Transcript The Fugitive

The Fugitive
Evidence on Public versus Private Law Enforcement from
Bail Jumping
Eric Helland and Alexander Tabarrok
Introduction
•
Most felony defendants are released before trial.
•
One quarter of all released felony defendants, some
200,000 a year, fail to appear at trial.
•
About 60,000 felony defendants a year skip town and
disappear for a year or more.
•
Defendants who fail to appear impose significant
costs on others.
• Types of Pretrial Release
– Own recognizance
– Surety bond
– Cash Bond
– Public bail (called deposit bond)
• We consider each form of release to be a “treatment”
and evaluate the effect of each treatment on the failure
to appear rate, the time to recapture and the fugitive
rate.
The Attack on the Commercial Bail System
“Bail bondsmen are a cancer on the body of criminal justice.”
Judge William Snouffer
“The commercial bail system is ‘offensive’ and ‘odorous.’”
Supreme Court Justice Harry Blackmun
The American Bar Association says “the commercial bond business
has been one of the most tawdry parts of the criminal justice
system.” And the ABA suggests that "The central evil of the
compensated surety system is that it generally delegates public
tasks to largely unregulated private individuals.”
American Bar Association, emphasis added.
“The present system of commercial surety bail should be simply and
totally abolished….It is not so much that bondsmen are evil –
although they sometimes are – but rather that they serve no useful
purpose.”
U.C. Davis law Professor, Floyd Feeney
Arguments Against Commercial Bail
• Unlike cash bond the fee for a surety bond is sunk, for
this reason it's often asserted that the commercial bail
system discourages appearance.
• Supreme Court Justice William O. Douglas argued that
the commercial bail system fails "to provide an incentive
to the defendant to comply with the terms of the bond.“
• Drimmer (1996, 742), says "hiring a commercial
bondsman removes the incentive for the defendant to
appear at trial."
Incentives to Appear
• The arguments misses the role of other
incentives.
• Lenders generally create incentives to
encourage repayment
– Collateral on loans
– Lenders create a “circle of responsibility” people who
have the same incentive as the bondsman to get the
defendant to court. E.g. cosigning by family members.
– Bond dealers will also monitor their charges and
require them to check in periodically.
• Thus it is an open question how these incentives
compare to the incentives under other systems.
Incentives for Recapture
• The “treatments” differ especially in how they respond to
those who abscond.
• Under own recognizance, deposit bond or cash release
the defendant who skips town is pursued by the police.
• But if you are released on commercial bail you are
pursued by…
Bounty Hunters
•
If you are released on commercial bail you are
pursued by the surety and his or her agents, “bounty
hunters.”
–
–
–
•
If the defendant is not returned within a short period of time
(typically 90-180 days), then the bail bondsman forfeits the
bond.
Bail bonding is competitive-95% of clients must show up to
turn a profit.
Bail bondsman has strong incentives to track down skips.
In some ways bounty hunters have greater powers
than the police because legally the bail bond agent
stands in for the jailer and thus inherits the rights of the
jailer.
Bounty Hunters versus the Police
• Public police bureaus are often strained for resources
and the rearrest of defendants who FTA is low priority.
• The flow of arrest warrants for failure to appear has
overwhelmed many police departments so that today
many counties are faced with a massive stock of
unserved arrest warrants.
• California has the largest backlog of arrest warrants in
the nation. The California Department of Corrections
estimated that as of December 1998 there were more
than two and a half million unserved arrest warrants
including 2,600 outstanding homicide warrants.
• Howe and Hallissy (1999) report that "local, state and
federal law enforcement agencies have largely
abandoned their job of serving warrants in all but the
most serious cases."
The Data
•
The data is a random sample of felony filings from
approximately 40 jurisdictions designed to represent the 75
most populous U.S. counties.
•
The data covers 58,585 felony defendants from 1988, 1990,
1992, 1994, 1996
•
The data contains detailed information on:
arrest charges
the criminal background of the defendant
sex and age of the defendant
release type (surety, cash bond, own recognizance etc.)
rearrest charges for those rearrested
whether the defendant failed to appear and if so whether the
defendant was still at large after one year
–
–
–
–
–
–
Table 2: Mean FTA Rates by Release Category, 1988-1996
Own
Recognizance
Own
Recognizance
Deposit Bond
Cash Bond
Surety Bond
Emergency
Release
26%
[20,944]
Deposit
Bond
Cash
Bond
Surety
Bond
Emergency Release
5%*
5%*
9%*
-19%*
21%
[3,605]
1%
4%*
-23%*
20%
[2,482]
3%*
-25%*
17%
[9,198]
-28%*
45%
[584]
Mean FTA rates for release categories are along the main diagonal with the number of
observations in square brackets. Off diagonal elements are the difference between the mean
FTA rate for the row category and the mean FTA rate for the column category.
* Statistically significant at the greater than 1% level.
In Search of Schrödinger's Defendant
•
One reason for lower FTA rates may simply be selection. Bail bondsmen "cream-skim" for those least likely to FTA.
•
•
Consider the simplest case where i denotes a particular individual
and there are two treatments. The treatment effect is - and
the average treatment effect
but we only ever observe (treated) or
(untreated).
•
Necessity of finding “comparables.”
–
–
•
Random assignment
Weaker assumption is that treatment assignment is random
conditional on some Xi
This suggests an intuitive estimator – match individuals across the
conditioning variable.
Matching on Propensity Score
• The matching method will quickly run into a problem because as we
expand the number of matching variables the number of matching
categories expands exponentially.
• Rosenbaum and Rubin showed how this problem could be
overcome. What they showed is that if treatment is independent of
outcome conditional on X
• then it is also true that:
• where p(X) is the probability that T=1 | X also called the propensity
score.
• Rosenbaum and Rubin showed that you could shrink the multidimensional problem of matching on many categories into a single
dimensional problem of matching on the probability of treatment.
Intuition on Propensity Score Matching
• We would like to have random assignment which means
that the X would be distributed identically across the two
treatments.
• What matching on the propensity score does is to
rebalance the sample so that X is balanced across the
treatments and thus we recreate a sample that looks as
if it were drawn randomly.
• The matching technique extends naturally to applications
with multiple treatments through the use of a multivalued propensity score with matching on conditional
probabilities (Lechner1999, Imbrens 1999).
Generating P-Scores
•
•
•
We use an ordered probit model which includes the crime the defendant is
alleged to have committed, whether the defendant had previous experience
with the criminal justice system including whether the defenant had failed to
appear in the past. Note that these are the sorts of variables that judges
used to make their treatment assignments.
Idea is that we will create an as if “random sample” based on these
characteristics for each treatment group.
We then match individuals across the treatment groups according to their pscores using a caliper of .0001.
Treatment Effects of Row versus Column Release Category on FTA
Rates using Matched Samples, 1988-1996
Own
Deposit
Recognizance Bond
Own
Recognizance
Cash
Bond
Surety
Bond
26%
Deposit Bond
-3.1*
(-12%)
21%
Cash Bond
-5.8*
(-22%)
-1.5
(-7%)
20%
Surety Bond
-7.3*
(-28%)
-3.9*
(-18%)
1.7
(8.5%)
17%
Mean FTA rates for release categories for the full sample are along
the main diagonal. Off diagonal elements are the estimated treatment
effects of the row category versus the column category.
* Statistically significant at the greater than 1% level (two sided).
Estimated Treatment Effects: The Fugitive
•
Not every failure to appear is the result of an attempt to escape
justice.
•
In some cases, a defendant fails to appear because he forgets
his trial date, gets sick, or is already in jail.
•
A telephone call usually resolves these issues and the defendant
is back on trial within a couple of weeks.
•
Commercial bail is most different from other release types
including pretrial release bureaucracies in tracking down
defendants who have purposively skipped town.
•
Recapture of purposeful skips is where we would expect to see
the largest effects from surety release.
•
Once a defendant has absconded, the efforts of the bail
bondsman relative to the police are what determines differences
in results.
Treatment Effect of Row versus Column Release Category on the
Fugitive Rate using Matched Samples, Conditional on FTA, 1988-1996
Own
Recognizance
Deposit
Bond
Cash
Bond
Own
Recognizance
32%
[5440]
Deposit Bond
-.2
(.6%)
33%
[766]
Cash Bond
11.9*
(37%)
-3.8
(-11%)
40%
[506]
Surety Bond
-17*
(-53%)
-15.5*
(-47%)
-25.6*
(-64%)
Surety
Bond
21%
[1537]
Mean fugitive rates, defined as FTAs that last longer than a year, for
release categories for the full sample are along the main diagonal with the
number of observations in that category conditional on an FTA in square
brackets. Off diagonal elements are the difference between the mean
fugitive rate for the row category and the mean fugitive rate for the
column category estimated using matching.
* Statistically significant at the greater than 1% level (two sided).
** Statistically significant at the greater than 5% level.
** Statistically significant at the greater than 10% level.
Effect of Alternative Treatment versus Surety Bond on FTA and Fugitive Rates
(conditional on FTA) Matching Individuals from States that have Banned Surety Bonds
with Similar Individuals Released on Surety Bond, 1988-1996
Own Recognizance v. Deposit v. Surety
Surety Bond
Bond
Cash v. Surety Bond
Treatment Effect on
FTA Rates
+7.8*
(30%)
+6.2*
(23%)
-1.6
(-6%)
Treatment Effect on
Fugitive Rates
+14.8*
(78%)
+19.8*
(85%)
+35.7*
(93%)
Standard errors are in parentheses.
Matching Caliper=.0001
* Statistically significant at the greater than 1% level (two sided).
** Statistically significant at the greater than 5% level.
*** Statistically significant at the greater than 10% level.
Kaplan-Meier Estimation of FTA duration
• We estimate a Kaplan-Meier model of the duration of an
FTA spell.
•
The analysis is conducted on days at large and is
truncated at one year.
• The Kaplan-Meier estimator is a non-parametric estimate
of the survivor function: the probability that an individual
survives in the current state. Note, however, that we
compare matched samples so our KM estimator controls
for covariates even though it is non-parametric.
Estimation of Time to Recapture
Looking for Unobservables
•
•
6 robustness tests in the main paper
controlling for a variety of fixed effects,
information observed by judge but not
the econometrician, as well as general
unobserved factors.
Results are consistent across a wide
variety of specifications.
A “Twins” Study
• We look at individuals who were released on their own
recognizance the first time they got arrested and who
were then arrested again.
• We then compare the fugitive rates on the first charge for
those who got own-own with those who got own-surety.
• The reasoning here is that if the bond dealers make the
defendant stick around for the second charge then this
creates a unintended benefit on the first charge – the
defendant can't skip town on one charge but not the
other.
• At the same time we are comparing individuals both of
whom were released on their own recognizance – thus
who should be especially comparable.
Table 12: Unconditional Fugitive Rates by Arrest-Rearrest Category, 1988,1990
Fugitive
Rate
1) Own and Not
Rearrested
2) Own-Own
3) Own-Surety
8.48
[17,828]
8.04
[191]
1.49
[134]
Own-Own indicates first release on own recognizance and second release on
own recognizance. Own-Surety indicates first release on own recognizance,
second release on surety.
Bounties for Justice: New Directions
• The success of the commercial bail system suggest that
we ought to look for other areas to apply the bounty
concept.
• Privatization of Warrant Service
– Recall that there are 2.5 million unservered arrest warrants in CA
– many for serious charges.
– Put a bounty on serving these warrants.
• Parole Bonds
– FTAs and Fugitives are common for parolees and probationiers
just as for felony defendants. Some 10% of probationiers have
absconded.
– Probation and parole officiers are strained and understaffed.
– Felony defendants can be required to pay bail to be released.
Why not require probationiers and parolee’s to pay parole bond?
Summary of Results
• Defendants released on surety bond are 28
percent less likely to fail to appear than similar
defendants released on their own recognizance.
• If they do fail to appear, they are 53 percent less
likely to remain at large for extended periods of
time.
• Deposit bonds perform only marginally better
than release on own recognizance.
• Requiring defendants to pay their bonds in cash
can reduce the FTA rate but conditional on FTA
the fugitive rate is much lower on surety bond.
• Given that a defendant skips town the probability
of recapture is much higher for those defendants
on surety bond.
• The bounty concept ought to be extended to
privatized warrant service and probation and
parole bonds.