Read an Excerpt
Why Crime Rates Fall and Why They Don't
By Michael Tonry
The University of Chicago PressCopyright © 2014 The University of Chicago
All rights reserved.
Why Crime Rates Are Falling throughout the Western World
Crime rates have moved in parallel in Western societies since the late Middle Ages. Homicide rates declined from 20 to 100 per 100,000 population in western Europe to one per 100,000 in most Western countries by the beginning of the twentieth century. Crime rates in major cities and in countries fell from the early nineteenth century until the middle of the twentieth. From the 1960s to the 1990s, rates for violent and property crimes rose in all wealthy Western countries. Since then, rates in all have fallen precipitately for homicide, burglary, auto theft, and other property crimes. The patterns appear in both police and victimization data. Rates for nonlethal violence have fallen sharply in the English-speaking countries and parts of continental Europe. In other parts of Europe, nonlethal violence has been stable or increasing, but the data are probably wrong. Interacting changes in rates of reporting and recording and in cultural thresholds of tolerance of violence that occurred earlier in the English-speaking countries are the likeliest explanation for the appearance of crime rate increases. Diverse explanations have been offered for both the long-and short-term declines. Most agree that, whatever the explanations may be, they do not include direct effects of changes in policing or sanctioning policies.
Almost no one except a handful of academic specialists seems to have noticed that crime rates are falling throughout the Western world. That is curious. It should be seen everywhere as good news. Fewer people are victimized. Fewer are arrested, prosecuted, convicted, and punished. Hospital emergency rooms handle fewer intentional injuries. Insurance companies compensate fewer losses. Politicians have less incentive to propose and policy makers to adopt severe policies aimed at pleasing, placating, or pacifying an anxious public.
No one has a really good explanation for why crime rates are falling. Since few people have noticed that they are, this is not entirely surprising. Why crime rates are falling may in any case be the wrong question. The better question may be, Why did crime rates rise in all developed Western countries during the 1960s, 1970s, and 1980s before peaking and turning downward? There is nearly incontestable evidence that homicide rates fell dramatically and more or less continuously from the late Middle Ages through the middle of the twentieth century (Eisner 2003; Spierenburg 2008, 2012; Muchembled 2011). Few historians question that rates for other crimes fell from the early nineteenth century to the middle of the twentieth (e.g., Gurr, Grabosky, and Hula 1977). Those findings support a hypothesis that the rising rates of recent decades were an anomaly. The centuries-long downward trend has been interrupted before by profound social and economic dislocations. Crime rates usually fall during wars, for example, rise for a short period afterward, and then resume long-term patterns (Archer and Gartner 1976, 1984; Gurr 1981, pp. 344–46). Well-known social and economic dislocations of recent decades affected all developed countries and may simply have caused another short-term disruption to the long-term trend.
My aims in this essay are to demonstrate that crime rates have moved in parallel in the English-speaking countries and western Europe since the 1960s and to assess the main explanations for why they rose in the 1970s and 1980s and fell afterward. The implications are important. One is that Western countries are much more closely linked in structural, social, and cultural ways than is commonly recognized. Many demographic patterns in recent decades have characterized most or all developed countries including rising life expectancies, female labor force participation, and income inequality and declining fertility, infant mortality, and marriage rates. It would be odd if crime trends were not also common. All Western countries (and many others) appear to march to the same distant drummers without realizing that is what they are doing.
A second implication is that many of the things that governments have done to reduce crime rates in recent decades have been largely epiphenomenal—normatively and politically important, and having major effects on many people's lives, but pretty much beside the point in terms of crime rates and patterns. It is at least a little parochial, for example, that the US National Academy of Sciences twice in recent years has sought to explain recent American crime trends primarily in terms of distinctively American developments such as mass imprisonment, policing initiatives, legalization of abortion, and reduction in children's exposure to lead paint (e.g., Goldberger and Rosenfeld 2008; Rosenfeld 2014).
Comparison of American and Canadian developments makes the parochialism clear. Canadian crime rate patterns have closely paralleled America's since 1960, but Canada's imprisonment patterns and criminal justice policies have been starkly different (Webster and Doob 2007). Since 1960, the Canadian imprisonment rate has fluctuated around 100 per 100,000 population, while America's rose from 150 per 100,000 in 1970 to 756 in 2007. Canadian agencies have not emphasized zero-tolerance and other aggressive forms of policing, and the Canadian Parliament did not enact three-strikes, truth-in-sentencing, and life-without-possibility-of-parole laws. Only in the last couple of years have significant mandatory minimum sentence laws been enacted; they do not, however, require sentences measured in decades and lifetimes and are meeting strong resistance from the appellate courts (R. v. Smickle, 2013 Ontario Court of Appeals 677 [CanLII]; R. v. Nur, 2013 Ontario Court of Appeals 678 [CanLII]; Doob and Webster 2013).
Yet crime rates moved in tandem. Figure 1 shows US and Canadian homicide, robbery, and imprisonment rates since 1960. The US imprisonment rate nearly quintupled; the Canadian rate barely changed. Homicide rates, however, moved almost in lockstep. Robbery trends also were closely similar. Whatever caused homicide rates in the two countries to change in parallel, it was something other than changes in imprisonment rates and sentencing laws. Yet both US National Academy of Sciences initiatives have sought to explain American crime trends as if the reasons for them are uniquely American.
Reasonable arguments can be made that recent drops in property crime rates were influenced by improved security technologies in motor vehicles, residences, and retail stores (e.g., van Dijk, Tseloni, and Farrell; Farrell, Tilly, and Tseloni 2014). Those developments may explain minor differences in trends between countries that adopted them more and less extensively or at different times, but all developed countries experienced substantial declines. Few plausible arguments can be made, however, that technological or crime control policy changes have caused sharp parallel declines in homicide and other violent and sexual offending. Something more fundamental is happening.
The evidence that crime rates have moved in parallel in Western countries between the 1960s and 1990s, with occasional lags in particular countries, is overwhelming. No informed person disagrees that homicide rates rose from various starting points in the 1950s and 1960s, peaked in the 1990s, and have since fallen (e.g., Eisner 2008, 2014). No informed person disagrees that nonviolent property crime rates—burglary, theft, motor vehicle theft—in all Western countries have been falling at least since the 1990s (van Dijk and Tseloni 2012). In some countries including the United States, there are good reasons from victimization data to believe that property crime declines began earlier. The declines in property crime rates are especially striking since they continued during the recessions that afflicted almost all Western countries beginning in 2008 and are continuing in many as these words are being written.
The disagreements concern trends in nonhomicidal violence and sexual offending since the early 1990s. In the English-speaking and some western European countries, both police and victimization data show that robbery, rape, and aggravated assault rates have fallen since the early to mid-1990s. In other western European countries, rates for nonhomicidal violence have not fallen since the early 1990s and in some have increased (e.g., Aebi and Linde 2010, 2012). The challenge is to explain those divergent patterns.
The explanation is that there is no divergence. The appearance of one is misleading. The incidence not only of homicide but of all violent crime is almost certainly falling in all developed Western countries, but this is obscured in three interrelated ways. First, victims have become more likely to report incidents to the police, but the timing of the increases varies between countries. Victimization surveys regularly ask respondents to indicate whether they reported incidents to the police. In the American national surveys, reporting rates for rape and domestic violence increased in the 1970s; after the mid-1980s, reporting rates for violent and sexual offenses increased substantially. Among the effects are that apparent increases in violent crime in the 1970s and 1980s were exaggerated and recent declines have been substantially underestimated. The most comprehensive analysis concludes that, when changes in victim reporting are taken into account, nonlethal serious violence decreased by 51 percent between 1991 and 2005. Police data indicate only a 27 percent decline (Baumer and Lauritsen 2010, p. 173). Similar patterns of increased levels of reporting to the police have been documented in England and Wales and Scandinavia, other places where annual victimization surveys have long been conducted (Chaplin, Flatley, and Smith 2011; Kivivuori 2014). As in the United States, apparent rate increases in official crime data have been exaggerated and decreases underestimated.
Second, net of changes in victim reporting, police recording practices have changed in ways that artificially increase crime rates. Partly this is the result of professionalization of police management and the related shift over 40 years from paper to electronic record keeping (Reiss 1992; Mastrofsky and Willis 2010). Recording rates for most crimes have increased, but more for some than for others. Domestic violence and sexual offenses are the paradigm cases, but other violent offenses are affected. No one who has lived through recent decades can fail to recognize that people in Western societies (at least) have become much less tolerant of intimate violence, violence against women, and sexual offenses generally (Pierotti 2013). Both for public relations reasons—to forestall criticism of police insensitivity—and because police are part of society and inevitably are affected by changing cultural attitudes, some kinds of incidents have become more likely to be officially recorded as crimes.
In some countries, policy decisions have been made to increase recording rates. Two major recent changes were made in England and Wales. In 1998–99, counting rules were changed to record more minor, or "summary," offenses particularly for less serious violent crimes, frauds, and drug offenses. The Home Office estimated that the change increased crime statistics overall by 14 percent. However, there were wide variations by type of offense. Burglary and robbery were estimated to have been little affected but other violence against the person to have increased by 118 percent (Povey and Prime 1999). Hough, Mirrlees-Black, and Dale (2005, p. 28) later observed that the changes "would have taken several years to bed in ... [and] will have artificially inflated the count of crimes each year."
In 2002, the National Crime Recording Standard was introduced. England and Wales shifted from operating an "output" system in which police apply established criteria before recording incidents as crimes to an "input" system in which citizens' reports are taken at face value. The aim was to make the process more victim-oriented. The Home Office estimated that the change increased total crime rates by 10 percent in 2002/3, but as with the 1998–99 change, the effects varied between types of crime and were predicted to be especially significant for violent crimes (Simmons and Dodd 2003).
Third, cultural thresholds of tolerance have been changing since at least the 1960s. Behaviors now regarded as appropriately reported to police were in earlier times often regarded as unpleasant, undesirable, or socially unacceptable but not as criminal. Well-known examples include violence within domestic relationships, fights between acquaintances or drinking partners, and unwanted sexual touching in public. If the cultural meanings of such behaviors change over time, people at different times will use the same words to describe wrongful acts but mean different things. Here, too, domestic violence and sexual assaults are the paradigm cases. Had there been national victimization surveys in the 1960s, for example, respondents would have been much less likely than now to have thought of a blow from a husband, wife, or partner as an assault. Many victims of domestic violence or of bar fights would not have considered themselves to be crime victims and said "no" if asked if they had been assaulted. Many more such people in the 1980s and even more in the 2000s would have answered "yes." Issues of reporting and recording by definition would not have been raised in the earlier period for behaviors not then thought of as crimes. The social meaning of assault would have changed even though respondents themselves did not realize that they now considered behaviors to be criminal that earlier they would not have.
Similar changes are likely to have occurred concerning a wide range of sexual incidents especially including allegations of nonstranger rapes and attempted rapes in dating and acquaintance relationships. Feminists worked hard in the 1970s and 1980s to expand cultural conceptions and legal definitions of rape (Estrich 1987). No reasonable person can believe that their efforts were unsuccessful. Something similar occurred in relation to child abuse. Rates of reported child physical and sexual abuse increased enormously in the United States in the 1980s as the result of a succession of new laws requiring social workers, psychologists, nurses, doctors, and psychiatrists to report suspected incidents to the police (Garbarino 1989; Brosig and Kalichman 1992; Kalichman 1999). After the early 1990s, recorded rates of child physical and sexual abuse plummeted. Between 1992 and 2004, reported physical abuse fell by 43 percent and sexual abuse by 49 percent (Finkelhor and Jones 2006, fig. 1).
Taken together, changes in reporting, recording, and cultural thresholds of tolerance in recent decades have increased official crime rates substantially in all developed countries. Those increases are, however, likely to have occurred at different times in different places. Their effects on official crime rates were largely manifest in the United States and some other countries by the early and mid-1990s. In other places, including many western European countries, they occurred and are occurring later. It is difficult, for example, not to hypothesize that all three mechanisms are major causes of recent increases in rates of sexual offending and assaults in official data in the Scandinavian countries (Selmini and McElrath 2014). Sexual crimes are the targets of major social and political movements there and the subjects of important changes in criminal laws and practices (Skilbrei and Holmström 2011, 2013; Tham, Rönneling, and Rytterbro 2011). It would be astonishing if those initiatives had no effects on cultural attitudes, victim behavior, and police policies and practices.
In this essay, I provide evidence that supports the preceding assertions. Section I summarizes the literature concerning a many-centuries-long decline in homicide rates through the mid-1950s, after which rates increased in all developed Western countries. Section II documents declines in homicide rates beginning in the 1990s, declines in rates since then for all crimes in many countries, and discordant evidence of increases in violence rates in a few. Section III canvasses reasons why the apparent increases in violence rates in some countries are much more likely to be the effects of changes in recording, reporting, and cultural thresholds of tolerance than of real increases. Section IV briefly explores the underdeveloped literature that attempts to explain both long-term and recent falls in rates of property and violent crimes. The explanations are simultaneously simple and complex. The simple explanations center on secular subjects including situational crime prevention initiatives and target hardening, the bureaucratization of modern life, and cultural shifts that are denunciatory of crime and violence. The complex explanations concern interactions among cultural and secular changes that have influenced capacities for informal social control and individual self-control (Baumer and Wolff 2014; Eisner 2014).
Excerpted from Why Crime Rates Fall and Why They Don't by Michael Tonry. Copyright © 2014 The University of Chicago. Excerpted by permission of The University of Chicago Press.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.