29%, compared to 82.59% for the United States, 75.25% for Japan, and 71.39% for Germany. It can be seen that the role of Chinese consumption in driving the economy is much lower than that of the world’s major developed countries, and it also shows that China still suffers from a serious lack of domestic demand at this stage. In order to strengthen its own economic resilience and achieve long-term economic growth, it is necessary for China to stimulate domestic consumption and raise the consumption awareness of its residents.

Consumption theory shows us that income is the most important factor influencing consumption. People’s reasons for wanting money may be divided into three categories under Keynes’ idea of liquidity preference from 1936: transactional, precautionary, and speculative motives [1]. Transaction-motivated money demand refers to people’s need to hold a portion of money to cope with daily transactions. Precaution-motivated money demand refers to people’s motivation to hold money in order to cope with uncertainty. Speculation-motivated money demand can be understood as people’s need for money for investment purposes. After the publication of Keynes’ theory of money demand, Friedman’s new quantity theory considered the demand for money as the real demand for money according to price levels [2]. Friedman also argued that household wealth consists of assets such as money, bonds, stocks, and consumer durables, thus expanding the fungible assets for speculative money demand from bonds alone to include bonds, stocks, and other real assets [3]. The concept of household consumption studied in this study is transaction-motivated money demand. Under conditions of a certain level of income in the population, an increase in the money demand in the categories of precaution- and speculation-motivated demand will reduce people’s transaction-motivated demand. In other words, an increase in the population’s precautionary and speculative motivations reduces the amount of household consumption.

This research selects appropriate variables from the perspective of money demand and uses a panel model to study demand motivation on residential consumption in 34 large and medium-sized cities in China, with a view to elucidating the characteristics of Chinese residents’ consumption in terms of the motivation underlying money demand, providing useful ideas for realizing sustainable growth in the scale and quality of China’s residents’ consumption, and thus supplying a stable foundation for the sustainable development of China’s economy as a whole. In general, the empirical findings show that the theory of the money demand motive can more than adequately explain household consumption in large and medium-sized cities in China. However, as the economic level increases, there is a possibility that the strength of the monetary demand motive in explaining consumption decreases. In addition, the strength of the monetary demand motive in explaining clothing is typically relatively small. During the period 2007–2019, there were only a small number of households where the CPI time series did not have an impact on the real money demand. Of greater concern is the significant adverse impact of price fluctuations on the consumption of health care and food for low-income households in that category. There is also the fact that the effect of financial products on the speculative money demand of households in eastern cities is much lower than that of households in central and western cities. Even for affluent households in large and medium-sized Chinese cities, precautionary money demand in the form of commercial insurance premiums can significantly squeeze their household consumption. Finally, the speculative money demand effect for real estate and the precautionary money demand effect for government social security spending is significantly affected by extreme values.

The remainder of this paper is structured as follows: Section 2 reviews the relevant literature. Section 3 and Section 4 describe the empirical methodology used. Section 5, Section 6 and Section 7 provide the empirical results. Section 8 discusses the robustness of the empirical results. Section 9 provides a discussion and conclusions.

2. Literature Review

Building on the research on consumption structure [4,5,6,7] and consumption inequality [8,9,10,11,12], there is a large body of work in the literature that examines the factors influencing people’s consumption.

2.1. Influence of Factors Motivating Transactions on Consumption

According to the related theories, the transactional motivation of money demand is mainly affected by income. There are a substantial number of studies in the literature on the effect of income on consumption. Examples include the Keynesian theory of absolute income consumption [1], Duesenberry’s theory of relative income consumption [13], Modigliani’s theory of life-cycle consumption [14], and Friedman’s theory of perpetual income consumption [15]. However, in recent years, a few scholars have used direct income data to study the impact on consumption; there are an increasing number of studies on the effects on consumption of various policies or shocks affecting income. For example, Agarwal and Qian [16] studied the effect of the growth dividend program announced by the Singapore government on household consumption. Rodriguez [17] examined the effect of a new welfare system targeting public pensions on household spending in Spain. Alonso [18] studied the effect of a rising minimum wage on the consumption of nondurable goods. Moreover, others have studied the impact of fiscal stimulus–tax rebates [19,20] as well as transfer payments [21,22] on household consumption. In addition to investigating the effect of income stimulus on consumption, some of the literature investigates the impact of various income disruptions on consumption. For example, Kueng [23] investigated the impact of Alaska Permanent Fund dividends on consumption, and Fagereng et al. [24] used large lottery rewards to investigate the effect of temporary income disruptions on family spending and saving. Barrett and Brzozowski [25] and Lepage-Saucier [26] studied the impact of labor market migration shocks, such as involuntary retirement and temporary layoffs, on food expenditure. As previously stated, the literature has investigated the effect of income on consumption from a variety of viewpoints. The results all point to the same conclusion: a positive revenue boost encourages people to increase their spending levels. In other words, the increase in income stimulates the transactional money demand of the population, which in turn raises the consumption level of the population.

The incentive to trade in money is also affected by price levels if we start from the actual demand for money. Taylor [27], Lieb, and Schuffels [28], and Binder and Brunet [29] examined the impact of inflation or inflationary expectations on residential consumption.

2.2. Influence of Factors Motivating Speculation on Consumption

Keynes considered the alternative asset to money to be bonds, the proxy variable of which is the interest rate. This means that residents would convert a portion of their currency holdings into bonds, thus satisfying the speculation-motivated demand for money. Subsequently, Friedman extended this alternative asset to equities and other tangible assets such as real estate.

The existing literature on the impact of investing in real estate on household consumption has developed in two ways. One is to study the impact of changes in housing wealth [30,31,32,33] on household consumption. Their findings are consistent, suggesting that a rise in household housing wealth has a significant positive effect on consumption. The second way is to study the impact of changes in house prices on household consumption. The empirical findings in the literature suggest that the effect of house prices on consumption does not show uniformity. Some scholars have argued that the impact of house price increases on household consumption is neutral and that it is co-causal; for example, economic growth leads to simultaneous increases in house prices and household consumption [34]. Some scholars believe that rising house prices can enhance household consumption [35,36,37]. There are also some findings in the literature that show that rising house prices have a significant negative impact on household consumption [38,39]. In terms of speculative motivation, scholars have not only studied the impact of real estate on household consumption but also much literature has examined the impact of investing in financial instruments on household consumption; for example, studying the impact of financial wealth on household consumption [40,41,42,43].

2.3. Influence of Factors Preventing Motivation on Consumption

Keynes argued that precautionary money demand is mainly influenced by income. However, more of the literature now examines the impact of products with preventive effects on household consumption, such as insurance. Studies have shown that the precautionary nature of insurance can boost household consumption. For example, Janzen and Carter [44] found that an innovative microinsurance program had a smoothing effect on the consumption of rural households in Kenya. Zou et al. [45] and He et al. [46] found that participation in old-age insurance and health insurance played a significant positive role in household consumption. Bai and Wu [47] concluded from their study that the new cooperative health insurance scheme in China increased rural households’ non-medical consumption.

The development of the Internet has changed the way people shop and pay, and there has been a relatively large number of studies in recent years examining the impact of informatization and digitization on household consumption; for example, studying the impact of Internet use [48,49], mobile and electronic payments [50,51], financial technology [52], and the digital divide [53] on household consumption. This impact on household consumption due to technological innovations is difficult to link directly to any kind of money-demand motive.

2.4. Research Gaps

The literature has focused on a particular motivation for money demand in the content section of this article, such as the precautionary saving motivation and the speculative motivation, and does not visualize this research perspective. Since the existing research does not focus on the impact of money demand motivation on household consumption, the indicators it selects do not fully fit the motivation for money demand. For example, the use of housing wealth and financial wealth as proxies for speculative money demand is inaccurate and reflects more of an income effect. In addition, the study of abnormal price fluctuations alone does not provide a good indication of the actual demand for money by the population. Based on this, our study reflects the following characteristics. This research puts the money demand motive at the center, completely covering the three money demand motives and selecting the corresponding indicators strictly according to the money demand motives. In this study, we used price time series to highlight the actual money demand in the theory of money demand; used the yield of financial instruments, house prices, and the house price–income ratio instead of the amount of wealth to study the impact of speculative demand on household consumption; and used the actual insurance premium paid by the residents, instead of whether to buy insurance and the number of insurance claims, to study the impact of precautionary money demand on household consumption.

3. Research Methodology

3.1. Sample Selection

The initial sample for this study was selected from 35 large and medium-sized cities in the Chinese real estate market. These 35 cities are mainly composed of provincial capitals in the eastern, central, and western regions of China. These cities are the most economically developed and dynamic cities in each region, and the results of this study using these cities as a sample are highly representative and can be used to illustrate and speculate on the current consumption status of urban residents in various regions of China. Based on the data collection, 34 large and medium-sized cities were finally identified. In total, 440 annual pieces of sample data from 2007 to 2019 were collected. Among these 34 large and medium-sized cities, there are 29 provincial capitals; according to the economic zone, there are 8 western cities, 9 central cities, and 17 eastern cities.

The data used in this study were mainly sourced from the statistical yearbooks of the 34 large and medium-sized cities, the China Financial Yearbook, the China Insurance Yearbook, the China Real Estate Statistical Yearbook, the Shanghai Stock Exchange, and the Wind database. We obtained household income, consumption, price index, and government social security expenditures from urban statistical yearbooks and the Wind database; we compiled data on commercial insurance premium expenditures of urban households from the China Financial Yearbook and the China Insurance Yearbook. Housing prices were obtained from the China Real Estate Statistical Yearbook; the data on financial instrument returns were obtained from the public data of the Shanghai Stock Exchange.

3.2. Empirical Models

The Chow (F) test showed that the panel data for all variables should reject the POOL model. In addition, the speculative money demand variables, Treasury yields, and stock return indicators use nationally uniform data, which means that each individual has the same time series. If the time effect of the panel data is considered, two years of data will be removed and have an impact on the significance of the variable. In China, bank savings, the stock market, and the housing market are the main investment options for Chinese households due to the existence of strict capital controls [38]. In addition, the flexibility and convenience of investing in financial instruments cannot be replaced by investing in housing. Therefore, financial instruments are indispensable variables in the research for this study. Considering the characteristics of consumption, residents’ consumption is much more affected by individual effects than by changes over time. For all the above reasons, this study only considered the individual effects of household consumption in the course of this study. The Hausman test methodology shows that some variables support the use of individual fixed effects and some variables support the use of individual random effects. The main objective of our study is to obtain the estimated parameters of the model, and to make this article more compact; therefore, the following panel fixed effects model is used throughout this research:

$$y_{it}=x_{it}^{\prime }\beta +z_t^{\prime }\delta +u_i+{\epsilon }_{it}\left(i=1,\cdots ,n;t=1,\cdots ,T\right),$$

(1)

where $i$ denotes the region, $t$ denotes the time, $y_{it}$ is the explanatory variable, and $z_t$ represents the time-characteristic explanatory variables that do not vary with individuals. Finally, $x_{it}$ refers to the explanatory variables that can vary between individuals and over time. The unobservable random variable $u_i$ represents the intercept term of individual heterogeneity.

4. Variable Definitions

4.1. Independent Variable

The explanatory variables of this study are total non-housing consumption of households and six sub-consumptions, including food, tobacco, and alcohol; clothing; household goods and services; health care; transport and communication; and education, culture, and entertainment. All the explanatory variables are logarithmically treated in the empirical evidence (see

Table 1. The descriptions of the variables.

Variable Name	Variable Symbol	Definition
Non-housing consumption	c	Total consumption excluding residential expenses
Food, tobacco, and alcohol	f&t&l_c	Expenditure on various foodstuffs and tobacco and alcohol
Clothing	c_c	Expenditure related to residents’ clothing, including clothing, clothing materials, footwear, other clothing and accessories, and clothing-related processing services
Household facilities, articles, and services	d_c	All kinds of household and personal living goods and household services, including furniture and interior decorations, household appliances, home textiles, household sundries, personal effects, and household services
Health care and medical services	m_c	The total cost of drugs, supplies, and services used for medical and health care
Transportation and communication	t&c_c	Expenses for transportation and communication tools and related various services, maintenance and vehicle insurance, and others
Education, culture, and recreational activities	e&c&r_c	Expenditures on education, culture, and recreation.

and

Table 2. The descriptive statistics of the variables.

Variable	N	Mean	Std. Dev	Min	Max
lnc	440	9.737	0.334	8.803	10.464
lnf&t&l_c	440	8.776	0.310	7.763	9.575
lnc_c	440	7.472	0.326	5.757	8.094
lnd_c	440	7.159	0.438	5.720	8.808
lnm_c	440	7.190	0.423	6.071	8.986
lnt&c_c	440	7.905	0.469	6.648	8.915
lne&c&r_c	440	7.788	0.435	5.901	8.904

for details).

As can be seen from Table 2, the fluctuation of development and enjoyment consumption is much higher than the fluctuation of basic consumption such as clothing and food during the sample period. It indicates that the enjoyment-oriented consumption of Chinese large and medium-sized urban households has grown rapidly between 2007 and 2019, and the consumption structure of Chinese large and medium-sized urban households has been enhanced.

4.2. Dependent Variables

Based on the above theory, in this study, we select disposable income per capita and the price index as factors influencing transaction-motivated money demand, Treasury bond yields, stock yields, and home prices as factors influencing speculation-motivated money demand, and government social security spending per capita on education and health care and commercial insurance premiums per capita as factors influencing precaution-motivated money demand. Considering the accessibility and applicability of the data, the statistics of commercial insurance premium expenditure in this research are adopted from commercial life insurance premium expenditure data. In the empirical evidence, disposable income per capita, housing prices, social security expenditure per capita, and commercial insurance premiums per capita are logarithmically treated. The details are shown in

Table 3. The classification of the explanatory variables.

Money Demand Motivation	Variable Name
Transaction motive	Disposable income per capita
	Price index
Speculative motive	Treasury bond yields
	Stock yields
	Home prices
Precautionary motive	Commercial insurance premiums per capita
	Social security spending per capita

and

Table 4. The descriptive statistics of the variables.

Variable	Sample	N	Mean	Std. Dev	Min	Max
Log of disposable income per capita	lnper_inc	440	10.266	0.417	9.237	11.210
Price index of non-housing consumption	t_cpi	440	2.767	1.704	−2.661	8.404
Price index of food, tobacco, and alcohol consumption	f&t&l_cpi	440	5.756	4.601	−2.100	21.700
Price index of clothing consumption	c_cpi	440	1.073	3.097	−13.663	11.569
Price index of consumer of household goods and services consumption	d_cpi	440	1.496	1.832	−4.400	9.300
Price index of healthcare consumption	m_cpi	440	2.887	2.789	−2.400	19.260
Price index of transportation and communication consumption	t&c_cpi	440	−0.643	1.955	−6.762	11.293
Price index of educational, cultural, and entertainment consumption	e&c&r_cpi	440	0.901	2.080	−6.000	7.821
Log of urban commercial housing residential prices	lnhp	440	8.924	0.585	7.641	10.929
Treasury bond yields	bond	440	3.665	2.471	−0.464	9.401
Stock yields	stock	440	9.575	42.756	−65.375	96.136
Log of per capita commercial life insurance premium expenditure	lnper_ins	440	7.621	0.882	5.429	10.187
Log of per capita social security benefits	lnper_sec	440	7.742	0.716	5.973	9.857

Notes: The treasury bond yields and stock yields are time-dependent explanatory variables that do not vary between individuals, and the rest are explanatory variables that vary between individuals and with time.

.

According to the results of the treatment in Table 4, comparing the price level fluctuations, it can be concluded that the price fluctuations of clothing, transport and communication, and education, culture, and recreation are relatively high, whereas the price fluctuations of foodstuffs and health care are relatively small.

5. Empirical Analyses

In this study, we process all the sample data via the panel fixed effect model.

Table 5. The regression results of the fixed effects model.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	lnc	lnf&t&l_c	lnc_c	lnd_c	lnm_c	lnt&c_c	lne&c&r_c
lnper_inc	0.885 ***	0.598 ***	0.561 ***	1.291 ***	1.143 ***	0.915 ***	1.180 ***
	(0.098)	(0.104)	(0.112)	(0.264)	(0.234)	(0.192)	(0.144)
Price index	−0.008 ***	−0.001	0.004	−0.005	0.005	0.009	−0.003
	(0.003)	(0.001)	(0.003)	(0.005)	(0.004)	(0.006)	(0.005)
bond	−0.008 ***	−0.006 **	−0.009 **	−0.010 ***	0.005	−0.009	−0.009 **
	(0.002)	(0.002)	(0.004)	(0.003)	(0.003)	(0.007)	(0.004)
stock	−0.001 ***	−0.001 ***	−0.001 ***	−0.001 ***	0.000	−0.000	−0.000 **
	(0.000)	(0.000)	(0.000)	(0.002)	(0.000)	(0.000)	(0.000)
lnhp	−0.036	−0.002	−0.027	−0.073	0.058	−0.141	−0.128
	(0.047)	(0.043)	(0.060)	(0.098)	(0.112)	(0.093)	(0.073)
lnper_ins	−0.074 ***	−0.074 **	−0.194 ***	−0.094	0.049	−0.007	−0.002
	(0.026)	(0.029)	(0.035)	(0.071)	(0.049)	(0.047)	(0.039)
lnper_sec	−0.023	0.075	0.166 **	−0.108	−0.303 **	0.101	−0.117
	(0.041)	(0.045)	(0.074)	(0.094)	(0.119)	(0.078)	(0.079)
_cons	1.778 ***	2.670 ***	2.178 ***	−3.845 **	−3.121 **	−0.922	−2.226 **
	(0.491)	(0.498)	(0.623)	(1.416)	(1.172)	(0.975)	(0.856)
Regional fixed effect	YES	YES	YES	YES	YES	YES	YES
adj. R2	0.92	0.89	0.63	0.77	0.76	0.78	0.83
N	440	440	440	440	440	440	440

Notes: the clustering robustness criteria errors are in parentheses; ** and ***, respectively, indicate significance at the 0.05 and 0.01 levels; R² represents the corrected goodness-of-fit value.

shows the regression results of the fixed effects model.

Based on the parameter estimation results exhibited in Table 5, the following conclusions about the average consumption situation in 34 large and medium-sized cities in China can be drawn: (1) Disposable income and price level are variables that affect the demand for transactional money. There are significant positive effects of disposable income on both total consumption as well as categorical consumption, similar to the findings from the literature on the effect of income on consumption. This also implies that an increase in disposable income will enhance residents’ transactional demand for money, which in turn raises the amount of household consumption. The increase in disposable income has a higher degree of impact on development and enjoyment consumption and a lower degree of impact on the consumption of the most basic daily expenses of food and clothing. Looking specifically at the income elasticity of consumption, we find that an increase in disposable income has an impact on development and enjoyment consumption in the range of 0.915–1.291, whereas the impact on consumption of basic daily expenditure on food and clothing is 0.598 and 0.561, respectively. This shows that people are gradually moving toward higher levels of consumption. However, it is important to keep in mind that basic consumption still has a lot of room for improvement in basic household consumption in China’s 34 large and medium-sized cities. Over the whole sample, fluctuations in the price index only had a significant effect on total non-housing household consumption, with a value of −0.008. The results showed that the increase in the price level reduced the transactional money demand of the population, which in turn affected the amount of expenditure by the population on total household consumption. (2) Variables of speculative money demand are Treasury yields, stock yields, and housing prices. The effect of Treasury bond yields and stock yields on total non-housing household consumption was −0.008 and −0.001, respectively, and both were highly significant. In terms of sub-consumption, when Treasury bond and stock yields rise, residents significantly reduce the amount of money spent on food, clothing, household goods and services, education, culture, and entertainment, and increase the amount of money they spend on speculation in the Treasury bond and stock markets. Although the effect of government bond yields and stock yields on household health care consumption is 0.005 and 0.000, showing a positive effect, this positive effect is not significant. The above results suggest that in 34 large and medium-sized cities in China, on average, financial investment shows a strong crowding-out effect on both total and sub-consumption of household non-housing expenditure, and also suggest that a booming financial market can stimulate residents’ desire to speculate. However, the findings of the existing literature suggest that financial products have a significant wealth effect on household consumption, similar to the effect of income on household consumption. The main reason for the discrepancy in the findings, according to the authors, is the use of different indicator data. The existing literature uses the number of financial assets owned by households or the amount of return on financial assets, exploring the impact of different sizes of financial assets [40,41,43] or increases [42] on household consumption. In contrast, in this study, we use an indicator of financial market returns and explore the impact of rising and falling yields on financial products on the speculative psychology of the population. In general, residential price increases have a negative effect on total consumption (−0.036) and other consumption categories, apart from having a positive effect on the amount of household health care consumption (0.058), but all effects are insignificant. The empirical results indicate that the increase in housing prices showed neither a wealth effect [35], nor a precautionary saving effect [39], or speculative effect [38]. Speculative money demand for real estate shows a neutral effect on Chinese household consumption. This can be explained by the following reasons. Shared causality drives both house prices and Chinese household consumption growth [34]. In the alternative, the positive and negative wealth effects of house prices over the sample period offset each other, thus exhibiting a neutral effect of house prices on household consumption. That may also be due to the effect of extreme data in the sample data. (3) Expenditure on commercial life insurance premiums and government expenditure on social security are precaution-motivated money demands. Commercial life insurance premiums are a household expense, which boosts the population’s precautionary demand for money, whereas social security premiums are a government subsidy to households for education and health care, reducing the population’s precautionary demand for money. It can be seen from Table 5 that the effect of commercial life insurance premiums on food, clothing, and total non-housing consumption is significant, with values of −0.074, −0.194, and −0.074, respectively. The empirical results suggest that residents need to reduce their consumption of food and clothing, and thus reduce their total consumption in order to meet the need to pay for commercial insurance premiums. This study utilized commercial insurance premiums to derive a significant crowding out effect of precautionary money demand on household consumption, whereas the existing literature uses the presence or absence of insurance [45,46,47] and the number of insurance claims [44] to conclude that the precautionary feature of insurance increases household consumption. For government spending on social security, theoretically, it has a crowding-out effect on the consumption of targeted subsidies; on the other hand, it has an income effect on other consumption. The results show that there is a significant crowding-out effect of social security payments by the government on the amount of health care consumed by households, with a value of −0.303, and a non-significant negative effect on total non-housing consumption by households, with a value of −0.023. It can be said with certainty that the government’s targeted spending on health care has played a targeted role in 34 large and medium-sized cities in China. However, there may also be cases where government spending crowds out private spending.

The treatment results reported in Table 5 are analyzed in terms of the row variables and the empirical results subsequently are further analyzed in terms of the column variables. The following can be determined from the results: (1) With the exception of clothing consumption (0.63), the regression relationship explains more than 70% of the variation in the dependent variables, especially for total consumption and food consumption, where the corrected R² value exceeds 0.85. It can be concluded that Keynes’s theory of money demand is still a strong theoretical guide to household consumption in contemporary China. (2) It is assumed that residents will move to a higher level of demand only after satisfying a lower level of consumption demand. When crowding-out effects occur, residents will reduce consumption that is already well-satisfied and consumption at a more advanced level. Based on these assumptions and comparison of coefficients and significance of each demand motivation factor, it can be found that China has sufficiently solved people’s food and clothing problems on the whole, whereas household goods and services, health care, and transportation and communication are still in the process of being satisfied. Moreover, education, culture, and entertainment consumption are in the early stages of satisfaction and have not yet had a substantial impact on residents’ lives.

6. Quantile Regression

To obtain a more comprehensive understanding of the status of the independent variables on the dependent variables, and eliminate the effect of extreme values on the results of the empirical treatment, in this study, we conducted a quantile regression operation based on panel data, and the results of the treatment are shown in

Table 6. Results of quantile model regression.

	lnc			lnf&t&l_c			lnc_c			lnd_c
	0.25	0.50	0.75	0.25	0.50	0.75	0.25	0.50	0.75	0.25	0.50	0.75
lnper_inc	0.805 ***	0.774 ***	0.745 ***	0.623 ***	0.473 ***	0.554 ***	0.507 ***	0.507 ***	0.578 ***	1.011 ***	1.114 ***	1.184 ***
	(0.077)	(0.034)	(0.012)		(0.037) (0.055)	(0.009)	(0.049)		(0.034) (0.007)	(0.081)	(0.092)	(0.031)
Price index	−0.005 ***	−0.003 ***	−0.001	0.001 ***	0.001	−0.000	0.003	0.004 ***	0.006 ***	−0.009 ***	−0.004 **	0.001
	(0.002)	(0.001)	(0.001)	(0.000)	(0.001)	(0.000)	(0.002)	(0.001)	(0.001)	(0.002)	(0.002)	(0.002)
bond	−0.007 ***	−0.007 ***	−0.007 ***	−0.005 ***	−0.002	−0.001	−0.005 ***	−0.007 ***	−0.017 ***	−0.007 ***	−0.011 ***	−0.011 ***
	(0.001)	(0.001)	(0.000)	(0.001)	(0.001)	(0.001)	(0.002)	(0.002)	(0.001)	(0.002)	(0.002)	(0.002)
stock	−0.001 ***	−0.001 ***	−0.000 ***	−0.001 ***	−0.000 ***	−0.000 ***	−0.001 ***	−0.001 ***	−0.001 ***	−0.001 ***	−0.001 ***	−0.000 ***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
lnhp	−0.042 ***	−0.087 ***	−0.064 ***	0.041	0.049	0.067 ***	−0.130 ***	−0.075 **	−0.236 ***	−0.132 ***	−0.077	−0.052 ***
	(0.014)	(0.024)	(0.020)	(0.031)	(0.042)	(0.006)	(0.045)	(0.035)	(0.014)	(0.043)	(0.045)	(0.017)
lnper_ins	−0.056 ***	−0.018 **	−0.011 ***	−0.074 ***	−0.030 **	−0.050 ***	−0.062	−0.094 ***	−0.044 ***	0.032	−0.048	−0.049 ***
	(0.020)	(0.008)	(0.003)	(0.008)	(0.012)	(0.004)	(0.034)	(0.029)	(0.005)	(0.028)	(0.054)	(0.008)
lnper_sec	0.029	0.054 ***	0.057 ***	0.048 ***	0.098 ***	0.077 ***	0.119 ***	0.121 ***	0.142 ***	0.002	−0.043	−0.111 ***
	(0.041)	(0.014)	(0.015)	(0.004)	(0.025)	(0.005)	(0.029)	(0.025)	(0.009)	(0.034)	(0.041)	(0.022)
N	440	440	440	440	440	440	440	440	440	440	440	440
Regional fixed effect	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
	lnm_c				lnt&c_c				lne&c&r_c
	0.25	0.50		0.75	0.25	0.50		0.75	0.25	0.50	0.75
lnper_inc	1.299 ***	1.204 ***		1.110 ***	0.788 ***	0.879 ***		0.920 ***	1.041 ***	0.977 ***	1.062 ***
	(0.036)	(0.058)		(0.055)	(0.025)	(0.033)		(0.068)	(0.070)	(0.046)	(0.011)
Price index	0.013 ***	0.002		0.006	0.001	0.004 **		0.007 ***	−0.000	0.002	0.001
	(0.001)	(0.002)		(0.003)	(0.003)	(0.002)		(0.002)	(0.001)	(0.002)	(0.001)
bond	0.011 ***	0.000		0.002	−0.013 ***	−0.011 ***		−0.016 ***	−0.010 ***	−0.007 **	−0.006 ***
	(0.002)	(0.002)		(0.002)	(0.003)	(0.002)		(0.002)	(0.001)	(0.003)	(0.001)
stock	0.000 ***	0.000		−0.000	−0.000 **	−0.000 ***		−0.000 ***	−0.000 ***	−0.000	−0.000 ***
	(0.000)	(0.000)		(0.000)	(0.000)	(0.000)		(0.000)	(0.000)	(0.000)	(0.000)
lnhp	−0.463 ***	−0.044		−0.001	−0.090 **	−0.071		−0.029	−0.064	−0.074 ***	−0.071 ***
	(0.008)	(0.133)		(0.047)	(0.039)	(0.056)		(0.044)	(0.039)	(0.012)	(0.017)
lnper_ins	0.048 ***	0.044 ***		0.097 ***	−0.012	−0.014		−0.040 ***	−0.002	0.017	−0.026 ***
	(0.010)	(0.017)		(0.014)	(0.017)	(0.011)		(0.014)	(0.015)	(0.024)	(0.003)
lnper_sec	−0.064 ***	−0.285 **		−0.326 ***	0.184 ***	0.108 ***		0.053 ***	−0.090	−0.073	−0.060 **
	(0.020)	(0.120)		(0.057)	(0.033)	(0.013)		(0.020)	(0.068)	(0.059)	(0.023)
N	440	440		440	440	440		440	440	440	440
Regional fixed effect	YES	YES		YES	YES	YES		YES	YES	YES	YES

Notes: The quantile regression is performed using the qregpd command. The clustering robustness criteria errors are in parentheses; ** and ***, respectively, indicate significance at the 0.05 and 0.01 levels.

.

Firstly, the empirical results of the transactional money demand variable are analyzed. From the results of disposable income, it can be seen that as the quartile rises, the strength of the influence of the increased motivation of transactional money demand on the amount of food consumption is 0.623, 0.473, and 0.554, respectively, which shows a tendency of decreasing firstly and then rising later on, which indicates that only a relatively small portion of households in 34 large and medium-sized Chinese cities have qualitative requirements for food consumption. In addition, as the quartile rises, the strength of the impact of disposable income on the consumption of household goods and services and transport and communications rises from 1.011 to 1.184 and from 0.788 to 0.920, respectively, suggesting that an increase in the motivation of residents’ transactional monetary demand will lead to a significant rise in the consumption of these two areas. The rising price level reduces the transactional money demand of the population, and the processing results show that the negative effect of the reduction of transactional motivation caused by the rise in price level on the total consumption is very significant at the 0.25 and 0.5 quartiles. In particular, it is important to note that, even if the price level rises, the coefficient of transactional money demand on food and health care consumption remains significantly positive at the 0.25 quantile, indicating that these households are still at the stage of consumption of food and health care necessities. Secondly, the empirical results of speculative money demand are analyzed. The coefficients of Treasury bond yields and stock yields are almost negative and significant at the quantile level, suggesting that households at low, medium, and high consumption levels try to reduce their consumption of all kinds so as to satisfy the speculative monetary demand for financial instruments. In terms of the total effect on consumption, the negative effect of stock returns on households at the low, medium, and high consumption levels decreases sequentially with 0.00062, 0.00055, and 0.00047, respectively. In contrast, the negative effect of Treasury yields on households at the low consumption level (0.0067) is lower than that of households at the medium (0.0071) and high (0.0071) consumption levels, indicating that households with lower income levels are keener on risky investment instruments. After eliminating the extremes through quantile regression, the empirical results show that the effect of rising house prices on total consumption has values of −0.042, −0.087, and −0.064, respectively. Thus, for households in 34 large and medium-sized cities in China, there is a strong demand for speculation in real estate regardless of income. The empirical results from the quantile regressions eliminating extreme values validate the findings of the existing literature [38,39], related to the fact that rising house prices significantly reduce the consumption of Chinese households. Finally, the empirical results of the precautionary money demand variable are analyzed. At the 0.25 quantile, commercial life insurance premiums have a significant crowding-out effect only on food consumption, with a value of −0.074. At the 0.50 quantile, commercial life insurance premiums have a significant crowding-out effect on food and clothing consumption, with corresponding coefficient values of −0.030 and −0.094. At the 0.75 quantile, commercial life insurance premiums have a significant crowding-out effect on food (−0.050), clothing (−0.044), household goods and services (−0.049), transport and communications (−0.040), and education, culture, and entertainment (−0.026). The above results indicate that, given that household consumption needs are met, households with high consumption levels can reduce the amount of consumption in several areas, thus increasing the precautionary monetary demand for the purchase of commercial insurance. In terms of the protection function of commercial insurance, at the 0.25, 0.50, and 0.75 quartiles, commercial insurance premiums can all significantly increase the amount of household consumption in the area of healthcare, with values of 0.048, 0.044, and 0.097, respectively. The results of this treatment show that commercial insurance has a precautionary savings effect on household consumption of healthcare, but the effect is greater for households with high levels of consumption. In terms of the combined effect, the total impact effect of commercial insurance on household consumption is a significant crowding-out effect in all quartiles. The impact effect values are −0.056, −0.018, and −0.011, respectively, indicating that this crowding-out effect tends to decrease as the total household consumption level rises. It also shows that even for high-income households in 34 large and medium-sized cities in China, commercial insurance premiums are still expensive. For government expenditure on social security, the effect of government social security expenditure on household health care expenditure is −0.064, −0.285, and −0.326 at the 0.25, 0.50, and 0.75 quartiles, and all coefficients are significant at the 0.05 level of significance. The results indicate that government spending on social security significantly reduces the amount of health care spending at all quartiles, but the effect increases as the quartile rises. In terms of the combined effect, the coefficient values of government social security expenditure on total household consumption at the 0.25, 0.50, and 0.75 quartiles are 0.029, 0.054, and 0.057, respectively, indicating that government social security expenditure raises total household consumption, especially for households with medium and high consumption. Unlike the results for the full sample, the quantile regression results suggest that there is no crowding out of private spending by government social security expenditure on overall consumption. The results of the treatment after removing the extremes suggest that government spending on social security does have an inverse effect on the precautionary money demand of the population.

7. Further Discussions

The sample cities are now divided into three regions, east, central, and west, to investigate the question of whether there is regional heterogeneity in the impact effects of consumption motives. Figure 1 illustrates the development of urban disposable income per capita in these three regions.

Table 7. The empirical results for different regions.

	lnc			lnf&t&l_c			lnc_c			lnd_c
	Eastern	Central	Western	Eastern	Central	Western	Eastern	Central	Western	Eastern	Central	Western
lnper_inc	0.658 ***	1.156 ***	1.098 ***	0.371 ***	0.954 ***	0.786 ***	0.413	0.796 ***	0.797 ***	0.951 ***	0.622	1.985 ***
	(0.104)	(0.114)	(0.131)	(0.115)	(0.182)	(0.143)	(0.213)	(0.196)	(0.145)	(0.269)	(0.425)	(0.315)
Price index	−0.005	−0.007	−0.013	−0.003 **	0.001	−0.001	0.004	0.012 **	−0.006	−0.005	0.001	−0.008
	(0.003)	(0.003)	(0.006)	(0.001)	(0.004)	(0.002)	(0.003)	(0.005)	(0.006)	(0.007)	(0.009)	(0.010)
bond	−0.004	−0.014 ***	−0.010 **	−0.001	−0.012 ***	−0.009 ***	−0.004	−0.017 **	−0.014	−0.006	−0.016	−0.002
	(0.003)	(0.004)	(0.003)	(0.003)	(0.003)	(0.002)	(0.006)	(0.007)	(0.006)	(0.006)	(0.008)	(0.009)
stock	−0.001 ***	−0.001 ***	−0.001 ***	−0.001 ***	−0.001 ***	−0.001 ***	−0.001	−0.001 **	−0.001 **	−0.001	−0.001 **	−0.001
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.001)	(0.000)	(0.000)	(0.001)
lnhp	−0.007	−0.084	0.086	0.013	−0.007	0.049	−0.037	−0.060	0.111	−0.046	0.096	0.022
	(0.052)	(0.086)	(0.075)	(0.059)	(0.078)	(0.050)	(0.100)	(0.169)	(0.065)	(0.110)	(0.252)	(0.242)
lnper_ins	−0.059	−0.074 ***	−0.141 ***	−0.055	−0.104 **	−0.117 **	−0.217 ***	−0.148 **	−0.293 ***	−0.027	0.091	−0.339 ***
	(0.032)	(0.021)	(0.030)	(0.031)	(0.031)	(0.046)	(0.066)	(0.055)	(0.075)	(0.087)	(0.085)	(0.079)
lnper_sec	0.069	−0.161 **	−0.158	0.176 **	−0.117	−0.039	0.300 **	−0.050	0.042	0.004	−0.057	−0.279
	(0.053)	(0.053)	(0.068)	(0.067)	(0.096)	(0.055)	(0.135)	(0.081)	(0.110)	(0.121)	(0.170)	(0.196)
_cons	3.003 ***	0.429	0.134	3.972 ***	0.682	1.534	2.837 **	1.450	0.367	−2.010	−0.263	−8.487 ***
	(0.538)	(0.535)	(0.539)	(0.474)	(1.004)	(0.848)	(1.107)	(1.096)	(0.678)	(1.344)	(2.175)	(1.980)
N	220	117	103	220	117	103	220	117	103	220	117	103
adj.R2	0.91	0.92	0.95	0.89	0.86	0.92	0.53	0.69	0.79	0.84	0.70	0.80
Regional fixed effect	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
	lnm_c			lnt&c_c					lne&c&r_c
	Eastern	Central	Western	Eastern		Central	Western		Eastern	Central	Western
lnper_inc	1.104 ***	1.091 ***	1.035 **	0.613 **		1.360 ***	1.146 ***		0.944 ***	1.479 ***	1.330 ***
	(0.358)	(0.225)	(0.330)	(0.279)		(0.380)	(0.290)		(0.094)	(0.342)	(0.344)
Price index	0.009	0.002	0.004	0.014		0.003	0.012		0.004	−0.014	0.010
	(0.005)	(0.004)	(0.009)	(0.010)		(0.007)	(0.012)		(0.005)	(0.013)	(0.011)
bond	0.007	0.005	0.000	0.004		−0.021	−0.018 **		−0.007	−0.012	−0.010 ***
	(0.004)	(0.005)	(0.007)	(0.013)		(0.010)	(0.006)		(0.006)	(0.009)	(0.002)
stock	0.000	−0.000	−0.000	0.001		−0.001	−0.001 **		−0.000	−0.001	−0.000
	(0.000)	(0.000)	(0.001)	(0.001)		(0.001)	(0.000)		(0.000)	(0.001)	(0.000)
lnhp	0.113	−0.143	0.235	−0.096		−0.108	−0.040		−0.032	−0.323	0.073
	(0.163)	(0.142)	(0.164)	(0.108)		(0.146)	(0.198)		(0.064)	(0.169)	(0.238)
lnper_ins	−0.030	0.134	0.159	0.056		−0.053	−0.155		−0.004	−0.030	−0.028
	(0.072)	(0.067)	(0.068)	(0.087)		(0.052)	(0.071)		(0.052)	(0.035)	(0.093)
lnper_sec	−0.266	−0.272	−0.348 **	0.137		−0.080	0.034		−0.116	−0.070	−0.288
	(0.184)	(0.159)	(0.122)	(0.101)		(0.157)	(0.122)		(0.056)	(0.170)	(0.259)
_cons	−3.011	−1.637 **	−3.868 **	1.034		−4.077	−2.523		−0.653	−3.736 **	−3.960
	(1.807)	(0.668)	(1.436)	(1.333)		(1.804)	(1.580)		(0.592)	(1.486)	(2.311)
N	220	117	103	220		117	103		220	117	103
adj. R2	0.67	0.81	0.87	0.69		0.85	0.89		0.85	0.80	0.87
Regional fixed effect	YES	YES	YES	YES		YES	YES		YES	YES	YES

Notes: The clustering robustness criteria errors are in parentheses; ** and ***, respectively, indicate significance at the 0.05 and 0.01 levels; R² represents the corrected goodness-of-fit value.

presents the empirical results of the heterogeneity of the three economic zones. Comparing Table 5, the values of the variable coefficients as well as the significance of each region are significantly different from the overall results.

The following can be concluded from Table 7: (1) According to the overall regression effect (R²), money demand motivation explains significantly weaker consumption of clothing, health care, and transport and communication in eastern cities than in central and western cities. The explanatory strength of the consumption of household goods and services in the central cities is significantly lower than that in the eastern and western cities. In addition, on the whole, as the economic level increases, there is a possibility that the explanatory strength of the money demand motive for consumption will decline. (2) Disposable income is an important variable affecting transactional money demand. For households in the eastern cities, with the increase in disposable income, the stimulus of transactional money demand has been significantly lower than that of households in the central and western cities in terms of total consumption, basic consumption (for example, food, clothing), and developmental enjoyment consumption (for example, transport and communication, education, culture, and recreation). However, for health care consumption by urban households in the east, center, and west, the effect of transactional money demand induced by the increase in disposable income is not significantly different. (3) There is a crowding-out effect of low-risk government bonds and high-risk stocks on the total consumption of household residents in eastern, central, and western cities. The impact coefficients of Treasury bonds on total household consumption are −0.004, −0.014, and −0.010. The impact coefficients of stocks on total household consumption are −0.0005, −0.001, and −0.001. From the absolute values of the impact coefficients, the effect of financial products on the speculative demand for money of urban households in the east is much lower than the effect on urban households in the center and west. (4) The crowding-out effect of commercial insurance premiums on total household consumption in the eastern, central, and western cities increases in turn, with values of −0.059, −0.074, and −0.141, respectively. Moreover, the coefficient values in the central and western cities become significant at the 0.01 level of significance. All previous results show that the positive effect of expenditures on commercial insurance premiums on the precautionary demand for money of urban households in the east, center, and west of the country increases in turn. Additionally, the coefficients of the impact of government social security expenditure on total household consumption in the east, central, and west were 0.069, −0.161, and −0.158, respectively. The results indicate that there is a weak boost to total household consumption in the eastern cities and a crowding-out effect on total household consumption in the central and west. It can be argued that the impact of government social security spending on the population’s precautionary money demand has not yet translated into income effects. In addition, the coefficient values of government social security expenditure on the amount of health care consumption of households in the eastern, central, and western areas were −0.266, −0.272, and −0.348, respectively. It can be seen that government social security expenditure reduced the health care consumption of households in the eastern, central, and western cities, but the effect on the west was significantly higher than that of the east and central areas.

8. Robustness Test

The house price-to-income ratio indicator [54], which is used in some of the literature studying real estate, can be used to measure the housing affordability of households and also to measure the prosperity of the real estate market. In this study, we will use this indicator to replace residential prices to perform robustness tests with the full sample. The results of the treatment are shown in

Table 8. The regression results of robustness test.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	lnc	lnf&t&l_c	lnc_c	lnd_c	lnm_c	lnt&c_c	lne&c&r_c
lnper_inc	0.853 ***	0.599 ***	0.536 ***	1.230 ***	1.193 ***	0.793 ***	1.070 ***
	(0.089)	(0.090)	(0.114)	(0.238)	(0.220)	(0.170)	(0.151)
Price index	−0.008 ***	−0.001	0.004	−0.004	0.005	0.008	−0.003
	(0.003)	(0.001)	(0.003)	(0.005)	(0.004)	(0.006)	(0.005)
bond	−0.008 ***	−0.005 **	−0.009 **	−0.009 ***	0.005	−0.009	−0.009 **
	(0.002)	(0.002)	(0.004)	(0.003)	(0.003)	(0.007)	(0.004)
stock	−0.001 ***	−0.001 ***	−0.001 ***	−0.001 ***	0.000	−0.000	−0.000 **
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
h/e	−0.103	0.024	−0.116	−0.156	0.166	−0.423 **	−0.336 **
	(0.116)	(0.123)	(0.136)	(0.243)	(0.304)	(0.206)	(0.154)
lnper_ins	−0.075 ***	−0.075 **	−0.194 ***	−0.097	0.050	−0.010	−0.006
	(0.025)	(0.029)	(0.034)	(0.070)	(0.050)	(0.046)	(0.037)
lnper_sec	−0.025	0.074	0.166 **	−0.111	−0.301 **	0.096	−0.121
	(0.041)	(0.046)	(0.074)	(0.095)	(0.120)	(0.079)	(0.078)
_cons	1.825 ***	2.656 ***	2.233 ***	−3.786 **	−3.196 **	−0.743	−2.084 **
	(0.489)	(0.491)	(0.625)	(1.395)	(1.201)	(0.974)	(0.866)
Regional fixed effect	YES	YES	YES	YES	YES	YES	YES
adj. R2	0.92	0.89	0.63	0.77	0.76	0.79	0.83
N	440	440	440	440	440	440	440

Notes: The clustering robustness criteria errors are in parentheses; ** and ***, respectively, indicate significance at the 0.05 and 0.01 levels; R² represents the corrected goodness-of-fit value.

below.

Comparing the results of the empirical treatments in Table 5 and Table 8, it can be seen that the magnitude, direction, significance level, and overall goodness of fit of the equation do not change for each indicator coefficient value except for the replacement indicator, which exhibits excellent robustness. However, when the housing price indicator is replaced by the house price-to-income ratio, the monetary demand for the real estate market has a significant crowding-out effect on the household’s higher level consumption of transportation (−0.423), communication, and education, culture, and entertainment (−0.336). This also suggests that when the house price-to-income ratio rises, households significantly reduce their higher level consumption to meet their speculative money demand for real estate, whereas there is no such crowding-out effect of real estate investment on household consumption when the residential price indicator is used. The reason for this significant variability is that the relative number indicator reduces the effect of extreme values on the treatment results to some extent. In addition, the house price-to-income ratio indicator is a better indicator of the actual impact of property prices on residents than the house price indicator.

9. Conclusions

Inadequate household consumption is one of the main economic problems in China. Some of the literature has attributed this to insufficient income, some to high house prices that discourage consumption expenditure, and others to insufficient future security that deters residents from spending. Based on the Keynesian money demand motive, this study examined consumption in 34 large and medium-sized cities in China in general, in quartiles of consumption, and in different regions, using panel data for the period 2007–2019, with the help of a fixed effects model. The aim is to extend this study and provide new evidence.

We found that, first, although Chinese households have moved from basic consumption to developmental enjoyment consumption, there is still much room for improvement in basic consumption in China’s 34 large and medium-sized cities. In addition, the regression of quartiles also shows that only the top 25% of households in China’s 34 large and medium-sized cities have a qualitative demand for food consumption. There is reason to believe that the general awareness of consumption among the population is to achieve a structural upgrading of consumption on the primary level before moving to a demand for quality in consumption. Therefore, in order to enhance consumption more efficiently, the government can intervene in certain aspects of consumption to improve the quality of consumption. Second, price volatility can have a significant negative impact on households whose total non-housing consumption is at a low-to-moderate level of consumption. According to the relevant theory, moderate inflation can give residents an expectation of economic development and prosperity, thus stimulating consumption. However, price changes in China do not provide people with this expectation. In response to this reaction of Chinese households to price changes, we believe that the benchmark value of price volatility should be lowered in order to regulate prices more effectively. Third, financial instruments have a significant and robust crowding-out effect on consumption, suggesting that the rate of return on financial investment products has a negative effect on the transactional demand for money and a positive effect on the speculative demand for money, a finding that validates the impact of interest rates on the demand for money in the theory of money demand. In addition, the treatment results show the interesting phenomenon that equity returns have a greater crowding-out effect on consumption in low-consuming, low-income households, suggesting that low-consuming, low-income households have a stronger speculative mentality. Fourth, the payment of commercial insurance premiums still has a significant crowding-out effect on the consumption of relatively high-income households, suggesting that it is not universally meaningful for households to smooth their consumption by purchasing commercial insurance at this stage and that it is necessary for the government to provide some social benefits to reduce the impact of precautionary monetary demand on household consumption. The empirical results of the research also support this view to some extent, as government spending on social security significantly reduces household health care consumption and is highly robust, but this reduction does not translate into income effects in the total sample and regional empirical results, and there is a crowding out of private spending via government spending. Fifth, the results of various empirical treatments show that the income elasticity of clothing consumption is relatively small and tends to decrease further with the overall development of the economy. In addition, when there is a crowding-out effect of money demand, residents also tend to reduce clothing consumption. The above empirical results show that Chinese residents’ consumption of clothing shows an environmentally friendly consumption concept.

Although we contend that our research results are of some significance, they could be followed up in the following areas. Firstly, the samples selected in the empirical evidence are the most developed cities in each region, and we need to further expand the research sample to validate the results achieved in this study; secondly, merging consumption categories in order to demonstrate the effect of each factor on the structure of consumption would clarify the results. Finally, we would also like to study the impact of commercial insurance and social security on household consumption from multiple perspectives so as to find the real significance of precautionary money demand for Chinese household consumption at the current stage.