Poverty aversion or inequality aversion? The influencing factors of crime in China

Song, Zhe; Yan, Taihua; Jiang, Tangyang Article Poverty aversion or inequality aversion? The influencing factors of crime in China Journal of Applied Economics Provided in Cooperation with: University of CEMA, Buenos Aires Suggested Citation: Song, Zhe; Yan, Taihua; Jiang, Tangyang (2020) : Poverty aversion or inequality aversion? The influencing factors of crime in China, Journal of Applied Economics, ISSN 1667-6726, Taylor & Francis, Abingdon, Vol. 23, Iss. 1, pp. 679-708, https://doi.org/10.1080/15140326.2020.1816130 This Version is available at: https://hdl.handle.net/10419/314112 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Journal of Applied Economics ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/recs20 Poverty aversion or inequality aversion? The influencing factors of crime in China Zhe Song, Taihua Yan & Tangyang Jiang To cite this article: Zhe Song, Taihua Yan & Tangyang Jiang (2020) Poverty aversion or inequality aversion? The influencing factors of crime in China, Journal of Applied Economics, 23:1, 679-708, DOI: 10.1080/15140326.2020.1816130 To link to this article: https://doi.org/10.1080/15140326.2020.1816130 © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. Published online: 02 Oct 2020. Submit your article to this journal Article views: 3178 View related articles View Crossmark data Citing articles: 12 View citing articles Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=recs20 ARTICLE Poverty aversion or inequality aversion? The influencing factors of crime in China Zhe Song, Taihua Yan and Tangyang Jiang Economics and Business Administration, Chongqing University, Chongqing, China ABSTRACT This paper aims to understand whether and how poverty aversion and inequality aversion affect the criminal behaviors. We analyze the relationship between the three variables through a theoretical model and an empirical model. The panel data of 27 provincial-level regions in China were collected for testing the hypothesis. The investigation revealed: 1. Inequality significantly increases crime, while the poverty reduction does not reduce crime. 2. The widening consumption gap between urban and rural residents may be the cause of crime, the effect is more significant for visible consumer goods. 3. The excessive consumption difference between the rich and ordinary people may lead to crime. 4. The increasing inequality of distribution between the state and the people has a positive impact on crime too. The research shows that the Chinese residents are not affected by poverty but by inequality in the choice of crime. ARTICLE HISTORY Received 24 February 2019 Accepted 24 August 2020 KEYWORDS Poverty; inequality; crime; China 1. Introduction For a long time, it has been accepted that social unrest is caused by material shortages. Historically, many periods of social unrest and uprisings have been associated with both famine and poverty. China’s state policy believes that if the people are rich, the state is easy to govern, and if the people are poor, it is rather difficult to govern. Therefore, China has been sparing no efforts to improve the income level of its residents. Especially, the income level of Chinese residents has risen sharply in the past 20 years. But oddly, rising incomes did not improve security in China during this period. Since 1988, with the improvement of people’s income level, criminal cases in China have been increasing year after year. In 2000, the average annual disposable income of urban residents in China was only 6,280 yuan, and the number reached 36,396 yuan in 2017; in the same period, the number of prosecuted criminal people nearly doubled from 708,836 to 1,663,975. Figure 1 depicts that the disposable income of Chinese residents shows a positive relation with the number of crimes. This makes us wonder whether the wealth of the people is related to the crime. So what are the real reasons for the increasing crime? The ancient Chinese sage Confucius pointed out the idea “Inequality and insecurity are more dangerous to the rulers than poverty” in his book the Analects of Confucius 2600 years ago. He believed CONTACT Taihua Yan [email protected] Economics and Business Administration, Chongqing University, Chongqing, China JOURNAL OF APPLIED ECONOMICS 2020, VOL. 23, NO. 1, 679–708 https://doi.org/10.1080/15140326.2020.1816130 © 2020 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. that people would pay more attention to the distribution of inequality rather than the amount of benefits gained, and worry more about the stability of their surroundings rather than poverty. In his philosophy, if wealth is distributed equally, the concept of abundance disappears. At present, China faces problems of unfair distribution of resources and a serious gap between the rich and the poor. In 2013, the China’s National Bureau of Statistics released the annual GINI coefficient from 2003 to 2012, all the numbers were greater than 0.4. According to the evaluation criteria of the United Nations Development Program (UNDP), China’s inequality is quite serious. In 1985, Chinese leader Deng Xiaoping put forth the idea of “let some people get rich first, and bring along the poor”. The differential development strategies in China between regions have resulted in some people being wealthy. However, the widening gap between the rich and the poor is far away from the second part of the goal. The resulting hatred of the inequality may lead to a rise in the crime. Discontent with poverty may lead to crime. The Strain Theory (Merton, 1938) suggests that if efforts to achieve personal goals by legal means are hindered, people may try various illegal means to achieve their goals. Economists believe that the realization of goals leads to maximize individual utility; and this process is often achieved by increasing income and consumption. If the above viewpoints are true, then abundant personal wealth can obviously make the vast majority of goals to achieve, and poor people are more likely to become criminals. Therefore, some scholars believe that there is a positive correlation between poverty and crime. For example, Flango and Sherbenou (2010) regarded crime as the joint resultant of the individual propensity to crime and situational factors which determine inducements to crime, and income level is the key to solve this problem. Berk and Others (1980) believed that poverty caused by less property will increase property crime cases. In his research, poverty is apparently causally related to crime at the individual level. Other studies have come to similar conclusions. Bignon, Caroli, and Galbiati (2017) took the crime rates of France in the nineteenth century as the research object, and found that phylloxera crisis caused a large decrease in the income of fruit farmers thus, resulting in a substantial increase in the property crime rate. Patterson (2010) divided poverty indicators into indirect poverty and direct poverty, and found that direct poverty had a stronger correlation with community crime. The above studies defined poverty as the absolute poverty or material poverty, which can be explained as the lack of wealth and income of low-income groups. The relationship between two variables Figure 1. Disposable income and crime. Source: China Statistical Yearbook. 680 Z. SONG ET AL. is not difficult to understand. Crime is mainly caused by the constraint of living conditions and the inability to meet the basic consumption intention of the individuals. Whereas, the purpose of crime is to directly improve the living standard of individuals, and even to maintain the most basic living conditions (Fafchamps & Minten, 2006; Ludwig, Duncan, & Hirschfield, 2000). The conclusion that dissatisfaction with one’s poverty leads to crime is also agreed by many scholars (Crawford, Whitbeck, & Hoyt, 2011; Hannon, 2002; Kleck & Jackson, 2016). If crime is caused solely by material poverty, then the problem may be dealt with simply by addressing the scarcity of material wealth. But the essence of poverty is not simply a change in one’s condition. Many scholars have found that the unfair income distribution may also be an important cause of crime. Hsieh and Pugh (1993) found that both absolute poverty and income inequality had a positive impact on the crime rate. Blau (1982) drew a similar conclusion after studying the relevant data of 125 largest American metropolitan areas in the United States. He found that absolute poverty in general will increase the incidence of criminal violence, but poverty will no longer affect these rates once economic inequality is controlled. Demombynes and Özler (2016) examined the effects of local inequality on property and violent crimes in South Africa, and found there are many factors in regional inequality, but the unequal distribution has the most significant influence on crimes. Clark and Senik (2010) analyzed the mechanism by which inequality affects happiness. He argued that people with lower incomes are more likely to compare themselves with others, and want the government to intervene when the injustice is serious. These people are most likely to commit crimes. The above studies generally believed that crime is affected by inequality, and many scholars agreed with this conclusion too (Enamorado, Lopezcalva, Rodriguezcastelan, & Winkler, 2016; Hooghe, Vanhoutte, Hardyns, & Bircan, 2011; Pratt & Eisentraut, 2014; Rauma & Berk, 1982). Some scholars have also studied whether China conforms to this conclusion. For example, Li, Wan, Wang, and Zhang (2018) used different indicators of distribution inequality to analyze the interaction between them and criminal acts. In fact, the conclusion that unequal distribution leads to crime has not been reached. Some researchers believed that inequality may have nothing to do with the occurrence of crime. For example, Pare and Felson (2014) found that inequality is unrelated to assault, robbery, burglary, and theft when poverty is controlled. From the perspective of the impact mechanism of unequal distribution on crime, the aversion to inequality is also a behavior of hatred toward the rich, and hatred leads to crimes. After reviewing the literature, we found that most of the existing studies believed that both poverty and unequal distribution have an impact on the rise of crime, and these two variables are generally regarded as important causes of social unrest. But the existing conclusions are less convincing in explaining the phenomenon that rising incomes accompany with rising crimes in China. Therefore, this paper attempts to explain the issue. The subsequent arrangement of the article is as follows: In the second part, we will construct a theoretical model to analyze the relationship between poverty, unequal distribution and crimes, and then make assumptions. In the third part, we will establish the econometric model and explain the empirical model and data. In the fourth part, we will conduct empirical tests and analyze the regression results. Finally, we will draw conclusions and put forward relevant suggestions based on the empirical research results. The contribution of this paper can be summarized as follows. First, we use provincial JOURNAL OF APPLIED ECONOMICS 681 macro data to study crime, avoiding the problem of sample inhibition and standard difference. Second, we try to explain the reasons behind the phenomenon of the simultaneous growth of income and crime in China. Finally, the paper addresses some hot issues such as poverty aversion and hatred of the rich in China and its practical relevance in today’s context. 2. Theoretical model Referring to Fehr and Schmidt’s research (1999), we established an ultimatum game model to analyze the influence of poverty aversion and inequality aversion on criminal behaviors. In the model, we assume that the participants are rich A and poor B. There is a sum of money E to be allocated, and rich A is responsible for the right of distribution. The decision made by A is to give S shares to the poor, leaving (E-S) shares for themselves; the poor have a choice between accepting distributive decisions or rejecting and committing crimes against the rich. The ultimate goal of both sides is to maximize the utility. We need to add three assumptions before we establish the model: 1) each participant prefers a fair outcome, that is, everyone wants to be treated equally as well as others, 2) when faced with unfair situations, the anger effects caused by the damage to own interests are greater than the guilt effects caused by damage to others’ interests, 3) the amount needs to be allocated without surplus. The model is set as follows: UiX1;X2 ð Þ ¼ Xiαimax XjXi;0 � �βimax XiXj;0 � � (1) i¼1;2;j¼3i Where, αimax XjXi;0 � �represents the anger effects caused by damage to own interests; βimax XiXj;0 � �represents the guilt effects caused by damage to others, αi and βican be considered as anger and guilt coefficients. At the same time, refer to the points of Di Tella, Perez-Truglia, Babino, and Sigman (2015), when A is faced with the uneven distribution state in which he is dominant, the guilt coefficient of A is relatively low, while the anger coefficient of B is relatively low. This means that inequality may be more likely to occur in the distribution game. According to the hypothesis above, it is certain that αi>βi>0, and βi<1/2. 1 Therefore, it can be understood that when people choose to damage their own interests or the interests of others, they always choose the latter. We now analyze and solve the model. Optimal reaction of B can be calculated by considering the following situations: (1) If poor B refuses to accept the allocation plan proposed by rich A and opts to resist, then A’s benefits will be taken away, but at the same time B will be punished, assuming that the benefits of A and B benefits are 0 at this time. (2) If B accepts the distribution plan, X2 1¼ES; X2 2¼S Obviously, the reject option by B is not an optimal solution for both A and B. So when B accepts the allocation S given by A, the utility function of B can be divided into two cases. When, S ≥ E/2, the utility function of B is expressed as: 1 When βi>1/2, the coefficient of X i in the utility function is not positive. 682 Z. SONG ET AL. U2sð Þ ¼ Sβ22S Eð Þ (2) Since βi<1/2, so U2Sð Þ≥0. This shows that if the amount given to B in the distribution plan is greater than the amount given to A itself, then B must accept the proposal of A. When S ≤ E/2, the utility function of B is expressed as: U2sð Þ ¼ Sα2E2Sð Þ (3) Let U2sð Þ � 0, and we get the lowest acceptable value of S: S�� S α2 ð Þ ¼ α2E 1þ2α2 (4) From the above analysis, we get the optimal reaction of B. It can be seen that only when S�� S α2 ð Þ, B would accept the distribution plan proposed by A. We can know from Equation (4), the acceptance condition of B is affected by the anger coefficients α2 and the total distribution E. Let us calculate the optimal reaction of A. For A, as the decision maker of the distribution scheme, he only needs to consider the situation accepted by B. When A knows the information of α2, his share of S given to B is also going to be in either case: 1) when S ≥ E/2, the utility function of a is: U1sð Þ ¼ ESα22S Eð Þ (5) When S = E/2, the effect of A is maximized. 2) when S ≤ E/2, the utility function of A is: U1sð Þ ¼ 1β1  �Eþ2β11  �S (6) However, the lowest value acceptable to B is analyzed above, so the optimal decision of A at this time is: S�¼� S α2 ð Þ ¼ α2E 1þ2α2 (7) It can be seen that the value of Equation (7) is strictly less than 2/E, and is an increasing function of the anger coefficient α2 and the total allocation E. The above analysis of A strategy was based on the premise that A was aware of α2, but generally speaking, A may not know this specific value, but its distribution value. The distribution value of anger coefficient for B can be expressed as � α and α: � α¼minfαjFαð Þ ¼ 1g(8) α¼maxfαjFαð Þ ¼ 0g(9) When S ≥ E/2, the utility function of A is the same as before U1sð Þ ¼ 1β1  �Eþ2β11  �S; when S ≤ E/2, the probability that B accepts the allocation plan of A is r: JOURNAL OF APPLIED ECONOMICS 683 r¼f xð Þ ¼ 0;S�� Sαð Þ FS E2S  �;S�� S� αð Þ 1;S�� S� αð Þ 8 > > > > < > > > > : (10) The expectation effect of A is known as E (U1): E U1 ð Þ ¼ 0;S�� S� αð Þ FS E2S  ��1β1  �Eþ2β11  �S � �;S�� S� αð Þ 1β1  �Eþ2β11  �S;S�� S� αð Þ 8 > > > > < > > > > : (11) Therefore, we can draw decision of A as follows: max � Sαð Þ � S�� S� αð ÞFS E2S � ��1β1  �Eþ2β11  �S � � (12) After analyzing the above model, we can draw the following conclusions: (1) A would never come up with a plan that would hurt his interests; the share given to B will always be less than the share given to himself. In other words, there is always an imbalance in the distribution. (2) For too small S, B will refuse the allocation plan and choose to commit a crime. (3) The S share acceptable to B is positively correlated with the anger coefficient and the total allocation amount. (4) The probability that B accepts the allocation plan of A increases with the amount of S. We can find from the theoretical model that the reason influencing people to commit crimes is not the absolute value of the distribution amount S, but whether the distribution is fair and reasonable. They may not commit a crime if S is small; and they may commit a crime although S is large. In real life, the distribution process is not a pure benefit distribution experiment, but there are reasons to consider that social progress benefits the population as a whole. The decision makers who make the assignment are usually the rich. Behind the efforts to ensure fair distribution lays not only the hope of fairness, but also the consideration of self-interest. To some extent, the behavior whether residents agree to accept the distribution scheme or choose to commit crimes can be explained by the above theoretical model. It is noteworthy that the above model assumes that the people who choose to commit crimes only are the poor, while the theory that “the rich do not commit crimes” is obviously not true. However, the data used in this article can solve this problem. The data used in this paper to measure crimes are criminal cases, which can be divided into violent crimes and property crimes. Although the people involved in crimes are usually spread across all the levels. But robbery, theft and other cases, which account for the largest number of criminal cases, are all crimes of property assault, and most violent crimes are often accompanied by the purpose of property assault too (Fajnzlber, Lederman, & Loayza, 2002). So our model is reasonable in the large sample case. Combined with the 684 Z. SONG ET AL. theoretical model, the following hypotheses were proposed and tested in subsequent empirical analysis: Hypothesis1: Rising absolute incomes of the poor can not reduce crime. Hypothesis2: Changes in economic aggregates may be related to criminal behavior. Hypothesis3: Unbalanced distribution has a positive impact on crime. 3. Data and model 3.1. Data sample and collection We collected 14 years’ data, extending from 2004 to 2017. Moreover, we selected the data of 27 provinces, municipalities, and autonomous regions of China mainland; however, Beijing, Chongqing, Xinjiang and the Tibet Autonomous Region were excluded from this study. The main reason for excluding the four regions is that the relevant data of these regions may be based on following accounts: 1) Tibet is located in the western part of China. It has the harsh natural environment and a relatively small population. In addition, a large number of people have a nomadic lifestyle, which makes it extremely difficult to collect relevant data. 2) Whereas, the capital of China, Beijing is also the economic and administrative hub of China, with most of the headquarters of Chinese institutions. As compared to the single local government model in other regions, Beijing has two organ systems, namely the central government and the local government. However, these features cannot be apparent while using provincial data. 3) The gang crackdown activity of Chongqing started in 2009 and ended in 2012, a large number of people were arrested and prosecuted by Chongqing procuratorate during these time, and most of them were punished for criminal offences. Therefore, the data of Chongqing may be particular. 4) Due to the continuous years of violent and terrorist activities associated with region belief in Xinjiang, the local criminal crimes are more special. In this experiment, we used crime data which was reported in China inspection yearbook; besides the crime data reported in the annual work reports of local people’s procuratorates; and the other data were obtained from National Bureau of Statistics of China, China urban statistics yearbook, China Real Estate Yearbook, etc. 3.2. Descriptive of data 3.2.1. Explained variables and main explanatory variables Crime (Crime): The crime data used in this paper are the number of criminal cases prosecuted by the procuratorate per 100,000 people. It mainly includes serious violent crimes such as intentional homicide, rape and arson, and property crimes such as robbery, drug trafficking and theft. We do a logarithm of the number of crimes. Poverty aversion (Income): According to the theoretical analysis, we believed that lowincome groups are more likely to commit crimes. China’s statistics bureau uses the sampling method to divide residents into five equal groups according to their income. JOURNAL OF APPLIED ECONOMICS 685 own houses. And people without houses are likely to be potential criminal groups, so we find that the coefficient of housing prices is positive. According to the research of Meng, Gregory, and Wang (2005), the problem of unequal income distribution in Chinese society had already become serious for a long time. From the survey data of China statistical yearbook, although the actual income level of rural residents increased year by year, the urban incomes are rising at a faster rate. However, due to geographical restrictions, class solidification and other factors, the uneven distribution exists, but the ability to obtain the income level of urban residents is limited for low-income groups. Therefore, even though the gap between the two classes is prominent, few people (low-income group) perceive the information, so there is not too much social conflict happening. However, with the development of urbanization and the prosperity of the real estate market in China (Cao, Chen, & Zhang, 2018), more and more rural residents are entering cities and towns, the large income gap that already exists is gradually observed. The widening consumption gap shows that China’s rural residents have been benefited from the economic growth, but more for urban residents. As mentioned above, we believed that people usually get the information of uneven distribution by observing consumption. So we should pay attention to the improvement of people’s ability to observe information through scientific and technological progress. After entering the internet era, the number of internet users in China has soared. According to the Statistical Reports on Internet Development in China, the number of internet users in China has increased from 16.9 million in 2000 to 772 million in 2017. The invention of smart phones has made the dissemination of information more convenient, among which mobile internet users have gradually become the main group of the internet. By 2017, China’s mobile internet group has reached 753 million, accounting for 97.5% of the total number of internet users. During the high-speed flow of information through the new carriers, a large number of behaviors such as flaunt wealth and wealth displays are widely known. Due to the lack of channels, the spread of these messages was limited in the past. The wide application of the internet and smart phones enabled low-income groups to break through the barriers of the original class, obtain the information of unfair distribution indirectly through other people’s consumption behaviors, and feel the unfair phenomenon. Thus, it appears that use of the internet has a significant impact on increasing the crime rate in China. The wide application of smartphones has brought obvious changes to people’s life; Public at large can access more information through mobiles and communications in society. According to the analysis from the report, the rapid growth of the Internet and mobile phone networks in recent years is mainly due to the rapid increase in usage by rural residents. With the gradual decrease in prices of computers and mobile phones, communication tools are more affordable in rural areas. As can be seen from Figure 3, the growth rate of Chinese internet users was above 20% before 2010. But the growth rate gradually slowed down, which indicates that Chinese internet users basically reached a stable stage in 2010. At the same time, the number of mobile internet users began to increase since 2006 and smartphones gradually replaced the original functional machines and became the mainstream after 2010. For the reasons, we use 2010 as a time node to analyze the impact of uneven distribution on crime in different time periods. The first half of the node is the time interval of slow message propagation, and the second half of 692 Z. SONG ET AL. the node is the time interval with fast message propagation. Tables 3 and 4 show the relevant regression results. According to the regression results, the consumption gap coefficient before 2010 was 0.0545, but the results are not significant; and the number was 0.1152 in 10% confidence level after 2010. This shows that the positive impact of consumption gap on crime is gradually increasing with the passage of time. This proves that it is reasonable for us to use consumption gap to measure the inequality of distribution. In other words, people often evaluate the inequality of income distribution by observing consumption information, and then choose crime. As mentioned above, an important difference between consumption and income is visibility, consumer information is more transparent, but not all consumption behaviors are observable. Although we verified the correlation between the overall consumption gap and crime from the previous regression results, the overall consumption is actually composed of both observable consumption and unobserved consumption. It is necessary to classify these consumption types to obtain more information. So, we calculated the consumption of different consumer goods to find out which consumption differences will have a more significant impact on the criminal behaviors. Categories of consumption include Food (Food, tobacco and liquor), Clothing, Equipment (Household facilities articles and services), Transport (Transport and communications), E&E (Education, culture and entertainment), Health (Health care and medical services), Residence (Housing maintenance, rental hotel and the charge of water and electricity) and Others. The above consumer goods data and consumer goods interpretation are from the National Bureau of Statistics of China. According to the information of Table 5, we can find that the widening gap between education and entertainment (E&E) consumption, health consumption and residence consumption does not significantly increase criminal behaviors; the expansion of the ratio of food consumption, clothing consumption, equipment consumption and transportation consumption will significantly promote the occurrence of crime, among which the coefficient of food consumption gap is 0.0744, the clothing consumption gap is 0.028, and the equipment consumption gap is 0.0245, the transportation consumption gap is 0.0232. By observing its characteristics, we can find that on comparing with the nonsignificant consumption categories, the significant consumer goods are generally actual goods that can be purchased, in other words, it is usually visible. The food consumption gap has the highest impact on crime, and followed by clothing consumption, equipment 0 0.5 1 1.5 2 0 20000 40000 60000 80000 Internet Users Mobile Internet Users(MIU) MIU ratio Figure 3. Statistical reports on internet development in China. Source: Statistical Reports on Internet Development in China (2018). JOURNAL OF APPLIED ECONOMICS 693 Table 3. Crime and inequality before 2010. M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 VARIABLES Crime Crime Crime Crime Crime Crime Crime Crime Crime Crime Crime Income 0.636** 0.634** −0.117 −0.0184 −0.059 −0.145 −0.082 −0.0664 0.0399 0.00936 0.0125 (0.244) (0.291) (0.310) (0.310) (0.315) (0.321) (0.320) (0.323) (0.318) (0.319) (0.320) Gap1 0.000647 0.0774 0.0809* 0.0818* 0.0905* 0.062 0.0567 0.0545 0.0546 0.0545 (0.049) (0.048) (0.047) (0.048) (0.048) (0.050) (0.052) (0.051) (0.051) (0.051) GDP 0.445** 0.375* 0.243 0.256* 0.282* 0.244 0.243 0.256 0.259* (0.207) (0.241) (0.245) (0.247) (0.249) (0.247) (0.248) (0.253) (0.245) Density −0.785** −0.758** −0.659* −0.605* −0.630* −1.012*** −0.924** −0.924** (0.355) (0.358) (0.365) (0.362) (0.370) (0.385) (0.395) (0.396) Judicial −1.176 −1.308 −1.967 −1.998 −1.612 −1.532 −1.42 (1.563) (1.562) (1.584) (1.591) (1.560) (1.562) (1.587) Welfare 0.548 0.562 0.539 0.758* 0.788* 0.775* (0.417) (0.413) (0.419) (0.416) (0.417) (0.419) Education −0.0765* −0.0755* −0.107*** −0.112*** −0.111*** (0.039) (0.040) (0.040) (0.041) (0.041) Housing 0.0386 0.00153 0.000543 −0.00721 (0.103) (0.102) (0.102) (0.104) Age 2.605*** 3.055*** 3.017*** (0.917) (1.018) (1.025) Floating −0.338 −0.351 (0.333) (0.335) Urban 0.181 (0.412) Constant −4.081* −4.062 −4.26 1.11 1.254 1.129 1.174 0.923 1.886 1.546 1.525 (2.385) (2.782) (2.590) (3.528) (3.538) (3.531) (3.499) (3.573) (3.507) (3.523) (3.533) Observations 189 189 189 189 189 189 189 189 189 189 189 R-squared 0.469 0.469 0.543 0.557 0.559 0.564 0.574 0.575 0.597 0.6 0.6 Number of id 27 27 27 27 27 27 27 27 27 27 27 Standard errors in brackets *** p < 0.01, ** p < 0.05, * p < 0.1. 694 Z. SONG ET AL. Table 4. Crime and inequality after 2010. M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 VARIABLES Crime Crime Crime Crime Crime Crime Crime Crime Crime Crime Crime Income 0.666** 0.676** 0.662** 0.444 0.403 0.532* 0.534* 0.539 0.513 0.508 0.555 (0.271) (0.274) (0.276) (0.286) (0.288) (0.291) (0.292) (0.288) (0.285) (0.218) (0.202) Gap1 −0.0138 −0.00967 0.0495* 0.1213** 0.1252* 0.1222* 0.1272* 0.1193** 0.1185** 0.1152* (0.049) (0.050) (0.052) (0.069) (0.069) (0.068) (0.068) (0.070) (0.068) (0.069) GDP 0.501* 0.513*** 0.531*** 0.565*** 0.543*** 0.536*** 0.406** 0.399** 0.398** (0.152) (0.168) (0.170) (0.171) (0.170) (0.172) (0.174) (0.174) (0.174) Density 0.0363** 0.0341** 0.0451*** 0.0450*** 0.0447*** 0.0446*** 0.0443*** 0.0460*** (0.015) (0.015) (0.016) (0.016) (0.016) (0.016) (0.016) (0.016) Judicial 2.607 1.481 1.482 1.509 1.324 1.281 1.804 (2.285) (2.321) (2.329) (2.338) (2.350) (2.367) (2.429) Welfare −0.756** −0.757** −0.760** −0.738** −0.732** −0.727** (0.358) (0.359) (0.360) (0.362) (0.364) (0.364) Education 0.00252 0.00132 −0.00225 −0.00143 0.00113 (0.027) (0.027) (0.028) (0.028) (0.028) Housing −0.0368 −0.033 −0.0278 −0.0428 (0.131) (0.131) (0.134) (0.135) Age 0.197 0.208 0.32 (0.229) (0.236) (0.263) Floating −0.0389 −0.0484 (0.192) (0.193) Urban −0.968 (1.001) Constant −4.546 −4.617 −5.034* −4.137 −4.065 −4.209 −4.253 −3.917 −3.35 −3.255 −3.612 (2.820) (2.840) (3.024) (2.997) (2.994) (2.961) (3.008) (3.244) (3.313) (3.357) (3.378) Observations 189 189 189 189 189 189 189 189 189 189 189 R-squared 0.322 0.322 0.323 0.349 0.355 0.373 0.373 0.374 0.377 0.377 0.381 Number of id 27 27 27 27 27 27 27 27 27 27 27 Standard errors in brackets *** p < 0.01, ** p < 0.05, * p < 0.1. JOURNAL OF APPLIED ECONOMICS 695 Table 5. Classification of consumer goods regression. Variables Crime Crime Crime Crime Crime Crime Crime Crime Crime Income 0.113 0.214 0.251 0.165 0.123 0.185 0.205 0.232 0.262 (0.211) (0.193) (0.200) (0.200) (0.202) (0.204) (0.196) (0.196) (0.195) GDP 0.446*** 0.377*** 0.404*** 0.450*** 0.431*** 0.431*** 0.417*** 0.430*** 0.400*** (0.083) (0.078) (0.082) (0.083) (0.080) (0.083) (0.080) (0.083) (0.080) Density 0.0555*** 0.0489*** 0.0469*** 0.0475*** 0.0468*** 0.0493*** 0.0487*** 0.0470*** 0.0473*** (0.016) (0.015) (0.016) (0.016) (0.016) (0.016) (0.016) (0.016) (0.016) Judicial 0.537 0.278 0.324 0.165 0.24 0.109 0.449 0.317 0.289 (1.335) (1.314) (1.352) (1.325) (1.321) (1.334) (1.330) (1.329) (1.344) Welfare 0.144 0.0916 0.118 0.146 0.0769 0.131 0.114 0.18 0.116 (0.266) (0.263) (0.267) (0.266) (0.265) (0.266) (0.265) (0.271) (0.267) Education −0.0521** −0.0556*** −0.0539** −0.0507** −0.0540*** −0.0563*** −0.0561*** −0.0559*** −0.0546*** (0.021) (0.021) (0.021) (0.021) (0.021) (0.021) (0.021) (0.021) (0.021) Housing 0.203** 0.246*** 0.239*** 0.215*** 0.242*** 0.230*** 0.231*** 0.221*** 0.239*** (0.082) (0.079) (0.080) (0.081) (0.080) (0.080) (0.080) (0.081) (0.080) Age 0.415 0.392 0.373 0.354 0.448* 0.393 0.373 0.348 0.363 (0.254) (0.251) (0.256) (0.253) (0.255) (0.254) (0.253) (0.254) (0.254) Floating 0.216 0.16 0.222 0.185 0.206 0.191 0.188 0.222 0.229 (0.158) (0.158) (0.162) (0.159) (0.157) (0.161) (0.160) (0.158) (0.159) Urban 0.0962 0.156 0.0582 0.00109 0.0361 0.0496 0.167 0.11 0.0539 (0.358) (0.356) (0.361) (0.357) (0.356) (0.358) (0.363) (0.361) (0.361) Gap1 0.0711** (0.041) Food 0.0744*** (0.025) Clothing 0.0280* (0.016) Equipment 0.0245* (0.013) Transport 0.0232** (0.010) E&E 0.0112 (0.009) Health 0.00329 (0.011) Residence 0.0252 (0.020) Others 0.00132 (Continued) 696 Z. SONG ET AL. Table 5. (Continued). Variables Crime Crime Crime Crime Crime Crime Crime Crime Crime (0.007) Constant −5.856*** −6.444*** −6.893*** −6.307*** −5.907*** −6.413*** −6.576*** −6.866*** −6.936*** (2.036) (1.935) (1.960) (1.969) (1.989) (1.993) (1.954) (1.949) (1.953) Observations 378 378 378 378 378 378 378 378 378 R-squared 0.672 0.678 0.669 0.673 0.674 0.671 0.672 0.671 0.669 Number of id 27 27 27 27 27 27 27 27 27 Standard errors in brackets *** p < 0.01, ** p < 0.05, * p < 0.1. JOURNAL OF APPLIED ECONOMICS 697 consumption and finally transport consumption. The information presented by the above four kinds of consumption is easier to be found due to the public visibility. The consumption of food, clothing, equipment and transport also includes relevant income information of consumers, but their consumption behaviors are less likely to be known. This finding is consistent with Mejía and Restrepo’s (2016) conclusion, and also fits our hypothesis that visual consumption is related to crime. We verified the interaction between the urban-rural consumption gap, which exists in rural and urban groups as an indicator of income distribution inequality and criminal acts. So does the unequal distribution of income within the same group affect crime as well? According to Li et al. (2018), income distribution inequality has a Matthew effect. According to the statistical data, it is not difficult to find that the large gap of income distribution in China is not only reflected in different categories of residents, such as rural residents and urban residents mentioned above, but also reflected in the residents of the same dimension. The difference between the average annual consumption expenditure ratio of urban residents and rural residents in 2017 is 2.3times, while the consumption ratio between the highest group and the lowest group is 5.32 times. We collected relevant data of five groups’ income among urban residents, used the ratio between the highest income group and the average income as an indicator to measure the income distribution inequality, and analyzed the relationship between it and crime. 2 The regression results in Table 6 reveal that by using the ratio between the highest income group and the average income as inequality also has a positive impact on crime; its value is 0.128 at 10% level of confidence. The high-income group’s consumption level is higher than the average level of consumption per unit, crime will increase by 12.8%. This result shows that the influence of inequality on crime not only exists in people living in different regions, but also in people living in the same region, the excessive consumption difference between the rich and ordinary people may lead to crime. This inequality can be interpreted as flaunt wealth phenomenon at some moments. Just as Wu and Lan (2018) found that flaunt wealth behaviors not only increase people’s desire for wealth, but also cause resentment against the rich, resulting in social security turbulence. In the above empirical research, we mainly analyzed the consumption differences among residents and the problems between criminal crimes. However, China’s unique economic system and political system also created the situation of national division in essence, resulting in the uneven distribution between the government and the people in the process of economic development. National consumption is usually composed of government consumption and residents’ consumption. The main channels of government consumption focus on providing public services to the whole society, including expenditure on science, education, culture, health and administration. In most countries of the world, government consumption is mainly service-oriented. However, China’s government consumption is mainly aimed at promoting construction, and its characteristics are dominant rather than service-oriented (Wang & Wen, 2019). For example, the research of Alonso Carrera, Caballe, and Raurich (2015) shows that in the short term, the increase of government consumption will stimulate the growth of residents’ consumption, but when the economy enters the normal development stage, the excessive increase 2 According to the local yearbook, there are only 20 districts have reported consumption data for different income groups in some years. 698 Z. SONG ET AL. Table 6. People grouping regression. M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 VARIABLES Crime Crime Crime Crime Crime Crime Crime Crime Crime Crime Crime Income 0.616*** 0.544** −0.0296 −0.0969 −0.0762 −0.137 −0.171 −0.152 −0.15 −0.0974 −0.0964 (0.225) (0.226) (0.234) (0.237) (0.240) (0.262) (0.257) (0.254) (0.255) (0.258) (0.258) Gap2 0.159** 0.117 0.132* 0.135* 0.136* 0.131* 0.132* 0.132* 0.130* 0.128* (0.075) (0.071) (0.071) (0.072) (0.072) (0.070) (0.070) (0.070) (0.069) (0.070) GDP 0.566*** 0.592*** 0.584*** 0.593*** 0.602*** 0.571*** 0.552*** 0.567*** 0.556*** (0.101) (0.102) (0.103) (0.104) (0.102) (0.102) (0.104) (0.105) (0.108) Density 0.0518 0.0523 0.0476 0.0535 0.0628* 0.0633* 0.0588* 0.0578* (0.034) (0.034) (0.035) (0.034) (0.034) (0.034) (0.034) (0.034) Judicial −0.866 −0.963 −2.374 −2.338 −2.158 −2.517 −2.411 (1.572) (1.583) (1.614) (1.599) (1.611) (1.634) (1.656) Welfare 0.218 0.239 0.0147 0.07 0.0451 0.0336 (0.372) (0.364) (0.375) (0.380) (0.380) (0.381) Education −0.0937*** −0.0954*** −0.0987*** −0.0983*** −0.0984*** (0.030) (0.030) (0.030) (0.030) (0.030) Housing 0.236** 0.221** 0.220** 0.210* (0.107) (0.108) (0.108) (0.111) Age 0.378 0.226 0.204 (0.415) (0.431) (0.435) Floating 0.274 0.27 (0.216) (0.217) Urban 0.221 (0.516) Constant −3.9646* −3.498 −3.690* −3.719* −3.803* −3.261 −2.079 −4.062 −3.834 −4.499* −4.416* (2.353) (2.344) (2.185) (2.178) (2.187) (2.378) (2.358) (2.502) (2.515) (2.566) (2.578) Observations 238 238 238 238 238 238 238 238 238 238 238 R-squared 0.595 0.599 0.653 0.657 0.658 0.658 0.674 0.682 0.683 0.686 0.686 Number of id 20 20 20 20 20 20 20 20 20 20 20 Standard errors in brackets *** p < 0.01, ** p < 0.05, * p < 0.1. JOURNAL OF APPLIED ECONOMICS 699 of government consumption expenditure may produce a crowding out effect on residents’ consumption. It is geneally believed that the economic growth should mainly rely on residents’ consumption. The excessive proportion of government consumption is likely to lead to inefficient and corrupt resource allocation, and the social atmosphere of corruption is often the root cause of crime. If the dividend of economic growth is seen as another form of distribution of benefits between the state and the people, it is necessary to use the ratio of government consumption to residents to test whether such inequality also causes crime. For this reason, we chose to use the ratio of government consumption to residents’ consumption (Gap3) as another measure of consumption difference to further study its impact on criminal crimes. The regression results in Table 7 show that in the process of adding control variables, the coefficient of Gap3 value is significantly negative. After controlling all variables, the coefficient was 0.156, indicating that for every unit of increase in Gap3, crime increased by 15.6%. At the same time, the poverty aversion was still insignificant in model 11. The above results show that the widening consumption gap between the government and residents also increase the value of crimes. In fact, the absolute income of Chinese residents has increased quickly in recent years, the relative income of Chinese residents is declining against the background of overall economic growth, and the positive effects brought by the increase in income may be offset. Although residents’ income level and standard of living are improving, their living standard does not fully enjoy the dividends brought by economic development. The situation of a wealthy country and poor people makes residents dissatisfied with their status, and they may choose to commit crimes. 4.1. Robustness check In order to ensure the robustness of the experimental results, we did two parts of jobs. First, we used the number of criminal arrests approved by the procuratorial organs as the explained variable and put it into the model for regression verification. According to China’s criminal procedure law, an arrest must meet the following three conditions: 1) there is evidence to substantiate the crime, 2) the crime carries a penalty of imprisonment or more and, 3) arrest is necessary. As one of the most severe coercive measures in criminal proceedings, arrest deprives suspects of their personal freedom. And punishment is stronger than prosecution, so the number of arrests is usually less than prosecution. Currently, only procuratorates and courts have the right to approve arrests in China. The arrest rate index for the robustness test in this paper is a logarithm of the number of criminal suspects arrested per 100,000. We still use the urban-rural consumption gap to measure inequality. The results from Table 8 show that the coefficient of each variable does not appear big difference, the regression of this paper is basically robust. Second, the second robustness work of this paper was to deal with the endogeneity problem. To control endogenous correlation, the core explanatory variable of Income and Gap was lagged in this experiment. Considering that the main work of this paper focused on the consumption gap between urban and rural residents, we deliberately selected the representative index of Gap1. According to robustness results in Table 9, the regression results are still robust after the lagged terms of Income and Gap. 700 Z. SONG ET AL. Table 7. People and country grouping regression. M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 VARIABLES Crime Crime Crime Crime Crime Crime Crime Crime Crime Crime Crime Income 0.878*** 0.906*** 0.538*** 0.453** 0.435** 0.368* 0.314 0.253 0.247 0.288 0.288 (0.176) (0.175) (0.182) (0.184) (0.185) (0.195) (0.195) (0.193) (0.193) (0.194) (0.194) Gap3 0.229** 0.240** 0.200* 0.196* 0.213** 0.190* 0.151* 0.137* 0.155** 0.156* (0.107) (0.103) (0.103) (0.103) (0.105) (0.104) (0.109) (0.117) (0.121) (0.121) GDP 0.383*** 0.391*** 0.395*** 0.406*** 0.429*** 0.428*** 0.385*** 0.408*** 0.404*** (0.072) (0.072) (0.072) (0.072) (0.072) (0.072) (0.075) (0.076) (0.079) Density 0.0391** 0.0375** 0.0335** 0.0380** 0.0433*** 0.0415*** 0.0433*** 0.0430*** (0.015) (0.016) (0.016) (0.016) (0.016) (0.016) (0.016) (0.016) Judicial 1.068 1.038 0.375 0.286 0.28 0.162 0.194 (1.324) (1.324) (1.338) (1.322) (1.318) (1.317) (1.328) Welfare 0.294 0.27 0.14 0.212 0.171 0.17 (0.267) (0.265) (0.265) (0.267) (0.268) (0.268) Education −0.0532** −0.0478** −0.0531** −0.0521** −0.0526** (0.021) (0.021) (0.021) (0.021) (0.021) Housing 0.240*** 0.237*** 0.224*** 0.222*** (0.080) (0.080) (0.080) (0.081) Age 0.420* 0.338 0.323 (0.239) (0.244) (0.254) Floating 0.256 0.259 (0.158) (0.159) Urban 0.0748 (0.358) Constant −6.756*** −7.135*** −7.508*** −7.013*** −6.909*** −6.354*** −5.526*** −7.069*** −6.567*** −7.176*** −7.151*** (1.834) (1.833) (1.764) (1.760) (1.766) (1.835) (1.849) (1.898) (1.913) (1.945) (1.951) Observations 378 378 378 378 378 378 378 378 378 378 378 R-squared 0.605 0.611 0.641 0.648 0.648 0.65 0.657 0.666 0.669 0.671 0.671 Number of id 27 27 27 27 27 27 27 27 27 27 27 Standard errors in brackets *** p < 0.01, ** p < 0.05, * p < 0.1. JOURNAL OF APPLIED ECONOMICS 701 Pare, P. P., & Felson, R. (2014). Income inequality, poverty and crime across nations. British Journal of Sociology, 65(3), 434–458. Patterson, E. B. (2010). Poverty, income inequality, and community crime rates. Criminology, 29 (4), 755–776. Phillips, J. A. (2006). Explaining discrepant findings in cross-sectional and longitudinal analyses: An application to U.S. homicide rates. Social Science Research, 35(4), 948–974. Pratt, T. C., & Eisentraut, B. D. (2014). Poverty, inequality, and area differences in crime.New York: Springer. Qian, X., & Smyth, R. (2010). Measuring regional inequality of education in China: Widening coast–inland gap or widening rural–urban gap? Journal of International Development, 20(2), 132–144. Rauma, D., & Berk, R. A. (1982). Crime and poverty in California: Some quasi-experimental evidence. Social Science Research, 11(4), 318–351. Sampson, R. J., & Wilson, W. J. (1995). Toward a theory of race, crime, and urban inequality. Race, 35(4), 486–494. Soh, M. B. C. (2012). Crime and urbanization: Revisited Malaysian case. Procedia - Social and Behavioral Sciences, 42, 291–299. Song, Z., Yan, T., & Jiang, T. (2019). Can the rise in housing price lead to crime? An empirical assessment of China. International Journal of Law, Crime and Justice, 59, 100341. Wang, X., & Wen, Y. (2019). Macroeconomic effects of government spending in China. Pacific Economic Review, 24(3), 416–446. Wu, X., & Lin, L. (2018). Resentment against the rich: Conceptualization, measurement, and empirical evidence from China. International Journal of Conflict Management, 31(4), 529–558. Ziesemer, T. (2016). Gini coefficients of education for 146 countries, 1950–2010: United Nations University - Maastricht Economic and Social Research Institute on Innovation and Technology (MERIT). 708 Z. SONG ET AL.