Measuring police performance: what do the Dutch know

1
Measuring police performancewhat do the Dutch
know?

• Ben A. Vollaard
• RAND Graduate School
• CPB Netherlands Bureau for Economic Policy
  Research

2
Good cop, bad cop

• Police force Rotterdam-Rijnmond was furious when
  the Interior Department stated that Rotterdam had
  the worst police force of the Netherlands.
• (Algemeen Dagblad, January 7, 2004)
• After a clarifying discussion with the Interior
  Secretary, police force Rotterdam-Rijnmond
  received their performance bonus after all.
• (de Telegraaf, January 7, 2004)

3
Outline

• State of the art
• (assessment of police as organization, not of
  specific methods of policing)
• Research into police performance at CPB
• First results
• Conclusion what do we know?

4
State of the art (1)Assessing individual police
forces

• Key question forces actions ? outcomes (crime,
  traffic safety, )
• UK that link can only be made subjectively
• Relation output measures-outcomes often ambiguous
• Weighing multiple tasks
• NL performance assessment in its infancy
• Poor availability, reliability comparability of
  data
• No methods for fair comparisons between forces
• No structured subjective assessment (peer review,
  police inspection)

5
State of the art (2)Assessing the police as a
whole

• So far assessment output-based (solved crimes,
  etc.) ? very informative
• Ambiguous link to outcomes more output better?
• Many outputs cannot be measured (only 5 of time)
• Output mix changes over time (? policing methods)
• Need for direct link police actions - outcomes
• Current stage do the police have any impact at
  all?
• Challenges
• Endogeneity non-experimental research design
• Data often incomplete, unreliable
• Power for identification need large panel data
  set

6
Outline

• State of the art
• Research into police performance at CPB
• First results
• Conclusion what do we know?

7
Research at CPB ? police resources ? ? crime

• Is more police the solution?
• How approach challenges?
• Endogeneity problem non-experimental design,
  natural experiment ? resources by 12-52
• 1st Step how does crime affect budgetary
  decisions? (February 2004)
• 2nd Step how do budgetary decisions affect crime
  using IV from 1st step (October 2004)
• Data region-specific, bi-annual victimization
  survey
• Power source of variance 25 police
  regions/1993-2003

8
Outline

• State of the art
• Research into police performance at CPB
• First results
• Conclusion what do we know?

9
1st step crime ? regional distribution of police
budget

• High crime rate ? high level of resources
• High crime rate ? high growth in resources
• Policy makers trade off
• Shifting resources to rural areas to achieve more
  equality across regions in assistance to the
  public
• vs.
• Shifting resources to urban, crime prone areas to
  achieve more equality across regions in crime
  fighting

10
The burden of crime vs.The burden of distance
crime ? budget (2)
11
Equality in police assistancepreferred to
equality in crime fighting
crime ? budget (3)
Rural regions receive more resources (given their
crime rate) to assure equality in assistance
Dyfed Powys
Chance of having your crime solved depends on the
region you live in
12
Trading off inequalities police assistance vs.
crime fighting
crime ? budget (4)
Netherlands, 1993-2003 swings in budget
allocation 1993-1998 urban regions better
off 1998-2003 countryside better off
Relation rurality dissatisfaction with
timeliness of police response
High dissatisfaction in rural regions
Relation rurality detection rate
No urban/rural inequality
13
2nd step police resources ? crime

• 1st step endogeneity can go either way
• 2nd step relate regional variation in growth of
  police resources to development of crime using
  the budget formula as instrumental variable

14
Outline

• State of the art
• Research into police performance at CPB
• First results
• Conclusion what do we know?

15
Conclusion measuring police performance, what do
the Dutch know?

• Assessment individual police forces in its
  infancy
• organize subjective assessment for link to
  outcomes
• develop methods to compare forces as input
• improve availability, reliability, comparability
  of data
• Assessment police as a whole
• direct link to outcomes necessary
• current stage do the police have any impact at
  all?