Reference
Article 1
Rosenfeld, R., Vogel, M., & McCuddy, T. (2019). Crime and Inflation in U. S. Cities. Journal of Quantitative
Criminology, 35(1), 195–210. https://doi-org.su.idm.oclc.org/10.1007/s10940-018-9377-x. Retrieved
from https://web-b-ebscohost-com.su.idm.oclc.org/ehost/pdfviewer/pdfviewer?vid=5&sid=e513cc831ae3-4732-8d88-7fad707ca89b%40sessionmgr101
Article 2
Walker, J., Wilson, P. R., Chappell, D., & Weatherburn, D. (1990). A Comparison of Crime in Australia and
Other Countries. (cover story). Trends & Issues in Crime & Criminal Justice, 23, 1–8. Retrieved from
https://web-b-ebscohost-com.su.idm.oclc.org/ehost/pdfviewer/pdfviewer?vid=9&sid=e513cc83-1ae34732-8d88-7fad707ca89b%40sessionmgr101
Article 3
Schönteich, M. (2002). 2001 CRIME TRENDS: A turning point? SA Crime Quarterly, 1, 1–6. Retrieved from
https://web-b-ebscohost-com.su.idm.oclc.org/ehost/pdfviewer/pdfviewer?vid=11&sid=e513cc83-1ae34732-8d88-7fad707ca89b%40sessionmgr101
Article 4
Byrne, J. M., Pattavina, A., & Taxman, F. S. (2015). International Trends in Prison Upsizing and
Downsizing: In Search of Evidence of a Global Rehabilitation Revolution. Victims & Offenders, 10(4),
420–451. https://doi-org.su.idm.oclc.org/10.1080/15564886.2015.1078186. Retrieved from
https://web-b-ebscohost-com.su.idm.oclc.org/ehost/pdfviewer/pdfviewer?vid=14&sid=e513cc83-1ae34732-8d88-7fad707ca89b%40sessionmgr101
J Quant Criminol (2019) 35:195–210
https://doi.org/10.1007/s10940-018-9377-x
ORIGINAL PAPER
Crime and Inflation in U. S. Cities
Richard Rosenfeld1
•
Matt Vogel1,2 • Timothy McCuddy1
Published online: 3 March 2018
Springer Science+Business Media, LLC, part of Springer Nature 2018
Abstract
Objectives The current study replicates prior national-level research on the relationship
between crimes committed for monetary gain and inflation in a sample of 17 U. S. cities
between 1960 and 2013.
Methods A random coefficients model is used to estimate the effects of inflation on the
change in acquisitive crime over time, controlling for other influences.
Results The estimates yield significant effects of inflation on acquisitive crime rates in the
17 cities. City-specific coefficients reveal nontrivial variation across the cities in the significance, size, and impact of inflation on acquisitive crime.
Conclusions Continued low inflation rates should restrain future crime increases in many
US cities. U. S. monetary policy should be evaluated with respect to its effect on crime.
Keywords Acquisitive crime Inflation Random coefficient models
Recent studies have found that the change in consumer prices over time is a robust
predictor of crimes committed for monetary gain in the United States (Nunley et al. 2016;
Rosenfeld and Levin 2016). Over the past half century, increases in acquisitive crime have
coincided with rising prices and decreases in acquisitive crime have coincided with slowed
inflation. It appears that inflation is an important part of the story of acquisitive crime
trends during the contemporary era, not only in the United States but in other nations as
well (Rosenfeld 2014).
But the story of crime and inflation is far from finished. Nearly all of the research on
crime and inflation, including the Nunley et al. (2016) and Rosenfeld and Levin (2016)
& Richard Rosenfeld
Richard_Rosenfeld@umsl.edu
1
University of Missouri-St. Louis, St. Louis, USA
2
OTB - Research for the Built Environment, TU Delft, Delft, The Netherlands
123
196
J Quant Criminol (2019) 35:195–210
studies, has been conducted at the national level. It remains unclear whether similar results
hold for subnational units, such as cities, that differ in both prices and crime rates. This
paper reports the results of a study that addresses this unresolved research issue. We
estimate the effects of inflation on annual rates of acquisitive crime in 17 US cities for
which both inflation and crime data were available between 1960 and 2013. Our model
yields estimates of inflation effects on city crime rates, controlling for income, unemployment, police strength, and other conditions associated with macro-level crime rates.
We find significant inflation effects on crimes committed for monetary gain for the sample
as a whole, but the significance and size of the effects differ across the cities.
Background
Several prior studies have reported a significant relationship between inflation and crime
rates in the United States (e.g., Allen 1996; Cohen and Felson 1979; LaFree and Drass
1996; Ralston 1999). As Rosenfeld and Levin (2016) point out, however, past research
lacks a compelling rationale for why the effect of inflation on crime rates should differ
from that of other economic indicators, such as unemployment or economic growth.
Rosenfeld and Levin contend that inflation increases robbery, burglary, and other crimes
committed for monetary gain by bolstering consumer demand for cheap stolen goods. As
prices rise, stolen goods become more attractive to consumers in comparison with the same
goods sold by legitimate retailers, including bargain outlets such as the thrift stores
operated by the Salvation Army and Goodwill Industries. Increased demand, in turn,
strengthens incentives for the robbers, burglars, and thieves who supply underground
markets with stolen merchandise. As a result, acquisitive crime rates increase. By the same
logic, lower inflation diminishes the attraction of stolen goods to consumers and criminals
alike, and acquisitive crime rates decline (Rosenfeld and Levin 2016:431–432). Nunley
et al. (2016) posit a somewhat similar economic theory connecting acquisitive crime rates
to inflation.
Subnational Effects of Inflation
Rosenfeld and Levin (2016) acknowledge the absence of direct empirical evidence for their
theory of inflation effects on crime, owing to the paucity of data on the price of stolen
goods over time. But even if the requisite evidence were available, previous studies of the
impact of inflation on crime have been based almost exclusively on nation states as units of
analysis.1 That leaves open the question of whether the criminogenic effects of inflation are
similar across subnational units such as cities. It would be an obvious error, akin to the
ecological fallacy, to assume that results from national-level analysis necessarily hold at
the local level. Most macro-level etiological theories of crime assume that substantively
meaningful variation exists among subnational units in crime and its covariates. Indeed,
subnational variation is the basis on which these theories are commonly tested (e.g.,
Peterson and Krivo 2010). Crime policy is also set and evaluated primarily at the state and
local level in the United States. In short, for both theoretical and policy purposes, it would
be a mistake to assume that results obtained from national-level data on crime and inflation
ipso facto are replicated at the local level.
1
An exception is a city-level study of crime and economic conditions by Baumer et al. (2013). In that study,
however, inflation is measured at the level of US Census regions.
123
J Quant Criminol (2019) 35:195–210
197
It is well known that both crime and commodity prices differ substantially across U. S.
cities. The robbery rate in Houston, for example, is nearly twice that in Seattle.2 A pair of
boy’s jeans costs $17.45 in Cincinnati and $22.07 in Chicago. An apartment in Cincinnati
rents for $856 on average and $1236 in Chicago.3 These large between-city differences in
crime and prices might prompt skepticism about whether the relationship between crime
and inflation observed at the national level also exists at the local level.
But between-city differences in crime and price levels need not correspond with withincity changes in crime rates and prices – inflation – the issue of primary concern in the
current study. In fact, it appears that city crime trends have converged on a common
national trend over the past several decades in the United States (McDowall and Loftin
2009). The same is true for city inflation trends (Cecchetti et al. 2002; Clark 1984).
Even if the national and local trends are similar, however, the relationship between
crime and inflation trends may differ across cities. For one thing, despite their general
similarities, some variation does exist in city crime and inflation trends.4 Cities also differ
in other respects, for example in income levels and police strength, that may affect crime
rates, inflation, or both. Cities may also differ in the significance and size of the effects of
inflation on crime rates. Finally, even if the relationship between crime and inflation
observed at the national level is found to hold at the city level, it is unlikely to exist in all
cities, and the deviant cases may contain important information about the dynamics of the
relationship unavailable from either national-level or pooled city analyses. For all of these
reasons, the national-level research leaves an important part of the crime and inflation story
untold. The study of individual cities can help to fill in some of the remaining research
gaps.
Current Study
This study analyzes the relationship between acquisitive crime rates and inflation in 17 U.
S. cities between 1960 and 2013. Based on the forgoing discussion, we hypothesize that
inflation has a significant and positive effect on city-level acquisitive crime rates. We
evaluate this hypothesis in a random coefficients panel model that yields estimates of
inflation effects on city acquisitive crime rates cities over the 54-year observation period.
The model produces individual regression coefficients for each of the 17 cities, which
indicate the range of variation across the cities in inflation effects on acquisitive crime. In
addition to inflation, our models include measures of income, socioeconomic disadvantage,
police per capita, and age composition. The specification of our city-level models is similar
to the national-level models in Rosenfeld and Levin (2016).
2
The robbery data are from the 2012 Uniform Crime Reports (https://www.bjs.gov/ucrdata/).
3
The price comparisons, which should be viewed as rough approximations, are from Bankrate (http://www.
bankrate.com/calculators/savings/moving-cost-of-living-calculator.aspx).
4
Cecchetti et al. (2002) report, for example, that although city inflation trends converge to a common mean,
convergence may take several years to occur. To our knowledge, a similar assessment has not been conducted for city crime rate trends.
123
198
J Quant Criminol (2019) 35:195–210
Data and Methods
The outcome in our analyses is the acquisitive crime rate per 100,000 population for 17 US
cities between 1960 and 2013. The measure of acquisitive crime consists of the combined
rates of robbery, burglary, larceny, and motor vehicle theft. The crime data are crimes
known to the police from the FBI’s Uniform Crime Reporting program (https://ucr.fbi.
gov).5
The independent variable of primary interest is the inflation rate for the metropolitan
areas of the 17 cities (see ‘‘Appendix’’). Inflation is defined as the year-over-year percentage change in the Consumer Price Index (CPI). The CPI is based on the price paid by
urban consumers for a representative market basket of goods and services, with 1982–1984
average city prices assigned the value of 100 (https://www.bls.gov/cpi). The current study
is based on 17 cities for which annual inflation data were available from the Bureau of
Labor Statistics between 1960 and 2013. The inflation data are for the metropolitan areas in
which the 17 cities are situated; all other data are for the 17 central cities.
In order to distinguish the effects of inflation on acquisitive crime rates from those of
other economic and social conditions, several additional socioeconomic and demographic
indicators are included in our analyses. Median family income, measured in nominal
dollars,6 is from the 1960–2010 decennial censuses and is interpolated between census
years. Data for 2011–2013 are from the American Community Survey (ACS) (https://
www.census.gov/programs-surveys/acs/). The annual civilian unemployment rate is from
the Current Population Survey (CPS) (http://www.census.gov/programs-surveys/cps.html).
The poverty rate, the unemployment rate, the percentage of families with children under
the age of 18 headed by females, and the percentage of the population that is black or
African-American are highly correlated (a = .74) and have been combined by factor
analysis into a single measure of socioeconomic disadvantage.7 The measures of poverty,
female-headed families, and racial composition are from the decennial censuses and ACS.
The analyses also include, from the CPS, the percentage of the city population between the
ages of 15 and 24. Finally, the number of police officers per 10,000 population, from the
FBI’s Uniform Crime Reports, is included in the analyses as a measure of deterrence.
These substantive indicators represent many of the major economic and social conditions shown in prior research to be associated with city crime rates (Land et al. 1990;
McCall et al. 2010). In addition, we include a measure of time in our models to capture the
effects of unmeasured time-varying conditions common to the 17 cities. The measure
consists of 11 five-year intervals from 1960–1964, 1965–1970 … 2010–2013 (the final
interval contains four years). The effects of inflation and the control variables could not be
reliably estimated with single year or shorter intervals in the panel model used in the
study.8
5
We thank Roland Chilton for sharing crime data for this analysis.
6
We measure income in nominal dollars because the inflation rate controls for price changes.
7
The factor scores are based on an orthogonal rotated solution. A single factor combining the four measures
(eigenvalue = 2.82) was retained that explains 99% of the variance. The rotated and unrotated solutions are
very similar.
8
With single year effects included, the model would not produce estimates of the effects of the explanatory
variables based on the pooled data. Time intervals shorter than five years yielded missing estimates in the
city-specific results. The likely reason is that these time intervals absorbed degrees of freedom required to
estimate the pooled and city-specific coefficients. The Chi square test of parameter constancy, for example,
is computed on 288 degrees of freedom (see Table 2).
123
J Quant Criminol (2019) 35:195–210
199
Analytic Strategy
We estimate the effect of inflation on acquisitive crime in a random coefficients linear
panel model that yields an effect for the sample as a whole and separate effects for each of
the 17 cities. We have selected this estimator precisely because it models panel variation in
the effects of independent variables on the outcome, a key research objective. The specific
model we use is from Swamy (1970), implemented in the xtrc routine in Stata 13.1 (see Poi
2003). The model assumes that the parameter estimates, both intercepts and slopes, are
drawn from a random distribution common to the panels. Given this assumption, the model
produces unbiased and efficient estimates of the outcome for the sample and for each of the
panels. A Chi square test of the significance of the difference in the regression coefficients
across the panels (termed ‘‘parameter constancy’’) is also provided.
Multiple explanatory variables can be included in the equation, and lags of the outcome
can be added to the righthand side to address serial correlation in the error term. Inspection
of the pooled data indicated serial correlation (rho = .970, DW = .186). We therefore
added the first lag of the acquisitive crime rate to the explanatory variables. The lagged
outcome eliminated first-order serial correlation in the error term (rho = .059,
DW = 1.984). The inclusion of the lagged outcome reduced the number of observations
from 918 (54 years 9 17 cities) to 901. Finally, with the lagged outcome on the righthand
side, the model estimates the residual change in the acquisitive crime rate – the effect of
the explanatory variables on the change in the acquisitive crime rate from the prior year to
the current year.9
Results
We first present descriptive statistics for each of the covariates, followed by a graphic
presentation of the acquisitive crime and inflation time series for the 17 cities. The
descriptive statistics include the pooled mean, and the standard deviation, minimum, and
maximum values for each variable between cities and within cities over time (see Table 1).
The mean acquisitive crime rate for the sample is approximately 7141 crimes per 100,000
city population. Somewhat more of the variation in acquisitive crime occurs within than
between the 17 cities. The mean inflation rate is just below 4%, and nearly all of the
variation in inflation occurs within the cities over time, from a minimum inflation rate of
approximately - 2.5% to a maximum of approximately 16.4%. The within-city variation
in median family income is also far larger than the variation between the cities. The sample
variation in the socioeconomic disadvantage factor is roughly similar between and within
cities. The same is true for city age composition. Finally, the number of police officers per
10,000 city residents varies more between than within cities.
The between- and within-city variation in the crime measure and the inflation rate are
graphically displayed in Figs. 1 and 2. Figure 1 shows the acquisitive crime time series
between 1960 and 2013. We observe notable differences across the cities in levels of
acquisitive crime over the entire period. Despite the differences in acquisitive crime levels,
9
Lagged dependent variables are controversial. Many analysts discourage their use because they can induce
downward bias in the coefficients on the explanatory variables (e.g., Allison 2015). In the present case,
however, including the lagged acquisitive crime rate in the model slightly increases the coefficient on
inflation. The coefficients on the other explanatory variables are non-significant regardless of whether the
lagged outcome is included (see Table 2).
123
200
J Quant Criminol (2019) 35:195–210
Table 1 Descriptive statistics
Mean
St. Dev.
Minimum
Maximum
Acqu. crime
Overall
2860.389
1388.057
18,332.500
Between
7140.605
1670.461
4452.553
10,785.350
Within
2356.409
782.004
15,694.500
16.500
Inflation
2.926
- 2.600
Between
Overall
3.958
.114
3.817
4.137
Within
2.924
- 2.459
16.376
Income
Overall
17,685
5029
95,725
Between
27,807
5381
20,494
39,878
Within
16,896
5129
83,653
Soc. disad.
N (observations) = 918; n
(cities) = 17; T (years) = 54
Overall
Acqu. crime = acquisitive crimes
per 100,000 population;
Inflation = yearly percentage
change in consumer prices;
Income = median family income
in dollars, Soc. disad. = factor
score for unemployment, percent
black, percent poor families,
percent female-headed families
with children; Age 15–
24 = percent age 15–24;
Police = sworn officers per
10,000 population
.945
- 1.647
3.047
Between
.000
.792
- 1.332
1.447
Within
.549
- 1.985
1.600
Age 15–24
Overall
2.383
11.499
24.286
Between
16.330
1.611
13.805
20.605
Within
1.799
8.500
20.316
Police
Overall
8.686
6.807
52.635
Between
30.689
7.537
18.374
41.834
Within
4.683
13.731
44.217
20000
18000
16000
14000
12000
10000
8000
6000
4000
2000
0
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
2010
Fig. 1 National and city acquisitive crime rates per 100,000 population, 1960–2013
however, the cities exhibit similar change in acquisitive crime over time. Acquisitive crime
rates in the 17 cities generally rose from the 1960s to a peak in 1980, fell for a few years
and then increased to a second peak in the early 1990s, and fell thereafter, a temporal
123
J Quant Criminol (2019) 35:195–210
201
20
15
10
5
0
-5
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
2010
Fig. 2 National and city inflation rates, 1960–2013 (%)
pattern that corresponds with annual change in the national acquisitive crime rate during
the 54-year period. Acquisitive crime rates are higher on average in the 17 cities than the
nation as a whole. The average city-level and national trends are nearly identical, however
(r = .982). As shown in prior research, city crime trends appear to be offshoots of a
common national trend (McDowall and Loftin 2009).
There is little question that the inflation rates of the 17 cities follow a national trend. As
shown in Fig. 2, the metro area inflation rates bunch tightly around the national inflation
rate. The correlation between the average inflation rate for the 17 cities and the national
inflation rate is .970. Inflation rose from the mid-1960s to peaks in the mid-1970s and in
1980. After cresting again in 1990 at just over 5%, inflation rates hovered between 2 and
4% until the Great Recession; in 2009 the CPI fell for the first time in nearly 60 years. The
national acquisitive crime rate was associated with the ebb and flow of inflation during this
period, rising with increases and falling with decreases in inflation (Nunley et al. 2016;
Rosenfeld and Levin 2016). We now consider whether acquisitive crime and inflation rates
are associated at the city level and whether this relationship withstands controls for other
economic and social conditions.
Panel Model Results
Table 2 presents the estimates for the residual change in city acquisitive crime rates from
the random coefficients panel regression. In Model 1 of the table, the lagged acquisitive
crime rate and the inflation rate have significant effects on change in the acquisitive crime
rate. None of the coefficients on the other variables in the model is statistically significant
at the 5% level. Table 2 displays both the unstandardized regression coefficients (b) and
the standardized coefficients (b). In Model 1, a one unit increase in inflation produces
approximately 96 additional acquisitive crimes per 100,000 population. A one standard
deviation increase in inflation is associated with a .099 standard deviation increase in the
123
202
J Quant Criminol (2019) 35:195–210
Table 2 Effect of inflation and covariates on residual change in acquisitive crime rate (N = 901)
Model 1
Acqu. crime lag
Model 2
b
b
b
.604*
.605
.623*
(.056)
Inflation
96.348*
.099
- 10.254
- 701.826
- .063
86.496
- .245
19.198
.030
–
- 772.452
28.110
.007
11.089
- .270
.010
.004
(37.435)
–
- 3.571*
–
(1.754)
Wald
v2(16) = 625.140*
v2(17) = 886.600*
Parameter constancy
v2(288)
v2(304) = 566.700*
= 565.370*
- .166
(163.218)
(36.625)
Inflation 9 income (000)
- 12.663
(1085.204)
(149.142)
Police
.074
(21.782)
(936.087)
Age 15–24
171.400*
(45.576)
(21.711)
Soc. disad.
.623
(.060)
(22.810)
Income (000)
b
- .065
Random coefficient estimates. Standard errors in parentheses. Income in thousands. Period effects not
shown. See Table 1 for variable definitions
*p \ .05
contemporaneous acquisitive crime rate (approximately 283 acquisitive crimes per 100,000
population).10
These appear to be small effects, but because the estimates are conditioned on the
lagged acquisitive crime rate they reflect the predicted change in acquisitive crime from
one year to the next. The substantive import of inflation was especially pronounced during
the first half of the observation period, when acquisitive crime was increasing rapidly. The
acquisitive crime rate rose by .086 units of standard deviation per year between 1960 and
1985. All of this observed change in acquisitive crime was accounted for by the .099
standardized effect of inflation. (The combined effect of the other variables produced a
small decrease in acquisitive crime.)
Evaluating the effect of inflation on acquisitive crime should take account of the level of
income over time and across cities. If income keeps pace with or exceeds inflation, we
should expect the criminogenic effect of inflation to diminish. For that reason, we have
included income in our regression model. But income should not only have an additive
effect, it should condition the effect of inflation on acquisitive crime. Based on the logic
model underlying the analysis, the effect of inflation on acquisitive crime should be
stronger in less affluent, more disadvantaged cities where consumers may be especially
likely to search for cheap stolen goods when prices rise. To evaluate this hypothesis, we
10
With the lagged outcome omitted from the model, the unstandardized and standardized coefficients on
inflation are 78.0 and .080, respectively (p \ .05). The other explanatory variables remain non-significant
(see fn9).
123
J Quant Criminol (2019) 35:195–210
203
included interaction terms in the model for inflation 9 income and inflation 9 socioeconomic disadvantage.
We found no evidence that the measure of socioeconomic disadvantage conditions the
effect of inflation on the change in acquisitive crime rates. The interaction term including
socioeconomic disadvantage is non-significant in estimations with and without the term
including income (results not shown).11 By contrast, the inflation 9 income interaction
term is statistically significant and, as hypothesized, negative (b = - 3.571, p \ .05).
Inflation has a smaller effect on acquisitive crime in cities where the median family income
is comparatively high than in less affluent cities.
We inspected the random coefficients models for multicollinearity and nonlinearity. The
variance inflation factors for the regressors show no evidence of multicollinearity.12 The
model assumes that the outcome is a linear function of the predictors. We checked this
assumption for the relationship between the change in the city crime rates and inflation by
estimating a quadratic specification of the panel models incorporating the square of
inflation. In results not shown, we found no evidence of a nonlinear relationship between
the acquisitive crime rate and inflation.
City-Specific Results
We have characterized the panel regression results as indicating the effect of inflation on
the change in the city acquisitive crime rate. In fact, however, the results pertain to the
effects of inflation on the change in acquisitive crime in 17 individual cities. On average,
the effect of inflation is positive and significant, but the variability in those effects could be
quite large, and in some cities inflation may have no significant effect on the change in
acquisitive crime. It is therefore informative to inspect the effects of inflation on the change
in acquisitive crime rates in each of the cities.
The random coefficients model returns a separate slope coefficient for each of the 17
cities in the sample. The Chi square test of parameter constancy indicates that the slope
coefficients vary significantly across the cities (see Table 2). To determine whether the
variance in the coefficients is attributable to inflation and not simply the other explanatory
variables or the period effects, we re-estimated the model retaining only the lagged
acquisitive crime rate and inflation on the right-hand side. The test for parameter constancy
remained significant (v2 = 93.6, p = .0001).13
Figure 3 graphically summarizes the variance in the coefficients. To compare the effect
sizes across the cities, the standardized coefficients (b) are shown. The standardized
inflation effects vary from zero (Portland) to .184 (New York). In addition to New York,
four cities cluster at the top of the display, with coefficients C .168 (Kansas City, Boston,
St. Louis, Atlanta). In the remainder of the cities where the inflation effects are statistically
significant, the standardized effects range from .125 to .062. Finally, in five of the 17 cities
the effect of inflation on the change in acquisitive crime is not significant (Houston, San
Francisco, Los Angeles, Chicago, and Portland).
11
All results not shown are available from the first author by request.
12
In Model 1, the mean VIF = 1.43 and the max VIF = 1.74. The comparable values in Model 2 are
increased by the interaction term, but remain within an acceptable range (mean VIF = 3.28, max
VIF = 6.62).
13
The variance in the estimated effects is not due to the lagged acquisitive crime rate. With only the lagged
crime rate in the model, the parameter constancy v2 = 30.7, p = .531.
123
204
J Quant Criminol (2019) 35:195–210
0.2
New York Boston
0.18
Kansas City
p < .05
p > .05
Atlanta
St. Louis
0.16
0.14
Pittsburgh
Philadelphia
0.12
0.1
Pooled mean = .099
Cleveland
Milwaukee
0.08
Detroit
Seattle
0.06
0.04
0.02
San Francisco
Cincinnati
Houston
Los Angeles
Chicago
Portland
0
Fig. 3 Inflation effects on city acquisitive crime rates in units of standard deviation
The variance in inflation effects across the cities is not simply a consequence of the
presence in the sample of the five cities where inflation has no significant effect on
acquisitive crime. With those cases omitted, the test for parameter constancy remains
significant in models with and without the other explanatory variables and period effects
(results not shown).
The variance across the cities in inflation effects on acquisitive crime is both statistically
significant and, in certain cases, of substantive significance as well. Consider the difference
in the effects of inflation on the acquisitive crime rate in Cincinnati and New York, the
cities with the smallest and largest inflation effects that are statistically significant. Figure 4 displays the predicted effect of inflation on the acquisitive crime rate in Cincinnati
and New York, with the other explanatory variables set to their mean values. At the sample
minimum and maximum inflation rates (- 2 and 16%), the predicted acquisitive crime rate
in New York varies from 4228 to 7335 per 100,000 population, a difference of 73.5%. The
comparable difference in Cincinnati is from 4789 to 5773 acquisitive crimes per 100,000
population, a difference of just 16.8%.
The greater impact of inflation on acquisitive crime in New York than in Cincinnati is
evident when the impact is evaluated at the extreme bounds of inflation during the 54-year
observation period. But a marked difference between the two cities in the impact of
inflation also exists when evaluated within a more modest range. Figure 4 displays the
predicted inflation rates in Cincinnati and New York between minus one and one standard
deviation from the sample mean, or inflation rates of approximately one and seven percent.
Within this range, New York’s predicted acquisitive crime rate varies from 4789 to 5773
per 100,000 population, a difference of 20.6%. By contrast, the acquisitive crime rate in
Cincinnati varies from 6373 to 6545 per 100,000 population, a difference of just 2.70%.
These results pertain to the two cities in the sample with the greatest variance in
significant inflation effects on acquisitive crime; other comparisons would have yielded
smaller differences in impact. Nonetheless, there is considerable variability in the impact
of inflation on the change in acquisitive crime, even in those cities where the inflation
effects are statistically significant. Differences of the magnitude shown in Fig. 4 probably
would attract the notice of crime control analysts and policymakers and, were they to
develop an interest in crime, financial economists. We emphasize that these results serve
123
J Quant Criminol (2019) 35:195–210
205
8000
Mean
Acquisitive Crime Rate
7500
-1sd
7000
7335
7149
6500
6000
Cincinnati
+1sd
6545
6373
6122
New York
5773
5500
5000
4500
4789
4228
4000
3500
3000
-2%
0%
2%
4%
6%
8%
10%
12%
14%
16%
Inflation
Fig. 4 Predicted acquisitive crime rate in Cincinnati and New York at intervals of inflation
only to illustrate the differing impacts of inflation on the change in acquisitive crime. But
we believe such individual case comparisons, where feasible, add informative context to
the results of pooled analyses that distill the effects of explanatory variables into a single
point estimate.
Where Inflation Has No Effect
Inflation has no statistically significant effect on acquisitive crime in Houston, San Francisco, Los Angeles, Chicago, and Portland. These non-significant effects are not simply a
function of large standard errors of the estimates; the estimates themselves are comparatively small, as shown in Fig. 3. It is not immediately obvious what demographic or
socioeconomic characteristics the five cities share in common that might explain the null
effects of inflation. We compared the cities with null inflation effects to those where the
effects on acquisitive crime are significant with respect to each of the explanatory variables. The results are presented in Table 3.
Table 3 Comparison of cities
where inflation effects on
acquisitive crime are significant
and non-significant (N = 918)
Inflation
b
Odds ratio
.066
1.068
(.035)
Income (000)
.013*
1.013
(.005)
Soc. disad.
- .957*
.384
(.140)
Age 15–24
Logistic regression estimates.
Cities with ns inflation
effects = 1, else = 0. Standard
errors in parentheses. Income in
thousands. See Table 1 for
variable definitions
*p \ .05
- .217*
.805
(.045)
Police
- .040*
.961
(.012)
Likelihood ratio
v2(5) = 229.740*
Pseudo
R2 = .207
123
206
J Quant Criminol (2019) 35:195–210
Table 3 displays logistic regression coefficients and odds ratios for the difference in
inflation, income, socioeconomic disadvantage, age composition, and police strength
between cities with and without significant inflation effects on acquisitive crime (assigned
the values of zero and one, respectively). The cities differ significantly on each of the
explanatory variables, except for the rate of inflation (p = .058). Cities where inflation is
not significantly associated with acquisitive crime tend to be more affluent and less disadvantaged. They also have smaller youth cohorts and somewhat fewer police per capita.
As noted above, less disadvantaged cities with higher median incomes are places where we
might expect inflation effects on acquisitive crime to be muted. Even among the 12 cities
with significant inflation effects on acquisitive crime, we found that the effects are weaker
in locales with higher median incomes (results not shown).
This assessment represents only a small step toward specifying the conditions under
which the effects of inflation on acquisitive crime may vary in magnitude and significance.
But the city-specific analyses do disclose the increments to knowledge to be gained from
moving beyond standard panel analyses that distill the causes and consequences of social
phenomena into a single pooled estimate.
Discussion
Prior studies of the influence of inflation on US crime rates have been conducted at the
national level. Those studies have consistently found that inflation has a statistically significant and positive effect on crimes committed for monetary gain. National acquisitive
crime rates rise with increases in inflation over time and fall with decreases in inflation.
Prior research, however, leaves unanswered whether a similar relationship between
inflation and acquisitive crime rates exists at the city level, arguably a more relevant unit of
analysis for both criminological research and crime control policy. The current study
replicates at the city level prior national-level research on the relationship between crimes
committed for monetary gain and inflation in the United States.
Inflation trends over the past half century are remarkably similar across our sample of
17 cities for which the requisite data on both inflation and crime were available. The citylevel trends also correspond closely to the national inflation trend. Nonetheless, cities may
differ with respect to the effect of inflation on local crime rates. We therefore analyzed the
relationship between inflation and acquisitive crime with random coefficient models that
yield separate regression estimates for each of the 17 cities.
In an analysis that pools the city-level data, we found a significant association between
inflation and the residual change in city acquisitive crime rates between 1960 and 2013 that
withstands controls for other influences on the change in acquisitive crime. The pooled
results show that inflation increases acquisitive crime rates, particularly in less affluent
cities. These results support the hypothesis that price increases prompt consumers to search
for cheaper goods. Stolen goods cost less than those available from legitimate sources and
should be especially attractive to consumers in areas where average incomes fail to keep
pace with inflation.
These results are based on panel regressions that pool into a single value the effects of
inflation on the change in acquisitive crime in the 17 cities. The pooled regression results
are generally confirmed by separate estimates for each of the 17 cities. We also discovered,
however, considerable variability across the cities in the significance and size of the
inflation effects. In five of the cities, we found no significant effect of inflation on
123
J Quant Criminol (2019) 35:195–210
207
acquisitive crime. The null effect of inflation and the modest significant effect in many
cities may result, we have suggested, from higher average incomes that reduce the demand
for cheap stolen goods. In addition, both supply and demand may vary according to other
characteristics of underground markets, including police effectiveness in stemming
acquisitive crimes and the interpenetration of local drug markets and the market for stolen
goods. The former would increase the risks associated with trafficking in stolen goods,
while the latter could increase demand apart from changes in consumer prices. These and
other sources of local variation in the dynamics of underground markets deserve attention
in future research.
The city-specific estimates reveal nontrivial variation in some of the inflation effects that are
statistically significant. These are not ‘‘deviant’’ cases as usually understood. They are part of
the normal covariation of crime and inflation across cities in the United States. The information
they contain about the spatial dynamics of the relationship between crime and inflation would
be lost if the relationship were examined merely by means of standard pooled cross-section
time-series methods. An important, though by no means original, takeaway from this study is to
learn how a model fits each panel unit whenever feasible in cross-section time-series analyses.
That way, the investigator also learns how each unit influences the pooled results (see DeFina
and Arvanites 2002).
Several caveats about our analyses and results are in order. The 17 cities in the study are not
necessarily representative of all large US cities. Caution should be exercised, therefore, in
generalizing the results beyond the sample. Further, the variability in inflation effects found in
the city-specific estimates should be viewed as merely illustrative of city differences in the
significance, size, and impact of inflation on acquisitive crime. Finally, although we assume
that the causal direction in the relationship between crime and inflation is from inflation to
crime, some feedback is certainly possible such as, for example, when retailers raise prices in
response to shoplifting.14 National level research, however, has shown that the causal
direction of the relationship between crime and inflation is from inflation to crime (Rosenfeld
and Levin 2016; Tang and Lean 2007). Within the confines of the current sample we conclude
that inflation has a significant influence on the change over time in acquisitive crime rates in
many, perhaps most, U. S. cities.
In addition to extending the current analysis to a larger number of cities, which will
depend on whether the Bureau of Labor Statistics increases the number of metropolitan
areas for which it tracks inflation rates over time, future research should investigate the
relationship between inflation and violent crime. Prior research has disclosed inflation
effects on the homicide rates of several European nations and the United States (Rosenfeld
2014). There are good criminological reasons for assuming that inflation has both direct
and indirect effects on violent crime. Studies have shown that violent crime rates are
connected to changes in institutional legitimacy and other sources of social stress that have
accompanied the rise and fall of inflation over time (Fischer 1996; LaFree 1998). In
addition, inflation may influence violent crime through its effect on acquisitive crime.
Disputes that arise among buyers and sellers in illegal markets cannot be settled by
recourse to the police, courts, or other formal regulatory agencies (Venkatesh 2006). In
such ‘‘stateless’’ social locations, violence becomes a potent enforcement mechanism
(Black 1983). Burglars and thieves are also vulnerable targets for street robbers. Even if
only a small fraction of acquisitive crimes end in violence, expansion in the market for
stolen goods driven by increases in acquisitive crime can produce appreciable increases in
violent crime (see Rosenfeld 2009).
14
We thank Eric Baumer for this insight.
123
208
J Quant Criminol (2019) 35:195–210
U. S. cities have experienced a steep drop in street crime rates since the latter part of the
twentieth century (Blumstein and Wallman 2006; Zimring 2007). Inflation rates have been
running at or near historic lows for several years, at times well below the US Federal
Reserve Board’s goal of 2%.15 The crime drop has been particularly pronounced in New
York, where homicide rates have returned to the levels of the 1950s (Southall 2017). Many
analysts attribute New York’s exceptional crime reductions to effective policing (e.g.,
Zimring 2007). Without discounting the role of the police, inflation has also likely played a
part in New York’s crime decline. Recall that inflation has a greater effect on acquisitive
crime in New York than in any other city in our sample, although a few other cities come
close (see Fig. 3). Falling rates of inflation merit a prominent place among the factors
responsible for the historic drop in U. S. crime rates (Rosenfeld and Levin 2016).
Fischer (1996) has charted four ‘‘great waves’’ of price increases since the Middle Ages,
the last of which occurred throughout the developed world during the final quarter of the
twentieth century. Each wave of inflation was followed by an extended period of price
stability and relatively low crime rates. Crime forecasts are always hazardous and prone to
upset by exogenous shocks.16 But low inflation rates, which appear to be at least partly
responsible for the storied crime decline in many American cities, should continue to curb
crime increases in the foreseeable future.
The change in consumer prices over time is driven by national and international forces –
including exogenous shocks in the supply of energy and long term expansion and contraction in the money supply (Nunley et al. 2016)—over which individual cities have little
control. Subnational settings are appropriate criminal justice policy units; monetary policy
is a federal responsibility. But what are the broad policy objectives that should guide the
quest for the optimal inflation rate? Two economists have written that ‘‘policymakers
should aim for an inflation rate that maximizes the economic well-being of the public’’
(Billi and Kahn 2007:6). We agree but would expand public wellbeing to encompass the
considerable costs of crime to victims and society (McCollister et al. 2010). It is time for
the unheralded consequences of low inflation for crime to become one of the guideposts for
evaluating monetary policy in the United States.
Appendix
Sample of 17 Metropolitan Areas and Central Cities
Atlanta, GA
Boston-Brockton-Nashua, MA-NH-ME-CT
Chicago-Gary-Kenosha, IL-IN-WI
Cincinnati-Hamilton, OH-KY-IN
Cleveland-Akron, OH
Detroit-Ann Arbor-Flint, MI
Houston–Galveston-Brazoria, TX
Kansas City, MO-KS
15
See
https://www.federalreserve.gov/publications/annual-report/2015-monetary-policy-and-economicdevelopments.htm#xsubsection-14-c4f66975.
16
For example, some economists expect inflation rates to increase if ‘‘trade wars’’ break out under the
Trump Presidency (Zumbrun 2016). See, also, Rosenfeld (2016; Rosenfeld et al. 2017) on unexpected
homicide increases in American cities during 2015 and 2016.
123
J Quant Criminol (2019) 35:195–210
209
Los Angeles-Riverside-Orange County, CA
Milwaukee-Racine, WI
New York-Northern New Jersey-Long Island, NY-NJ-CT-PA
Philadelphia-Wilmington-Atlantic City, PA-NJ-DE-MD
Pittsburgh, PA
Portland-Salem, OR-WA
San Francisco-Oakland-San Jose, CA
Seattle-Tacoma-Bremerton, WA
St. Louis, MO-IL
References
Allen RC (1996) Socioeconomic conditions and property crime. Am J Econ Sociol 55:293–308
Allison P (2015) Don’t put lagged dependent variables in mixed models. https://statisticalhorizons.com/
lagged-dependent-variables. Accessed 21 Feb 2018
Baumer EP, Rosenfeld R, Wolff KT (2013) Are the criminogenic consequences of economic downturns
conditional? In: Rosenfeld R, Edberg M, Fang X, Florence CS (eds) Economics and youth violence:
crime, disadvantage, and community. New York University Press, New York
Billi RM, Kahn GA (2007) What is the optimal inflation rate? https://www.kansascityfed.org/PUBLICAT/
ECONREV/PDF/2q08billi_kahn.pdf. Accessed 21 Feb 2018
Black D (1983) Crime as social control. Am Sociol Rev 48:34–45
Blumstein A, Wallman J (eds) (2006) The crime drop in America, Revised edn. Cambridge University Press,
New York
Cecchetti SG, Mark NC, Sonora RJ (2002) Price index convergence among United States Cities. Int Econ
Rev 43:1081–1099
Clark GL (1984) Does inflation vary between cities? Environ Plan 16:513–527
Cohen LE, Felson M (1979) Social change and crime rate trends: a routine activity approach. Am Sociol
Rev 44:588–607
DeFina RH, Arvanites TM (2002) The weak effect of imprisonment on crime: 1971–1998. Soc Sci Q
83:635–653
Fischer DH (1996) The great wave: price revolutions and the rhythm of history. Oxford University Press,
New York
LaFree G (1998) Losing legitimacy: street crime and the decline of social institutions in America. Westview,
Boulder
LaFree G, Drass KA (1996) The effect of changes in intraracial income inequality and educational
attainment on changes in arrest rates for African Americans and whites, 1957–1990. Am Sociol Rev
61:614–634
Land KC, McCall PL, Cohen LE (1990) Structural covariates of homicide rates: are there any invariances
across time and social space? Am J Sociol 95:922–963
McCall PL, Land KC, Parker KF (2010) An empirical assessment of what we know about structural
covariates of homicide rates: a return to a classic 20 years later. Homicide Stud 14:219–243
McCollister KE, French MT, Fang H (2010) The cost of crime to society: new crime-specific estimates for
policy and program evaluation. Drug Alcohol Depend 108:98–109
McDowall D, Loftin C (2009) Do city crime rates follow a national trend? The influence of nationwide
conditions on local crime patterns. J Quant Criminol 25:307–324
Nunley JM, Stern ML, Seals RA, Zietz J (2016) The impact of inflation on property crime. Contemp Econ
Policy 34:483–499
Peterson RD, Krivo LJ (2010) Divergent social worlds: neighborhood crime and the racial-spatial divide.
Russell Sage Foundation, New York
Poi BP (2003) Swamy’s random-coefficients model. Stata J 3:302–308
123
210
J Quant Criminol (2019) 35:195–210
Ralston RW (1999) Economy and race: interactive determinants of property crime in the United States,
1958–1995. Am J Econ Sociol 58:405–434
Rosenfeld R (2009) Crime is the problem: homicide, acquisitive crime, and economic conditions. J Quant
Criminol 25:287–306
Rosenfeld R (2014) Crime and inflation in cross-national perspective. Crime Justice 43:341–366
Rosenfeld R (2016) Documenting and Explaining the 2015 Homicide Rise: Research Directions. NCJ
249895. National Institute of Justice, Washington, DC
Rosenfeld R, Levin A (2016) Acquisitive crime and inflation in the United States: 1960–2012. J Quant
Criminol 32:427–447
Rosenfeld R, Gaston S, Spivak H, Irazola S (2017) Assessing and Responding to the Recent Homicide Rise
in the United States. NCJ 251067. National Institute of Justice, Washington, DC
Southall A (2017) Crime in New York City plunges to a level not seen since the 1950s. New York Times
(December
27).
https://www.nytimes.com/2017/12/27/nyregion/new-york-city-crime-2017.html.
Accessed 21 Feb 2018
Swamy PAVB (1970) Efficient inference in a random coefficient regression model. Econometrica
38:311–323
Tang CF, Lean HH (2007) Will inflation increase crime rate? New evidence from bounds and modified
Wald tests. Global Crime 8:311–323
Venkatesh SA (2006) Off the books: the underground economy of the urban poor. Harvard University Press,
Cambridge
Zimring FE (2007) The Great American crime decline. Oxford University Press, New York
Zumbrun J (2016) GDP, inflation and interest rates forecast to rise under Trump Presidency. Wall Street
Journal (November 13). http://www.wsj.com/articles/gdp-inflation-and-interest-rates-forecast-to-riseunder-trump-presidency-1479054608. Accessed 21 Feb 2018
123
Journal of Quantitative Criminology is a copyright of Springer, 2019. All Rights Reserved.
AUSTRALIAN INSTITUTE
OF CRIMINOLOGY
trends
No.23
&
A Comparison of Crime
in Australia and Other
Countries
issues
in crime and criminal justice
J. Walker, P. R. Wilson, D. Chappell
and D. Weatherburn
Since Australia was first used by the British as a penal colony, there has
always been considerable interest in comparisons between crime and
punishment in this country and elsewhere. From the early days of
colonisation the principle comparisons made were between the convict
society, which was viewed by some as a kind of social laboratory, and the
English society which had created it. The purpose of such comparisons was
then, as now, to judge the relative merits or demerits of the two societies
and their criminal justice systems.
In more recent years it has become fashionable to broaden the scope of
comparison to include other nations. However, until now, such comparison
has proved surprisingly difficult, due largely to the variety and complexity
of the criminal justice systems themselves, and differences in the way they
record and define their activities.
With the advent of crime surveys, utilising standard sampling
techniques and questions to obtain information about victimisation
experiences, many of these dilemmas have been overcome as this unique
International Survey of Crime Victims vividly demonstrates. The Survey
findings reveal that Australia has crime rates which are much more
comparable to those of North America than Europe. In a number of cases
these rates are the highest recorded among the 14 jurisdictions included in
the survey - a result which must surely give cause for concern to all
Australians and reinforce the need for the establishment of a national crime
prevention strategy.
Duncan Chappell
Director
June 1990
ISSN 0817-8542
ISBN 0 642 15435 X
An Overview
• How does Australia compare with other countries in terms of its rates of
crime and offending? Are we more, or less, likely to become a victim of a
crime than people who live in other countries?
• How do Australians feel about their police? Are police in Australia more
respected or less respected than their counterparts overseas? Do
Australians get the services they need from their police?
• Is there anything about the Australian way of life which makes us
particularly vulnerable to crime, or does our
Australian Institute
of Criminology
GPO Box 2944
Canberra ACT 2601
Australia
http://www.aic.gov.au
Australian Institute of Criminology
supposedly free and easy lifestyle
reduce the impact of crime?
• Do Australians take appropriate
preventative measures to avoid or
deter criminals?
• How do our opinions on sentencing
compare with those of people
overseas?
During 1989, over two thousand
Australians took part in a survey
designed to answer these and other
questions about crime. At the same
time, similar sized samples of the
population of ten other countries and
the three major administrative units of
the United Kingdom were being asked
identical questions. Questions asked
in the survey included details of
incidents occurring to respondents
over the previous five years, and their
responses to them, including degree of
satisfaction with police efforts.
Surveys of people's experience of
crime offer an alternative view of
crime levels to that based on police
statistics. It is extremely difficult to
compare police figures from one
country to another as different police
jurisdictions often use diverse methods
of recording and defining particular
offences. Though ‘victim surveys', as
they are called, are far from foolproof,
they at least allow researchers to
estimate the risk of crime for citizens
in particular countries as well as
citizens’views regarding crime and
sentencing matters.
So how did Australia emerge from
these comparisons? Unhappily, the
report concludes that Australian rates
of crime are in some respects very
high compared with most other
comparable countries. Australia is
ranked third highest of the fourteen
countries in terms of overall
victimisation, behind the other two
non-European countries, the USA and
Canada. In assaults involving force, in
the less serious types of sexual
incidents, in burglary and in motor
vehicle thefts, Australia ranked
highest of all the countries surveyed.
Only in the categories of motorcycle
or bicycle theft, and pickpocketing did
we fail to finish in the top half of the
list.
It is perhaps surprising, then, to
find also that Australians appear to be
very much happier with the efforts of
their police forces than residents in
most other countries in the survey.
Almost eighty percent of victims of
crime expressed satisfaction with the
police response to their problem, and
around three quarters of all
respondents thought the police were
doing a reasonable job controlling
crime in their neighbourhood.
This apparent paradox is not easily
explained. There is evidence that some
aspects of the Australian lifestyle may
actually be responsible for our high
crime rates. For example, we may
make ourselves easy targets for
burglars with our detached houses in
big, impersonal cities. Both these
features are found to be associated
with high burglary rates. Likewise, it
2
can be thought that we make life easy
for car thieves by driving to work leaving the car parked all day
unattended. Ironically, however, those
countries which choose public
transport for their journeys to work
have much higher rates of pickpocketing than we do. Different
lifestyles give different opportunities
to commit crime. Other features of
Australian society appear to be
associated with high risks of crime,
but the in-depth analysis of the data
needed to confirm these suspicions has
not yet been completed.
The report concludes, for example,
that big city dwellers are over fifty
percent more likely to be victimised
than those who live in towns of under
10,000 people. Australia is, of course,
Australian Institute of Criminology
one of the most highly urbanised
nations of all, and this single feature
appears to explain most of the
difference between the risk of crime in
Australia and elsewhere. Other factors
likely to adversely affect our risk of
victimisation include our demography,
our high living standards, and our high
labour participation rates, especially
for women. Most intriguing of all,
however, is that the report shows that
the risk of all major categories of
crime clearly decreases the further a
country is from the equator - possibly
because a colder climate imposes an
informal curfew on both offenders and
potential victims alike, reducing the
frequency with which people leave the
comforts of home and reducing the
opportunities for a range of criminal
activities. In all these respects,
Australia would probably stand out as
a 'high risk' country.
Despite the risks, Australians do
not appear to be unduly fearful of
violent crime, by international
standards. Their most common fear
was of being burgled, which, as the
responses showed, was supported by
the relatively high incidence of these
crimes in Australia.
There are some clear lessons in
these comparisons, in the need for
continued and improved crime
prevention activities in Australia. But
the figures should not necessarily be
cause for alarm at our position vis-a-
3
Australian Institute of Criminology
vis other countries - rather they should
be used to identify the special needs of
a country such as ours, in terms of
crime prevention, policing and the
criminal justice system in general.
The Survey
This is the first time an attempt has
been made to conduct parallel surveys
of crime in several countries. It fills a
long-felt need for genuinely
comparable estimates of crime and
patterns of victimisation in different
countries. Researchers have
principally wanted to test theories
about the social causes of crime by
means of cross-national comparisons.
Policy makers have principally
wanted to understand better their
national crime problems by putting
these in an international perspective.
Comparisons based on numbers of
offences recorded by police have
presented severe difficulties, not least
of which are caused by differences in
offence classifications, counting rules,
and the frequencies with which
offences are actually brought to the
notice of police.
Crime victim surveys, of one form
or another, have been developed in
many countries in recent years, as a
means of estimating the 'true' level of
crime, the extent and nature of
offences which are never reported to
police, and the distribution of risks
across different sections of the
community. This particular survey
was first proposed to a meeting of
local and regional authorities of the
Council of Europe in 1987,1 but the
idea of extending it to other, non-
European, countries was readily
accepted by the coordinators of the
survey. Eventually, Australia, through
the support of the Australian Institute
of Criminology and the New South
Wales Bureau of Crime Statistics and
Research, joined Belgium, Canada,
England and Wales, West Germany,
Finland, France, the Netherlands,
Northern Ireland, Norway, Scotland,
Spain, Switzerland and the USA in the
venture. (Local surveys were also
4
conducted in regions of Poland, Japan
and Indonesia, but in each case the
sample was not drawn from the whole
population, and would therefore be
unrepresentative.)
Each survey was based on a
sample of the population and, for cost
reasons, was conducted by telephone.
Both of these features may clearly
limit the degree to which the survey
responses adequately represent the
actual level of crime in each country.
Australian Institute of Criminology
However, great care was taken to
ensure that, as far as possible, fair
comparisons between nations could be
made from the results.
The offences about which people
were asked included :
• thefts of cars, motor cycles and
pushbikes
• thefts from, and damage to, cars
• burglary and attempted burglary
• robbery
• thefts of personal property
• sexual assaults and offensive
behaviour
• other assaults and threats
Respondents who had been
victimised were asked for details of
when, where and how the incident
happened, its material consequences,
whether the police were involved (and
if not, why not), and satisfaction with
the police response and any victim
assistance given. Questions were also
asked relating to fear of crime, general
satisfaction with local policing, crime
prevention behaviour, and attitudes to
sentencing
The Results in More Detail
Crime in Australia, Compared to
Other Countries
Table 1 shows estimates of the
percentages of the sampled population
aged 16 or over in each country who
had been victimised at least once
during 1988.
Australia is ranked third highest of
the fourteen countries in terms of
overall victimisation, behind the other
two non-European countries, the USA
and Canada. In assaults involving
force, sexual incidents, burglary and
motor vehicle thefts, Australia ranked
highest of all countries surveyed. Only
in such categories as motorcycle or
bicycle theft, and pickpocketing did
we fail to finish in the top half of the
list.
The risk of having a car stolen in
1988 was equal highest with France
(2.3 per cent of respondents), even
after taking Australia's high rate of
car-ownership into account. During
1988 6.9 per cent of Australian
respondents also said they had had
things stolen from their cars during
1988 (e.g. luggage, radios, car mirrors
etc.), and 8.7 per cent had their car
damaged in some way.
Burglary affected 4.4 per cent of
Australian respondents during 1988
compared with 3.8 per cent in the
USA and 3.0 per cent in Canada.
European rates of burglary were only
around half the Australian figure.
These figures appear to be related to
the percentage of detached or semidetached houses, which is very high
(85 per cent) in Australia.
Australia on 5 per cent ranked
equal second with Spain for its
frequency of personal thefts, with
Canada (5.4 per cent) in the
unenviable top-spot. These crimes
consisted of thefts of personal items
such as purses, jewellery or shopping
while at work, school or in some
public place. Interestingly, however,
Australians were at very low risk of
pickpocketing.
Australia also had very high rates
of sexual incidents, including sexual
assaults, even when other factors,
such as the high female labour
participation rate and our high
propensity to go out in the evenings,
are taken into account. The
researchers suggest that an additional
factor may be differences in
willingness to talk about such
incidents. Issues such as sexual
harassment have been the subject of
heated debate in some countries in
recent years, but are still taboo
subjects in others.
It is relevant to note that 80.1 per
cent of all sexual incidents were for
‘offensive behaviour' (grabbing,
touching) while 7.8 per cent were for
rape or attempted rape and 11.3 per
cent for indecent assault. Of all
sexual incidents, 92.4 per cent were
not reported to the police. Half the
respondents who failed to report any
sexual incident to the authorities said
that they did not do so because the
incident was not serious enough or
because they would ‘solve it
themselves'.
The survey results also revealed
that a much smaller proportion of
incidents of a sexual nature are
5
described as actual ‘assaults' by
Australian respondents in comparison
with those surveyed in other countries.
This result tends to support the view
that sexual incidents are understated
in other countries by current
Australian standards.
Our Reaction to Crime.
Reporting Crime: At first sight these
are frightening results. In several of
these types of crime, in particular
sexual incidents, Australia appears to
have higher rates even than the USA.
Previous analyses, based on police
statistics of crime reported, have
concluded that the general level of
Australian crime is much lower than
that of America, although higher than
comparable European states.
Table 2 helps to explain the
phenomenon. In terms of the
percentage of incidents reported to
police, Australia ranks only twelfth
out of the sixteen countries. Basically,
this could be either because
Australians have so little confidence
in the police that they frequently
prefer not to ask for their assistance,
or because a large proportion of
offences are of such a minor nature
that they are not worth reporting.
For example, around two thirds of
incidents of assault were not reported
to police, mostly because they were
'not serious enough', or the victims
'solved it themselves', or believed that
it was 'inappropriate for police'. Fewer
than 10 per cent of sexual incidents
were reported to police in Australia less than average - and the reasons for
non-reporting were largely similar to
those for other assaults. Australians
were not more likely than average to
avoid reporting incidents because of
fear or dislike of the police.
Satisfaction with Police: Police in
Australia actually appear to be more
popular with the communities they
serve than their counterparts overseas.
All respondents who reported
incidents to police in 1988 were asked
whether they were satisfied with the
way police dealt with their problem.
Australia ranked first with almost 80
Australian Institute of Criminology
per cent of respondents satisfied,
compared with an average across all
countries surveyed of around 66 per
cent.
In more general terms, Australian
police also appeared to receive
relatively high community support. All
respondents, whether victims or not,
were asked 'Taking everything into
account, how good do you think the
police in your area are at controlling
crime. Do you think they are doing a
good job or not?' Australia in this case
ranked third behind Canada and the
USA, still scoring over 70 per cent
compared to an average of around 66
per cent. As in many of the countries
polled, young people were the most
likely to be critical of police.
Crime Prevention Measures: Fear
of crime induces people to take
precautions including fitting burglar
alarms, taking out insurance policies,
or avoiding going out alone after dark.
To provide information about fear of
violent crime in public places,
respondents were asked if they had to
avoid certain areas after dark and if
they go out with companions to avoid
crime. Australia was quite close to
average in these questions. As
elsewhere, women and people who
live in large cities were more likely to
take precautions. The high incidence
rates measured for violent crimes in
Australia did not seem to translate
into high levels of fear of violence.
On the other hand, burglary in
Australia does seem to worry people.
A questions was asked 'What would
you say are the chances that over the
next twelve months someone will try
to break into your home?' Forty four
per cent of Australian respondents
thought that this was 'likely' or 'very
likely' - ranking third behind West
Germany and Switzerland. Despite
this, Australians are moderately, not
exceptionally, high users of such
precautions as burglar alarms,
caretakers and insurance. The fact that
Australia has a very high percentage
of people living in large cities, and our
preference for detached houses, which
are comparatively easy targets for
burglars, appears more than likely to
explain both Australia's high rate of
burglary and Australians' reactions to
it.
Attitudes to Sentencing:
Respondents were asked which of five
types of sentence they thought was
most appropriate for 'a twenty-one
year old man found guilty of burglary
for the second time. This time he stole
a colour TV'. Table 3 shows the
principal results of this question.
Contrary to conventional wisdom,
imprisonment is not the public’s
preferred sentence for a recidivist
burglar, in Australia as in most
other countries. Community service is
seen as most appropriate by almost
half of all respondents. Less than 10
per cent opted for a fine. Just over one
in three Australian respondents would
send the offender to prison.
Comparison with actual imprisonment
rates in these countries showed that
public opinion in each country was
broadly in line with actual sentences.
Respondents from countries with high
imprisonment rates tended to support prison as
the best option. Whether this means
that the courts in each country are
6
actually expressing the wishes of the
public, or that the public is
conditioned by its knowledge of what
the courts actually do, is impossible to
decide from this study.
Who is at Risk
As Table 4 shows, men are marginally
more at risk than women, except in
'contact' crimes. Even here, however,
men are more at risk of violence than
are women, particularly of robbery.
Because only women were asked
about sexual incidents, women appear
to be at greater overall risk than men.
These results confirm the findings
published in Violence: Directions for
Australia (National Committee on
Violence 1990).
In addition, those aged 16-24
years were at least three times more at
risk than those aged over 65 years.
Because of Australia's post-war babyboom and high levels of immigration,
we have a particularly high proportion
of our population in this younger age
group. The fact that so many of us
crowd into large cities also
Australian Institute of Criminology
predisposes us to crime. Those living
in towns of less than 50,000 people
had lower risks of crime than those
living in the capital cities. Those in
cities of more than 50,000 people
were over 50 per cent more likely to
be victimised than those who live in
towns of under 10,000 people.
Employed people are more at risk
than the unemployed. Women in the
labour force were particularly at risk,
however. Those in higher income
groups were at greater risk than the
less well-off.
Opportunity is an important
feature of criminal offending. Clearly,
communities with high levels of
vehicle ownership are more at risk of
crimes involving theft of, or from,
vehicles or damage to them. Vehicles
left unattended in public places are
particularly at risk, so Australia's high
car-ownership and our tendency to use
private cars for journeys to work,
rather than public transport, probably
combine to explain in part, at least,
our high rates of motor vehicle theft.
On the other hand, this tendency of
ours to drive to work could help to
explain our comparatively low rates of
pickpocketing, since public transport
offers considerable opportunities for
such crimes.
People who frequently go out at
night for recreational purposes are
more at risk, in all categories of
offence. This could be because they
were more exposed to crimes
committed in public places, such as
pubs or public transport, but also
because they were more frequently
leaving their cars and houses
unattended. Our climate, which for
most of the year gives us pleasant
evenings and longer hours of daylight,
is likely to be particularly culpable in
this respect, increasing the range of
opportunities for all manner of crimes.
(See Table 5.) Australian respondents
appear to be about 30 per cent more
likely than, for example, the Dutch to
go out in the evening.
Most intriguing of all, however, is
the discovery that countries with
colder climates have lower levels of
all types of crime than countries
nearer the equator. The most
significant statistical effects of this
sort were found in burglary and sexual
incidents, in which over half of the
statistical variation between countries
can be explained by the countries'
distance from the equator. (See Figure
4.) The most likely explanation for
this appears to be that cold or wet
weather acts as an informal curfew,
keeping both offenders and potential
victims at home, thereby reducing the
opportunities for crime. Most
burglaries, for example, take place
while the occupants are away. Most
sexual incidents take place away from
the victim's home.
Summary and Conclusions
Australia is a relatively high risk
7
country as far as crime is concerned
as this cross-national survey indicates.
Happily, we are not in the same
league as the USA, but we do appear
to have higher crime rates than most
European countries. In terms of
serious crimes, Australia does not
differ substantially from most other
countries surveyed. We do appear,
however, to have higher rates of
relatively minor incidents than most
other countries. Our lifestyles appear
to create opportunities which
predispose us to much of this minor
crime.
Among the elements of the
Australian way of life which
determine how much crime we have to
put up with are:
Australian Institute of Criminology
• we live in large impersonal cities,
with few of the informal social
safeguards which are available in
smaller communities; we
compound this by living in houses
separated from our neighbours by
large gardens which actually
make a burglar's job easier;
• we travel to work across the city
by car, leaving behind us
residential suburbs with little
protection against opportunist
burglars, and leaving our cars and
their contents in public places all
day at the mercy of joy-riders,
vandals and professional thieves;
•
our long hours of daylight and
pleasant evenings tempt us to enjoy
outdoor recreational activities in
public places, including pubs and
sports clubs, again leaving the
home with little protection other
than perhaps the odd light left on,
and subjecting ourselves to risk of
all sorts of misfortunes.
Many of the incidents that affect
us are so minor that we do not bother
to report them to the police. When we
do call in the police we are generally
satisfied with their efforts to assist us.
This survey also confirms previous
surveys conducted by the Australian
Institute of Criminology in finding that
we are not exceptionally punitive in
our attitudes to offenders.
Although the recent growth in the
private security industry, and the
continuing spread of the
Neighbourhood Watch concept,
suggests that many Australians are
adopting greater precautionary
measures because of their concern
about crime, we suspect that the
majority will continue to live an
outgoing lifestyle. Efforts to reduce
the levels of violence through better
education and family support and
measures to reduce alcohol
consumption, as proposed by the
National Committee on Violence,
clearly need to be maintained,
however. At least, at this point in
time, it does not appear that our fear
of crime and the actual crime rate
should fundamentally affect the way
we live.
There is a considerable amount of
research still to be performed on the
data derived from this survey, and it is
likely that the results will shed even
more light on the nature and extent of
crime in Australia, compared with
other countries.
Note: This Trends and Issues summarises the
results published in:
Van Dijk J., Mayhew P., and Killias M. (1990)
Experiences of Crime Across the World: Key
Findings from the 1989 International Crime
Survey, Kluwer, Deventer and Boston.
Inquiries about the Trends and Issues
series should be forwarded to:
The Director
Australian Institute of Criminology
GPO Box 2944
Canberra ACT 2601 Australia
8
Copyright of Trends & Issues in Crime & Criminal Justice is the property of Australian Institute of Criminology
and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for individual use.
2001 CRIME TRENDS:
A turning point?
Martin Schönteich
Institute for Security Studies
Published in SA Crime Quarterly No 1, July 2002
Between 1997 and 2000 there was a steady increase in the annual number of recorded crimes. The first nine
months of 2001 saw an end to this trend. During 2001 the rate of increase in recorded crime diminished
markedly. Moreover, for the first time in years, violent crime increased at a lower rate than most other crime
categories. Given these encouraging figures, is South Africa beginning to win the war against crime? While the
latest numbers look promising, talk of a victory is premature. There has to be a sustained decline in the levels
of most serious crimes to reduce the country’s crime levels to acceptable levels.
At the time of writing, statistics reflecting crimes recorded by the South African Police Service (SAPS) have
been released covering the period up to the end of September 2001.
The most recent set of crime statistics, covering the July-September 2001 period, was released at the end of
2001. These third quarter figures for 2001 are significant as they are the first set of crime statistics collected by
the police’s Crime Information Analysis Centre (CIAC), making use of a new but undisclosed methodology,
whereby crime statistics are said to be collected in a more accurate manner. Readers will recall that a
moratorium on the release of crime statistics was imposed by the late Safety and Security minister Steve
Tshwete in July 2000, and only lifted almost a year later.
The July-September 2001 figures were released as part of the statistics for the first three quarters of 2001.
This does not allow for a comparison of only the third quarter 2001 figures with the same three-month period
during previous years. Nevertheless, it would appear that the moratorium has not resulted in unexpected
changes to crime figures. The available 2001 crime statistics do not contain any surprises or significant
changes to long term trends in individual crimes.
South Africa’s official crime statistics are widely regarded as comprehensive and certainly as the most detailed
and reliable of all countries on the African continent. Statistics on the number of crimes recorded by the SAPS
at national level right down to station level have been publicly available since 1994. A remarkable
achievement, considering the history of policing in the country.
Interpreting crime statistics
When analysing the crime statistics that follow below, it is important to remember that recorded crime levels
undercount the real levels of crime, as unrecorded crimes are not reflected. For crime to make it on to the
official police records two things need to happen. Firstly, victims or witnesses must report it to the police.
Secondly, the police must record the crime in their records.
According to Statistics South Africa’s 1997 national Victims of Crime Survey, crimes involving valuable and
insured property are mostly reported. For example, 95% of vehicle thefts, 60% of vehicle hijackings and 59%
of burglaries are reported. Less serious property crimes and interpersonal violent crimes are more often not
reported than reported. Thus, only 41% of robberies, 38% of assaults and 28% of theft of personal property
are reported (see Improved crime reporting, in this issue, for comments about how reporting affects police
crime statistics).
National picture
During the first nine months of 2001, 1 844 000 crimes were recorded by the SAPS; up from
1 464 000 over the same nine month period in 1994 (Figure 1). If the January-September period for 2001 is
compared with that of 1994, the number of recorded crimes increased by 26%.
Figure 1: Number of crimes recorded by the police, Jan-Sept 1994-2001
While the number of recorded crimes has increased since 1994, there has been little change in the relative
proportions of the different crime categories. For example, in the period January-September 1994, violent
crimes comprised 30.1% of all crimes recorded. During the January-September 2001 period this had increased
to 32.5%. All other crime categories experienced a slight decline as a proportion of the overall number of
recorded crimes. Property crimes make up the largest proportion of recorded crimes – approximately 55% of
the total.
Change over time
While recorded crime increased between 2000 and 2001, the rate of increase is slowing down. If the first nine
months of 1997 are compared with 1998, recorded crime increased by 4%. Thereafter, recorded crime
increased by 7% (1998/99) and 8% (1999/00). The January-September 2001 period experienced a 2%
increase compared to the same period in 2000 – the lowest year-on-year increase since 1996/97 (Figure 2).
Figure 2: % change in number of crimes recorded, Jan-Sept 1994-2001
While the figures indicate a slowing down in the rate of increase of recorded crime in 2000/01, it needs to be
remembered that this is occurring at a point where recorded levels of violent crime are extraordinarily high. The
time period is too short, moreover, to draw a firm conclusion whether the trend will continue to the extent that
recorded crime levels will enter a sustained period of decline.
In the January-September period between 1994 and 2001 the number of recorded violent crimes increased by
36%, more than any other crime category (Figure 3).
Figure 3: % change in the number of crimes recorded, Jan-Sept 1994-01 and 200-01
However, between 2000 and 2001 (again, for the period January-September) violent crime increased by 2%.
Both the number of property crimes and violent crimes against property such as arson and malicious injury to
property, increased to a greater extent (3%). Drunk driving offences, and drug and firearm related offences –
categorised as ‘other’ also increased by 3%, while commercial crimes (primarily fraud) decreased by 10%.
Not all crimes increased or decreased at the same rate for the January-September period between 2000 and
2001. Robbery with aggravating circumstances and common robbery experienced the greatest increase of 10%
and 7% respectively (Figure 4). Robberies with aggravating circumstances are robberies involving a dangerous
weapon, including bank robberies, cash-in-transit heists and vehicle hijackings. Common robberies involve
mainly muggings, as well as the snatching of handbags, jewellery and cell phones.
Figure 4: % change in selected crimes recorded, Jan-Sept 2000-2001
Recorded murders decreased by 3%. This is a continuation of a trend whereby the yearly number of murders
has been declining since 1994. Disconcertingly, however, recorded attempted murders are up by 5%. Assault
with the intent to inflict grievous bodily harm (assault GBH) decreased by 2% and car theft by 1%.
Encouragingly, vehicle hijackings – a crime that instills a high level of fear amongst the public – did not
increase over this period.
Violent crime in the provinces
Measured on a per capita basis, the violent crime rate for the January-September 2001 period is highest in the
Northern Cape. In that province 2 151 violent crimes were recorded for every 100 000 people resident (Figure
5). In other words, the average resident of the Northern Cape stood a 2.15% chance of becoming a victim of a
recorded violent crime during the first nine months of 2001. By comparison, the average South African faced a
1.3% chance of becoming a violent crime victim.
Figure 5: Violent crime rate in the provinces, Jan-Sept 2001
Recorded violent crime is spread unevenly over the country’s nine provinces. Residents of the Northern Cape,
Western Cape and Gauteng, for example, were more than two-and-a-half times as likely to become a victim of
a recorded violent crime during the first nine months of 2001 than Limpopo residents.
However, this provincial disparity might slowly be changing. While the violent crime rate was lowest in
Limpopo, the province experienced the greatest increase (6%) in the rate of recorded violent crime during
January-September 2001, compared to the same period in 2000 (Figure 6). Mpumalanga, another province with
a low violent crime rate, was the only other province that experienced an increase (1%) in the rate of recorded
violent crime over the abovementioned period.
Figure 6: % change in violent crime rate, Jan-Sept 200-2001
The Eastern Cape recorded a 3% decrease in its violent crime rate, followed by the Northern Cape, Western
Cape and KwaZulu-Natal (all 2%).
Violence in the Western Cape
Contrary to popular belief, the per capita rate of violent crime has since 1994 been consistently higher in the
Western Cape than in Gauteng. That is, since 1994 the average resident of the Western Cape has been at
greater risk of becoming a victim of a recorded violent crime than the average Gauteng resident (Figure 7).
Figure 7: Violent crime rate in Western Cape and Gauteng, Jan-Sept 1994-2001
The changing levels of the violent crime rate in the Western Cape between 1994 and 2001, JanuarySeptember period, are closely paralleled with those in Gauteng and the country as a whole. The late 1990s
experienced a significant increase in violent crime rates in all three geographical areas, followed by a
stabilisation or decline in the 2000/2001 period.
While national murder and attempted murder rates have declined since 1994, the opposite has been the case
in the Western Cape. For the period January-September 1994 the Western Cape had lower rates of murder
and attempted murder than the country as a whole (Figure 8).
Figure 8: Murder and attempted murder rate in South Africa and Western Cape, Jan-Sept 1994-2001
However, in 2001 (January-September) the Western Cape’s murder rate was 67% above the national average,
while the province’s attempted murder rate was 38% higher than the national average.
The Western Cape has experienced a significant increase in the rate of robberies with aggravating
circumstances. During January-September 1994 the Western Cape’s rate of aggravated robberies was 24%
lower than the national average. In 2000 the province’s aggravated robbery rate was roughly the same as that
of the country as a whole. In 2001, the aggravated robbery rate in the Western Cape was 16% higher than in
the country as a whole (Figure 9).
Figure 9: Aggravated robbery rate in South Africa and Western Cape, Jan-Sept 1994-2001
Similarly, the recorded vehicle hijacking rate in the Western Cape has more than doubled between 1994 and
2001, January-September, from six incidents per 100 000 residents to 13 incidents. Nationally, vehicle
hijacking rates are higher but much more stable. The national vehicle hijacking rate increased marginally
between 1994 and 2001, from 24 to 25 incidents per 100 000 of the population.
Conclusion
The recorded crime figures for the first three quarters of 2001 contain good news. Recorded crime is
increasing at its lowest rate in years. When measured on a per capita basis violent crime – long a trademark of
crime in South Africa – decreased between 2000 and 2001.
While the country needs good news about crime, South Africans should not be fooled into believing that the
country is about to enter an era of low crime levels. The first nine months of 2001 witnessed an average of 55
murders, 137 rapes, 558 robberies and 1 350 assaults a day – and that is before unrecorded crime is taken
into account.
Clearly crime levels have to actually decrease – and decrease for sustained periods of time – before the
country reaches acceptable levels of crime. Moreover, levels of serious violent crime in the Western Cape are
showing disturbing trends.
Source documents
All statistics in the graphs are provided by the SAPS CIAC, see www.saps.gov.za
Copyright of SA Crime Quarterly is the property of Institute for Security Studies and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.
Victims & Offenders, 10:420–451, 2015
Copyright © Taylor & Francis Group, LLC
ISSN: 1556-4886 print/1556-4991 online
DOI: 10.1080/15564886.2015.1078186
International Trends in Prison
Upsizing and Downsizing: In
Search of Evidence of a Global
Rehabilitation Revolution
James M. Byrne
School of Criminology and Justice Studies, University of Massachusetts, Lowell,
Lowell, Massachusetts, USA
April Pattavina
School of Criminology and Justice Studies, University of Massachusetts, Lowell,
Lowell, Massachusetts, USA; Wellesley Centers for Women, Wellesley College,
Wellesley, Massachusetts, USA
Faye S. Taxman
George Mason University, Fairfax, Virginia, USA
Abstract: Our review of global trends in the use of incarceration reveals that both
the number of people incarcerated worldwide and the global rate of incarceration continue to increase in every global region. In addition, there have been sizable global
increases in the use of pretrial detention, the proportion of female prisoners, the proportion of foreign prisoners, prison privatization, and prison crowding. After reviewing
these global trends, we examine the available data on correctional performance, focusing on global crime trends in the subgroup of countries (n = 20) where three-quarters
of the world’s prison population is currently located. While most (17/20) of these countries continued to increase their prison population despite decreases in most crime
categories, three countries—the Russian Federation, South Africa, and Pakistan—have
reduced their reliance on incarceration (number and rate) while also reducing their
crime rates. Compared to other countries with large prison populations, the United
States’ overall increase in both the prison population (+14.4% since 2000) and incarceration rate (+2.2) can be described as modest. Our review of global trends in corrections
and sentencing policy reforms revealed an emerging focus on offender rehabilitation
Address correspondence to James M. Byrne, School of Criminology and Justice
Studies, University of Massachusetts, Lowell, 1 University Ave, Lowell, MA 01854.
E-mail: profbyrne7@gmail.com
Color versions of one or more of the figures in the article can be found online at www.
tandfonline.com/uvao.
420
International Trends in Prison Upsizing and Downsizing
in both prison and community corrections settings and a growing recognition of the
link between community development and crime. Focusing on the potential impact of
rehabilitation-rich policy reforms in the United States, we highlight the results of a
recent simulation modeling study (Taxman & Pattavina, 2013) designed to estimate
the impact of expanding our institutional corrections treatment infrastructure. This
study reported significant, but relatively small, long-term reductions in prison populations when treatment capacity and treatment quality are increased. These findings
suggest that even major improvement in the quantity and quality of prison programming—by itself—will not fundamentally change the life course of prisoners, resulting
in continued high return-to-prison rates for these offenders. We conclude that for the
United States to significantly reduce its reliance on incarceration, we will need to
rethink our approach to crime and punishment. Our global review of corrections and
sentencing strategies identified a number of possibilities. Using the global incarceration rate and the global crime rate as benchmark performance “tipping points,” we
call for the development of corrections and sentencing policies—and the hiring of corrections personnel—based on the new goal of corrections: individual and community
rehabilitation.
Keywords: correctional
rehabilitation
policy,
international
corrections,
prison
downsizing,
To gain a better understanding of the nature and extent of crime, criminologists have studied global crime patterns using country-level data as the
unit of analysis. This body of research has identified a set of factors (e.g.,
income inequality, poverty/economic development) that distinguish high-crime
from low-crime countries, and leaders in several countries have designed
crime reduction initiatives utilizing this information (Commission on Crime
Prevention and Criminal Justice, 2014; United Nations Office of Drugs and
Crime, 2014). A similar effort is now underway to understand the global
response to crime, focusing on the utilization of prison and alternatives to
incarceration (Allen, 2015; Walmsley, 2015). While every country’s crime
problem—and each country’s criminal justice response—is unique, there is
value in the examination of global variations in corrections and sentencing
practices and attempts to identify (and rank) both high-performance and lowperformance countries, based on the evidence of the effectiveness of their corrections systems and sentencing strategies. Unfortunately, the global performance measurement data base needed to systematically review and rank the
performance of countries’ corrections and sentencing strategies does not currently exist, but it is possible—due largely to the work of Walmsley (2015) and
Allen (2015)—to provide a global snapshot of recent corrections and sentencing
trends and developments1 and to review the available country-specific evaluation research on the effectiveness of these strategies (Byrne, 2016, 2015).
The following review highlights recent global trends in the use of prison
internationally (2000 to 2015) and then examines the changes in prison populations in the subgroup of countries (20) that currently house over 75% of
421
422
J. M. Byrne et al.
the world’s prison population.2 Seventeen of these 20 countries increased their
prison populations between 2000 and 2015. Prison populations have continued
to rise globally at a rate slightly higher than world population growth during
this period; at the same time, declines have been reported in every major crime
category (except drugs). There are exceptions to the global upsizing trends
identified here. Three of the 20 countries with the largest prison populations—
the Russian Federation, South Africa, and Pakistan—reported declines in
their prison populations, with the Russian Federation’s prison population
decline particularly noteworthy (a 37% reduction in the total prison population between 2000 and 2015). While a detailed review of the reasons for the
Russian Federation’s prison decline was beyond the scope of our review,3 we
have identified the most commonly used prison downsizing strategies globally.
Strategies designed to reduce the rate of incarceration include new sentencing
legislation (e.g., federal and state sentencing guideline reform), the expansion
of alternative sanctions, and a renewed focus on offender treatment in both
prisons and community corrections. Evidence to support these strategic initiatives can be found in several recent systematic evidence-based reviews (Byrne,
2016, 2015).
Using the results of a recent simulation modeling study conducted in the
United States (Taxman & Pattavina, 2013), we consider the potential impact
of policy reforms designed to simultaneously downsize prisons and upsize
the treatment infrastructure of our institutional and community corrections
system in the United States. These analyses reveal that while long-term reductions in prison populations are possible, only incremental changes in our prison
population are likely unless we consider the need for fundamental (some might
say revolutionary) changes in policies, programs, and personnel. The following
three recommendations are offered for consideration:
1. New corrections and sentencing policies that result in rates of incarceration
at or below the average global incarceration rate.
2. A new commitment to the provision of high-quality treatment programs in
the institutional and community corrections system.
3. The recruitment, training, and retention of a new generation of corrections
personnel with skills in counseling, mentoring, and advocacy similar to
corrections personnel in countries with a pronounced rehabilitation orientation (e.g., Australia, Norway, Sweden, and the United Kingdom). Changes
in each of these areas will require a new commitment to an old strategy:
rehabilitation at the individual and community level.
International Trends in Prison Upsizing and Downsizing
GLOBAL PRISON POPULATION TRENDS: UPSIZING VERSUS
DOWNSIZING
It is estimated by the International Center for Prison Studies (ICPS) that
there are currently over 10.2 million prisoners living in prisons around the
world today, and if the reports on the number of offenders in pretrial or
administrative detention in China and North Korea are accurate, the world
prison population total is actually over 11 million (see the 2013 population
estimate included in the World Population List by Walmsley, 2015). Based on
data compiled by ICPS on current prison populations, we have examined the
prison population data for the countries with the 20 highest prison populations globally. Seventy-five percent of the world’s prisoners (7,689,884) reside
in these 20 countries, including 8 countries from Asia, 5 from the Americas,
4 from Europe, and 3 from Africa. Table 1 presents changes in the prison populations in these 20 countries since 2000. In 17 of 20 countries, the prison
population has increased at a rate slightly greater than the rate of increase
in the world’s population. According to the recent review by Walmsley, “Prison
populations are growing in all five continents. In the 15 years since the first
edition of the World Population List the estimated world prison population has
increased by some 20% to 25% but at the same time the world population has
risen by over 20%. The world’s prison population rate has risen by about 6%
from 136 per 100,000 of the world population to the current rate of 144 [per
100,000]” (2015, p. 1). Despite the calls for reform and for the downsizing of
our global prison population (see, e.g., Allen, 2015; Cullen, 2013), correctional
populations continue to be upsized in every global region.
When examining recent trends in our global prison population, it is worth
considering the fact that almost half of the world’s prison population is held in
only three countries: the United States, China, and the Russian Federation.
About 20% of the world population currently resides in these three countries. Two of these countries (United States, China) have continued to increase
their prison/pretrial detention populations—and their incarceration rate—
since 2000. By contrast, the Russian Federation has significantly decreased
both the number of prisoners and the rate of incarceration without increased
reported crime rates. Some observers will be quick to dismiss these findings for the Russian Federation with glib reference to potential bad data,
while pointing to the notable, documented deficiencies in prison conditions
and prison labor in the Russian Federation (Amesty International, 2015;
Moran, Pallot, & Piacentini, 2011; U.S. Department of State, 2013). However,
there certainly appears to be significant prison downsizing movement in the
Russian Federation, and further investigation is needed on whether—and if
so, how—this large-scale downsizing strategy was accomplished.
423
424
United States of
America (2013)
China (2014)
Russian Federation
Brazil (2014)
India (2013)
Thailand
Mexico
Iran (2014)
Indonesia (2014)
Turkey
South Africa
Vietnam (2014)
Colombia
Philippines (2013)
Ethiopia (2011)
United Kingdom: England
and Wales
Poland
Pakistan (2012 data)
Peru
Morocco (2013 data)
Title
Europe
Asia
South America
Africa
Asia
Europe
South America
Asia
Asia
Central America
Asia
Asia
Europe
Africa
Asia
South America
Asia
Africa
Europe
Northern America
Region
198 (68)
41 (206)
236 (53)
221 (60)
119 (123)
463 (9)
301 (34)
33 (212)
452 (10)
214 (64)
290 (36)
179 (66)
212 (65)
292 (35)
154 (94)
242 (50)
113 (131)
111 (134)
148 (99)
698 (2)
76,145
74,944
73,255
72,816
1,657,812
667,546
607,730
411,992
304,916
256,941
225,624
167,163
165,033
159,241
142,636
119,378
110,925
93,044
85,743
2,217,000
2015 Prison
Pop
70,544
78,938
27,734
54,288
1,427,407
1,060,404
232,755
272,079
223,406
154,765
158,000∗∗
53,399
49,512
171,462
88,414∗∗∗
51,518
79,299
55,209
64,602
1,937,482
2000 PP
%
Change
+14.4%
+16.1%
−37.0%
+161.1%
+51.4%
+36.5%
+66.0
+42.8%
+213.1%
+233.3%
−7.1%
+61.3%
+131.7%
+39.9%
+68.5%
+32.7%
+7.9%
−5.1%
+164.1%
+34.1%
PP
Change
+279,518
+230, 405
−392,858
+374,975
+139,913
+81,510
+102,176
+67,624
+113,764
+115,521
−12,221
+54,222
+67,860
+31,626
+37,835
+21,141
+5,601
−3,994
+45,521
+18,528
∗ 2015 data unless noted above; includes pretrial detention population. For details, go to http://www.prisonstudies.org/world-prison-brief.
Data included above retrieved July 14, 2015.
∗∗ 2001 data.
∗∗∗ 2005 data....
Purchase answer to see full
attachment