Presentation Slides - Centre for Aboriginal Economic Policy Research

Offender Mobility
Presented by Boyd Hunter
Centre for Aboriginal Economic Policy Research
Seminar series on 31 August 2011.
Acknowledgement:
Research partially supported by ARC Linkage grant (LP0776958) titled ‘An inter-disciplinary
analysis of the dynamics of Aboriginal interactions with the criminal justice system’ that
involves the collaboration between the ANU and NSW Attorney Generals Department
Motivation
• Study important because:
– OID 2011 shows gap between indigenous and nonindigenous offence/imprisonment rates increasing
– Migration has been raised as possible solution to
Indigenous socio-economic disadvantage and we can
get some insight about short to medium term mobility for
an important group of indigenous people, offenders
– Interaction with the Criminal Justice System is directly
and negatively associated with socio economic
outcomes, but does it directly affect the likelihood of
mobility and where offenders move to.
Age standardised imprisonment rates (2011 OID Report)
Motivation
• Study important because:
– OID 2011 shows gap between indigenous and nonindigenous offence/imprisonment rates increasing
– Migration has been raised as possible solution to
Indigenous socio-economic disadvantage and we can
get some insight about short to medium-term mobility
for an important group of Indigenous people, offenders
– Interaction with the Criminal Justice System is directly
and negatively associated with socio economic
outcomes, but does it directly affect the likelihood of
mobility and where offenders move to.
Hunter and Biddle (AJLE 2006)
• Human capital model of migration (partially)
supported
– Qualified Indigenous people more likely to have
moved
– Migration mostly follows standard lifecycle patterns
– Unemployment of area has expected ‘push’ effect
– However mixed effect of average regional income
• But other factors also important
– Proportion of population who are Indigenous reduces
out-flow of indigenous migrants
– In very remote areas, cultural factors appear to
dominate lifecycle factors
• Article recommended that selective migration
needs to be analysed
Most common destination - Non-Indigenous
Note: Percentage change 2001-2006
Most common destination - Indigenous
Note: Percentage change 2001-2006
Theory
A two step process is assumed for the migration decision
• First individuals make the decision to move to a different
LGA based on the characteristics of the LGA in which they
live
• Once the decision to migrate has been made, the decision of
migrants to move to a particular LGA is made based on the
characteristics of the potential destinations (following
Greenwood 1997). Empirics based on the ‘gravity’ model.
• Migration/mobility observed when the anticipated level of
ultility is greater than a given threshold representing the cost
1 M 5* > µ5
of moving
M 5 = if
0
M 5* ≤ µ5
Theory: Space, crime, mobility and migration
Crime can be related to either the mobility and migration decision
• Difference between location of offender and where the crime takes place
•
Routine activity theory suggests crime will take place within everyday patterns
of social interactions (Weir-Smith 2004)
• Routine influences the amount of exposure of potential victims to potential offenders.
Hesseling (1992) violent crime and vandalism are the more locally committed crimes as
compared to property crimes. Residential location can affect opportunity for crime
•
Social Disorganization Theory: Offenders live in economically depressed, less
stable and social capital depleted communities as a result of being relegated to
such locations, not due to choice (limited or downward social mobility)
•
Social reaction theory: extent of adverse social reaction to a person's criminal
records inversely relates to the distance of his current residence from the site of
his former misconduct (some crimes may provoke more reaction than others)
•
Repeat offenders, Indigenous identity and mobility (Kate Sullivan’s thesis)?
• Living in a racist community may bind people together and strengthen Aboriginality (Repeat
offence leads to prison where disproportionate indigenous presence also reinforce identity)
Repeat Offender Database (ROD)
• BOCSAR’s compilation of court data sets
– Sex
– Age
– ATSI status
– Current location for the particular court
appearance (postcode and LGA)
– Offence type
– Date of court appearance
– Penalty
ROD Data, 1994-2008
• Almost 2 million offences documented in ROD between
1994 and 2008
• Geographic information provided for post code and LGAs,
but latter used with concordance files for 2001 boundaries
to maximise comparability over time
– 118,255 indigenous offenders
– 411,283 non-indigenous offenders
• Movers were imputed when the LGA changed between
offences
• LGA characteristics imputed from last census before
current offence
– distance between areas estimated between population weighted
centroids
– Employment demand and housing payment indexes calculated in
AIFS RP 50
Descriptive statistics for Individual model
Non-indigenous
Mean
Indigenous
Std. Dev.
Mean
Std. Dev.
Move
0.302
(0.459)
0.291
(0.454)
Male
0.869
(0.338)
0.784
(0.412)
PO: Acts intended to cause injury
0.132
(0.338)
0.188
(0.391)
PO: Sexual assault and related offences
0.004
(0.067)
0.004
(0.061)
PO: Dangerous/negligent acts endangering persons
0.030
(0.170)
0.013
(0.111)
PO: Abduction and related offences
0.000
(0.018)
0.000
(0.018)
PO: Robbery, extortion and related offences
0.008
(0.091)
0.011
(0.106)
PO: Unlawful entry, burglary, break & enter
0.026
(0.160)
0.052
(0.221)
PO: Theft and related offences
0.132
(0.338)
0.155
(0.362)
PO: Deception and related offences
0.023
(0.151)
0.012
(0.110)
PO: Illicit drug offences
0.084
(0.278)
0.058
(0.233)
PO: Weapons and explosives offences
0.009
(0.096)
0.005
(0.070)
PO: Property damage & environmental pollution
0.046
(0.210)
0.065
(0.247)
PO: Public order offences
0.075
(0.263)
0.135
(0.342)
PO: Road traffic & vehicle regulatory offences
0.336
(0.472)
0.184
(0.388)
PO: Offences against justice procedures, government security and
operations
0.078
(0.268)
0.100
(0.300)
PO: Miscellaneous offences
0.015
(0.121)
0.018
(0.131)
Time since previous offence (in Months)
3.979
(8.723)
4.122
(7.939)
Number of prior convictions in the ROD data
2.736
(4.474)
5.189
(6.284)
Modeling whether offenders move: Individual level
• Probit model of whether to move estimated
separately for indigenous and non-indigenous
populations (similar incidence of mobility)
• Small effects in the expected direction for most
variables
Age (in months) + 0 *
Age squared 00
Male - Time since finalisation of previous offence (in months)++
Number of prior offences in ROD (<1994)++ *
Type of principal offence 00
Modeling where to move: Gravity Model
• Negative binomial model that accounts for overdispersion in the count data of movers between LGAsd
(off-set variable is the offender population in source LGA)
– Complicated by separate controls of source and
source/destination information. Notwithstanding main findings:
•
•
•
•
•
•
•
•
•
•
Housing payments (PCA index) - Employment demand 1996-2001++ *
Net migration 1996-2001 (%) ++
Year 12 completion (%) -- *
Employed in government sector
Families with children (%) -- *
Indigenous residents (%) -+ *
Accessibility of LGA (average ARIA) + - *
Sydney effect + - *
Distance between LGAs (in kilometres)+-*
Concluding reflections
• Mobility of offenders not associated with particular crimes
• Indigenous offenders are affected by local economic
conditions, albeit somewhat smaller affects apparent than
for non-indigenous offenders
– Not surprising given the interaction between arrest and economic
outcomes
• Indigenous specific factors in offender mobility are
important, but direction of causation ‘indeterminate’
• Reppetto (1976) examines whether crime prevention
programs that stress opportunity reduction or increased risk
to offenders merely displace crime (shifting its incidence to
other forms, times, and locales)
– programs least subject to displacement would be those based on
large areas rather than on individual targeted locations.
– But offenders appear to be opportunistic, territorial bound and lack
skills; hence crime prevention in one area will not lead to their
recurrence of crimes in another area
End of Presentation
Age standardised imprisonment rates (2011 OID Report)
Percentage of ROD data identified as ATSI
in local court data that year
Two methods of allocating unknown ATSI status
25.0
20.0
Unknowns randomly allocated as ATSI or non-ATSI
Unknowns treated as non-ATSI
15.0
10.0
5.0
0.0
1994
1996
Source: Hunter & Aayar (2011)
1998
2000
Year
2002
2004
2006
Longitudinal Information & data quality
Percentage Indigenous in local court data
that year
20
18
16
14
12
10
8
6
4
2
DSE based on previous appearance(s) in ROD
Consolidated Indigenous status with unknowns randomly allocated as
non-ATSI or ATSI
Consolidated Indigenous status with unknowns treated as non-ATSI
0
1994
1996
Source: Hunter & Aayar (2011)
1998
2000
Year
2002
2004
2006