The Canadian population, like that of other Western countries, has been aging since the 1970's, and is expected to continue aging for the foreseeable future. As the proportion of crime-prone young people in the Canadian population falls, and that of older people rises, the overall crime rate should fall. This paper explores the implications of demographic change for crime in Canada in the next forty years.
The role of demographic change as a determinant and predictor of crime rates was established by studies of the rise in American crime rates in the 1960's (Sagi and Wellford 1968; Ferdinand 1970; Wellford 1973), and their fall in the late 1970's and 1980's (Cohen and Land 1987; Steffensmeier and Harer 1987, 1991). These writers showed that a substantial part of these politically sensitive crime trends was explained simply by the changing age composition of the American population, primarily the aging of the baby boomers. Overall crime rates rose as this group reached their late teens -- the most crime-prone time of life -- in the 1960's, and then fell as they began to reach their thirties, a decade later. Thus, the age-standardized crime rate in the USA varied somewhat less than the reported crude rate.
The same reasoning can be used to make crime forecasts based on projected future changes in the age structure. In the mid-1990's, some American writers used demographic reasoning -- specifically, an expected upturn around the year 2000 in the number of young people in the USA -- to make rather apocalyptic crime predictions. According to Wilson (1995: 507), the USA could expect by the year 2000, "30,000 more young muggers, killers and thieves than we have now," and Fox (1996: 306) predicted that, as a result of increases in the number of young Americans, the homicide rate in the USA in 2005 would resemble "a blood bath." Other writers argue that demographics will have only a limited effect on American crime (Zimring 1998; Levitt 1999). Recent research in Britain has forecast a short-term rise in crime, due to, inter alia, a projected short-term rise in the number of males between 14 and 25 (Field 1999).
The possible impact of demographic change on crime has also begun to be noticed in Canada. A recent crime report by the Canadian Centre for Justice Statistics mentions the decrease in the proportion of 15 to 24 year olds as one of the reasons for the fall since 1991 in the recorded crime rate (Tremblay 1999: 5). Similarly, Foot gives demographic explanations for recent changes in the crime rate and the types of crime in Canada, and makes some demographically-based crime forecasts for the near future: more crime in general, and especially more white-collar crime (Foot 1998: 190-195; Foot, Loreto, and McCormack 1998: 38). John Howard Society of Ontario predicts "an increase, likely modest, in the incidence of crime generally, with property crime driving the trend" because of an expected increase from 1996 to 2016 in the numbers of Canadians aged 15 to 24, and "relative stability [in] violent crimes" because of smaller increases in the numbers of adults aged 25 to 49 (John Howard Society of Ontario 1999: 2-3). These demographically-based explanations and predictions of future crime in Canada are highly informal, even impressionistic. So far, there has been no systematic attempt to assess the likely impact of demographic change on the future amount and type of crime in Canada.
Demographic change and crime rates in Canada
Figure 1 shows the actual age structure of the Canadian population for 1999, and Statistics Canada's projections to the year 2041. The anticipated aging of the population is indicated by the growth in the group aged 40 or more, from 44 percent of the population in 1999 to 58 percent in 2041, and the corresponding decline in the relative size of the three younger age groups. As a result, the median age of Canadians, which increased from 26 years in 1971 to 36 years in 1999, is projected to increase to 46 years by 2041.
The tendency to commit crime rises sharply with age to a peak in the mid- to late-teens, and then falls with increasing age. This relationship between age and crime is "one of the oldest and most widely accepted in criminology" (Steffensmeier, Allan, Harer, and Streifel 1989: 803). It has been replicated in many different nations, at different times in recent history, in different social groups, and using various types of data. So strong and widespread is the evidence for this relationship that Hirschi and Gottfredson (1983: note 9) have made the claim that "the age distribution of the tendency to commit criminal acts is invariant" across social and cultural conditions. Figure 2 depicts the age-crime relationship for recorded crime in Canada in 1999. The highest recorded crime rate is for 15-19 year olds, and is about three times the overall crime rate. Age groups from 15 to 39 years old have substantially above-average recorded crime rates.
Figure 2 Recorded crime rates (incidents per 100,000 population) by age group, Canada (parts), 1999 Age group 0-4 8 5-9 335 10-14 7364 15-19 25708 20-24 20022 25-29 15363 30-34 14553 35-39 12083 40-44 8686 45-49 5604 50-54 4053 55-59 2755 60-64 1975 65-69 1281 70-74 791 75-79 503 80-84 230 85+ 132 Sources: Custom tabulations provided by Canadian Centre for Justice Statistics, UCR2 Survey, for age/crime data; Statistics Canada (2001a: Matrix 6367) for population data. Note: Table made from bar graph.
The likelihood of being a victim of crime is also strongly related to one's age (Fattah 1991: 119). Various explanations have been offered for this observation, the most plausible being that the more public, outgoing life-style of teenagers and young adults exposes them to greater risks of crime, while the more private, sedentary lives of middle-aged and older people reduce their exposure (Fattah 1991: 126-128). According to the 1999 General Social Survey, 25 percent of Canadians over 14 years old were victims of crime in the past twelve months. This proportion decreased with the respondent's age from a high of 40 percent of 15-19 year olds to a low of 5 percent of those aged 80 or older (Figure 3). Since most crime(2) involves both an offender(s) and a victim(s), changes in the age structure of the population will affect the crime rate via their effect on the supply of potential victims as well as on the supply of potential offenders (Cohen and Land 1987: 174-175; Marvell and Moody 1991: 248-249).
Figure 3 Self-reported victimization rates per 100,000 population by age group, Canada, 1999 Age group 15-19 39748 20-24 39202 25-29 33760 30-34 31336 35-39 27184 40-44 25531 45-49 23282 50-54 21776 55-59 17670 60-64 11966 65-69 9983 70-74 8729 75-79 6322 80+ 2848 Source: Statistics Canada (2001b) Note: Table made from bar graph.
Although the shapes of the age-criminal and age-victim curves (Figures 2 and 3 respectively) are not identical, they are remarkably similar, considering that they are derived from such different sources, with such different sources of measurement error: police records on detected offenders, and self-reported victimization in response to a social survey. Because of their similarity in salient respects -- high rates in 15 to 39 year olds, and very high rates in 15 to 24 year olds -- both indicators of crime may be expected to respond similarly to the projected changes in the age structure of the population.
The idea behind demographically-based forecasts of crime, as of any social phenomenon, is to examine the impact of expected changes in the age structure of the population, while ignoring all other factors that might affect the phenomenon (Foot 1998: 305). Thus, we are not attempting to predict future crime rates, a project which would require knowledge of future levels of all factors thought to affect crime, and a model of the way that they affect crime: rather, we are estimating the extent to which future crime rates will be affected by projected demographic change alone. Furthermore, we consider here only the direct effect of demographic change: that is, the simple effect of changes in the proportions of the population in each age group. We do not attempt to incorporate possible indirect effects on crime; for example, economic or social changes that may be precipitated by demographic change, and that may in turn cause a change in the age-standardized crime rate.(3)
In order to focus on the effect of demographic change, we assume that the only thing that affects the propensity to commit crime and the likelihood of being a victim of crime is one's age. That is, we assume that people of each age will continue to have the same probabilities of committing and being victims of crime in the future, but the number and proportion of people of each age in the population will change. Thus, the overall crime rate will change as the result of changes in the age composition of the population, while people of any given age will continue to commit and suffer from crime at the same rate as in the past. In general, if age groups with relatively high rates of offending and victimization decline in numbers, and are displaced by age groups characterized by lower offending and victimization rates, then the crime rate will decline. The formulae used to forecast recorded crime and victimization rates are given in the Appendix.
In order to forecast crime rates, we need age-specific crime rates for some base year, and projected numbers of people in each age group in the future. Of course, actual numbers of criminal incidents, persons committing crimes, or persons victimized by crime in a given year are not known. Two substitutes are available for Canada: recorded crime and self-reported victimization.
The most commonly used is recorded crime: that is, numbers of criminal incidents recorded by the police and remitted to the Uniform Crime Reporting Survey ("UCR"). In the base year, 1999, the UCR recorded 2,613,348 criminal incidents in Canada.(4) The population of Canada in 1999 was 30,493,433, so the aggregate recorded crime rate was 2,613,348/30,493,433 = 0.08570 incidents per capita, or 8,570 incidents per 100,000 population.
The police are able to record the age of persons involved in recorded crime only if the crime is solved, or "cleared" -- that is, if the perpetrator(s) is (are) identified. In 1999, 40 percent of recorded incidents in Canada were cleared. In relying on age-crime data from incidents that were cleared, we must assume that the age distribution of persons implicated in incidents that were cleared is the same as the age distribution of those involved in unsolved (but recorded) incidents.(5)
Unfortunately, the Canadian UCR does not record the ages of persons apprehended in incidents that are cleared. However, this information is provided in a new version of the Canadian UCR -- the "Revised UCR Survey" (UCR2), which is currently being operated in parallel with the existing, or "aggregate" UCR. This new survey of police-reported crime is being phased in over a number of years, beginning in 1988, with additional police forces participating each year. By 1999, most police forces in Canada were participating, with the major exceptions of the OPP and the RCMP, and a few municipal forces (Canadian Centre for Justice Statistics 1997; 2000: 6). The 1999 UCR2 included 1.17 million incidents, or 45 percent of the 2.61 million incidents recorded by the aggregate UCR Survey. The ages of the alleged perpetrators were recorded in 388,937 of these 1.17 million incidents.(6) Thus, there were about 7 times as many (2,613,348/ 388,937) criminal incidents recorded by police in Canada as those for which we have data on the age of the alleged offender(s). If we assume that the known age distribution of persons apprehended in the cleared incidents in the UCR2 was the same as the unknown age distribution of persons involved in all recorded incidents in Canada, then we can estimate the age distribution of the offenders implicated in all incidents recorded in Canada by multiplying the known age distribution by a factor of approximately 7.(7)
The result of this extrapolation is shown in Figure 2. For example, persons aged 15-19 years were implicated in 78,890 incidents recorded in the 1999 UCR2.(8) This is extrapolated to 530,076 incidents (78,890*2,613,348/388,93) in all of Canada. The population of 15-19 year olds in Canada in 1999 was 2,061,942. Therefore, the age-specific recorded crime rate for 15-19 year olds in 1999 was 530,076/2,061,942 = 0.25708 per capita, or 25,708 incidents per 100,000 population -- slightly more than one recorded incident for every four 15-19 year olds (Figure 2).(9)
Data on the age distribution of persons victimized by crime in Canada are available from the Victimization Module of the General Social Survey (GSS), a survey of a large representative sample of Canadians.(10) This victimization module has been incorporated in the GSS approximately every five years: in 1988, 1993, and 1999 (Besserer and Trainor 2000; Gartner and Doob 1994; Statistics Canada Housing, Family and Social Statistics Division 1994; Sacco and Johnson 1990). It asks detailed questions about eight types of criminal victimization suffered by respondents aged 15 years or older.
As a source of data on the amount and type of crime in Canada, the GSS Victimization Survey has certain advantages and disadvantages relative to the UCR. Its big advantage is its inclusion of incidents that are omitted from the UCR because they are not reported to police -- the "dark figure of crime". It is also not subject to the upward or downward pressures generated by the bureaucratic/political contingencies of police work and budgets. Also, extrapolation from attributes of the GSS sample to the adult population of Canada is straightforward, since the samples are chosen to be representative within statistical bounds; whereas, extrapolation from apprehended offenders in the UCR to all offenders in Canada requires a series of questionable assumptions (see above).
This survey of self-reported criminal victimization necessarily omits much crime that is included (though not in its entirety) in the UCR: "victimless" crime, such as drinking-driving, morals, public order and weapons possession/storage/handling offences, and much corporate crime; crime against organizations, such as shoplifting, much fraud, and other white-collar crime; homicides; and crime with victims aged less than 15 years, which includes a high proportion of sexual assaults, other sexual offences, and abduction (Canadian Centre for Justice Statistics 2000: Table 4.12). It also relies on respondents' memories of incidents occurring as much as twelve months ago, and their understandings of what constitutes a crime, although the questions specifically prompt for the most common predatory crimes -- robbery, assault, sexual assault, burglary, personal theft, household theft, motor vehicle theft, vandalism, and attempts, and include explanations of these;(11) whereas the UCR relies on interpretations of relatively recent events by police officers, who are trained to distinguish between criminal and non-criminal incidents. Finally, even in its much-vaunted role as the discoverer of unreported crime, the victimization survey has its limitations: according to the Director of the Bureau of Justice Statistics (USA), many sexual assaults are omitted from the National Crime Victimization Survey (the American equivalent of the GSS Victimization Survey), either because women are reluctant to report them to the NCVS interviewers, or because they do not consider them to be crimes (Chaiken 2000: 12-13). In general, however, as a source of data specifically on the age distribution and other characteristics of identifiable, human victims of crime, the victimization survey is excellent, except for its omission of victims under 15 years old.
Projections of the future population of Canada and its age and sex distribution are published periodically by Statistics Canada. This paper uses projections released in March, 2001, based on population estimates for the year 2000 (George, Loh, Verma and Shin 2001).(12) Four main alternative projections are given, which differ in the assumptions that are made concerning the components of demographic change: births, deaths, and migration. Projections 3 and 4 differ in their assumptions as to internal migration but very little with respect to the overall Canadian population. The total population of Canada in 2026 under each projection is (in millions): 34.2 (Projection 1: "low growth"), 36.2 (Projection 2: "medium growth"), and 38.6 (Projection 4: "high growth").
We used all three population projections (low, medium, and high growth) to forecast future crime and victimization rates, but the forecast crime rates based on these three projections differed by very little: less than 1 percent by 2026 and approximately 2 percent by 2041. Therefore, only the results based on the "medium growth" population projection are reported here. Of course, the implication is that crime rate forecasts are very robust with respect to varying population change scenarios.
Projected changes in the sex ratio (the number of males per 100 females) of the population could be significant for forecasting crime rates, because males are responsible for approximately 4 times as much recorded crime per capita as females, although their victimization risks are approximately equal. The sex ratio in the Canadian population -- particularly in the age range that is responsible for most recorded crime -- is projected to change very little between 1999 and 2041. Among 15-39 year olds, the sex ratio is projected to change from 103.1 in 1999 to 104.0 in 2041: a change of less than 1 percent.(13) Sex ratios are projected to change substantially in the over-65 age group, but this group is involved in so little recorded crime or victimization that changes in its sex structure have no significant implications for crime projections. This means that forecasts of crime rates are not affected materially by changes in the sex ratio, and can therefore be estimated using only age-specific rather than age/sex-specific rates.
Forecasts of crime and victimization rates
Based on the population projections described above, the recorded crime rate in Canada will decline from 8,570 incidents per 100,000 population in 1999 to 7,261 in 2026: a drop of 15 percent in 27 years, or a little more than 0.5 percent per year on average. By 2041, it will have declined to 6,978 incidents per 100,000 population, or 81 percent of its 1999 level (Figure 4).
The rate(14) of criminal victimization is projected to decline from 24,830 per 100,000 in 1999 to 21,661 in 2026 -- a drop of 13 percent -- and to 20,782 in 2041, a total decrease of 16 percent from its 1999 level. The forecast trajectory of the rate of criminal victimization (Figure 4) is similar to that of the recorded crime rate, since the age distributions of recorded offenders and self-reported victims are similar.
The reason for this projected decline in crime rates is, ex hypothesi, the changing age structure of the population (Figure 5).(15) The low crime age groups (0-14 and over 39 years old) are projected to increase gradually in relative size, from 63 percent of the population in 1999 to 70 percent in 2026 and to 72 percent in 2041; while the high and very high crime age groups are projected to decline from 24 to 18 percent and from 14 to 11 percent, respectively, by 2041. The projected rates of decline in the high and very high crime age groups vary over time, with the result that the crime rate is forecast to fall a little faster during the period between about 2011 and 2026.
The high crime age group (25-39 years old) is projected to experience relatively rapid decline in relative numbers from 1999 to about 2006, reflecting the aging of the swollen numbers of baby-boomers, born between 1946 and 1966.(16) The baby boomers had already left the very high crime age range (15-24) by about 1990. They began leaving the high crime age range (25-39) in about 1986 and will finish that process by about 2006.
The very high crime age group (15-24 years old) is projected to decline in relative size over the entire period 2000-2026. It is projected to increase in absolute numbers until 2011. Presumably, it is this projected increase in absolute size, and its counterparts in the USA and Britain, that are the basis of published predictions of increased youth and young adult crime during that period. The demographic projections suggest a 7 percent increase from 1999 to 2011 in the absolute number of incidents involving that age group,(17) while the total number and overall per capita rate of incidents are falling.
There will be a rapid decline between about 2011 and 2022 in the relative size of the very high crime age group, reflecting the aging-out of the "baby-boom echo" generation (children of the baby-boomers), born during the period from about 1980 to about 1995. This smaller replica of the baby boom population bulge began entering the very high crime age range in about 1995, and will be moving out of it and into the high crime age range during the period from about 2005 to about 2019. This group will then leave the high crime age range between about 2020 and 2035: this accounts for the more rapid rate of decline evident in Figure 5 for the high crime age group during the period between about 2023 and 2036.
The relative contribution of each age group to the projected decline in recorded crime rates is assessed in Table 1. In 1999, 15-24 year olds made up 14 percent of the population and were implicated in 36 percent of the recorded crime (3,092 incidents per 100,000 overall population); by 2026, their proportion of the population is forecast to fall to 11 percent, and their crime to 34 percent (2,435 incidents) of the considerably lower overall crime rate. The relative decline in this group's numbers accounts for 40 percent of the forecast change in the overall crime rate by 2026.(18)
Table 1 Breakdown of forecast changes in recorded crime rates by age group, 2026 and 2041 Age group 0-14 15-24 25-39 40-49 50+ total 1999 population in thousands ([Qi..sub.1999]) 5915 4124 7158 4819 8478 30493 Proportion of 1999 population ([Pi..sub.1999]) 0.19 0.14 0.23 0.16 0.28 1.00 Contribution to overall crime rate (Ci[Pi..sub.1999]) 513 3092 3253 1145 567 8570 Proportionate contri- bution to crime rate (Ci[Pi..sub.1999]/ [C.sub.1999]) 0.06 0.36 0.38 0.13 0.07 1.00 2026 population in thousands ([Q.sub.i.2026]) 5382 3862 7083 4903 14960 36191 Proportion of 2026 population ([P.sub.i.2026]) 0.15 0.11 0.20 0.14 0.41 1.00 Contribution to overall crime rate ([C.sub.i] [P.sub.i.2026]) 395 2435 2725 970 736 7261 Proportionate contribution to crime rate ([C.sub.i] [P.sub.i.2026]/ [C.sub.2026]) 0.05 0.34 0.38 0.13 0.10 1.00 Change from 1999 -118 -657 -528 -175 168 -1309 Proportionate contribution to decrease from 1999 0.07 0.40 0.32 0.11 0.10 1.00 2041 population in thousands ([Q.sub.i.2041]) 5088 3998 6559 4927 16532 37105 Proportion of 2041 population ([P.sub.i.2041]) 0.14 0.11 0.18 0.13 0.45 1.00 Contribution to overall crime rate ([C.sub.i] [P.sub.i.2041]) 374 2455 2470 940 740 6978 Proportionate contri- bution to crime rate ([C.sub.i][P.sub.i. 2041]/[C.sub.2041]) 0.05 0.35 0.35 0.13 0.11 1.00 Change from 1999 -139 -637 -783 -205 173 -1592 Proportionate contribution to decrease from 1999 0.07 0.33 0.40 0.11 0.09 1.00 Sources: Custom tabulations provided by Canadian Centre for Justice Statistics, UCR2 Survey, for 1999; formulae given in the Appendix for 2026 and 2041.
An additional 32 percent of the change in recorded crime to 2026 is due to the shrinking of the 25-39 year old group from 23 percent of the population in 1999 to 20 percent by 2026. The accelerated projected decline in relative numbers of the 25-39 year old group from about 2023 to 2036 means that it accounts for 40 percent of the forecast change from 1999 to 2041 in the recorded crime rate; whereas the 15-24 year old group is responsible for only 33 percent. This analysis demonstrates, inter alia, that in the demographic explanation of trends in crime rates, changes in the proportion of 25-39 year olds in the population can be more important than those of 15-24 year olds. Although the 25-39 year olds are implicated in less crime per capita, there are many more of them.
Forecasts of rates of different types of recorded crime
Different types of crime have different offender age distributions: rates of property crime decline rapidly with offender age after peaking in the late teens; whereas rates of crime against the person ("violent" crime) and of other ("victimless") crime decline more slowly with age: therefore, rates of different types of crime should respond differently to projected changes in the age structure of the population. Figure 6 shows forecast recorded rates of these three broad categories of crime. To facilitate comparison, the rates of crimes against the person and other crime have been scaled up by factors of 4 and 2 respectively.
All three categories of crime are projected to decline in similar ways. There is a period of more rapid decline in all three rates during the period between about 2011 and 2026, which was noted above for the projected overall crime rate. The period of more rapid decline in the rate of property crime begins a little earlier -- in about 2009 -- and the decline is a little steeper until about 2018, than in the other two types of crime, reflecting the passage of the tail end of the echo generation out of their late teens, an age characterized by elevated rates of property crime.
By 2026, the rate of recorded property crime is projected to have fallen to 84 percent of its 1999 level, and rates of crime against the person and other crime to 86 and 85 percent, respectively, of their 1999 levels (Table 2). By 2041, recorded property crime is projected to be at 81 percent, and the other two categories of crime at 83 and 82 percent, respectively, of their 1999 levels. Although there are slight variations over the 42 year period in the projected rates of decline of these three categories of crime, their relative contributions to the overall recorded crime rate remain constant, within half a percentage point, at 62 percent property crime, 11 percent crimes against the person, and 27 percent other crime.
Table 2 Actual and forecast recorded rates of specific offences, 1999, 2026, and 2041 Incidents per 100,000 Ratio 1999 2026/ 2041/ (actual) 2026 2041 1999 1999 Homicide, attempt 4 3 3 0.86 0.83 Assault and sexual assault, levels 2 and 3 129 111 107 0.86 0.82 Robbery 92 75 72 0.82 0.78 Sexual assault level 1 74 67 64 0.91 0.87 Assault level 1 578 499 477 0.86 0.82 Other person 79 69 66 0.88 0.84 Total person 955 825 789 0.86 0.83 Break and enter 1044 860 828 0.82 0.79 Other indictable property 645 531 515 0.82 0.80 Theft under 2227 1908 1838 0.86 0.83 Other summary and hybrid property 1416 1187 1138 0.84 0.80 Total property 5332 4487 4318 0.84 0.81 Administration of justice 287 237 228 0.83 0.79 Public order, morals, weapons 321 278 267 0.87 0.83 Drugs 265 216 209 0.81 0.79 Criminal Code traffic 455 404 388 0.89 0.85 Miscellaneous 953 813 780 0.85 0.82 Total other 2283 1949 1872 0.85 0.82 Total 8570 7261 6978 0.85 0.81 Sources: Custom tabulations provided by Canadian Centre for Justice Statistics, UCR2 Survey, for 1999; formulae given in the Appendix for 2026 and 2041.
Table 2 shows actual recorded levels in 1999, and forecasts to 2026 and 2041, of specific types of crime. There are some variations from the overall declines of 15 percent by 2026 and 19 percent by 2041. Crimes that are most characteristic of adolescents and young adults, such as robbery, break and enter, other indictable property crime, and drug offences, show above-average declines; while crimes that are more characteristic of adults, such as sexual assault, Criminal Code traffic offences (mainly drinking-driving offences), and miscellaneous offences against the person, are least affected by the aging of the population.
This paper has forecast levels of recorded crime and criminal victimization in Canada to the year 2041. The forecasts are based on the assumption that age-specific rates of these phenomena are constant over time: that the only factor causing changes in their overall levels is change in the age structure of the population. This assumption is made, not because it is believed to be justified, but in order to explore the implications of anticipated demographic change, while ignoring possible changes in all other criminogenic factors.
All forms of crime in Canada that were examined are forecast to fall slowly between 1999 and 2041, with a period of slightly faster decline between about 2011 and 2026. The forecast decline in the crime rate is due to the projected aging of the Canadian population, and is due in about equal parts to the declines in relative size of the 15 to 24 year old age group -- which has the highest crime rate -- and of the 25 to 39 year old group, which also has a high crime rate, and is larger than the 15-24 year old group. Variations over time in the rate of decline of the crime rate are related mainly to the passage of the baby boom echo generation through and out of the crime-prone years of late adolescence and early adulthood. By 2026, the recorded crime rate and risk of criminal victimization are forecast to be at 85 and 87 percent, respectively, of their 1999 levels; by 2041, at 81 and 84 percent, although these latter forecasts are so far in the future that they must be treated with considerable caution. The patterns of decline in these two indicators of crime in Canada are very similar.
Forecast declines in recorded rates of particular types of crime are related to their age profiles: crimes such as robbery, break and enter and other indictable property crime, and drug offences, which are characteristic of teenagers and young adults, are forecast to decline somewhat faster and farther; whereas crimes such as sexual assault, Criminal Code traffic offences, and miscellaneous offences against the person, which are more characteristic of older adults, are forecast to decline less.
It must be emphasized that these are not predictions of the future: these forecasts take no account whatsoever of possible future changes in other factors that affect measured crime. These other factors include changes in social and economic conditions, changes in the role of women which may affect female crime rates, legislative changes such as the creation of new offences, changes in the administration of justice, such as police enforcement, diversion, and recording practices, and changes in the reporting of crime by the public. Not all likely demographic change has been included in these projections; for example, the projected change in age structure of the First Nations population is very different from that of other Canadians, and this may have a substantial effect on future local crime rates in areas with large First Nations populations.(19) The forecasts reported here represent estimates of changes in levels of crime that can be expected to occur as the result of demographic change alone; these constitute a baseline on which trends caused by changes in the many other criminogenic factors will be overlaid.
(1.) I am indebted to John Fleischman for suggesting this line of research and for his help in obtaining support for an earlier version of it (Carrington 1999) through a contract with the Department of Justice Canada. Custom tabulations from the UCR2 Survey were prepared by Rebecca Kong, Paul de Souza, and Al Harding, Canadian Centre for Justice Statistics. Shabiran Rahman of UW Dana Porter Library was most helpful in obtaining recent Statistics Canada population projections. J.E. Curtis, M.V. George, J.V. Roberts, R. Verma, and the Journal's two anonymous referees provided helpful comments on earlier drafts. Preparation of this article was supported by SSHRC Standard Research Grant No. 410-2000-0361.
(2.) Except "victimless" crime.
(3.) The following analyses are also limited to linear effects of demographic change. Easterlin (1978, 1980) proposed that changes in the age structure may have non-linear effects on crime: for example, a very large cohort of young people (such as baby boomers in the 1960's) may experience unusual economic strain or opportunities for criminal activity, so that the age-specific crime rate temporarily rises. Using data from the Ontario juvenile court, Maxim (1985) found support for the Easterlin hypothesis, but on a very limited sample. Subsequent American research has found little support for this hypothesis (Steffensmeier, Streifel, and Shihadeh 1992; Steffensmeir, Streifel, and Harer 1987; O'Brien 1989; Levitt 1999; Savolainen 2000).
(4.) These include all incidents allegedly involving federal statute offences: offences under the Criminal Code (including Criminal Code traffic offences), the Controlled Drugs and Substances Act, and other federal statutes. Unless otherwise noted, crime statistics reported in this and the following paragraph are from Canadian Centre for Justice Statistics (2000): Tables 3.3 and 4.19.
(5.) Property crimes have a below-average likelihood of being solved: in 1999, the clearance rate for incidents of property crime was 23 percent, compared with an average clearance rate for all types of recorded crime of 40 percent (Canadian Centre for Justice Statistics 2000: Table 3.3). Since young persons are disproportionately involved in property crime, they are probably over-represented in recorded unsolved incidents, and under-represented in recorded incidents that have been cleared. Therefore, an offender age distribution based on UCR data probably understates the crime rates of children and teenagers and overstates adult crime, as a proportion of total recorded crime. This bias may be offset to some extent by "reporting bias": property crime is more likely than crimes against the person to be reported to police (Gartner and Doob 1994). The fact that young persons who are apprehended by police are less likely than adults to be charged, and more likely to be dealt with informally, does not bias the age-crime data used here, since they include all apprehended persons, whether charged or processed by other means (see note 8 below).
(6.) Calculated from custom tabulations prepared for the author by Canadian Centre for Justice Statistics.
(7.) See note 5 above. Note that it is not necessary to assume that the crime rate in the part of Canada covered by the UCR2 was the same as the overall Canadian crime rate, since the extrapolation to Canada used the overall number of incidents (2.61 million) from the UCR.
(8.) In allocating incidents to age groups, I included all persons apprehended in connection with the incident. In the UCR2, these are termed "charged suspect/chargeable": "... any person who has been identified by police as being involved in a criminal incident and against whom an information could be laid as a result of sufficient evidence/information ... [including] those charged as well as those not charged for a variety of reasons, including diplomatic immunity, use of alternative measures or diversion, and death ... [but excluding persons] involved only for investigative purposes and subsequently released without being charged or formally processed in some other official manner" (Canadian Centre for Justice Statistics 2000: 73). Some incidents involve more than one identified alleged offender. For such incidents, I allocated a fraction of an incident to each relevant age group. For example, if an incident implicated a 10 year old and a 17 year old, then I counted 0.5 incidents for each of the 10-14 and 15-19 year old age groups. This fractional allocation procedure resulted in the correct relative allocation of incidents among age groups, while maintaining the correct total number of incidents. Persons aged 15-19 years were actually implicated in more than 78,890 incidents in the 1999 UCR2, some of which involved multiple apprehended offenders.
(9.) This does not necessarily mean that one in four 15-19 year olds was involved in a recorded crime in 1999, since some persons are implicated in more than one incident in a given year, and each of those incidents contributes to the per capita crime rate.
(10.) All statistics reported in this paper from the GSS have been weighted to yield population (of Canada) estimates.
(11.) For example, "Now I'm going to ask you a question about being attacked. An attack can be anything from being hit, slapped, pushed or grabbed, to being shot or beaten. Please remember to include acts committed by family and non-family.... were you attacked by anyone at all? ... [If so,] How many times?" (Question C8(a), GSS Personal Risk Questionnaire) (Statistics Canada Housing, Family and Social Statistics Division 1994).
(12.) The detailed projections contained on the CD-ROM included with the publication were used. (These are also available -- but only to 2026 -- in Statistics Canada 2001: Matrix 6900.) Forecasts based on demographic projections for 2027-2041 should be treated with particular caution, since "the further one forecasts into the future, the more the assumptions are likely to be wrong" (Newell 1988:181).
(13.) These sex ratios are based on the medium-growth population projection, but projected sex ratios are practically the same under the low-and high- growth projections.
(14.) To facilitate comparison with the recorded crime rate, the risk of criminal victimization has been converted to a rate: defined as the number of victims per 100,000 population.
(15.) The forecasts of crime rates used all 18 and 14 age groups shown in Figures 2 and 3; these have been aggregated into three groups in Figure 5 in order to simplify the presentation.
(16.) A small decrease in the absolute size of this age group is projected during the period 1999-2006, but the decline in its proportion of the population is considerably greater, since the total population is increasing.
(17.) From 941,871 incidents in 1999 to 1,007,876 in 2011.
(18.) The formula used to calculate the contribution of the change in the crime rate of age group i to the change in the overall crime rate is:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
(19.) I am indebted to an anonymous referee for pointing this out.
(20.) This development is adapted from Levitt (1999).
(21.) As offender, or as victim, depending on which manifestation of crime is being examined. The problem of incidents involving multiple offenders or victims is dealt with in note 8 above.
Besserer, Sandra and Catherine Trainor 2000 Criminal victimization in Canada, 1999. Juristat 20 (11). Catalogue No. 85-002-XPE. Ottawa: Canadian Centre for Justice Statistics, Statistics Canada.
Canadian Centre for Justice Statistics 1997 National coverage for UCR II - as of Jan. '97. Personal communication, Feb. 10, 1997.
Canadian Centre for Justice Statistics 2000 Canadian Crime Statistics, 1999. Catalogue No. 85-205-XPE. Ottawa: Canadian Centre for Justice Statistics, Statistics Canada.
Carrington, Peter J. 1999 Demographic Projections of Crime in Canada, 1999-2016. Report to the Department of Justice Canada.
Chaiken, Jan M. 2000 Crunching numbers: Crime and incarceration at the end of the millennium. National Institute of Justice Journal. January, 2000: 10-17.
Cohen, L.E. and K.C. Land 1987 Age structure and crime: Symmetry versus asymmetry and the projection of crime rates through the 1990s. American Sociological Review 52: 170-183.
Easterlin, Richard A. 1978 What will 1984 be like? Socioeconomic implications of recent twists in age structure. Demography 15: 397-421.
Easterlin, Richard A. 1980 Birth and Fortune. The Impact of Numbers on Personal Welfare. New York: Basic Books.
Fattah, Ezzat A. 1991 Understanding Criminal Victimization. Scarborough, Ont.: Prentice-Hall Canada.
Ferdinand, Theodore N. 1970 Demographic shifts and criminality. British Journal of Criminology 10: 169-175.
Field, Simon 1999 Trends in Crime Revisited. London: Home Office.
Foot, David K. 1998 Boom, Bust & Echo 2000. Rev. ed. Toronto: Macfarlane Walter & Ross.
Foot, David K., Richard A. Loreto, and Thomas W. McCormack 1998 Demographic Trends in Canada, 1996-2006: Implications for the Public and Private Sectors. Ottawa: Industry Canada.
Fox, James A. 1996 Trends in Juvenile Violence: A Report to the United States Attorney General on Current and Future Rates of Juvenile Offending. Boston: Northeastern University Press.
Gartner, Rosemary and Anthony N. Doob 1994 Trends in criminal victimization: 1988-1993. Juristat 14 (13). Catalogue No. 85-002-XPE. Ottawa: Canadian Centre for Justice Statistics, Statistics Canada.
George, M.V., S. Loh, R. Verma, and Y.E. Shin 2001 Population Projections for Canada, Provinces, and Territories, 2000-2026. Catalogue No. 91-520-XPB. Ottawa: Statistics Canada.
Hirschi, Travis and Michael Gottfredson 1983 Age and the explanation of crime. American Journal of Sociology 89: 552-584.
John Howard Society of Ontario 1999 Population Trends and Crime: What should we be planning for? Fact Sheet #13. Toronto: John Howard Society of Ontario.
Levitt, Steven D. 1999 The limited role of changing age structure in explaining aggregate crime rates. Criminology 37: 581-597.
Marvell, Thomas and Carlisle Moody 1991 Age structure and crime rates: The conflicting evidence. Journal of Quantitative Criminology 7: 237-273.
Maxim, Paul S. 1985 Cohort size and juvenile delinquency: A test of the Easterlin hypothesis. Social Forces 63: 661-681.
Newell, Colin 1988 Methods and Models in Demography. London: Pinter.
O'Brien, Robert 1989 Relative cohort size and age-specific crime rates: An age-period-relative-cohort-size model. Criminology 27: 57-77.
Sacco, Vincent F. and Holly Johnson 1990 Patterns of Criminal Victimization in Canada. General Social Survey Analysis Series. Catalogue No. 11-612E, No. 2. Ottawa: Statistics Canada.
Sagi, Phillip C. and Charles F. Wellford 1968 Age composition and patterns of change in criminal statistics. Journal of Criminal Law, Criminology, and Police Science 59: 29-36.
Savolainen, Jukka 2000 Relative cohort size and age-specific arrest rates: A conditional interpretation of the Easterlin effect. Criminology 38:117-136.
Statistics Canada 2001a CANSIM. CANadian Socio Economic Information Management System. In Statistics Canada's Internet site: http:// www.statcan.ca:80/english/CANSIM/index.html. March 1, 2001.
Statistics Canada 2001b General Social Survey. Cycle 13. Victimization. At: http:// www.drc.uoguelph.ca/cgi-bin2/drc.cgi/. March 5, 2001.
Statistics Canada Housing, Family and Social Statistics Division 1994 Tables in Victimization. 1993 General Social Survey Series. Product No. 12F0042XPE. Ottawa: Statistics Canada.
Steffensmeier, Darrell J., Emilie A. Allan, Miles D. Harer, and Cathy Streifel 1989 Age and the distribution of crime. American Journal of Sociology 94: 803-831.
Steffensmeier, Darrell J. and Miles D. Harer 1987 Is the crime rate really falling? An "aging" U.S. population and its impact on the nation's crime rate, 1980-1984. Journal of Research in Crime and Delinquency 24: 23-48.
Steffensmeier, Darrell J. and Miles D. Harer 1991 Did crime rise or fall during the Reagan presidency? The effects of an "aging" U.S. population on the nation's crime rate. Journal of Research in Crime and Delinquency 28: 330-359.
Steffensmeier, Darrell, Cathy Streifel, and Miles D. Harer 1987 Relative cohort size and youth crime in the United States, 1953-1984. American Sociological Review 52: 702-710.
Steffensmeier, Darrell, Cathy Streifel, and Edward S. Shihadeh 1992 Cohort size and arrest rates over the life course: The Easterlin Hypothesis Reconsidered. American Sociological Review 57: 306-314.
Tremblay, Sylvain 1999 Crime statistics in Canada, 1998. Juristat 19 (9). Catalogue No. 85-002-XPE. Ottawa: Canadian Centre for Justice Statistics, Statistics Canada.
Wellford, Charles F. 1973 Age composition and the increase in recorded crime. Criminology 11: 61-70.
Wilson, James Q. 1995 Crime and public policy. In James Q. Wilson and Joan Petersilia (eds.), Crime. San Francisco, CA: Institution for Contemporary Studies Press.
Zimring, Franklin E. 1998 American Youth Violence. New York: Oxford University Press.
Demographic forecasts of crime rates(20)
In base year y, let the number of criminal incidents in which a person in age group i is involved(21) be represented by [T.sub.iy]. Let the number of persons in age group i in the Canadian population in year y be [Q.sub.iy]. Then the age-specific crime rate for age group i_is:
[C.sub.i] = [T.sub.iy]/[Q.sub.iy].
(Note that [C.sub.i] does not need the year subscript y because it is assumed to be the same every year.)
For example, in the base year, 1999, an estimated 530,076 recorded criminal incidents involved 15-19 year old offenders, and the number of 15-19 year olds in Canada was 2.062 million: then the age-specific crime rate for 15-19 year olds was 530,076/ 2.062m = 0.25708 incidents per capita (or 25,708 incidents per 100,000 population). This age-specific rate is assumed to remain constant in the future, so the projected number of incidents involving 15-19 year olds in a future year z is a function of the projected number of 15-19 year olds in year z:
[T.sub.iz] = [C.sub.i][Q.sub.iz].
Under the medium growth projection, the number of 15-19 year olds in Canada is projected to increase to 2.175 million by the year 2011, so they will be implicated in 559,149 (2.175 million* 0.25708) incidents. Thus, the number of incidents involving 15-19 year olds is projected to increase by about 29,000, while their crime rate remains 0.25708 per capita. The total number of incidents projected for any future year z is the sum of the age-specific projections:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII],
and the projected overall crime rate in year z is the projected total number of crimes divided by the projected total population:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII],
If [P.sub.iz] is defined as the proportion of the total population in year z represented by a given age group i:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII],
then the formula for the projected aggregate crime rate in year z can be simplified to:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII],
that is, the weighted sum of age-specific crime rates. Since the age-specific crime rates ([C.sub.i]) are assumed to be fixed, this formula implies that the overall crime rate ([C.sub.z]) changes from year to year as a result of changes in the proportions of the population accounted for by the different age groups ([P.sub.iz]).
Peter J. Carrington University of Waterloo Waterloo, Ontario…