Read Gluckman’s report on crime with scepticism
Gluckman report on the use of evidence in the criminal
justice system has become the go to document for the new
government and the media in recent months but it failed a
basic sniff test. This report in March by the chief
scientific advisor to the Prime Minister was too good to be
Professor Gluckman claimed that all the evidence
showed that tough on crime policies do not work. No nuances
here, not a two-handed economist to be seen. As a bonus,
Gluckman found that some easy to adopt early interventions
were available to significantly reduce incarceration rates
with no risk to the public of increased crime. Finland was
his model for interventions that reduced prison populations
without increasing crime rates. What more would a prison
Every social science literature has mixed
That tough on crime policies do not work fails the
sniff test that if a social sciences literature is said to
have so clear-cut empirical results, then PhD students and
academics are just not working hard enough to win promotion
and research grants. Academic careers are built on raising
Academics must publish or perish. Journal editors
publish new or conflicting findings. There is a relentless
thirst for the new and the controversial. No academic wins a
PhD or promotion or even survives an annual performance
review by replicating the results of others in the social
Does the death penalty deter?
In 1983, Edward
Leamer shook the foundations of econometrics so hard that
young economists despaired for the possibility of a fruitful
career in applied economics. In Let’s Take the Con Out
of Econometrics, Leamer showed that competing
researchers using slightly different but still plausible
approaches to the causes and effects of crime and punishment
had shown that the death penalty either greatly reduced the
murder rate or increased it and anything in between.
death penalty literature has not progressed far beyond that
ambiguity of the 1980s. But my point here is there are still
plenty of warring econometricians making reputations and
careers by casting doubt on the earlier findings be it that
the death penalty works or does not work. We are talking
honest disagreements about research design. We have not even
got to data mining and publication bias: too many
researchers publish only those results that suit their
If a policy works, why wasn’t John Key
already all over it?
The 2nd sniff test that Gluckman
fails is John Key, Tony Blair and their ilk are always on
the lookout for a policy that reduces crime. Blair was
elected on being tough on crime, tough on the causes.
the early intervention and rehabilitation programs
championed in the Gluckman report worked, John Key and Tony
Blair would have implemented them years ago. John Key, Tony
Blair and Bill Clinton were notorious triangulators
interested in whatever works to get them re-elected. There
is little under the policy sun that is new or has not been
tried and tried again.
We are not America
cited by Gluckman on childhood interventions to reduce
criminal offending were from America. Professor Gluckman,
for example, noted that 65% of the payoff from the Perry
preschool scheme established in the 1960s was from reduced
Much of the payoff from the Perry
preschool program and similar US interventions were from
fewer murders in a country with a far higher murder rate
than ours. Many of the children in the pioneering early
childhood interventions grew up in the crack cocaine
epidemic which doubled murder rates.
There is an
intersection in the 75th precinct of the New York Police
Department known as the “4 corners of death”. The 75th
precinct did not have a murder in the first 4 months of this
year; there were 11 last year. Back in the early 1990s, when
the crack cocaine epidemic peaked, there were over 100
murders a year in the 75th police precinct. Countries differ
in important respects.
Do you want to bring an early
intervention to New Zealand from the inner cities of the US
and expect fewer murders when we have so few street crime
murders and stranger murders to start with? An astounding 26
percent of black males in the US report seeing someone shot
before turning age 12; and a total of 43% by their 18th
birthday. This is two to three times the average for
American boys. On average, black males report hearing one
gunshot per week in their neighbourhoods, more than twice
the average for white males. Import American social findings
Europe has overtaken the US in crime
Gluckman’s strong finding that tough on crime
policies, longer and longer prison sentences, do not reduce
crime is easy to tear down if you look on each side of the
Atlantic, and not just in Finland.
It is well-known that
murder rates and crime in general in the USA halved since
the early 1990s. What is less well known is violent and
property crime have increased so much so in Europe despite
their generous welfare states that crime rates apart from
homicide are worse in Europe than the USA. London recently
overtook New York City in murder rates despite Americans
having plenty of guns and drug gangs armed to the teeth.
London had a knife crime explosion.
Galbiati and Zanella (2011) found that in 1970, violent
crime in Europe was 62% of the corresponding rate in the US.
By 2008, violent crime aside from homicides was more than
twice the US figure. The European property crime rate in
2007 was 20% above the US rate, while in 1970 it was
one-third of the US. Europe is no longer a model of low
Crime waves follow prison population
Fortunately for Professor Gluckman, better
evidence than Finland is available on the impact of reduced
incarceration rates on crime. Nine Italian mass pardons
allow us to test precisely what happens when we reduce
The most recent Italian mass pardon
was in 2006; a surprise response to prison overcrowding that
released 1/3rd of inmates on condition they will serve the
remainder of their pardoned sentence if they reoffended.
Some of the released inmates faced as little as one month
extra to serve if caught again, others up to 3 years if they
reoffended. Did the released Italian inmates with the longer
commuted sentences hanging over them pending good behaviour
Buonanno and Raphael (2013) found sizable
increases in crime after the 2006 Italian mass pardon. Each
prison-year served prevented between 14 and 46 crimes
reported to the police, mostly theft and to a lesser degree,
robbery. The mass pardon may have led to a slight increase
in violent crime.
There were eight earlier Italian
collective pardons between 1962 and 1990 releasing up to 35%
of the prison population. Barbarino and Mastrobuoni (2014)
found that the post-pardon crime waves suggested a reduction
of total crime with respect to more imprisonment of 17 to 30
percent. (Same thing after school teachers’ strikes. There
are juvenile crime waves when schools out all day).
criminal justice review by the Obama White House found a
smaller number than for Europe; a 10% increase in
incarceration rates reduces crime by 2% or less. The Obama
White House said the crime reduction from longer prison
terms was lower because the USA sends criminals away for so
long already that diminishing returns set in. More European
prisoners finish their sentences while they are still young
and wild rather than more mellowed-out middle-aged or older
prisoners as is the case in the USA because of their
savagely long prison terms.
The French and Korean
presidents grant mass pardons for speeding and parking
offences in the six months to one year before their
election. Naturally, in the months prior to this anticipated
pardon, it is more dangerous to drive in Korea and France.
The 1995 vote won by president Jacques Chirac is blamed for
an extra 300 road deaths. An acquaintance who worked in
Paris saved up his parking tickets in anticipation of the
pardon. When 35% of the Oregon State Highway Patrol were
laid-off, road deaths and injuries rose by 10-20%.
Incentives work on lawbreaking, great and small.
argument against 3-strikes is deterrence works
reformers are at one with the tough on crime crowd when it
comes to the effects of 3-strikes and you are out. Tough on
crime advocates emphasise the incentive not to offend in the
first place. Prison reformers stress the other side of the
same coin; if a crime is committed, the offender has nothing
to lose on their 3rd strike from greater violence and
killing witnesses and then nothing to lose once they are in
prison because they cannot be sentenced to further time.
What was found in the USA was offenders serve the same
25-to-life anyway so third strikers were more violent. But
if offenders had something to lose, if their punishment was
more tailored to their latest crime, they chose to be less
Criminals think CSI is real
scares many criminals straight. The reduction in reoffending
is far greater than any reasonable impact DNA science could
be said to have had on police work. DNA science certainly
helps with current and old cases but rarely does a
prosecution rely solely on a DNA identification. Prosecutors
have even introduced expert witnesses to remind juries to
not expect scientific evidence of the calibre they see on
CSI and the poetic license on TV about forensic
DNA profiling was rolled out slowly over the 50
American states for the convicted and then in 30 American
states for arrestees. States initially swabbed serious
violent offenders, then later added robbers, burglars, and
more minor offenders. This phased roll-out by state and by
the seriousness of offences left plenty of variation in the
data to tease out the effects of DNA profiling.
on reoffending rates by prisoners released before and after
a state introduced DNA swabs was startling. Doleac (2017)
found that DNA profiling reduced the probability of future
convictions by 17% for serious violent offenders and by 6%
for serious property offenders. Her later study of Danish
DNA profiling showed similar large drops in reoffending. In
another study, Doleac found that an extension to daylight
saving reduced robberies by 7%. Criminals prefer the
Criminals mind getting arrested
sociologists and criminologists unite on the notion that
criminals do not like to have their collars felt; offending
drops when the probability of arrest is higher. Criminals
have a lot of trouble weighing up the deterrent effect of
longer terms in prison because of illiteracy, addiction,
mental illness and low IQs. Deterrence works because
criminals hate getting caught by police. Their sentence if
convicted is hazier in the criminal mind’s calculations
and impulse control.
The review by the Obama White House
concluded that a strong empirical result is a 10% increase
in police hiring reduces crime by at least 3% and sometimes
much more. Police tactics such as hotspot policing work well
but a broken windows policy of cracking down on petty
offences is more dubious.
But there is still growing
evidence in studies by Steve Machin that robberies and
burglaries bend to factors ranging from recent harsher
sentencing to rises and falls in the price of petrol, scrap
metal and jewellery. Crime fell when long sentences were
handed down in the aftermath of the 2011 London riots. Not
only did robbery and violent disorder drop in the six-months
after the 2011 riots, other crimes increased because
sentences had not stiffened for not riot-type crimes.
Criminals are also put off their calling by the spread of
security devices. Car thefts about halved when the Lockjack
car immobilisation device spread. Dutch burglary rates
dropped by one-quarter when burglar-proof doors and windows
were made mandatory for new houses. Most of the fall in
crime was for houses and cars with no extra security.
Criminals just did not bother in the first place.
reoffending in the age of three-second
Soundbites in the 21st century leave no room
for nuance. No talk of on balance and on the other hand like
the two-handed economist. These days, snappy messages that
all the evidence supports you and no evidence supports your
opponent is all the journalists give you time to say.
prison muster is always too small because there are unsolved
murders, rapes and robberies. It is only too high in the
limited sense there will be a few prisoners who were wrongly
convicted and yet to be cleared on appeal. In addition, 20%
of prisoners on remand are not convicted at trial. Remanded
prisoners who are convicted have the time served taken off
their sentence so stricter bail laws just bring forward the
same time behind bars, so they do not increase the prison
Explaining crime rates in your own country is
challenging enough. Cross-country differences are a puzzle.
Crime has fallen so quickly in the US that Europe is now
more crime ridden. Crime is low in Scandinavian with their
nicely-appointed prisons and even lower both in Japan which
has petrifying prisons and in Singapore which uses the Rotan
to whip you back literally to the straight and narrow.
Singapore had over 1,200 canings with the Rotan in 2016,
down from 6,000 per year 10 years prior.
Work both ends of
making crime not pay
Rising reoffending rates are bad
because there are more new victims of crime. Too many lose
their moral compass to go on about the unfortunate criminal
going back to prison and forget their victims.
prison sentences reduce crime. Teaching prisoners to read
and cope with their addictions reduces reoffending rates.
These policies work at opposite ends of the same decision
calculus; they make crime less rewarding because longer
prison sentences deter, and literacy and addiction
programmes make a life of crime less tempting because other
more legitimate options have opened.
policy for thee but not for me
Genuine criminal justice
reformers must be willing to swallow what are for many of
them are a few ideological dead-rats. Both 90-day trials for
recruits and charter schools give another chance to
teenagers and adults who have fallen off the rails.
Evidence-based policy is for everyone. You cannot expect
others to give up cherished beliefs if your own are equally
immune to inconvenient evidence.
Jim Rose is an economic