Transcript Crime (Further Empirical Studies)

Crime (Further Empirical Studies)
• Anderson Mutemererwa, 1 March 2005
Is Capital Punishment a Deterrent
to Murder?
• H. Dezhbakhsh, P.H. Rubin & J.M.
Shepherd (2003), “Does Capital
Punishment Have a Deterrent Effect? New
Evidence from Postmoratorium Panel
Data”, American Law and economics
Review, Vol 5(2), pp344-376.
Background to the Study
• A huge debate about the merits of capital crime
in the US. So there are profound policy
implications.
• Important question is whether the capital
punishment deters murders. Anecdotal evidence
from psychologists and criminologists initially
found no deterrence.
• Yet studies by economists (Ehrlich, 1975, 1977)
found a significant deterrent effect a study which
was used by the Solicitor General of the USA in
a submission to the US Supreme Court in
support of capital punishment.
Background to the Study
• Ehrlich’s study spurred a huge interest by
econometricians using his data but different
model specifications. Inevitable that different
results would ensue.
• Capital crime imposed for different crimes
(murder, treason, heresy) throughout history
and by different authorities (religious and civil).
Most Western civilisations have since abolished
the death penalty.
• US an exception. Changed positions in the
1970s, first banning it in 1972 and then
reversing this in 1976 (under certain carefully
specified conditions). 38 states currently have it.
Some that do not have it are debating whether to
reinstate it and those that do whether they
should abolish it.
Background to the Study
• Debate on capital punishment in US involves
moral principles and social welfare
considerations. Moral arguments tend to be
theoretical while welfare ones are empirical.
• Becker’s (1968) economic model tends to
provide the theoretical foundations for much of
the empirical work.
• Ehrlich extended Becker’s work; Hence
increases in perceived probabilities of
apprehension, conviction given apprehension or
conviction given conviction will reduce an
individual’s incentive to commit murder.
• Changes in legitimate and illegitimate income
opportunities have a similar crime-reducing
effects.
Research Questions
• Statistics show that although executing states
had higher rates of murder than non-executing
ones, the rates are converging. Is there a
deterrent effect from capital punishment?
• What about certain murders that are not
“deterrable” such as non-premeditated crimes of
passion and non-negligent manslaughter (see
Shepherd (2004))?
• Looked at the functional form of the model
bridging the gap between an individual’s
behaviour and that of county, state or nation
and suggest improvements.
Model Specification
• A murder supply equation (macabre as it
may sound!) is specified for a county and
denoted by
• Mi,t= αi+β1Pai,t β2Pc|ai,t + β2Pe|c +γ1Zi,t+TDt+εi,t
• The murder supply equation is built from an
individual’s decision rule of the following form
• Ψt=f(Pat, Pc|at, Pe|ct, Zt, ut) where
Ψt =1 when an individual commits murder during
period t and
Ψt = 0 otherwise
(Economic) Model of Murder
• The above is called the murder supply equation.
• The decision (for an individual) whether or not to
commit murder (ψj,t) is a function of the
individuals subjective probability (P) conditional
on apprehension (a) conviction (c) and
execution (e), the individual’s own economic and
demographic characteristics (Z) and other
random effects (u).
• TD is a time trend dummy which captures
national trends such as violent TV programs etc.
• ai is a county-specific fixed effect.
• Summing up all individuals in a county and
dividing by the county population allows the
calculation of the murder rate for a county, Mi,t.
(Economic) Model of Murder
• Added to the murder supply equation will be three
expenditure models for the law enforcement
agencies: police, judicial-legal and prisons
• Pai,t=φ1,t+φ2Mi,t+ φ3PEi,t+ φ4TDt+ζi,t
• Pc|ai,t=φ1,t+φ2Mi,t+ φ3JEi,t+ φ4PIi,t + φ5PAi,t +φ6TDt+ζi,t
• Pe|ai,t=φ1,t+φ2Mi,t+ φ3JEi,t+ φ4PIi,t + φ5TDt +ζi,t
• JE is expenditure on judicial and legal system, PE
is the police payroll and PI is peer influence such
as a Republican candidate’s percentage of statewide votes whil TD is the same as in the murder
supply model.
• If police and prosecutors want to minimise the
social costs of crime, they must balance the
marginal cost of enforcement with the marginal
benefits of prevention. Hence more
Results and Conclusions
• The data came from 3 054 counties for period
1977-96.
• The results of the models reflect strong deterrent
effects of execution.
• The execution probability is negative and highly
significant. This suggests that the Results and
Conclusions perceived probability of execution
given that one is sentenced to death leads to a
lower murder rate.
• The sentencing effect is weak and does not
seem to have a deterrent effect.
Results and Conclusions
• The murder rate appears to increase with
aggravated assault and robbery. This suggest
that these crimes are partly caused by the same
factors which lead to murder. Also, some
murders are by-products of robbery and
aggravated assaults.
• Demographic effects are also important; a larger
percentage of African-Americans (as opposed to
other minorities) is associated with higher
murder rates while an increase in teenage
population appears to lower the murder rates.
Results and Conclusions
• Finally in those counties with a high members of
the NRA, the murder rates appears to be high--high ownership of handguns, leading to guns
being the weapons of choice in confrontations.
• Police expenditure has a consistently positive
effect on arrest
Pai,t=φ1,t+φ2Mi,t+ φ3PEi,t+ φ4TDt+ζi,t
• The same is true for the judicial-legal system
Pc|ai,t=φ1,t+φ2Mi,t+ φ3JEi,t+ φ4PIi,t + φ5PAi,t +φ6TDt+ζi,t
Pe|ai,t=φ1,t+φ2Mi,t+ φ3JEi,t+ φ4PIi,t + φ5TDt +ζi,t
• Joanna M. Shepherd (2004), “Murders of
Passion, Execution Delays, and the
Deterrence of Capital Punishment”,
Journal of Legal Studies, Vol. 33, pp 283321.
Research Question?
• Which murders are deterred and what of
the effect of a wait on death row on
deterrence?
• What types of murders are deterred?
Data and Model
• The data consists of monthly state level data for
the period 1977-99.
• During this time 7 416 inmates served on death
row and 598 of them were executed.
• The model is specified as
M = β1 DETERi,t + β2 ECONi,t +β3 DEMOi,t +
β4 si + β5 γt + β6 zt+εi,t
• DETER is a vector of deterrent variables
(probability of sentencing to death row and
execution)
Model Specification
• ECON is a vector of economic variables
(real per capita income and monthly
unemployment rate)
• DEMO is a vector of demographic variables;
age (specifically the percentage of
population which is between 10 and 29
years of age), percentage of population
which is African-American, percentage of
population which is minority other than
African-American.
• γ is a series of year dummies, z is a series
of monthly dummies and the s is a series of
state dummies (three strikes, non-death
penalty for murder)
Results
• Probability of a death row sentence is a
deterrent to all types of murder.
• Shorter waits on death row before execution
also deter murder– the longer the wait, the
higher the murder rate and vice versa (given that
the average wait on death row has risen from 1
year in 1981 to 12 years in 1999).
• Probability of execution also deters all murders
(by intimates and acquaintances or strangers,
murders of white or blacks).
Results and Conclusions
• In addition, the death sentence appears to
deter both blacks and whites (in answer to
conjecture that the death sentence is
racist and disproportionately falls on
blacks)
Result and Conclusions
• The Shepherd study confirms and extends the
Dezhbakhsh et al study, namely that the death
penalty deters crime.
• She also goes further by showing that all
murders are deterred by capital punishment and
this deterrence cuts across race. Recall that
Dezhbakhsh et al had not given an unequivocal
answer to this question.
• The key conclusion is that the expected penalty
(the probability of arrest, conviction and a
lengthy prison sentence/execution) affects the
potential criminal’s desire to commit a crime.
Other studies
• Look at John R. Lott, Jr., And David B. Mustard
(1997), “Crime, Deterrence, And Right-to-carry
Concealed Handguns”, Journal of Legal Studies, vol.
XXVI, pp 1-68.
• This study looks at the issue of carrying
concealed hand guns as a deterrent to crime
across the whole USA.
• The main conclusion is that allowing citizens [who
have no history of crime or mental illness] to carry
concealed handguns deters violent crime. Makes
the claim that, as a result, if the whole USA
allowed the carrying of concealed weapons in
1992, 1 414 murders and 4 177 rapes would not
have taken place. An estimated social benefit of
$5.74 billion would have been realised.
Other studies
• However, the study also notes that this
resulted in property crimes involving stealth
where the contact between the criminal and
the victim is low or no-existent. This is
especially in large population areas.
• The study also supports the economic notion
of deterrence. High arrests rates (irrespective
of eventual conviction) consistently reduce
crime rates. The economic reason of which is
that increased arrest rates significantly
increases the economic penalty on criminals.