When the Elderly Turn to Petty Crime

Abstract

Declines in offending in older age have been consistently observed in nearly every criminological study to date. Because of this, theories that address offending in older life focus exclusively on explaining decreases in crime or “desistance.” However, recent increases in elderly arrest rates in some aging societies provide a unique opportunity to forward theories of older age offending with an empirical touchstone. Using Japan as a case study, this article draws from a social integration perspective to examine whether changes in family and economic integration are associated with increasing arrest rates. Using prefectural fixed-effects models with elderly arrests from 1995 to 2004, the findings suggest that weakened family integration is associated with elderly arrest rates, particularly for petty crimes. This article situates these results within the context of Japan and discusses how social integration and later life offending may be related in aging societies like Japan.

Keywords

elderly crime Japan social integration

One of the core foundations of criminological theory is the observed relationship between age and crime. As Steffensmeier and colleagues (1989) write, “the proposition that involvement in crime diminishes with age is one of the oldest and most widely accepted in criminology” (p. 803). Contemporary theories of crime and aging were developed within this context, and they propose various explanations for declines in older age offending. Individuals desist from crime because they develop bonds to conventional institutions such as marriage or employment (Laub & Sampson, 2003; Sampson & Laub, 1995); yield to social pressures toward conformity (Steffensmeier, Allan, Harer, & Streifel, 1989); or experience maturation involving psychological, sociological, and biological mechanisms (Moffitt, 1993, 1997). Because declines in older age have been consistently observed, it has proven difficult to differentiate among these processes, and criminological theory and research seldom consider offending at the oldest ages.

A recent trend of increasing arrests among the elderly in some aging societies provides an unprecedented opportunity to revisit conceptualizations of offending in older age. Starting in the mid-1990s, Japan has witnessed a near tripling of the arrest rate among people 65 years and older (Ministry of Justice, 2008). In 2007, 13% of all arrestees were 65 years and older and the majority was arrested for the first time in elderly age (Ministry of Justice, 2008). For comparison, in the United States, 0.7% of arrestees were elderly (Federal Bureau of Investigation, U.S. Department of Justice, 2008). Although the case of Japan has received the most publicity, there is some indication that elderly arrest rates may be also increasing in other aging societies, including Korea, China, and Taiwan (Chen, 2015; Hyun-chae, 2015; Ji-sook, 2008; Ota, 2009; Taipei Times, 2012).

These trends are not easily explained by contemporary criminological theories on crime and aging; however, recent studies suggest that social isolation, economic need, and lenient punishments may be contributing to increasing crime rates among the elderly (Chen, 2015; Ezaki, 2014; Ministry of Justice, 2008; Ota, 2009). Although this research provides important insight into individual-level explanations for elderly offending, it is based on selected samples (Ezaki, 2014; Ota, 2009) or does not consider confounding factors in regression analyses (Ministry of Justice, 2008; Ota, 2009). These limitations result from the very difficult challenges of studying elderly offending at the individual level, where administrative data are rarely available (Brinton, 2003) and the stigma around criminal offending has resulted in a complete absence of crime-related questions in nationally representative social surveys. Apart from these difficulties, individual-level explanations overlook social and economic dislocations that are more broadly experienced by older populations in rapidly aging societies (Ezaki, 2014; Ministry of Justice, 2008; Ota, 2009). The context of population aging suggests a macrolevel approach, where regional changes in social integration potentially explain increases in elderly arrest rates.

In this article, I analyze unique administrative data on all prefectural-level elderly arrests in Japan from 1995 to 2004, in order to examine increases in older age offending and to identify social integration factors that are associated with rising rates. I distinguish between female and male arrest rates, and I find that the relationships between arrest rates and social integration are gendered. I then draw on in-depth interviews with a small number of individuals living in Tokyo and arrested in older age to help situate the prefectural-level findings within the Japanese context. Although the use of administrative records to study crime, particularly petty crimes like theft and shoplifting, is not without limitations, this article contributes one of the first empirical examinations of rising elderly arrest rates in Japan.

Elderly Offending and Social Integration

The case of increasing elderly arrest rates in Japan and other aging societies seems to defy most contemporary theories on crime and aging. Indeed, declines in offending in older age, and the shape of the age–crime curve more generally (with a peak in offending in adolescence and decrease over age), are perceived to be so commonplace that they are considered a “social fact” (Greenberg, 1985; Hirschi & Gottfredson, 1983; Steffensmeier et al., 1989; Tittle & Grasmick, 1997). Even apart from elderly crime, however, recent trends in Japan appear to challenge beliefs about the universality of these patterns. Research on homicide rates among men in Japan finds that the age–crime curve has flattened over the past several decades (Hiraiwa-Hasegawa, 2005).

Notwithstanding these recent trends in Japan, the basic shape of the curve has been generally uncontested, although the reasons for the age–crime patterns have triggered much debate. A main distinction among perspectives is the comparative role of early life experiences as opposed to later life factors for explaining crime declines over age. On the one hand, maturation perspectives focus on childhood factors and social bonds for shaping criminal propensities over an individual’s lifetime (Gottfredson & Hirschi, 1990; Moffitt, 1993, 1997). Changes in offending over age do not reflect adult transitions, such as getting married or finding a good job; rather, declines in offending result from aging processes that are universally experienced. In contrast, life-course perspectives on crime propose that social bonds and attachments shape offending not only in childhood but also in throughout adulthood (Laub & Sampson, 2003; Sampson & Laub, 1995; South & Messner, 2000). Adult social bonds have the potential to strengthen attachments to normative lifestyles, reduce negative peer influences, and establish conventional routines that cause declines in offending. Although aging plays some role in desistance, sociological factors over the entire life course are important for declines.

Prior research on elderly offending in Japan suggests that life-course perspectives on crime may help to explain the trends. Consistent with the theory’s emphasis on later life social bonds, it may be that the unexpected weakening or loss of attachments, such as divorce or forced retirement, place individuals at higher risk of offending. A small survey study of elderly offenders in Japan found that higher proportions of individuals lived alone and were unmarried prior to their arrest, compared to the general elderly population (Ota, 2009). Another recent study in Japan found differences in the strength of social bonds of elderly individuals who were arrested for shoplifting, as compared to those that had no history of prior arrest (Ezaki, 2014). Although these studies consider selected samples and often do not adjust for other potentially endogenous factors, such as economic hardship, they provide some of the only research on increasing elderly arrests in Japan.

In this article, I move from these individual-level explanations to examine whether contextual levels of social integration explain changes in elderly arrest rates over the period of increase from 1995 to 2004. Sometimes described as the “social integration-regulation thesis” (Kubrin, Wadsworth, & DiPietro, 2006, p. 1560), this idea has its roots in Durkheimian concepts of social integration and social regulation. Both aspects—integration and regulation—are intimately related, where integration is the attachment people feel toward others, which provides meaning beyond personal troubles, and regulation is the behavioral norms and rules that facilitate integration. Durkheim suggested that social integration and regulation together enable societies to exert social control over deviant tendencies, such as suicide (Durkheim, 1951). Because of the close linkages between integration and regulation, contemporary scholarship typically focuses on social integration.¹ As Crutchfield and colleagues (1982) explain, “an integrated social system provides 1) a high degree of consensus in norms, values, and goals; 2) cohesiveness or social solidarity; and 3) a sense of belonging or “we feeling” among persons living in the community in question” (p. 468). In the United States, weakened social integration is thought to manifest through increasing proportions of divorced households, single-parent families, and unemployed individuals, as well as higher rates of residential mobility. Macrosocial associations between integration and crime are the result not only of aggregated individual experiences but also of broader regional changes that affect all residents in an area (Stockard & O’Brien, 2002). By focusing on social integration, the key question is not what types of elderly individuals are more likely to commit crimes, but rather, what social and economic conditions are more likely to lead to elderly offending (Blau & Blau, 1982). Social integration is often a central concept in theories such as social disorganization and control (Crutchfield, Geerken, & Gove, 1982), as well as macrolevel perspectives such as the institutional anomie theory (Messner & Rosenfeld, 1994) and the general strain theory at the community level (Agnew, 1999). These latter theories differ with regard to their intervening mechanisms² and are fruitful areas for future research.

In the case of Japan, recent trends in elderly arrest rates are occurring within a rapidly aging population that has recently undergone major social and cultural transitions. Japan is the first country to experience population decline, the result of below replacement-level fertility since 1973. The persistence of very low fertility rates is unprecedented and has resulted in a multitude of social problems and anxieties about the future (Coulmas, 2007; Morgan & Taylor, 2006; Ogawa & Retherford, 1997; Onishi, 2004). Along with population aging, the country has experienced a broad range of social structural shifts, including a weakened economy and changes in cultural norms, which have created dramatic changes to the family and the labor market. In the next sections, I discuss these various aspects of social integration within the context of Japan and population aging.

Family integration

Family integration is considered a key dimension of social integration. Weakened family integration not only compromises social norms, attitudes, and feelings of social solidarity but also reduces the effectiveness of informal social controls for deterring criminal behavior (Crutchfield et al., 1982; Sampson, 1987; Stockard & O’Brien, 2002). In the U.S. context, family integration is often measured as the proportion of female-headed households or the percent of individuals divorced. In Japan, the issue of female-headed households is less relevant; however, divorce is increasingly a social issue of concern even though the overall rates are quite low (1.84 per 1,000 population in 2013) and have been declining in recent years (Statistical Handbook of Japan, 2014). Despite these recent trends, divorce rates among longer marriages have been increasing over the long term,³ and the media has focused on the negative impact of divorce among older men, suggesting that their families are “discarding” them (Fuess, 2004). Moreover, as interviews from this study will describe, divorce in Japan can often entail drastic family disruption, such as cutting off all ties to family members. For these reasons, the social meaning of divorce in Japan may indicate a more severe form of weakened family integration compared to the U.S. context, despite low rates of marital dissolution in Japan overall.

Apart from marital status, the living situation of older adults may also be an important indicator of family integration in Japan. During the same period of increasing elderly arrest rates, the prevalence of three-generational households and adult children caring for their elderly parents has declined (Ogawa & Retherford, 1997; Traphagan & Knight, 2003). This decrease is related to changing norms regarding the responsibilities of adult children and is connected to the dynamics of population aging, particularly the difficulties of caring for a growing and disproportionate number of elderly individuals who are living longer. In the context of population aging and in societies that are transitioning away from three-generational households, changes in the living situation of older adults, such as whether they live alone or with a spouse as opposed to with their children, may be a particularly relevant measure of family integration.

Economic integration

Economic integration is considered another important dimension of social integration. Living in an impoverished or highly disadvantaged area is thought to lead to despair and a sense of hopelessness about the future (Burr, Hartman, & Matteson, 1999; Kubrin et al., 2006). In the case of the elderly in Japan, the country’s decadelong recession and anxieties about the viability of the government pension system (Onishi, 2004, 2005) provide an important backdrop to economic integration measures, such as poverty and unemployment rates. The shifting age structure has made it difficult to find work not only among the elderly, who are increasing the pool of available older workers, but also for young people, who are facing a shrinking labor market economy. In a society where social norms have traditionally dictated that aging elders are revered and cared for, contemporary experiences of economic uncertainty likely conflict with individual expectations.

Other factors related to social integration

Apart from familial and economic factors, social integration is thought to depend on the extent of mobility in a region. Mobility not only disrupts feelings of social cohesion and a sense of belonging, but it also weakens the effectiveness of informal social controls (Crutchfield et al., 1982). Studies in the U.S. context operationalize the concept of mobility as the frequency of recent residential moves (Crutchfield et al., 1982; Sampson, 1985); however, in the case of older individuals in Japan, measures of commuting regions may be more appropriate for several reasons. Areas with high levels of daytime commuters reflect a lack of guardianship and informal social controls during the day (Cohen & Felson, 1979) that are particularly relevant to mostly retired, older individuals. Feelings of social integration and cohesion are likely quite different for older individuals living in areas with growing numbers of outbound commuters during the day.

Social Integration, Crime, and Gender

Because the meaning of social integration factors and the nature of criminal offending are both very different among older men and women, it is likely that social integration and crime are differentially associated by gender. This expectation is particularly relevant to a highly gendered society, such as Japan. Across time and place, women engage in much lower rates of crime across all crime types, with the exception of prostitution, and they are much less likely to be involved in serious offenses (for reviews, see Kruttschnitt, 2013; Steffensmeier & Allan, 1996). The situation in Japan follows these patterns, where older women have lower levels of arrest and are more likely to be arrested for petty offenses as compared to serious crimes.

Theoretically, some integration factors have been discussed as differentially important to women and men. Integration related to the family, and particularly to childbearing, may be especially salient to women (Kubrin et al., 2006; Steffensmeier & Allan, 1996). For women, who are socialized in the “ethic of care” and caretaking of others (Steffensmeier & Allan, 1996), social status and identity are more closely tied to family. The living situation of elderly adults, such as the prevalence of three-generational households, may be a particularly important integration factor for older women in Japan. Whereas elderly men more often rely on their wives for caretaking in older age, women often outlive their spouses and must depend more on their adult children (Ogawa & Retherford, 1993). Among men, economic integration factors are thought to be more important for crime (Kubrin et al., 2006; Steffensmeier & Allan, 1996). Men are typically socialized to value status and identity based on economic achievement and occupation. Their contributions to the family are often through these channels, where financially supporting a family is regarded more highly than caretaking practices. In Japan, where men dominate the country’s “permanent employment system” (Brinton, 1989), employment may be particularly salient to men. In addition, integration factors that are related to the labor force may be more important for men compared to women. For example, weakened feelings of belonging in commuter regions may be exacerbated among older men, who are more likely to have recently retired from work, as compared to older women in Japan.

Although the theoretical arguments for gendered differences in social integration and crime seem convincing, research has found few variations by gender at the aggregate level (for a review, see Kruttschnitt, 2013). In these studies, factors such as family and economic integration have had largely similar associations, although with some differences in magnitudes, for male and female crime rates. Moreover, research on elderly offending at the individual level indicates few gender differences in feelings of social integration, such as social distrust (Ezaki, 2014). Notwithstanding these findings, the evidence regarding gendered differences at the macrolevel is limited (Kruttschnitt, 2013), and both the theoretical reasons and the specific context of Japan, as a highly gendered society, warrant analyses that distinguish elderly crime among men and women.

Data, Method, and Measures

Data

In this article, I use prefecture-level arrest records to examine whether increases in elderly arrests are associated with changes in social integration. In Japan, there are 47 prefectures with an average geographic area of 8,000 km², which is slightly larger than the U.S. state of Delaware. The prefectural unit of analysis is comparably larger than most U.S. studies of aggregate crime rates, which typically examine counties, but it is relatively smaller both in geographical area and in population size compared to U.S. states. The data come from several sources. The main outcome measure—elderly arrest rates—is based on prefecture-level information on all elderly arrests from 1995 to 2004. Although much of this information is publicly available, it is seldom used and involves a long and complicated process to compile. Each prefectural police department publishes an annual report with detailed arrest information, including age-specific arrest tables. The National Diet Library in Tokyo collects some of these reports; however, there is not a single repository for all prefectures and years studied here. I therefore contacted each of the 47 prefectural police departments by letter or e-mail, and most departments were responsive to requests for their reports. After collecting most of the data in this manner, the Tokyo Metropolitan Police Department (TMPD) provided me with complete information on the number of elderly arrests by prefecture from 1995 to 2004. Information from the TMPD includes all prefectures and additionally distinguishes by gender and offense type, which is a level of detail that is not always included in the prefectural police department reports. Information on all other measures is publicly available and comes from Japanese government agencies such as the Census and Statistics Bureau, as well as an annual government publication entitled Social Indicators by Prefecture (see the Measures section and Table 1 for more details).

Table 1.
Description, Source, and Year for All Variables Used in Fixed-Effects Panel Models.

Variable Source and Observed Year

Dependent variables

Older arrest rate (per 100,000 persons) Tokyo Metropolitan Police and the Statistics Bureau (population): 95, 98, 01, 04

Independent variables

Marital status of older adults (%) Japan census: 90, 95, 00, 05

Living situation of older adults (%) National institute of population and social security research: 95, 98, 01, 04

Labor force status of older adults (%) Japan census: 90, 95, 00, 05

Receipt of cash assistance among older adults (per 1,000) Social indicators by prefecture: 95, 98, 01, 04

Commuters to other prefectures (% of employed) Japan census: 90, 95, 00, 05

Number of police per 1,000 residents Social indicators by prefecture: 95, 98, 01, 04

Density: Number of people per 1 km² inhabitable area Social indicators by prefecture: 95, 98, 01, 04

Note. The term “older adults” refers to people 65 years and older.

The data have several advantages for studying elderly offending. First, they include the entire universe of all elderly arrestees in Japan. Second, they concern the period of increase, prior to national attention to the issue.⁴ Third, they differentiate between male and female arrestees, enabling an analysis by gender. Fourth, Japanese criminal statistics and arrest information are generally viewed as more consistent compared to administrative records from other countries, such as the United States (Finch, 2000; Park, 1992). Compared to the Uniform Crime Reports, Japanese statistics reflect more standardized reporting and policing policies across regions.

However, despite these strengths, Japanese arrest records remain an imprecise measure of criminal offending, which reflects crime, police policies, and regional practices.⁵ I discuss these limitations throughout the rest of this article; however, since official records are currently the only available source of crime and offending information in Japan, these data provide a first step toward examining elderly offending in aging populations.

Method

I utilize prefecture-level regression models to investigate how changes in elderly arrest rates are associated with changes in marital status, living situation, employment, and other factors between 1995 and 2004. The models are fixed-effects panel models, which consider changes within Japan’s 47 prefectures over time, and they estimate associations between arrest rates and explanatory variables at four time points during the period of change: 1995, 1998, 2001, and 2004 (N = 188 prefecture years). Prefectural-based fixed-effects models are well suited to study changes in offending and social and contextual factors for several reasons. As opposed to pooled cross-sectional models, the advantage of prefectural-based fixed-effects models is that they account for any unobserved factors that are specific to the prefecture and do not vary over time. For example, fixed-effects models adjust for prefectures that consistently have higher crime rates, unique policing environments, or different practices for recording and processing arrests. Moreover, examining changes by prefecture, as opposed to pooling prefectural-level data to the national level, enables me to consider more fine-grained variation in social and contextual changes that may be importantly related to changes in arrest rates. Apart from these substantive advantages, fixed-effects models are most appropriate when data have large Ns (in this case, prefectures) across relatively few time points. Pooling data to the national level may be useful if researchers want to study dynamic or temporal patterns at a national level and have access to national-level data over a long period of time (Frees, 2004); however, the aim of this article is to understand changes in relationships between offending and social and contextual factors among prefectures during the period of increase.

For the prefectural fixed-effects models, I use the Huber/White/sandwich estimate of variance to adjust standard errors due to the presence of correlated errors (Wooldridge, 2002). In order to construct some independent variables (e.g., marital status, labor force status, and commuters, see Table 1) for years 1998, 2001, and 2004, I use ordinary least squares regression to estimate values based on information from the Japanese censuses for 1990, 1995, 2000, and 2005. For example, I estimate the marital status variable for each prefecture based on observed values in 1990, 1995, 2000, and 2005, allowing for nonlinearity for the construction of the value by incorporating squared and cubic terms for the years. I then use the regression estimates to predict values for the years 1998, 2001, and 2004. Using estimated values for these years introduces error, which should bias the resulting fixed-effects regression results for these coefficients downward; given this, it is possible that the estimates are conservative relative to the true associations. In addition, the models that I present in this article do not include fixed effects for years, which additionally control for period-based associations; however, the results are generally consistent with the more conservative approach and I describe any differences in the text and footnotes.

Using these models, I examine total arrest rates for individuals 65 years and older, and I also distinguish by crime type such as larceny and violent offenses.⁶ Because arrest rates reflect both offending and police practices, some researchers that use administrative records have focused on violent offenses that involve a victim, such as homicide or other violent crimes, in order to address the very salient concern that changes in arrest rates reflect police practices only. While I include this approach here and I discuss other methods to help address this concern later in the article, I suggest that focusing on violent offenses has considerable limitations in research on elderly offending. Partly due to physical restrictions of aging that preclude violent crime, research on elderly offending in the United States finds that older individuals typically engage in petty theft such as shoplifting and other minor crimes (Cullen, Wozniak, & Frank, 1985; Feldmeyer & Steffensmeier, 2007). As this article’s results will describe, elderly individuals in Japan are no exception and focusing on violent crimes therefore excludes the majority of offenses for which elderly individuals are arrested. Given that the recent increase in elderly arrests is driven mostly by larceny offenses, I examine those offenses as a separate category. In addition to distinguishing by crime type, I analyze separate models for men and women. Theoretically, social integration literature suggests that the correlates of offending may differ depending on gender. Although the limited empirical evidence does not typically find gender differences, as discussed above, there are reasons to expect that some social integration factors would be more or less applicable to men as opposed to women.

Following these main models, I present findings from two additional analyses that provide supplementary information to aid in the interpretation of results. The first helps to assess whether changes in police practices might explain the associations, by estimating a model of young adult arrest rates using the elderly specific social integration factors. If changes in police practices, as opposed to changes in elderly offending rates, were systematically associated with social integration factors, the associations would be similarly observed in models that consider other age-specific arrests that are presumably impacted by the police practices. The second analysis provides suggestive evidence about the social meanings of family and economic integration factors in the Japanese context. Using in-depth interviews with a small, selected sample of elderly offenders, I discuss how social and economic integration factors are perceived and experienced by these individuals living in Tokyo, Japan. Although circumscribed, this final section complements the main prefectural-level analyses by providing contextualization for the interpretation of social and economic integration factors (Sale, Lohfeld, & Brazil, 2002; Small, 2011), which is a particularly important consideration in non-Western research (Scrimshaw, 1990).

Measures

Elderly arrest rates

Elderly arrest rates by prefecture are the key dependent variables. For a given year, the number of age-specific arrests by prefecture and gender (from the TMPD) was combined with age-specific population estimates (from the Statistics Bureau). Rates are calculated as the number of elderly adult arrestees per 100,000 persons 65 years and older and are logged.⁷

Marital status

The measure of marital status of individuals 65 years and older is separated by gender and measured using five categorical variables: the percent of older men or women who are married, divorced, widowed, never married, and who do not report status.

Living situation

The living situation of older adults (aged 65 years and older) is measured using four categorical variables: the percent of older people living alone, living with their spouse, living with their children, and living with nonfamilial individuals. I expect that the decline of three-generational households and the increase of elderly people living alone will be associated with increasing elderly arrest rates.

Labor force status

The measure of labor force status for elderly adults is separated by gender and measured by four categorical variables: not in the labor force, employed, unemployed, and status not reported in the census. I expect that weakened labor force attachment, measured as the percent unemployed, will be associated with arrest rates.

Receipt of cash assistance

To measure the extent of prefecture-level poverty among elderly adults, I include the rate of people 65 years and older receiving cash assistance. While there are many determinants of receipt besides economic situation, such as the application procedures of the prefecture and the stigma attached to receiving aid, it is the best available measure to capture poverty rates among elderly adults.

Commuter regions

The percent of employed persons that commute to a different prefecture for work is used to capture whether the prefecture is a commuting suburb.

Police ratio

The number of police officers per 1,000 residents measures changes in police presence that may affect arrest rates.

Density

Population density accounts for demographic changes related to population concentration, which are commonly correlated with crime rates.

Results

Before presenting estimates from the prefecture-level fixed-effects regression models, I describe the nature and extent of changes in elderly arrest rates at a national level. From 1995 to 2004, the number of elderly arrests per 100,000 individuals (65 years and older) increased 136% from 62.6 to 147.5. The majority of the change was due to arrests for larceny and other minor crimes, although there were increases across all crime categories (see Figure 1). Although arrest rates for all age categories did increase over this period, the increase at the oldest ages is considerably higher compared to the younger ages. For comparison, among age groups under age 50, arrest rates increased about 35% over this period.

Figure 1.
Number of arrests per 100,000 older individuals (65 years and older), by offense type and year. Source. Adapted from Ministry of Justice (2008).

Table 2 describes the means and standard deviations for all measures across prefectures from 1995 to 2004. The average elderly arrest rate by prefecture was 53 per 100,000 in 1995 and 139 per 100,000 in 2004, representing a 162% increase by prefecture. Distinguishing by gender, the elderly male arrest rate increased by 171%, from 80 per 100,000 in 1995 to 217 in 2004. For elderly females, the average arrest rate across prefectures increased by 147%, from 34 in 1995 to 84 per 100,000 in 2004. The arrest rate for violent and serious crimes is comparably very low, where the average rate was two per 100,000 in 1995 and seven per 100,000 in 2004. These rates are slightly higher among men, where the average arrest rate was 15 per 100,000 in 2004; however, it is clear that the majority of elderly arrests, as well as the increase in arrests, can be attributed to larceny and other nonserious crimes. Although all prefectures experienced increases in both elderly male and female arrest rates, there is notable regional variation in the magnitude of the changes. Some prefectures saw increases of more than 200% in arrest rates, while others experienced increases of less than 50% (see Figure 2). Distinguishing by gender, the prefectures with the largest increases in male arrest rates are sometimes, but not usually, the same prefectures that also experience the highest increases in female arrest rates.

Table 2.
Means and Standard Deviations for All Prefectures in Japan, 1995 and 2004.

Variable 1995 2004

Mean (%) SD Mean (%) SD

Dependent variables

Older arrest rate: total 52.73 (26.79) 138.97 (33.31)

Men 80.03 (50.19) 216.72 (60.06)

Women 33.81 (15.60) 83.52 (23.95)

Older arrest rate: violent/felonious 2.09 (1.31) 6.95 (3.48)

Men 4.80 (3.15) 15.43 (7.73)

Women 0.25 (0.36) 0.92 (0.81)

Older arrest rate: larceny 42.29 (18.63) 102.82 (26.44)

Men 57.08 (26.71) 139.17 (35.26)

Women 32.03 (14.97) 77.16 (24.38)

Independent variables

Marital status of older adults (%): men

Married 84.37 (1.22) 82.89 (1.77)

Divorced 1.55 (0.51) 2.44 (0.62)

Widowed 12.53 (0.65) 11.43 (0.78)

Never married 1.27 (0.38) 1.98 (0.65)

Unreported 0.28 (0.31) 1.25 (1.09)

Marital status of older adults (%): women

Married 43.22 (2.39) 46.63 (2.11)

Divorced 3.01 (0.70) 3.56 (0.82)

Widowed 50.68 (2.11) 45.48 (2.43)

Never married 2.67 (0.74) 3.08 (0.79)

Unreported 0.43 (0.36) 1.25 (0.91)

Living situation of older adults (%)

Alone 11.45 (4.14) 13.63 (3.75)

With spouse 27.55 (6.98) 33.56 (6.84)

With children 57.32 (11.03) 48.67 (10.08)

With others 3.67 (0.59) 4.16 (0.79)

Labor force status of older adults (%): men

Not in labor force 57.46 (4.23) 64.75 (3.44)

Employed 40.35 (4.43) 31.68 (3.46)

Unemployed 1.91 (0.59) 1.61 (0.37)

Unreported 0.27 (0.33) 1.95 (1.99)

Labor force status of older adults (%): women

Not in labor force 83.38 (3.17) 84.62 (2.29)

Employed 16.18 (3.24) 14.20 (2.39)

Unemployed 0.18 (0.07) 0.21 (0.06)

Unreported 0.26 (0.17) 0.97 (0.99)

Receipt of cash assistance among older adults (per 1,000) 13.80 (8.37) 17.58 (9.92)

Commuters to other prefectures 5.55 (8.22) 6.01 (8.17)

Number of police per 1,000 residents 1.62 (0.36) 1.74 (0.33)

Density 1,363.56 (1,613.17) 1,370.313 (1,659.79)

N 47 47

Note. The term “older adults” refers to people 65 years and older.

Figure 2.
Percentage change in elderly arrest rates in Japan from 1995 to 2004, by prefecture and gender.

For the independent variables, the changes from 1995 to 2004 reflect the aforementioned decline in three-generational households and a worsening economic situation for elderly adults. In 2004, a larger percent of elderly adults were living alone or with their spouse rather than with their adult children. There was also a higher percent of elderly adults receiving cash assistance from the government. For marital status, a slightly higher percentage of elderly men and women report being divorced or never married. A smaller percent report being widowed, perhaps reflecting the gains in life expectancy for their partners over this period. For labor force status, the proportion of employed men and women decreased over this period; however, a higher percentage of elderly men and women report that they are not in the labor force, as opposed to being unemployed. Although the means across many of these measures have increased only marginally, the percent changes are more consequential; for example, the percent of older men who are divorced has increased 57% from 1995 to 2004. Because there have been changes across all of these measures, they are incorporated into the following regression models.

Table 3 reports the fixed-effects regression results for men and women aged 65 years and older. Considering the total arrest rate (column 1) four main findings are evident. First, marital status—and particularly, increases in the percent divorced compared to the percent married—is significantly associated with increasing elderly arrest rates.⁸ As Table 3 shows, a one-unit increase in the percent divorced is associated with a 119% increase in the elderly arrest rate. In contrast to the association with divorce, the percent widowed is negatively related to arrest rates. A one-unit increase in the percent widowed corresponds to a 21% decrease in the arrest rate. These patterns hold among arrests for larceny, but not among arrests for violent and more serious crimes. While the description of these results suggests that a relatively small change in marital status corresponds to a very large change in arrest rates, it is important to note that the magnitude of arrest rates is comparably quite modest and is measured as the number of arrests per 100,000 persons 65 years and older. Overall, taking together the estimates related to marital status, the associations suggest that it is not simply changes in the proportion living with a spouse that is consequential; rather, the circumstances around those changes—whether under the stigma of divorce as opposed to the status of widowhood—are particularly important.

Table 3.
Fixed-Effects Models of Prefectural Arrest Rates (Logged) for People 65 Year and Older in Japan.

Total Arrests Violent/Felonious Arrests Larceny Arrests

(1) (2) (3)

Marital status of older adults (ref: married)

Divorced 1.188 (0.364)** 0.803 (0.721) 1.158 (0.459)*

Widowed −0.209 (0.054)* −0.097 (0.115) −0.183 (0.060)

Never married −0.446 (0.464) −0.551 (0.589) −0.144 (0.537)

Unreported −0.164 (0.179) −0.111 (0.369) −0.034 (0.202)

Living situation of older adults (ref: with children)

Alone 0.021 (0.016) 0.041 (0.037) 0.013 (0.021)

With spouse 0.013 (0.013) 0.027 (0.024) 0.017 (0.014)

With others 0.005 (0.023) −0.065 (0.056) 0.010 (0.025)

Labor force status of older adults (ref: not in labor force)

Employed 0.016 (0.024) −0.128 (0.061)* 0.017 (0.026)

Unemployed 0.901 (0.453) 0.469 (0.708) 0.856 (0.533)

Unreported −0.049 (0.109) 0.031 (0.128) −0.131 (0.105)

Receipt of cash assistance among older adults (per 1,000) −0.031 (0.013)* 0.001 (0.024) −0.042 (0.021)*

Commuters to other prefectures 0.217 (0.054)* 0.128 (0.140) 0.209 (0.059)*

Police rate (per 1,000) −0.433 (0.755) −0.582 (1.247) −1.408 (0.901)

Density 0.000 (0.001) 0.000 (0.001) 0.001 (0.001)

Constant 7.024 (2.148)** 5.176 (4.925) 6.175 (3.055)*

Adjusted R ² (within) 0.855 0.625 0.793

N of observations 188 186 188

Note. The number of observations refers to 47 prefectures observed at four time points. For violent and felonious arrests, there are two prefecture years that have zero elderly arrests and are excluded from the logged arrest rate analyses.

p < .05. p < .01. p < .001.

In terms of economic integration, the results are more mixed. Receipt of cash assistance is negatively associated with total arrest rates and arrests for larceny, but labor force status is not significantly associated with these arrests. For cash assistance, the estimated relationship is modest but significant, where a one-unit increase in the rate of receipt correlates with a 3.1% decrease in the arrest rate. The direction of this association is perhaps counterintuitive if the measure is a reflection of poverty status. Instead, it may be that broadened access to cash assistance for the elderly provides an economic safety net for the poor, deterring shoplifting and other offenses. Among violent and serious crimes, the percent employed is negatively associated with arrest rates, where a one-unit increase in the percent employed is related to a 12.8% decrease in violent and serious arrest rates; however, this association is not significant in models that include fixed effects for years.

Looking to other indicators of integration, the percent of commuters to other prefectures is significantly associated with elderly arrests for total crimes and larcenies, where increases in the number of commuters that leave during the day are positively related to elderly arrest rates. Specifically, a one-unit increase in the percent of employed persons that commute to a different prefecture for work is associated with a 22% increase in total arrest rates.

Separately analyzing arrest rates by gender reveals several important differences. Among men, who make up the majority of elderly arrests, the association between commuter regions and arrest rates remains positive for total arrests and larcenies (see Table 4). It is possible that the lack of daytime informal controls characteristic of commuter regions is particularly relevant for men, who are more likely to have been employed workers and are now retired or unemployed. Among women, living situation has a modest but significant association with arrest rates for total offenses and larcenies, where the percent living with a spouse only as opposed to living alone or with children in a three-generational household is associated with higher arrest rates (see Table 5). Marital status remains consequential in both models for men and women. The percent widowed continues to be negatively associated with arrest rates for total crimes and larcenies. The percent divorced also continues to be positively associated with arrests for total offenses and larcenies; however, the magnitude of the association is larger among women and the association is not significant among men in more conservative models that additionally control for year fixed effects.

Table 4.
Fixed-Effects Models of Prefectural Arrest Rates (Logged) for Men 65 Year and Older in Japan.

Total Arrests Violent/Felonious Arrests Larceny Arrests

(1) (2) (3)

Marital status of older adults (ref: married)

Divorced 0.966 (0.301) 0.166 (0.706) 0.903 (0.326)

Widowed −0.363 (0.141)* 0.120 (0.241) −0.294 (0.140)*

Never married −0.128 (0.315) 0.012 (0.418) 0.108 (0.359)

Unreported −0.136 (0.184) 0.111 (0.285) 0.007 (0.207)

Living situation of older adults (ref: with children)

Alone 0.024 (0.015) 0.005 (0.040) 0.020 (0.019)

With spouse 0.013 (0.013) 0.026 (0.025) 0.016 (0.016)

With others 0.010 (0.029) −0.115 (0.054)* 0.014 (0.031)

Labor force status of older adults (ref: not in labor force)

Employed 0.003 (0.018) −0.091 (0.047) 0.000 (0.019)

Unemployed 0.266 (0.204) −0.148 (0.314) 0.225 (0.245)

Unreported −0.017 (0.083) −0.008 (0.091) −0.089 (0.092)

Receipt of cash assistance among older adults (per 1,000) −0.022 (0.015) −0.008 (0.024) −0.040 (0.017)*

Commuters to other prefectures 0.231 (0.057)* 0.199 (0.160) 0.208 (0.075)

Police rate (per 1,000) 0.333 (0.821) 0.957 (1.174) −0.830 (1.026)

Density −0.000 (0.001) −0.001 (0.001) 0.000 (0.001)

Constant 4.799 (2.112)* 1.791 (3.385) 5.399 (2.749)

Adjusted R ² (within) 0.827 0.595 0.738

N of observations 188 183 188

Note. The number of observations refers to 47 prefectures observed at four time points. For violent and felonious arrests, there are five prefecture years that have zero elderly arrests and are excluded from the logged arrest rate analyses.

p < .05. p < .01. p < .001.

Table 5.
Fixed-Effects Models of Prefectural Arrest Rates (Logged) for Women 65 Year and Older in Japan.

Total Arrests Violent/Felonious Arrests Larceny Arrests

(1) (2) (3)

Marital status of older adults (ref: married)

Divorced 1.796 (0.534) −1.051 (1.050) 1.746 (0.577)

Widowed −0.112 (0.045)* −0.074 (0.155) −0.105 (0.046)*

Never married −0.646 (0.578) −0.018 (0.840) −0.405 (0.655)

Unreported −0.177 (0.133) −0.103 (0.385) −0.156 (0.127)

Living situation of older adults (ref: with children)

Alone 0.013 (0.026) 0.060 (0.084) 0.004 (0.034)

With spouse 0.026 (0.013)* −0.018 (0.037) 0.029 (0.014)*

With others −0.006 (0.031) −0.104 (0.084) 0.011 (0.030)

Labor force status of older adults (ref: not in labor force)

Employed −0.018 (0.038) −0.006 (0.165) −0.013 (0.036)

Unemployed −0.340 (1.624) 1.348 (3.873) −0.864 (1.611)

Unreported −0.112 (0.137) 0.357 (0.231) −0.147 (0.114)

Receipt of cash assistance among older adults (per 1,000) −0.047 (0.020)* 0.058 (0.047) −0.046 (0.030)

Commuters to other prefectures 0.137 (0.080) −0.028 (0.169) 0.143 (0.090)

Police rate (per 1,000) −1.450 (0.700)* 2.990 (2.057) −1.891 (0.746)*

Density 0.000 (0.001) 0.001 (0.002) 0.001 (0.001)

Constant 6.587 (3.315) −1.092 (13.783) 5.617 (3.657)

Adjusted R ² (within) 0.792 0.370 0.751

N of observations 188 106 188

Note. The number of observations refers to 47 prefectures observed at four time points. For violent and felonious arrests, there are 82 prefecture years that have zero elderly arrests and are excluded from the logged arrest rate analyses.

p < .05. p < .01. p < .001.

Additional Analyses

Assessing limitations of using arrest data for offending

The above findings are based on the universe of elderly arrests between 1995 and 2004; these data are unique and involved extensive relationship-building efforts with Japanese government agencies to acquire. Despite their strengths, they are administrative measures of crime and are imperfect measures of offending since arrest rates are based on both crime and police practices. To help address this limitation, researchers often focus exclusively on violent and serious offenses. The models presented above distinguish by crime type, and very few factors are related to arrest rates for violent and serious offenses among elderly individuals. This could very well indicate that increases in elderly offending are driven by changes in police practices for petty crimes. However, as noted above, research on elderly offending more generally finds that most elderly individuals commit petty theft and other minor crimes (Cullen et al., 1985; Feldmeyer & Steffensmeier, 2007) and more serious crimes make up a very small number of arrests among elderly individuals in Japan (see Table 2). While cognizant of the limitations of arrest measures, focusing solely on arrests for violent and serious offenses would exclude the vast majority of elderly arrests.

A main concern of using arrest records to analyze offending patterns is that the observed changes are due to police practices only and that the estimated associations are spuriously driven by these practices. For this counterfactual to have credibility, changes in police practices must explain increases in arrests and the estimated associations with social and economic situations, such as divorce rates, multigenerational households, and receipt of cash assistance. These changes must be systematically associated with the observed social and economic factors and must occur over and beyond those policing practices already controlled for in the model, such as the number of officers per residents and unobserved time-stable differences in policing environments. To help assess the concern that arrest rates are only capturing policing changes and not elderly offending rates, I estimate a model of young adult arrests.⁹ If broad-based changes in police practices, rather than elderly offending rates, are systematically correlated with social and economic factors, these associations would be observed for models that consider arrest rates for other ages. Although this approach cannot comment on police changes that are targeted at elderly individuals, it provides information to help assess one alternative explanation. Findings from a fixed-effects model of logged young adult arrest rates that consider elderly specific covariates estimate no significant associations, with the exception of density and the percent divorced (results available upon request). The estimated association for divorce is smaller compared to the elderly arrest models (0.63 compared to 1.19); however, it suggests that prefecture-level increases in divorce among older people are related to increases in arrest rates among young people. These findings suggest at least two conclusions. First, marital stability is importantly associated with younger arrests, which is consistent with the idea that family disruption among older people has a salient influence on criminal offending among younger generations (Sampson, 1987). Second, the associations between arrest rates and social integration factors that are documented in Tables 3, 4, and 5 are largely specific to elderly arrest rates and are not similarly correlated with younger adult arrest rates. This provides some suggestive evidence that the findings regarding social integration and elderly arrest rates are not due to broad-based policing changes that affect arrest practices among people of all ages.

Social meanings of family and economic integration in Japan

The main analyses in this article examined associations among social integration factors and elderly arrest rates. The theoretical underpinnings concern macrosocial changes in integration, where estimated relationships reflect both aggregated individual experiences and consequences of changing regional contexts for residents. In this final section, I present suggestive evidence about the social meanings of family and economic integration in Tokyo, Japan, from the perspective of a small number of elderly individuals with recent histories of arrest. The methods for locating the participants and conducting the interviews are described in the Appendix. Although the findings cannot be generalized given the small and selected sample (N = 17), I suggest that the perspectives and experiences described here help to situate the prefectural-level findings regarding divorce, living situation, and poverty within the Japanese context.

The social meaning of divorce in Japan appears to be quite different from countries with high rates of divorce, such as the United States. Instead, the meaning of divorce in Japan may be most similar to East Asian countries, such as Korea, where divorce is still stigmatized even though it is becoming more prevalent (Jones, 2010). Among those interviewed, divorce was a common experience and 10 of the 17 individuals were divorced (59%), which is high compared to the overall older population in Japan (5.2%).¹⁰ Nearly all of the respondents had children, divorced after lengthy marriages, and described leaving behind their children, homes, neighborhoods, and employers as a result of the divorce. For these individuals, divorce was an experience that entailed leaving behind all remnants of prior life. A 62-year-old man who had not seen his daughter (now in her mid-30s) since the divorce explained: “My daughter thinks that I died. Because of the divorce, I told her [my wife] to tell my daughter that your father is dead.” Another man, 65-years-old, described his relationship with his adult daughter who he had not seen in over 12 years as bleak: “I am not satisfied, but I have already given up.” These experiences were common among those divorced. They illustrate the stigma attached to divorce for older adults, suggesting that the meaning of divorce in Japan is quite different compared to contexts where it is more prevalent, such as the United States.

Changes in living situation, such as the death of coresident parents, a daughter leaving home, or a move entailed by divorce, were a particular focal point among the women in the sample. A 66-year-old woman described spending days sitting on her daughter’s bed and constantly cleaning her daughter’s room, after she left for college. Eventually, the woman started playing pachinko, a type of gambling, during the day and began to shoplift groceries to make up for the lost money. “I was really lonely and sad after my daughter went to college—and this doesn’t excuse what I did—but I started to play pachinko…I would immediately use my husband’s salary to play pachinko and we were pressed for living expenses.” Another 66-year-old woman, Yamada-san, described how she felt after both of her parents, with whom she lived after her divorce, passed away within the same year. “More than other people, my relationship with my parents was strong…when my parents passed away, I became very insecure, both emotionally and financially. My life became very unstable.” Since her parents’ passing, Yamada-san was arrested twice, once for shoplifting a jacket that she claims she thought was her own and most recently, for stealing a woman’s purse from a department store. When asked to reflect on these events, she said, “I don’t think that it’s hard to make a living, but in terms of connecting with other people, I feel like I can’t impose on them…. In terms of committing this type of crime, I think that it is partly my fault but the more I think about it, the more I feel that the various important things in my life that disappeared these last 10 years played a big role.”

Although the quantitative findings regarding economic integration were mixed, the majority of men (12 of the 14) emphasized their strained economic situation, where being poor and unable to find work was a central source of frustration. Many perceived that their older age was the main contributing factor to their inability to find work. According to one 60-year-old man, “employers don’t hire me because of my age. I have tried hard to find a job but the companies worry about that. [Interviewer: Do the employers say you’re too old or do you have that impression?] Yes, they say it. After hearing my age, they say no.” A 62-year-old man also said he is never given a chance: “Not surprisingly, what they [employers] ask me is my age, that’s the first thing they ask me.” Many of the men (11 of the 14) described having stable employment histories until reaching older age, when they found it increasingly difficult to find work. Because they often felt that their situations would not improve as they continued to grow older, individuals often described these situations with despair. As one man stated bluntly, “Right now, my life is horrible, the worst. After [turning] 50 years old and now, at 60, I am at the worst place in life. I don’t have any hope for the future considering my age.”

The findings from the interviews illustrate individual experiences of divorce, changes in living situations, and labor market marginalization among older adults in Tokyo. These stories are based on a small, selected sample and are no more than suggestive; however, they provide some indication of the magnitude of consequences of divorce, changes in living situation, and unemployment for social attachments and feelings of belonging in the Japanese context. They also illustrate how increases in these factors may lead to weakened attachments, declining cohesiveness, diminished informal social controls, and despair for the future among older individuals in Japan.¹¹

Discussion

Recent increases in elderly offending in Japan and other aging societies provide a unique opportunity to forward social integration perspectives on crime in older age. Specifically, I suggest that later life changes in social integration, which are intimately tied to population aging trends, explain recent increases in elderly arrest rates. Findings from fixed-effects panel models support the role of family integration, where it is not simply increases in the percent without a spouse that are consequential for offending rates but instead, it is the specific form of weakened integration through divorce that is important, particularly among women. The results concerning economic integration were more mixed. Although labor force status was not associated with arrest rates, receipt of cash assistance was negatively related to rates. In the case of elderly offending, where a majority of older individuals are no longer working, labor force status may be less consequential than receipt of cash assistance. However, the mixed findings around economic factors are consistent with prior U.S. research, where family integration is found to be comparatively more important (Sampson, 1987). In addition to family and economic integration, changes in the rate of commuters were associated with elderly arrest rates among men, where increases in commuters were related to increases in arrests. This finding not only aligns with prior research on the importance of mobility for informal social controls and offending (Baller & Richardson, 2002; Crutchfield et al., 1982), but in using a measure of daytime mobility, it also suggests that alternative measures of mobility (apart from residential mobility, which is typically utilized) are valuable.

Importantly, the results suggest that the above associations are largely limited to elderly arrests for larceny and do not extend to more serious crimes. Although variations on the social integration framework have been applied to all types of crimes, including serious and violent offenses (Blau & Blau, 1982; Crutchfield et al., 1982; Sampson, 1987), the findings reported here indicate that integration may be less relevant for elderly arrests related to violent offenses.

In addition to the prefectural models, findings from in-depth interviews with elderly individuals with recent histories of arrest provide some indication, although extremely circumscribed, that the experiences of divorce, living alone or without adult children, and unemployment may have social meanings among older adults in Japan that are different from those in contexts like the United States. Instead, these experiences may be more comparable to those in other East Asian countries, which have similar dynamics of population aging and recent histories of rapid economic development. Although the interviews were based on a small and selected sample, they complement interpretations of the prefectural-level findings by adding cultural contextualization (Sale et al., 2002; Scrimshaw, 1990; Small, 2011). In Japan, divorce is relatively uncommon and stigmatized, often entailing cutting off ties to family members. The prevalence of elderly living alone or without adult children is not only quickly increasing but is also at odds with the cultural expectations of the older generation. It is within this cultural context and within this time period of population aging that weakened family and economic integration are associated with elderly arrest rates.

This article contributes a first step toward examining elderly arrest rates in aging societies and future work that studies these associations at both the macro- and microlevels are needed. For this article, there are several important limitations that may be improved by future research. First, the prefecture-level models utilize administrative records on arrests. As described earlier, arrest information reflects both criminal behavior and responses to it, including police practices. Common methods to address this issue, such as focusing on violent crimes only, exclude the majority of offenses committed by elderly individuals. This article employed several approaches, including the use of fixed-effects analyses, a control variable for changes in police officers, and a supplementary analysis with young adult arrest rates; as a result, I conclude that the associations documented here are saliently related to increasing elderly crime rates. At the same time, however, future research with self-report information is warranted, in order to further examine these associations and to assess whether and how changing police practices contributes to the increase. A second limitation is the use of data at the level of prefectures, which are relatively large geographic units. Compared to U.S. states, the average size of a Japanese prefecture is relatively small; however, U.S. research on social integration often considers smaller areas, such as metropolitan regions (Blau & Blau, 1982; Kubrin et al., 2006; Wadsworth, Kubrin, & Herting, 2014) and counties (Baller & Richardson, 2002). A third limitation is the use of estimated values for some of the independent variables, as described in the Method section and Table 1. This approach introduces measurement error for these variables, which may bias the regression coefficients for these variables downward relative to their actual associations. As such, it is possible that the estimates for these variables understate the magnitude of the true relationships.

Despite these limitations, this article contributes one of the first empirical studies on an important and unanticipated social issue, which offers a unique opportunity to examine criminological theories in a non-Western context. Although this article focused on Japan, similar elderly arrest patterns have been observed in some aging East Asian countries (Chen, 2015; Hyun-chae, 2015; Ji-sook, 2008; Ota, 2009) and may occur in other countries as aging populations increase in size and proportion in the future. In Japan, and perhaps in other East Asian countries that share similar social and cultural emphases on the family (Agnew, 2015), family integration is importantly associated with increasing elderly arrests in Japan. However, family integration may not play a dominant role in contexts where divorce and living apart from adult children are more commonplace, as in the United States. Instead, unemployment and economic insecurity may emerge as more consistent factors for elderly offending in societies where economic wealth is highly valued.

Appendix

Interviews With Elderly Individuals

Interviews with male (n = 14) and female (n = 3) individuals who were arrested at or after 60 years of age were conducted in the summer of 2010 in Tokyo, Japan. Before recruitment, I anticipated that it would be difficult to find elderly offenders since the social stigma around arrests and criminal activity is severe. To recruit participants, I used a variety of methods, including distributing flyers at postrelease centers throughout the city and at food distribution sites. My uncle—a Japanese native who is an older low-wage worker himself—was a critically important part of the recruiting process. While distributing flyers, he was able to engage potential participants in conversation about the research. We were able to recruit elderly male participants through these methods; however, it was much more difficult to locate female respondents, and a small number of interviews with elderly females (n = 3) were arranged with the assistance of the Tokyo Metropolitan Police and the Ministry of Justice.

The interviews were semistructured and covered a range of topics, including current life situation, daily schedules and routines, previous turning points and important life events, circumstances around arrest(s), and personal beliefs about the current increase in elderly arrest rates.¹² Participants appeared very willing—and some were quite eager—to share their stories. Several stated that they were unable to discuss these events with others and felt unburdened after the interview because they were able to freely describe their offending histories with us.

Variable	Source and Observed Year
Dependent variables
Older arrest rate (per 100,000 persons)	Tokyo Metropolitan Police and the Statistics Bureau (population): 95, 98, 01, 04
Independent variables
Marital status of older adults (%)	Japan census: 90, 95, 00, 05
Living situation of older adults (%)	National institute of population and social security research: 95, 98, 01, 04
Labor force status of older adults (%)	Japan census: 90, 95, 00, 05
Receipt of cash assistance among older adults (per 1,000)	Social indicators by prefecture: 95, 98, 01, 04
Commuters to other prefectures (% of employed)	Japan census: 90, 95, 00, 05
Number of police per 1,000 residents	Social indicators by prefecture: 95, 98, 01, 04
Density: Number of people per 1 km² inhabitable area	Social indicators by prefecture: 95, 98, 01, 04

Variable	1995	2004
Dependent variables
Older arrest rate: total	52.73	(26.79)	138.97	(33.31)
Men	80.03	(50.19)	216.72	(60.06)
Women	33.81	(15.60)	83.52	(23.95)
Older arrest rate: violent/felonious	2.09	(1.31)	6.95	(3.48)
Men	4.80	(3.15)	15.43	(7.73)
Women	0.25	(0.36)	0.92	(0.81)
Older arrest rate: larceny	42.29	(18.63)	102.82	(26.44)
Men	57.08	(26.71)	139.17	(35.26)
Women	32.03	(14.97)	77.16	(24.38)
Independent variables
Marital status of older adults (%): men
Married	84.37	(1.22)	82.89	(1.77)
Divorced	1.55	(0.51)	2.44	(0.62)
Widowed	12.53	(0.65)	11.43	(0.78)
Never married	1.27	(0.38)	1.98	(0.65)
Unreported	0.28	(0.31)	1.25	(1.09)
Marital status of older adults (%): women
Married	43.22	(2.39)	46.63	(2.11)
Divorced	3.01	(0.70)	3.56	(0.82)
Widowed	50.68	(2.11)	45.48	(2.43)
Never married	2.67	(0.74)	3.08	(0.79)
Unreported	0.43	(0.36)	1.25	(0.91)
Living situation of older adults (%)
Alone	11.45	(4.14)	13.63	(3.75)
With spouse	27.55	(6.98)	33.56	(6.84)
With children	57.32	(11.03)	48.67	(10.08)
With others	3.67	(0.59)	4.16	(0.79)
Labor force status of older adults (%): men
Not in labor force	57.46	(4.23)	64.75	(3.44)
Employed	40.35	(4.43)	31.68	(3.46)
Unemployed	1.91	(0.59)	1.61	(0.37)
Unreported	0.27	(0.33)	1.95	(1.99)
Labor force status of older adults (%): women
Not in labor force	83.38	(3.17)	84.62	(2.29)
Employed	16.18	(3.24)	14.20	(2.39)
Unemployed	0.18	(0.07)	0.21	(0.06)
Unreported	0.26	(0.17)	0.97	(0.99)
Receipt of cash assistance among older adults (per 1,000)	13.80	(8.37)	17.58	(9.92)
Commuters to other prefectures	5.55	(8.22)	6.01	(8.17)
Number of police per 1,000 residents	1.62	(0.36)	1.74	(0.33)
Density	1,363.56	(1,613.17)	1,370.313	(1,659.79)
N	47		47

	Total Arrests	Violent/Felonious Arrests	Larceny Arrests
Marital status of older adults (ref: married)
Divorced	1.188	(0.364)**	0.803	(0.721)	1.158	(0.459)*
Widowed	−0.209	(0.054)***	−0.097	(0.115)	−0.183	(0.060)**
Never married	−0.446	(0.464)	−0.551	(0.589)	−0.144	(0.537)
Unreported	−0.164	(0.179)	−0.111	(0.369)	−0.034	(0.202)
Living situation of older adults (ref: with children)
Alone	0.021	(0.016)	0.041	(0.037)	0.013	(0.021)
With spouse	0.013	(0.013)	0.027	(0.024)	0.017	(0.014)
With others	0.005	(0.023)	−0.065	(0.056)	0.010	(0.025)
Labor force status of older adults (ref: not in labor force)
Employed	0.016	(0.024)	−0.128	(0.061)*	0.017	(0.026)
Unemployed	0.901	(0.453)	0.469	(0.708)	0.856	(0.533)
Unreported	−0.049	(0.109)	0.031	(0.128)	−0.131	(0.105)
Receipt of cash assistance among older adults (per 1,000)	−0.031	(0.013)*	0.001	(0.024)	−0.042	(0.021)*
Commuters to other prefectures	0.217	(0.054)***	0.128	(0.140)	0.209	(0.059)***
Police rate (per 1,000)	−0.433	(0.755)	−0.582	(1.247)	−1.408	(0.901)
Density	0.000	(0.001)	0.000	(0.001)	0.001	(0.001)
Constant	7.024	(2.148)**	5.176	(4.925)	6.175	(3.055)*
Adjusted R ² (within)	0.855	0.625	0.793
N of observations	188	186	188

	Total Arrests	Violent/Felonious Arrests	Larceny Arrests
Marital status of older adults (ref: married)
Divorced	0.966	(0.301)**	0.166	(0.706)	0.903	(0.326)**
Widowed	−0.363	(0.141)*	0.120	(0.241)	−0.294	(0.140)*
Never married	−0.128	(0.315)	0.012	(0.418)	0.108	(0.359)
Unreported	−0.136	(0.184)	0.111	(0.285)	0.007	(0.207)
Living situation of older adults (ref: with children)
Alone	0.024	(0.015)	0.005	(0.040)	0.020	(0.019)
With spouse	0.013	(0.013)	0.026	(0.025)	0.016	(0.016)
With others	0.010	(0.029)	−0.115	(0.054)*	0.014	(0.031)
Labor force status of older adults (ref: not in labor force)
Employed	0.003	(0.018)	−0.091	(0.047)	0.000	(0.019)
Unemployed	0.266	(0.204)	−0.148	(0.314)	0.225	(0.245)
Unreported	−0.017	(0.083)	−0.008	(0.091)	−0.089	(0.092)
Receipt of cash assistance among older adults (per 1,000)	−0.022	(0.015)	−0.008	(0.024)	−0.040	(0.017)*
Commuters to other prefectures	0.231	(0.057)***	0.199	(0.160)	0.208	(0.075)**
Police rate (per 1,000)	0.333	(0.821)	0.957	(1.174)	−0.830	(1.026)
Density	−0.000	(0.001)	−0.001	(0.001)	0.000	(0.001)
Constant	4.799	(2.112)*	1.791	(3.385)	5.399	(2.749)
Adjusted R ² (within)	0.827	0.595	0.738
N of observations	188	183	188

	Total Arrests	Violent/Felonious Arrests	Larceny Arrests
Marital status of older adults (ref: married)
Divorced	1.796	(0.534)**	−1.051	(1.050)	1.746	(0.577)**
Widowed	−0.112	(0.045)*	−0.074	(0.155)	−0.105	(0.046)*
Never married	−0.646	(0.578)	−0.018	(0.840)	−0.405	(0.655)
Unreported	−0.177	(0.133)	−0.103	(0.385)	−0.156	(0.127)
Living situation of older adults (ref: with children)
Alone	0.013	(0.026)	0.060	(0.084)	0.004	(0.034)
With spouse	0.026	(0.013)*	−0.018	(0.037)	0.029	(0.014)*
With others	−0.006	(0.031)	−0.104	(0.084)	0.011	(0.030)
Labor force status of older adults (ref: not in labor force)
Employed	−0.018	(0.038)	−0.006	(0.165)	−0.013	(0.036)
Unemployed	−0.340	(1.624)	1.348	(3.873)	−0.864	(1.611)
Unreported	−0.112	(0.137)	0.357	(0.231)	−0.147	(0.114)
Receipt of cash assistance among older adults (per 1,000)	−0.047	(0.020)*	0.058	(0.047)	−0.046	(0.030)
Commuters to other prefectures	0.137	(0.080)	−0.028	(0.169)	0.143	(0.090)
Police rate (per 1,000)	−1.450	(0.700)*	2.990	(2.057)	−1.891	(0.746)*
Density	0.000	(0.001)	0.001	(0.002)	0.001	(0.001)
Constant	6.587	(3.315)	−1.092	(13.783)	5.617	(3.657)
Adjusted R ² (within)	0.792	0.370	0.751
N of observations	188	106	188

Footnotes

Acknowledgments

Thank-you to the many people who helped me to collect the data used in this article; these include but are not limited to Katherine Newman, Devah Pager, Tetsuji Ezaki, Colin McKenzie, David Johnson, David Leheny, Sharon Yoon, Seiji Sugie, Kazuko Sugie, Fukuji Sugie, the Tokyo Metropolitan Police Department, the Japanese Ministry of Justice, and colleagues at Princeton, Keio, and Tohoku universities.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Global Network on Inequality and East Asian Studies program at Princeton.

Notes

References

Agnew

(1999). A general strain theory of community differences in crime rates. Journal of Research in Crime and Delinquency, 36, 123–155.

Agnew

(2015). Using general strain theory to explain crime in Asian societies. Asian Criminology, 10, 131–147.

Baller

R. D.

Richardson

K. K.

(2002). Social integration, imitation, and the geographic patterning of suicide. American Sociological Review, 67, 873–888.

Blau

J. R.

Blau

P. M.

(1982). The cost of inequality: Metropolitan structure and violent crime. American Sociological Review, 47, 114–129.

Brinton

(2003). Fact-rich, data-poor: Japan as sociologists’ heaven and hell. In Bestor

Steinhoff

Bestor

V. L.

(Eds.), Doing fieldwork in Japan (pp. 195–213). Honolulu: University of Hawaii Press.

Brinton

M. C.

(1989). Gender stratification in contemporary urban Japan. American Sociological Review, 54, 549–564.

Burr

J. A.

Hartman

J. T.

Matteson

D. W.

(1999). Black suicide in US metropolitan areas: An examination of the racial inequality and social integration-regulation hypotheses. Social Forces, 77, 1049–1080.

Chen

(2015). Analysis of the causes of crime: Lenient punishment of the elderly. In 2015 International Conference on Social Science and Technology Education (pp. 180–182). Amsterdam, the Netherlands: Atlantis Press. doi:10.2991/icsste-15.2015.53

Cohen

L. E.

Felson

(1979). Social change and crime rate trends: A routine activity approach. American Sociological Review, 44, 588–608.

10.

Coulmas

(2007). Population decline and ageing in Japan: The social consequences. Oxford, England: Psychology Press.

11.

Crutchfield

R. D.

Geerken

M. R.

Gove

W. R.

(1982). Crime rate and social integration: The impact of metropolitan mobility. Criminology, 20, 467.

12.

Cullen

F. T.

Wozniak

J. F.

Frank

(1985). The rise of the elderly offender: Will a new criminal be invented? Crime and Social Justice, 15, 151–165.

13.

Durkheim

É.

(1951). Suicide: A study in sociology. New York, NY: The Free Press.

14.

Ezaki

(2014). The origins of misconduct by elderly offenders in repeated shoplifting: Applying Travis Hirschi’s social control theory. Conference Paper for Asia Association Police Society Conference, Tokyo, Japan.

15.

Federal Bureau of Investigation, U.S. Department of Justice. (2008). Crime in the United States: 2007. Retrieved from https://www2.fbi.gov/ucr/cius2007/

16.

Feldmeyer

Steffensmeier

(2007). Elder crime: Patterns and current trends, 1980–2004. Research on Aging, 29, 297–322.

17.

Finch

(2000). Criminal statistics in Japan: The White Paper on crime, Hanzai Hakusho, and Hanzai Tokeisho. Social Science Japan Journal, 3, 237–249.

18.

Frees

E. W.

(2004). Longitudinal and panel data: Analysis and applications in the social sciences. Cambridge, MA: Cambridge University Press.

19.

Fuess

(2004). Divorce in Japan: Family, gender, and the State, 1600–2000. Stanford, CA: Stanford University Press.

20.

Gottfredson

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

21.

Greenberg

D. F.

(1985). Age, crime, and social explanation. The American Journal of Sociology, 91, 1–21.

22.

Hiraiwa-Hasegawa

(2005). Homicide by men in Japan, and its relationship to age, resources and risk taking. Evolution and Human Behavior, 26, 332–343.

23.

Hirschi

Gottfredson

(1983). Age and the explanation of crime. American Journal of Sociology, 89, 552–584.

24.

Hyun-chae

(2015). More elderly S. Koreans committing violent crimes. The Korea Times, May 11. Retrieved from http://www.koreatimesus.com/more-elderly-s-koreans-committing-violent-crimes/

25.

Ji-sook

(2008). Elderly crime jumps. The Korea Times, July 30. Retrieved from https://www.koreatimes.co.kr/www/news/nation/2016/02/117_28514.html

26.

Jones

(2010). Changing marriage patterns in Asia (Asia Research Institute Working Paper No. 131). Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1716533

27.

Kruttschnitt

(2013). Gender and crime. Annual Review of Sociology, 39, 291–308.

28.

Kubrin

C. E.

Wadsworth

DiPietro

(2006). Deindustrialization, disadvantage and suicide among young black males. Social Forces, 84, 1559–1579.

29.

Laub

Sampson

(2003). Shared beginnings, divergent lives: Delinquent boys to age 70. Cambridge, MA: Harvard University Press.

30.

Messner

S. F.

Rosenfeld

(1994). Crime and the American dream. Belmont, CA: Wadsworth.

31.

Ministry of Justice. (2008). White Paper on Crime. Retrieved from http://www.moj.go.jp/HOUSO/2008/index.html (in Japanese)

32.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychology, 100, 674–701.

33.

Moffitt

T. E.

(1997). Adolescence-limited and life-course-persistent offending: A complementary pair of developmental theories. Developmental Theories of Crime and Delinquency, 7, 11–53.

34.

Morgan

S. P.

Taylor

G. T.

(2006). Low fertility at the turn of the twenty-first century. Annual Review of Sociology, 32, 375.

35.

Ogawa

Retherford

R. D.

(1993). Care of the elderly in Japan: Changing norms and expectations. Journal of Marriage and Family, 55, 585–597.

36.

Ogawa

Retherford

R. D.

(1997). Shifting costs of caring for the elderly back to families in Japan: Will it work? Population and Development Review, 23, 59–94.

37.

Onishi

(2004). Japan opposition leader quits in pension payment scandal. New York Times, May 11. Retrieved from http://www.nytimes.com/2004/05/11/world/japan-opposition-leader-quits-in-pension-payment-scandal.html?_r=0

38.

Onishi

(2005). Japan’s population fell this year, sooner than expected. The New York Times, December 24. Retrieved from http://www.nytimes.com/2005/12/24/world/asia/japans-population-fell-this-year-sooner-than-expected.html

39.

Ota

(2009). Recent trends in elderly offenders and its countermeasures in Japan. Security Science Review, 11, 126–161.

40.

Park

(1992). Trends in crime rates in postwar Japan: A structural perspective. Tallahassee: Florida State University.

41.

Sale

J. E. M.

Lohfeld

L. H.

Brazil

(2002). Revisiting the quantitative-qualitative debate: Implications for mixed-methods research. Quality and Quantity, 36, 43–53.

42.

Sampson

R. J.

(1985). Neighborhood and crime: The structural determinants of personal victimization. Journal of Research in Crime and Delinquency, 22, 7–40.

43.

Sampson

R. J.

(1987). Urban black violence: The effect of male joblessness and family disruption. American Journal of Sociology, 93, 348–382.

44.

Sampson

R. J.

Laub

(1995). Crime in the making: Pathways and turning points through life. Cambridge, MA: Harvard University Press.

45.

Scrimshaw

S. C. M.

(1990). Combining qualitative and quantitative methods in the study of intra-household resource allocation. In Rogers Schollsman (Eds.), Intra-household resource allocation: Issues and methods for policy development and planning (pp. 86–98). Tokyo, Japan: United Nations University Press.

46.

Small

M. L.

(2011). How to conduct a mixed methods study: Recent trends in a rapidly growing literature. Annual Review of Sociology, 37, 57–86.

47.

South

Messner

S. F.

(2000). Crime and demography: Multiple linkages, reciprocal relations. Annual Review of Sociology, 26, 83–106.

48.

Statistics Bureau, Ministry of Internal Affairs and Communications. (2014). Statistical handbook of Japan. Tokyo, Japan.

49.

Steffensmeier

Allan

(1996). Gender and crime: Toward a gendered theory of female offending. Annual Review of Sociology, 22, 459–487.

50.

Steffensmeier

D. J.

Allan

E. A.

Harer

M. D.

Streifel

(1989). Age and the distribution of crime. American Journal of Sociology, 94, 803–831.

51.

Stockard

O’Brien

R. M.

(2002). Cohort variations and changes in age-specific suicide rates over time: Explaining variations in youth suicide. Social Forces, 81, 605–642.

52.

Taipei Times . (2012). More old people turn to crime as population ages. November 3, p. 5. Retrieved from http://www.taipeitimes.com/News/taiwan/archives/2012/11/03/2003546791

53.

Tittle

C. R.

Grasmick

H. G.

(1997). Criminal behavior and age: A test of three provocative hypotheses. Journal of Criminal Law and Criminology, 88, 309.

54.

Traphagan

J. W.

Knight

(2003). Demographic change and the family in Japan’s Aging Society. New York, NY: SUNY Press.

55.

Wadsworth

Kubrin

C. E.

Herting

J. R.

(2014). Investigating the rise (and fall) of young black male suicide in the United States, 1982–2001. Journal of African American Studies, 18, 72–91.

56.

Wooldridge

J. M.

(2002). Econometric analysis of cross section and panel data. Cambridge, MA: The MIT press.

Variable	1995		2004
Variable	Mean (%)	SD	Mean (%)	SD
Dependent variables
Older arrest rate: total	52.73	(26.79)	138.97	(33.31)
Men	80.03	(50.19)	216.72	(60.06)
Women	33.81	(15.60)	83.52	(23.95)
Older arrest rate: violent/felonious	2.09	(1.31)	6.95	(3.48)
Men	4.80	(3.15)	15.43	(7.73)
Women	0.25	(0.36)	0.92	(0.81)
Older arrest rate: larceny	42.29	(18.63)	102.82	(26.44)
Men	57.08	(26.71)	139.17	(35.26)
Women	32.03	(14.97)	77.16	(24.38)
Independent variables
Marital status of older adults (%): men
Married	84.37	(1.22)	82.89	(1.77)
Divorced	1.55	(0.51)	2.44	(0.62)
Widowed	12.53	(0.65)	11.43	(0.78)
Never married	1.27	(0.38)	1.98	(0.65)
Unreported	0.28	(0.31)	1.25	(1.09)
Marital status of older adults (%): women
Married	43.22	(2.39)	46.63	(2.11)
Divorced	3.01	(0.70)	3.56	(0.82)
Widowed	50.68	(2.11)	45.48	(2.43)
Never married	2.67	(0.74)	3.08	(0.79)
Unreported	0.43	(0.36)	1.25	(0.91)
Living situation of older adults (%)
Alone	11.45	(4.14)	13.63	(3.75)
With spouse	27.55	(6.98)	33.56	(6.84)
With children	57.32	(11.03)	48.67	(10.08)
With others	3.67	(0.59)	4.16	(0.79)
Labor force status of older adults (%): men
Not in labor force	57.46	(4.23)	64.75	(3.44)
Employed	40.35	(4.43)	31.68	(3.46)
Unemployed	1.91	(0.59)	1.61	(0.37)
Unreported	0.27	(0.33)	1.95	(1.99)
Labor force status of older adults (%): women
Not in labor force	83.38	(3.17)	84.62	(2.29)
Employed	16.18	(3.24)	14.20	(2.39)
Unemployed	0.18	(0.07)	0.21	(0.06)
Unreported	0.26	(0.17)	0.97	(0.99)
Receipt of cash assistance among older adults (per 1,000)	13.80	(8.37)	17.58	(9.92)
Commuters to other prefectures	5.55	(8.22)	6.01	(8.17)
Number of police per 1,000 residents	1.62	(0.36)	1.74	(0.33)
Density	1,363.56	(1,613.17)	1,370.313	(1,659.79)
N	47		47