Less cash, less theft? Evidence from fintech development in the People's Republic of China

Jiang, Hongze; Liang, Pinghan Working Paper Less cash, less theft? Evidence from fintech development in the People's Republic of China ADBI Working Paper, No. 1282 Provided in Cooperation with: Asian Development Bank Institute (ADBI), Tokyo Suggested Citation: Jiang, Hongze; Liang, Pinghan (2021) : Less cash, less theft? Evidence from fintech development in the People's Republic of China, ADBI Working Paper, No. 1282, Asian Development Bank Institute (ADBI), Tokyo This Version is available at: https://hdl.handle.net/10419/249461 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-nc-nd/3.0/igo/ ADBI Working Paper Series LESS CASH, LESS THEFT? EVIDENCE FROM FINTECH DEVELOPMENT IN THE PEOPLE’S REPUBLIC OF CHINA Hongze Jiang and Pinghan Liang No. 1282 August 2021 Asian Development Bank Institute The Working Paper series is a continuation of the formerly named Discussion Paper series; the numbering of the papers continued without interruption or change. ADBI’s working papers reflect initial ideas on a topic and are posted online for discussion. Some working papers may develop into other forms of publication. The Asian Development Bank refers to “China” as the People’s Republic of China. Suggested citation: Jiang, H. and P. Liang. 2021. Less Cash, Less Theft? Evidence from Fintech Development in the People’s Republic of China. ADBI Working Paper 1282. Tokyo: Asian Development Bank Institute. Available: https://www.adb.org/publications/less-cash-less-theft-evidencefintech-development-prc Please contact the authors for information about this paper. Email: [email protected] Hongze Jiang is a PhD student at the School of Government of Sun Yat-sen University, Guangzhou, People’s Republic of China (PRC). Pinghan Liang is a professor at the Center for Chinese Public Administration Research/School of Government of Sun Yat-sen University, Guangzhou, PRC. The views expressed in this paper are the views of the author and do not necessarily reflect the views or policies of ADBI, ADB, its Board of Directors, or the governments they represent. ADBI does not guarantee the accuracy of the data included in this paper and accepts no responsibility for any consequences of their use. Terminology used may not necessarily be consistent with ADB official terms. Working papers are subject to formal revision and correction before they are finalized and considered published. The authors thank Stephen Gong, Peter Morgan, and the audience at the 8th Seminar on Asia and Pacific Economies (Suzhou) for their valuable comments. We gratefully acknowledge the financial support from the National Natural Sci ence Foundation (No. 71803149) and the Asian Development Bank Institute. Asian Development Bank Institute Kasumigaseki Building, 8th Floor 3-2-5 Kasumigaseki, Chiyoda-ku Tokyo 100-6008, Japan Tel: +81-3-3593-5500 Fax: +81-3-3593-5571 URL: www.adbi.org E-mail: [email protected] © 2021 Asian Development Bank Institute ADBI Working Paper 1282 Jiang and Liang Abstract This research investigates the impact of Fintech development on an important type of crime: theft. Based on Becker’s rational criminal theory, we suggest that Fintech development could mitigate theft activities by increasing the earnings from legitimate work, relaxing potential criminals’ financial constraints, and reducing the expected gains from theft. We established a unique dataset containing information on more than 1 million theft defendants during the period 2014–18, which we extracted from 874,000 judgment statements. Then, we aggregated them to construct a city-year panel of theft activities and matched it with the city-level economic activities and Fintech development level. The results show that a 1 standard deviation increase in the Fintech development level has a significant association with a 0.39 standard deviation decrease in thefts’ density. Robustness checks and instrumental variable estimation support the main results. Further, the development of Fintech reduces thefts’ density by reducing residents’ cash holding and providing more job opportunities. Finally, we utilized a nationally representative household survey to estimate the cost of theft for households, finding that victims suffer from more mental health problems, increasing their health expenditure. Our results suggest an unexpected source of welfare gain from the development of Fintech: an improvement in public security. Keywords: fintech, theft, crime, People’s Republic of China JEL Classification: G59, K14, K42, C81 ADBI Working Paper 1282 Jiang and Liang Contents 1. INTRODUCTION ................................................................................................... 1 2. LITERATURE AND THEORETICAL FRAMEWORK ................................................ 2 2.1 Literature Review ........................................................................................ 2 2.2 Theoretical Framework ............................................................................... 3 3. DATA .................................................................................................................... 4 4. RESULTS .............................................................................................................. 7 4.1 Baseline Results ......................................................................................... 7 4.2 Robustness Check ...................................................................................... 9 4.3 IV Results ................................................................................................. 10 4.4 Heterogeneity Analysis ............................................................................. 10 5. MECHANISM ANALYSIS ..................................................................................... 12 6. THE HEALTH COST OF THEFT ACTIVITIES ....................................................... 14 7. CONCLUSION ..................................................................................................... 16 REFERENCES ............................................................................................................... 18 ADBI Working Paper 1282 Jiang and Liang 1 1. INTRODUCTION This paper examines the relationship between Fintech development and theft activities. Theft is an important and frequent type of criminal activity worldwide, causing considerable cost to the society. A study in the US showed that, including the victim costs, criminal justice system costs, crime career costs, and intangible costs, the total per-offense cost for theft (larceny, motor vehicle theft, and household burglary) ranges from $3,532 to $10,772 in 2008 USD (McCollister, French, and Fang 2010). In Chicago, during the period 2001–12, larceny, burglary, and theft auto represented 89% of all 1.8 million property crimes (Herrnstadt et al. 2021). In the People’s Republic of China (PRC), theft accounted for 64%–71% of all criminal cases that the police filed between 1995 and 2010 (Chen and Liu 2013). Moreover, theft easily turns into other crime types that involve violence or the threat of violence, such as robbery, leading to even greater harm to the victims (Miller, Cohen, and Rossman 1993; McCollister, French, and Fang 2010). By definition, theft targets property and cash, so the way in which citizens use and hold cash in daily life affects the expected gains from theft activities. The PRC has witnessed the rapid development of Fintech, featuring the wide spread of mobile transfer and payment systems, in the past decade and has become a world leader in the adoption of Fintech services. This substantially reduces the need to hold cash for daily transactions, making theft less profitable for a rational decision maker who trades off the expected gains from theft against the opportunity cost. Hence, the PRC provides us with an ideal context in which to examine the impact of Fintech development on theft activities. Based on the rational criminal model à la Becker (1968), first, we analyzed the impact of Fintech on theft activities. The indications are that Fintech could contribute to decreasing theft by reducing its expected gains, relaxing financial constraints, and increasing the expected earnings from legitimate work. Then, we proposed our hypotheses. To test the hypotheses, we employed administrative data to construct a city-level measure of theft activities. We scraped all publicly available judgment statements on theft in the PRC during the period 2014–18 and used text recognition techniques to extract the key information about the defendants, courts, and cases. To the best of our knowledge, this is the most comprehensive nationwide measure of theft activities at the city level. Then, we matched this information with city-level economic activities, demographic structure, Fintech level, and government efforts to control crime to construct city-year paired panel data. The empirical analysis indicated that a 1 standard deviation increase in the Fintech level has a significant association with a 0.39 standard deviation decrease in theft activities. Various robustness checks supported our main results. To deal with the potential omitted variable problem, we interacted the geographic distance from the city to Hangzhou, the headquarters of Alipay (the leading Fintech service provider), with the national level of Fintech development as the instrumental variable for the regional Fintech level. The mechanism analysis showed that Fintech development reduces theft by facilitating mobile payments and activizing the local economy. However, we found no evidence that Fintech works by releasing financial constraints. Finally, utilizing a nationally representative household dataset, we showed that theft incurs substantial mental health problems for victims and increases their medical expenditure. This suggests the large unexpected social benefits of Fintech development. ADBI Working Paper 1282 Jiang and Liang 2 The organization of the rest of this paper is as follows: Section 2 reviews the literature and lays out the theoretical framework as well as the empirical hypotheses; Section 3 describes the data; Section 4 presents the empirical results; Section 5 examines the mechanism; Section 6 discusses the social cost of theft from the perspective of mental health; and Section 7 concludes. 2. LITERATURE AND THEORETICAL FRAMEWORK 2.1 Literature Review Fintech is the fusion of finance and technology, and the scope of Fintech activities ranges from mobile payments, money transfer, peer-to-peer loans, and crowdfunding to blockchain, cryptocurrencies, and robo-investing (Goldstein, Jiang, and Karolyi 2019). Fintech not only transfers the type of financial services, like the preceding ATMs and wire transfers, but also rapidly creates many more competitors outside the traditional sectors. Previous research has documented the beneficial role of Fintech development in the expansion of credit (Buchak et al. 2018; Hau et al. 2019a), the increase in household consumption (Xie et al. 2018), and the promotion of entrepreneurship (Fu and Huang 2018; Zhang et al. 2020a). Studies have also extensively investigated the relationship between Fintech and traditional financial services, and many have suggested that Fintech leads to more competition with the traditional service providers rather than broadening access to finance (Buchak et al. 2018; Fuster et al. 2019; Tang 2019; Vallée and Zeng 2019). Fintech is a wide area with an ever-expanding scope. We focused on the mobile payment side of Fintech. Starting with Baumol (1952), research has formally investigated the convenience of cash in transactions in economics. Lately, many studies have addressed the consumer choice between cash and non-cash payments and found that the share of cash use decreases with the transaction size (Borzekowski, Kiser, and Ahmed 2008; Ching and Hayashi 2010; Koulayev et al. 2016; Wang and Wolman 2016). This is due to the threshold of non-cash payments, such as the fixed per-transaction cost associated with credit cards, debit cards, and so on. On the other hand, no per-transaction cost arises from mobile payment as people use mobile money accounts to transfer money. Hence, as long as the Internet access is stable and there is wide use of mobile phones, there is no threshold for mobile payments, and mobile payments can substitute the use of cash even in small transactions. This has important implications for entrepreneurship growth and macroeconomic development (Beck et al. 2018; Huang and Huang 2018; Hau et al. 2019b; Zhang et al. 2020b). This paper relates the development of Fintech to criminal activities. Both theory and empirical studies have suggested that most criminals transit between legitimate jobs and illegitimate work, leading to high elasticity of the crime supply (Becker 1968; Freeman 1999). Research has documented well that the law enforcement effort could deter criminal activities (Corman and Mocan 2000; Di Tella and Schargrodsky 2004). The list of general economic and social conditions that affect crimes is long, and some are specific to the PRC, including unemployment (Raphael and Winter-Ember 2001, Zhang et al., 2018), education (Deming 2011), pressure in the marriage market in the PRC (Edlund et al. 2013; Cameron, Meng, and Zhang 2019), demographic structure (İmrohoroĝlu, Merlo, and Rupert 2006; Zhang et al. 2014; Guo et al. 2020a), urban–rural migration (Chen, Li, and Chen 2009), urban–rural income inequality (Zhang, Liu, and Liu 2011), social insurance (Zhang, Du, and Xu 2019), air pollution (Herrnstadt et al. 2021), and so on. Moreover, the characteristics of victims make a ADBI Working Paper 1282 Jiang and Liang 3 difference to criminals’ choice of target, especially in fraud, such as their demographic characteristics (Lee and Soberon-Ferrer 1997; Ross, Grossmann, and Schryer 2014; Lichtenberg et al. 2016) and their credit constraints and financial access (Gao, Ma, and Xu 2020; Liang and Jiang 2020). The development of Fintech in the PRC is to a large extent an initiative of private companies, such as Alibaba and WeChat; hence, it has no direct relationship with the deterrence of crimes. However, it may affect both the opportunity cost and the expected gains underlying the decision to commit theft. 2.2 Theoretical Framework We used the standard economic model of decision making (Becker 1968; Freeman 1999) to describe individuals’ tradeoff between theft and legal activity. Equation (1) compares the expected utility from those acts. (1 − 𝑝𝑝)𝑈𝑈(𝑊𝑊 𝑐𝑐)−𝑝𝑝𝑈𝑈(𝑆𝑆)>𝑈𝑈(𝑊𝑊) (1) in which 𝑊𝑊 𝑐𝑐 is the gain from successful theft, p is the probability of arrest, S is the extent of punishment once arrested, and W is the earnings from legitimate work. Hence, 𝑝𝑝𝑈𝑈(𝑆𝑆)+𝑈𝑈(𝑊𝑊) represents the expected cost to criminals, including the opportunity cost, punishment cost, and psychological cost. The decision maker will choose to commit theft in a given time period when the expected gains from theft exceed the expected cost. From this equation, we could derive three potential channels through which the Fintech level could mitigate theft activities. First, the diffusion of Fintech, especially the diffusion of mobile payment systems, directly reduces the gains from theft 𝑊𝑊 𝑐𝑐. From equation (1), it is clear that the expected gains from successful theft have a positive relationship with the likelihood of theft. According to the PRC’s criminal law, theft covers household burglary, larceny, theft of motor vehicles, and theft with weapons, and the statistics show that burglary and larceny account for the largest part (Zhang, Liu, and Liu 2011). Harbaugh, Mocar, and Visser (2013) provided experimental evidence that the probability of theft increases with the amount of money that it is possible to steal. Beck et al. (2018) used a general equilibrium model to show that mobile payments could reduce the probability of theft. The use of mobile payments through smartphones considerably reduces the need to hold cash in daily life and hence the money that is available to steal. Even though thefts could target smartphones, it is not easy for traditional burglars to steal the money in electronic accounts since the owners can easily lock their smartphone remotely. This considerably reduces the need to hold cash in daily life, either in public or at home. Second, the mitigating effect of Fintech varies with the local financial development level. Equation (1) implies that the risks and the attitudes toward risks influence the decision to commit theft. Previous research has shown that those individuals with a low risk aversion level or a lack of self-control and emotion control ability are more likely to steal (Arneklev et al. 1993; Zhang et al. 2014). One implication is that low-risk-averse individuals are more likely to make risky decisions when they face adverse shocks in daily life. The development of Fintech expands the access to credit and relaxes the financial constraints. Consequently, it is easier for individuals to smooth their consumption. Hence, the mitigating effect of Fintech is more salient in previously less financially developed areas as the financial constraints in these areas were more binding. ADBI Working Paper 1282 Jiang and Liang 4 Third, the mitigating effect of Fintech on theft depends on the local economy. It is noteworthy that many new service industries have arisen following the development of Fintech in the PRC. On the one hand, some new industries have substantially reduced the gains to theft; for instance, the emerging shared bikes (Mobile Bike) diminish the need to purchase private bicycles and consequently the valuable objects to steal. On the other hand, many new industries are labor intensive and create more job opportunities for low-skilled labor in urban areas, such as e-commerce (Alibaba), food delivery (Meituan), ride hailing (DiDi), and package delivery (SF express). Furthermore, the mobile transfer and payment system based on Alipay and WeChat reduces the market entry barriers and encourages entrepreneurship as commerce can operate without cash or credit card transactions and individuals can run a business without huge investment, such as live streaming on a short-form video platform (TikTok). Equation (1) also indicates that the earnings from legitimate jobs are the key to the decision to commit theft. In general, in the PRC, thieves are from younger cohorts with lower educational attainment (Zhang et al. 2014). Hence, they are low-skilled workers in the labor market, with fewer available job opportunities. However, the success of these new industries also depends on the local economy as scale of economy characterizes many of them. Hence, Fintech is more likely to activize developed regions and lead to a beneficial mitigating effect on theft. Hence, we proposed the following hypotheses for empirical examination: H1: The development of Fintech reduces theft activities. H2: The development of Fintech mitigates theft activities by facilitating mobile transfers and payments and reducing cash holding. H3: The mitigating effect of Fintech on theft is larger in the less financially developed areas. H4: The mitigating effect of Fintech on theft is larger in the regions with higher per capita income and a lower unemployment rate. 3. DATA For this research, we constructed panel data covering four province-level municipalities (Beijing, Shanghai, Tianjing, and Chongqing) and all 285 prefecture-level cities in the PRC during the period 2014–18. We extracted the data on city-level economic activities and the demographic structure from the statistical annuals of Chinese cities. We constructed the measure of Fintech development at Peking University under the name Peking University Digital Financial Inclusion Index of China, releasing it in 2020 (Guo et al. 2020a). The construction of this annual index took place in cooperation with Alipay, a leading mobile payment service provider in the PRC. This index covers approximately 2800 counties, 337 cities (municipal, prefecture, and county level), and 31 provinces over the period 2011–18. It consists of three sub-indices—coverage, depth of use, and extent of digitalization—and 33 variables. In general, the coverage index aggregates the number of Alipay accounts, the number of bank cards associated with an Alipay account, and so on, at the regional level. The depth index measures the use of digital finance services, including payments, mutual funds, loans, insurance, investment, and credit. The digitalization index considers the facilitation of payments, cost, and so on; in particular, it includes the size of mobile payments and the share of QR codes in the payment process. Guo et al. (2020a) presented the details of the index construction. In 2018, Hangzhou, Shanghai, Shenzhen, Nanjing, and Beijing ADBI Working Paper 1282 Jiang and Liang 11 Table 4: IV Result (1) (2) Second Stage First Stage Dependent Variable Theft Rate Fintech Fintech development –2.3197*** (0.4571) Instrumental variable –0.0001*** (0.0000) Control variables Yes Yes Year fixed effect Yes Yes City fixed effect Yes Yes RKF value 66.87 Observations 1383 1383 R-squared 0.0533 No. of cities 286 Note: The instrumental variable is the multiplier of the geographic distance from the city to Hangzhou and the national level of Fintech development. The control variables are the same as in Table 2. *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. The parentheses contain the standard errors clustered at the city level. In Columns (3)–(5), we divided the theft defendants according to their education level. Column (3) uses the number of theft defendants who received less than 6 years of education per 100,000 persons as the dependent variable. Column (4) uses the number of theft defendants who received 6 to 12 years of education per 100,000 persons as the dependent variable. Column (5) uses the number of theft defendants who received more than 12 years of education per 100,000 persons as the dependent variable. It is noteworthy that 88% of all the theft defendants in our sample received less than 12 years of education, that is, they did not finish high school. We ran OLS regression as equation (2) for these three subsamples. The results show that Fintech development has a similar mitigating effect for thieves with low and medium education levels, but the mitigating effect is the largest for those thieves with a relatively high education level. In other words, the development of Fintech increases the opportunity cost of theft action for highly educated individuals the most. Perhaps those highly educated individuals are more familiar with exploiting internet-complementary jobs. This is consistent with the finding that less-educated workers’ employment gains have been lower in Africa since the introduction of the fast Internet (Hjort and Poulsen 2019). We calculated the theft rate using the medium age of theft defendants (33 years) in Columns (6) and (7) and ran an OLS regression as equation (2) for these two subsamples. Fintech reduces the theft actions for all ages, but the mitigating effect of Fintech development is larger for relatively younger individuals, who are more adaptive to the development of Fintech and are more likely to take the new service jobs that the development of Fintech has created. ADBI Working Paper 1282 Jiang and Liang 12 Table 5: Heterogeneity Analysis (1) (2) (3) (4) (5) (6) (7) Dependent Variable Large Loss Value Theft Cases Small Loss Value Theft Cases Theft Education <=6 Years 6 Years <Theft Education <=Years Theft Education >12 Years Theft Age <=33 Years Theft Age >33 Years Theft Rate Fintech development –0.3910*** –0.1443 –0.3065*** –0.2914*** –0.4970* –0.3720*** –0.2982*** (0.1245) (0.0964) (0.0977) (0.1024) (0.2612) (0.1148) (0.1051) Control variable Yes Yes Yes Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes Yes Yes Yes City fixed effect Yes Yes Yes Yes Yes Yes Yes Observations 1,383 1,383 1,383 1,383 1,281 1,373 1,378 R-squared 0.1706 0.1650 0.2300 0.2392 0.1269 0.2495 0.2468 No. of cities 286 286 286 286 264 284 285 Note: The control variables are the same as in Table 2. *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. The parentheses contain the standard errors clustered at the city level. 5. MECHANISM ANALYSIS We first examined the heterogeneous impacts of different aspects of Fintech development. Table 6 considers the influence of the coverage index, depth index, and digitalization index, respectively. We used these three indices as the core explanatory variables and ran the regression as equation (2). It turned out that only the digitalization index, which measures the share of mobile payments, significantly reduces theft activities. This is consistent with our H2. The depth index has an insignificant correlation with theft activities. The coverage index, which measures the number of Alipay users and the number of bank cards associated with an account, even has a positive correlation with theft activities, though it is only significant at the 10% level. It might be due to this index having a high correlation with the number of mobile phone users, and mobile phones are a valuable target for theft activities. Moreover, it suggests that the relaxation of financial constraints is unlikely to be the reason for Fintech mitigating theft activities. Table 6: The Heterogeneous Impact of Sub-indexes (1) (2) (3) Dependent Variable Theft Rate Theft Rate Theft Rate Coverage index 0.3136* (0.1709) Depth index 0.0180 (0.1554) Digitalization index –0.1398*** (0.0303) Control variables Yes Yes Yes Year fixed effect Yes Yes Yes City fixed effect Yes Yes Yes Observations 1,383 1,383 1,383 R-squared 0.2606 0.2536 0.2873 Number of cities 286 286 286 Note: The control variables are the same as in Table 2. *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. The parentheses contain the standard errors clustered at the city level. ADBI Working Paper 1282 Jiang and Liang 13 To test H2 further, we used a regular expression to identify the cash-related and mobile phone-related cases from the text of judgment statements. In Table 7, Column (1) calculates the theft defendants of cash-related cases per 100,000 persons as the dependent variable, and Column (2) uses the theft defendants of mobile phone-related cases per 100,000 persons as the dependent variable. The mitigating effect of Fintech on cash-related theft activities is larger and more significant. This further supports our H2. Column (3) in Table 7 uses the deposit&loan/GDP to measure the regional financial development level. We added an interaction term between Fintech development and financial development in equation (2). The coefficient of the interaction term is insignificant, rejecting H3. Hence, Fintech does not work by expanding and substituting the access to credit. Table 7: Mechanism Analysis (I) (1) (2) (3) Cash-Related Mobile Phone -Related Dependent Variable Theft Rate Theft Rate Theft Rate Fintech development -0.4059** -0.3045* -0.3971*** (0.1893) (0.1682) (0.1152) Fintech development* financial development 0.0000 (0.0006) Control variables Yes Yes Yes Year fixed effect Yes Yes Yes City fixed effect Yes Yes Yes Observations 1,383 1,383 1,386 R-squared 0.2381 0.5157 0.2670 No. of cities 286 286 286 Note: The control variables are the same as in Table 2. *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. The parentheses contain the standard errors clustered at the city level. To test H4, we added the interaction term between the Fintech development and the log GDP per capita and the interaction term between the Fintech development and the unemployment rate, respectively, in equation (2). Columns (1) and (2) in Table 8 show the regression results. It turns out that the mitigating effect of Fintech is stronger when the GDP per capita is higher or the unemployment rate is lower, supporting our H4. Further, we used the median GDP per capita and the unemployment rate in 2014, respectively, to divide the cities into subsamples. Again, the results demonstrate that the mitigating effect of Fintech only appears in developed (high GDP) regions and low-unemployment regions, and the coefficient of Fintech is insignificant in underdeveloped (low GDP) and high-unemployment regions. The differences in coefficients are statistically significant between Column (3) and Column (4) and between Column (5) and Column (6), respectively. This supports our H4. Hence, the mitigating effect of Fintech on theft activities relies on the regional development level, indicating the possibility of broadening regional inequality. ADBI Working Paper 1282 Jiang and Liang 14 Table 8: Mechanism Analysis (II) (1) (2) (3) (4) (5) (6) Dependent Variable Developed Regions Underdeveloped Regions High Unemployment Low Unemployment Theft Rate Theft Rate Theft Rate Theft Rate Theft Rate Theft Rate Fintech 0.6903* –0.4189*** –0.5663*** –0.0561 0.0193 –0.6960*** (0.3544) (0.1191) (0.1640) (0.0931) (0.1165) (0.1743) Fintech* log GDP per capita –0.0026*** (0.0009) Fintech* unemployment rate 0.0431*** (0.0120) Control variables Yes Yes Yes Yes Yes Yes Year fixed effect Yes Yes Yes Yes Yes Yes City fixed effect Yes Yes Yes Yes Yes Yes Observations 1,383 1,383 700 683 702 681 R-squared 0.2822 0.2781 0.2347 0.4779 0.3747 0.2521 No. of cities 286 286 143 143 146 140 Note: The control variables are the same as in Table 2. *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. The parentheses contain the standard errors clustered at the city level. 6. THE HEALTH COST OF THEFT ACTIVITIES Finally, we attempted to estimate the cost of theft to victims to evaluate the social benefits of Fintech development in mitigating theft activities. Estimating the social cost of crimes is the key to designing public policies to fight against them. However, so far, there has been no systemic estimate of the cost of crimes in the PRC. The only exception is the study by Chen and Liu (2013), who utilized the statistical annuals to estimate the monetary cost of crime due to the loss of property, labor time loss of inmates, public security cost and lawyers’ cost, and so on. However, criminal activities not only entail tangible losses for the victims but also impose considerable intangible losses. Miller, Cohen, and Rossman (1993) estimated that the mental health cost accounts for more than half of the social cost of household burglary for victims. Therefore, here, we attempted to estimate the intangible cost of theft for victims. We employed the China Labor Dynamic Survey (CLDS) to investigate this issue. In 2016, the data covered 29 provinces and municipal cities, 158 cities, 11,631 households, and 21,086 adults. In this wave of the survey, the subjects were asked “in the past 12 months, did you have the experience of theft in the local area?” We used this information to construct a dummy variable, 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖, to indicate the individual recent experience of theft and employed the following equation (3) to examine the social cost of theft: 𝑦𝑦𝑖𝑖=𝛼𝛼+𝛽𝛽𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖+𝛾𝛾𝑋𝑋𝑖𝑖+𝜂𝜂𝑍𝑍𝑖𝑖+δ𝑢𝑢𝐶𝐶𝑢𝑢𝑆𝑆𝑢𝑢𝑖𝑖+𝜀𝜀𝑖𝑖 (3) 𝑦𝑦𝑖𝑖 is the dependent variable for individual i, corresponding to a series of questions in the 2016 CLDS, including the trust level (from 1 to 4, 4 being the highest), happiness (from 1 to 5, 5 being the highest), the perception of safety (from 1 to 4, 1 being the safest), and the frequency of mental problems that the individual experienced in the past month (from 1 to 5, 1 representing no problem). 𝑋𝑋𝑖𝑖 is a set of individual characteristics, including age, gender, marital status, income level, educational attainment, work status, hukou type, and self-reported physical health. 𝑍𝑍𝑖𝑖 is a set of household-level control variables, including the number of household members and the ADBI Working Paper 1282 Jiang and Liang 15 household income. 𝑢𝑢𝐶𝐶𝑢𝑢𝑆𝑆𝑢𝑢𝑖𝑖 is a dummy variable indicating that the household lives in an urban area, and 𝜀𝜀𝑖𝑖 is the robust standard error clustered at the community level. Table 9 reports the ordered probit regression results based on equation (3). To facilitate the interpretation of the coefficients, it presents the marginal effect of the ordered probit model. It demonstrates that the experience of having something stolen significantly decreases the trust level, happiness, and perception of safety and increases the frequency of mental problems. Other things being equal, the victims of recent theft are 1.81% less likely to agree that most people are trustworthy, 3.8% less likely to feel very happy, 24.59% less likely to agree that the community is very safe, and 10.72% more likely to have experienced mental problems in the past month. These indicate a considerable intangible cost of theft for victims. Table 9: The Cost of Theft for the Mental Health of Victims (1) (2) (3) (4) Dependent Variable Ordered Probit Trust=4 (Strongly Agree) Happiness=5 (Very Happy) Safety=1 (Very Safe) Mental Problem=1 (No) Whether stolen (yes=1) -0.0181*** -0.0380*** -0.2459*** -0.1072*** (0.0047) (0.0128) （ 0.0171 ） （ 0.0151 ） Individual controls Yes Yes Yes Yes Household controls Yes Yes Yes Yes Urban dummies Yes Yes Yes Yes Observations 19,913 19,914 19,914 19,914 Pseudo-R-squared 0.0629 0.0668 0.1073 0.0624 Note: The control variables include age, gender, marital status, income level, educational attainment, work status, hukou type, self-reported physical health, number of household members, and household income. *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. The parentheses contain the standard errors clustered at the city level. Table 10 further employs OLS regression on equation (3) to examine the impact of the experience of theft on household consumption. We used the total household consumption in the past year as well as the total household medical expenses, respectively, as the dependent variables.3 Columns (1) and (2) report the regression results. It is apparent that the experience of having something stolen has no significant impact on consumption, perhaps due to the need to compensate for the lost property, but significantly increases the medical expenses by 29.54%. Given that the average household medical expenditure in the 2016 CLSD sample is 8,167 RMB, this implies that the victims of theft increase their medical expenditure by as much as 2,412 RMB a year. Columns (3) and (4) employ panel data by merging the CLDS 2014 and 2016 data, which contain 7,870 repeated subjects. In addition to the control variables in equation (3), we added individual fixed-effect and city-level control variables as in equation (2). The magnitude of the coefficient is similar to that in Columns (1) and (2). To deal with the possible confounders, in Columns (5) and (6), we further used a PSM-DID estimate to identify the consequences of theft for victims. We used the individual and household control variables in equation (3), including age, gender, marital status, income level, educational attainment, work status, hukou type, and self-reported physical health, as the covariates to match the victims of theft with other households. Based on whether households had experienced theft in 2016, we divided the households into a control group and a treatment group. The nearest-neighbor 3 We have also examined other categories of consumption, but they are not significant. ADBI Working Paper 1282 Jiang and Liang 16 matching estimate results are similar, and the significantly increasing medical expenses further confirm the cost of theft for the mental health of victims. These results indicate the large intangible social benefits of Fintech development through the mitigating effect on theft activities. Table 10: The Consumption Response of Victims of Theft (1) (2) (3) (4) (5) (6) Dependent Variable Log Total Consumption Log Medical Expenses Log Total Consumption Log Medical Expenses Log Total Consumption Log Medical Expenses Model Multiple Sections Multiple Sections Panel Data Panel Data PSM-DID PSM-DID Whether stolen 0.0097 0.2954*** 0.0225 0.2541* (0.0387) (0.1394) (0.0398) (0.1353) Whether stolen*2016 0.0067 0.2471** (0.0288) (0.1015) Individual controls Yes Yes Yes Yes Yes Yes Household controls Yes Yes Yes Yes Yes Yes Individual fixed effect \ \ Yes Yes Yes Yes Urban dummies Yes Yes Yes Yes \ \ City controls \ \ Yes Yes \ \ Observations 19,729 19,855 15,226 15,237 17,696 17,710 R-squared 0.2334 0.1129 0.2412 0.1027 0.0039 0.0020 Note: The control variables include age, gender, marital status, income level, educational attainment, work status, hukou type, self-reported physical health, number of household members, and household income. The city control variables are the same as those in Table 2. *, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively. The parentheses contain standard errors clustered at the city level. 7. CONCLUSION This paper examined the impact of Fintech development on theft activities in the PRC. By utilizing a unique dataset on theft defendants during the period 2014–18, we showed that the development of Fintech has significantly mitigated the local theft activities, which we measured using the number of theft defendants in a particular year over the population. A 1 standard deviation increase in Fintech development is associated with a 0.39 standard deviation decrease in theft activities. This result held if we used the geographic distance from the city to Hangzhou to construct an instrumental variable for the diffusion of Fintech. The mitigating effect of Fintech is due to mobile payments and the new service jobs resulting from the development of Fintech. Finally, we showed that theft imposes a considerable mental health cost on victims. Our work indicated the unexpected social benefits from the mitigating effect of Fintech development on theft activities. An important implication of this research is that the rapid development and diffusion of new technology may generate unexpected welfare implications for the society. Fintech has a wide scope, and we mainly focused on the digital finance aspect, specifically mobile payments and the associated e-commerce. Further work could explore the impact of other parts of Fintech. On the other hand, we showed that the mitigating effect of Fintech varies across regions, and the public security in more developed areas benefits more from Fintech. This points to possible widening regional inequality stemming from the so-called “digital gap,” which would be worthwhile studying in the future. ADBI Working Paper 1282 Jiang and Liang 17 Moreover, research has observed that some crimes arise or become more salient in the Internet age, such as Internet fraud (Liang and Jiang 2020). This arouses curiosity about the relationship between the development of Fintech and other types of crime; for instance, does Fintech change the types of crime that a rational decision maker chooses? Though the demographic structure of theft defendants and inmates suggests that there are large differences between thieves and fraud criminals in terms of demographic characteristics and personalities (Zhang et al. 2014; Guo et al. 2020b), the question about criminals’ possible substitution decision among different types of crimes is interesting and awaits future research. ADBI Working Paper 1282 Jiang and Liang 18 REFERENCES Arneklev, B. J., H. G. Grasmick, C. R. Tittle, and R. J. Bursik. 1993. “Low Self-Control and Imprudent Behavior.” Journal of Quantitative Criminology 9 (3): 225–47. Baumol, W. J. 1952. “The Transactions Demand for Cash: An Inventory Theoretic Approach.” Quarterly Journal of Economics 66 (4): 545–56. Beck, T., H. Pamuk, R. Ramrattan, and B. R. Uras. 2018. “Payment Instruments, Finance and Development.” Journal of Development Economics 133: 162–86. Becker, Gary S. 1968. “Crime and Punishment: An Economic Approach.” Journal of Political Economy 76 (2): 169–217. Borzekowski, R., E. K. Kiser, and S. Ahmed. 2008. “Consumers’ Use of Debit Cards: Patterns, Preferences and Price Response.” Journal of Money, Credit and Banking 40 (1): 149–72. Buchak, G., G. Matvos, T. Piskorski, and A. Seru. 2018. “Fintech, Regulatory Arbitrage, and the Rise of Shadow Banks.” Journal of Financial Economics 130: 453–83. Cameron, L., X. Meng, and D. Zhang. 2019. “China’s Sex Ratio and Crime: Behavioural Change or Financial Necessity?” Economic Journal 129 (618): 790–820. Chen, G., S. Li, and Y. Chen. 2009. “The Research on the Impact of Population Migration on Criminal Rate.” Chinese Population Science (in Chinese) 4: 52–61. Chen, S., and F. Liu. 2013. “The Estimate of the Social Cost of Crime in China’s Transition.” World Economic Papers (in Chinese) 3: 42–64. Ching, A. T., and F. Hayashi. 2010. “Payment Card Rewards Programs and Consumer Payment Choice.” Journal of Banking and Finance 34 (8): 1773–87. Corman, H., and H. N. Mocan. 2000. “A Time-Series Analysis of Crime, Deterrence, and Drug Abuse in New York City.” American Economic Review 90 (3): 584–604. Deming, D. J. 2011. “Better Schools, Less Crime?” Quarterly Journal of Economics 126 (4): 2063–115. Di Tella, R., and E. Schargrodsky. 2004. “Do Police Reduce Crime? Estimates Using the Allocation of Police Forces after a Terrorist Attack.” American Economic Review 94 (1): 115–33. Edlund, L., H. Li, J. Yi, and J. Zhang. 2013. “Sex Ratios and Crime: Evidence from China.” Review of Economics and Statistics 95 (5): 1520–34. Freeman, R. 1999. “The Economics of Crime.” In The Handbook of Labor Economics, Vol. 3, edited by O. Ashenfelter and D. Card, 3529–71. New York: North Holland. Fu, Q, and Y. Huang. 2018. “The Heterogeneous Impact of Digital Finance on Rural Financial Demand: Evidence from China Household Finance Survey and the Digital Inclusive Finance Index of Peking University.” Journal of Financial Research (in Chinese) 11: 68–84. Fuster, A., M. Plosser, P. Schnabl, and J. Vickery. 2019. “The Role of Technology in Mortgage Lending.” Review of Financial Studies 32: 1854–99. ADBI Working Paper 1282 Jiang and Liang 19 Gao, N., Y. Ma, L. C. Xu. 2020. “Credit Constraints and Fraud Victimization—Evidence from a Representative Chinese Household Survey.” World Bank Policy Research Working Paper 9460. Goldstein, I., W. Jiang, and G. A. Karolyi. 2019. “To FinTech and Beyond.” Review of Financial Studies 32 (5): 1647–61. Guo, F., J. Wang, F. Wang, T. Kong, X. Zhang, and Z. Cheng. 2020a. “Measuring China’s Digital Financial Inclusion: Index Compilation and Spatial Characteristics.” China Economics Quarterly (in Chinese) 19 (4): 1401–18. Guo, S., P. Liang, and E. Xiao. 2020b. “In-Group Bias in Prisons.” Games and Economic Behavior 122: 328–40. Harbaugh, W. T., N. Mocan, and M. S. Visser. 2013. “Theft and Deterrence.” Journal of Labor Research 34 (4): 389–407. Hau, H., Y. Huang, H. Shan, and Z. Sheng. 2019a. “How FinTech Enters China’s Credit Market.” American Economic Review Papers and Proceedings 109: 60–4. ———. 2019b. “FinTech Credit and Entrepreneurial Growth.” Working Paper, University of Geneva. Herrnstadt, E., A. Heyes, E. Muehlegger, and S. Saberian. 2021. “Air Pollution and Criminal Activity: Microgeographic Evidence from Chicago.” American Economic Journal: Economic Policy forthcoming. Hjort, J., and J. Poulsen. 2019. “The Arrival of Fast Internet and Employment in Africa.” American Economic Review 109 (3): 1032–79. Huang, Y., and Z. Huang. 2018. “The Development of Digital Finance in China: Present and Future.” China Economics Quarterly (in Chinese) 17 (4): 1489–502. İmrohoroĝlu, A., A. Merlo, and P. Rupert. 2006. “Understanding the Determinants of Crime.” Journal of Economics and Finance 30 (2): 270–84. Kahn, M., and P. Li. 2020. “Air Pollution Lowers High Skill Public Sector Worker Productivity in China.” Environmental Research Letters 15: 1–11. Koulayev, S., M. Rysman, S. Schuh, and J. Stavins. 2016. “Explaining Adoption and Use of Payment Instruments by U.S. Consumers.” RAND Journal of Economics 47 (2): 293–325. Lee, J., and H. Soberon‐Ferrer. 1997. “Consumer Vulnerability to Fraud: Influencing Factors.” Journal of Consumer Affairs 31 (1): 70–89. Levitt, S. D. 1998. “Why Do Increased Arrest Rates Appear to Reduce Crime: Deterrence, Incapacitation, or Measurement Error?” Economic Inquiry 36 (3): 353–72. Liang, P., and H. Jiang. 2020. “Financial Access and Internet Financial Risk: An Empirical Study Based on Internet Pyramid Schemes Cases.” China Industrial Economics (in Chinese) 4: 116–34. Lichtenberg, P. A., M. A. Sugarman, D. Paulson, L. J. Ficker, and A. Rahman-Filipiak. 2016. “Psychological and Functional Vulnerability Predicts Fraud Cases in Older Adults: Results of a Longitudinal Study.” Clinical Gerontologist 39 (1): 48–63. McCollister, K. E., M. T. French, and H. Fang. 2010. “The Cost of Crime to Society: New Crime-Specific Estimates for Policy and Program Evaluation.” Drug and Alcohol Dependence 108: 98–109. ADBI Working Paper 1282 Jiang and Liang 20 Miller, T. R., M. A. Cohen, and S. B. Rossman. 1993. “Victim Costs of Violent Crime and Resulting Injuries.” Health Affairs 12 (4): 186–97. Raphael, S., and R. Winter-Ember. 2001. “Identifying the Effect of Unemployment on Crime.” Journal of Law and Economics 44 (1): 259–83. Ross, M., I. Grossmann, and E. Schryer. 2014. “Contrary to Psychological and Popular Opinion, There Is No Compelling Evidence that Older Adults Are Disproportionately Victimized by Consumer Fraud.” Perspectives on Psychological Science 9 (4): 427–42. Skogan, W. G. 1977. “Dimensions of the Dark Figure of Unreported Crime.” Crime & Delinquency 23 (1): 41–50. Supreme People’s Court of the People’s Republic of China. 2017. “Judicial Transparency by People’s Courts.” http://english.court.gov.cn/201703/14/content_28552039.htm. Tang, H. 2019. “Peer-to-Peer Lenders versus Banks: Substitutes or Complements?” Review of Financial Studies 32: 1900–38. Vallée, B., and Y. Zeng. 2019. “Marketplace Lending: A New Banking Paradigm?” Review of Financial Studies 32: 1939–82. Wang, Y. 2018. “Relative Capture: Quasi-experimental Evidence from the Chinese Judiciary.” Comparative Political Studies 51 (8): 1012–41. Wang, Z., and A. L. Wolman. 2016. “Payment Choice and Currency Use: Insights from Two Billion Retail Transactions.” Journal of Monetary Economics 84: 94–115. Xie, X., Y. Shen, H. Zhang, and F. Guo. 2018. “Can Digital Finance Promote Entrepreneurship? Evidence from China.” China Economics Quarterly (in Chinese) 17 (4): 1557–80. Zhang, D., L. Li, and C. Tong. 2018. “Minimum Wage, Rural Migrants’ Unemployment, and Crime.” China Economics Quarterly (in Chinese) 17 (3): 1035–54. Zhang, D., Y. Wang, X. Meng, and L. Cameron. 2014. “The Determinants of Migrant’s Crime: An Empirical Study Based on a Prison Survey.” China Economics Quarterly (in Chinese) 14 (1): 83–112. Zhang, X., Y. Tan, Z. Hu, C. Wang, and G. Wan. 2020a. “The Trickle-Down Effect of Fintech Development: From the Perspective of Urbanization.” China & World Economy 28 (1): 23–40. Zhang, X., J. Zhang, G. Wan, and Z. Luo. 2020b. “Fintech, Growth, and Inequality: Evidence from China’s Household Survey Data.” Singapore Economic Review 65 (S1): 75–94. Zhang, Y., J. Du, and Y. Xu. 2019. “The Dual Structure of Urban Social Security and Crime Rate: An Empirical Study Based on 227 Cities’ Data.” World Economic Papers (in Chinese) 1: 84–100. Zhang, Y., S. Liu, and L. Liu. 2011. “Can We attribute Increasing Criminal Rate to Enlarging Urban-rural Inequality in China?” Economic Research Journal (in Chinese) 2: 59–72.