LEC2 - Economics of Crime II Flashcards
What are the two Endogeneity problems
Endogeneity, in simple terms, refers to a situation where there is a relationship or correlation between an independent variable (the cause) and the error term
Reverse causality
The causal relationship goes the other way around than we think.
Third (confounding) variable
Both variables can be influenced by a third variable, so they actually have no
direct relationship.
Can the Endogeneity problems be solved with RCT?
randomised controlled trial (experiment)
An RCT is a set up where a researcher/policymaker is able to
randomly assign the
treatment (X). Since the treatment (X) is assigned randomly, the treatment cannot be influenced by Y and the correlation between treatment (X) and
confounding variables (Z) is absent
However,
launching RCTs at a large scale in the domain of criminal justice system
is difficult, because:
It is expensive to run RCTs.
It is often unethical/unconstitutional to run RCTs in this field.
Can endogeniety problems be solved with natural or quasi experiments?
A natural (quasi-) experiment follows the logic of an RCT such that a treatment is
applied to some subjects but is whitheld from others. Importantly, in these designs the treatment is as-if randomly assigned by the force of nature or as a
byproduct of policy reforms.
What are the findings of the most important studies testing deterrence theory?
Levitt (1997) researched the effect of police on crime and encountered reverse causality. So he tried to find a third variable that could effect crime other than probability.
He used this third variable to argue that there is a negative relationship between
police employment and crime, not reverse causility.
Evans and Owens (2007) found that the hiring new police officers policy was
unrelated to trends in crime rates, so
there was another variable that affected the degree of police employment
They found a negative causality between hired police and crime. Elasticity is |E| = 1, meaning that 10% increase in officers caused a 10% decline in crime.
Di Tella and Schargrodsky (2004) used
terrorist attacks as a natural
experiment. Police presence is usually enhanced, even though it has nothing to do with the
changing crime rates. So reverse causality would be ruled out from the start.
They found that increased police presence reduced car theft by 75%
But they didnt check for the displacement effect, so it could be that the car thieves just moved two blocks away.
Draca et al. (2011) studied potential displacement effects after bombing attacks in london.
Increased police presence caused decline in violent crimes with |e| = 0,4.
There was neither geographical nor temporal displacement of crime.
Klick and Tabarrok (2005) used terror alerts which provides a shock to police
presence, but has no relation to local crime rates.
They found that crime decreases during high-alert periods.
Corman and Mocan (2000) argued that
monthly data simply do not allow
enough time for crime rates to affect the size of the police force (it takes time—
about three months—to hire and train new officers).
Any concurrent relationship between these two variables must pick up the
causation running from the size of the police force to crime.
The study found that an increase in police force reduced crimes of robbery and burglary. Median 𝜺 = 0.4
MacDonald et al. (2016) used police patrols within defined geographic boundary.
This created a sharp drop in patrols at the boundary.
Ratcliffe et al. (2011) employed
RCT in police foot patrols.
12 weeks of randomised patrolling of crime hotspots in Philadelphia caused a
net prevention effect (including displacement effect) of 53 crimes.
Chalfin and McCrary (2017) argued that on the whole, the high degree of visibility
around police crackdowns or policing hotspots suggest a potentially greater role for
deterrence.
What is the formula for Probability (Certainty) of Punishment
Probability (certainty) of punishment =
probability of apprehension (police) x
probability of conviction (judicial system)
Atkinds and Rubin (2003) found that if the probability of conviction goes down,
criminals face a reduced certainty of punishment and crime rates go up.
Entorf and Spengler (2015) found that using policy to reduce conviction rates and less indictment led to increased crime.
What are the effects of imprisonment on crime rates?
Lower Crime Rates
- Incapacitation: a locked up criminal can no longer commit crimes in public and this
has a direct effect on reducing crime rates. - Deterrence: threat of punishment provides a disincentive to commit crimes.
The overall message of the literature is that deterrence works through both
increasing severity and certainty of punishment.
The effect of imprisonment on crime has been illustrated in some studies:
Deterrence
Levitt (1998) found that
since adults are punished harsher than minors, crime rates among older people drop more sharply.
Drago et al. (2009) finds that
a one-month increase in the prison sentence reduces the probability of committing a crime by approx. 1.3%.
Iyengar (2008) found that
the three-strikes reduces criminal activity in California by 28% for third strike eligible offenders.
The Three Strikes Law focuses on habitual criminals, greatly enhancing
their prison sentences for additional convictions.
Incapacitation
Owens (2009) found that the incapacitation effect is found to reduce crime. But
he also shows that
there is at least a slight net social gain (benefits>costs)
attributed to the incapacitation effect.
Vollard (2013) used the application of the sentence enhancement law to show
that
sentence enhancements reduced the rate of theft by 25%.
Is there a deterrent effect in war crimes?
Jo and Simmons (2016) argue that the International Criminal Court can also have a deterrence effect on states.
They find that the ICC has stronger deterrent effects on governments than on nongovernmental actors, such as rebels
What are the Risk Profile and Ambiguity challenges to Deterrence Theory?
- People have varying risk preferences
Three Profiles:
In economics we distinguish between three risk profiles:
A. Risk netural: Individual is indifferent between higher certainty or higher
severity.
B. Risk averse people: would prefer to go for the option with larger certainty,
less severity.
So they would be deterred by lower probability of detection and harsher
punishment.
C. Risk seeking people: would prefer to go for the option with lower certainty,
higher severity.
So they would be deterred by higher probability of detection and less
harsh punishment.
But it is very difficult to assess the risk preferences of offenders in practice. - People tend to be Ambiguity Averse
- The tendency to favor the known over the unknown
- The probability of punishment is often unknown or ambiguous, so we are deterred by unpredictability
Do potential offenders even know which actions are criminalised?
What is the problem with knowledge of law and deterrence?
Do potential offenders even know which actions are criminalised by law?
If they know what the law prescribes,
will this knowledge influence their choices at the moment of committing an offense?
- Crimes are often committed out of rage, under the influence or in groups,
causing people to not think about the consequences.
Will the percieved cost of non-compliance ever outweigh the perceived benefits?
Two elements: Probability and Severity
Probability
People have a tendency to perceive probabilities in an unequal way →
probabililty weighing. (Prospect theory)
People have a tendency to overweight smaller probabilities and
underweigh large probabilities.
People have a tendency to see very small probabilities as zero →
oversimplification of options. Which would mean that there is no
deterrence. (Prospect theory)
People have a tendency to be
overconfident, the certainty that people
express in their judgments tends to exceed the frequency with which those
judgments are correct.
Severity
- Discount Rate peope have a tendency to put more weight on present
actions than ones in the future.
-Hedonic Adaption means that people quickly adapt to changes and
return to their defaults. Thus, criminals may also do not perceive the
severity of crime as harsh as they adapt to it.
- Duration Neglect is a tendency of people to perceive outcomes and events not based on how they went and how long they were, but how they ended.
- Mental Accounting means that the outcomes are percieved and evaluated depending on context.
Criminals may fail to correctly consider opportunity costs of punishment
if they think about the outcomes in terms of ratios.
What biases are there in the Criminal Justice System and its effect on deterrence?
Police might have:
Confirmation bias: searching and interpreting evidence to confirm prior beliefs,
resulting in a high error probability.
Implicit bias: acting based on prejudice.
Prosecuters might have:
Self-serving bias: might interpret the facts (evidence) as favorable to them,
resulting in a higher perceived probability of a “win”.
Judges might have:
Anchoring effect: Higher requested sentence by a prosecutor may increase
the imposed sentence by a judge.
Contrast effect: Previously judged light case might increase the perceived
severity of the current offense, theft → murder, or other way round.
Thus, police, prosecutors and judges might act in an unpredictable way, meaning that a
criminal might not be able to perform their expected punishment calculations.
However, this might induce ambiguity, which in turn might have a deterrent effect by
itself
