Date posted: 23/06/2017 3 min read

Fighting crime with statistics

PwC has created a big data model for the New Zealand Ministry of Justice that aims to identify youths at risk of offending so they can be set on a better life path.

In brief

  • The New Zealand government's decision to use data to influence social policy is a more recent step to help disadvantaged youth.
  • This data modelling is being introduced with the support of PwC and a team of Australian actuaries – led by Emily Prior, an affiliate member of Chartered Accountants ANZ.
  • The Whole-of-Population Micro-Simulation Model will allow the government to chart every resident New Zealander from birth.

By Steve Lewis, photography by Mark Mitchell.

While its title would be right at home in the newspeak of George Orwell’s 1984, New Zealand’s Whole-of-Population Micro-Simulation Model is already making a difference in tackling serious crime and reducing the prison population.

In the latest rollout of New Zealand’s so-called social investment approach, PwC is working closely with Justice Minister Amy Adams and her Department to target those considered most at risk of committing serious crime.

While the use of data to shape economic policy has been commonplace, the influence of statistics on social policy is a more recent phenomenon.

Harnessing private capital for public good

An inside look at a global movement that is harnessing private capital and new solutions for the public good.

And Adams is a firm believer that the vast integrated data infrastructure (IDI) at the government’s disposal can be used for positive social outcomes.

The new approach uses data to identify particular groups of New Zealanders – including the very young – who are most likely to offend, with the aim of targeting schemes to try and pre-empt their spiral into a life of crime. Adams says the government has recognised that the crime and justice system is a much bigger pipeline that starts at birth, “some would say pre-birth”.

“And that yes, we can do things to deal better with offenders and to work better on rehabilitation and reintegration into society, and drug and alcohol treatment.”

She also identifies potentially significant economic gains that can be made from early identification and targeted intervention.

“Increasingly, we are talking a language of social investment which is really about saying ‘look it’s not rocket science but if we can use data to understand exactly who we need to target and with what interventions and in what way, then the cost of those interventions is significantly dwarfed by the value of the savings that are created from that’.”

Accounting for crime

“Until now, we really have not had the robust data modelling and actuarial basis to strongly predict and build those models to a level that supports a Treasury-style business case,” says Adams.

This data modelling is being introduced with the support of PwC and a team of Australian actuaries – led by Emily Prior, an affiliate member of Chartered Accountants ANZ – which is making the regular trek across the Tasman.

This allows the government to start planning on ways to prevent young New Zealanders from embarking on a life of crime.

Prior, who heads up PwC’s Criminal Justice Practice, says the IDI developed by Statistics NZ is among the best and most comprehensive data warehouses in the world.

“Which explains why you have a bunch of Aussie actuaries who are desperate to spend time in New Zealand,” she adds.

One year into its four-year rollout, the Whole-of-Population Micro-Simulation Model will allow the government to chart every resident New Zealander from birth and provide a comprehensive framework of an individual’s welfare, health and education record.

While the community is rightly concerned about privacy, Prior is adamant that strict safeguards are in place.

For example, Statistics New Zealand gives someone a unique ID that cannot then be linked to their name or date of birth or anything that might reveal their identity.

This anonymised data is instead used to try and identify cohorts who could be at risk of being criminalised.

“So we might say to the justice sector that we have identified 735 girls between the age of 15 and 16 who have been truant from school more than three times in the last year, they happen to be largely in Auckland and some in Wellington. And then Justice will share this information with, say, the education sector who will then put a programme in place and identify those individuals,” Prior says.

“What we are not doing is letting people like me run around with the names of individuals and have any individuals discussed and that information shared across agencies.”

It will also be used to track those most likely to embark on a life of crime, utilising the database to track individuals, allowing authorities to hopefully intervene before it’s too late.

“If you want to pick up someone in the system with a gang affiliation, I can see if they have ever had a job, if they had interaction with child services when they were young, if they have been on welfare, if their mother suffers from a mental health problem – so anything that you can imagine about someone that is tracked in government data, I can see that,” Prior says.

“Which means we can build up a very rich model and it helps us to distinguish the different life courses.”

The matrix

Minister Adams is an enthusiastic supporter of using data to improve justice outcomes, describing the model as being like a “predictive matrix” that provides insights into those most at risk of offending.

This allows the government to start planning on ways to prevent young New Zealanders from embarking on a life of crime.

“We can then say with some confidence that actually the single biggest investment that we could make for a particular cohort of, say, 14- to 16-year-old young offenders who have the highest future crime prediction is maybe an investment in cognitive family therapy, either delivered through the social welfare department, the schools or health system,” she says.

The data framework is also able to help judges and magistrates impose penalties that will have a more positive outcome.

“Interestingly and somewhat counter-intuitively, we have learned that a sentence of a fine leads to a lower prospect of re-offending and a lower prospect of future benefit dependence than sentencing the same person to a community-based sentence.”

Adams displays a good grasp of statistics as she outlines how particular cohorts could benefit from the social investment approach.

For instance, there are 1,800 New Zealanders between ages 14 and 16 who have committed a serious offence such as assault with intent in the past two years. These young people have already been dealt with by police for six offences each, on average.

If nothing changes, the statistical model shows these same young New Zealanders will commit ten further offences each over the next 30 years, including 2,800 serious property offences and 1,600 serious violent offences.

Risks of profiling

While these statistics make for sobering news, so does the fact that 65% of this cohort is Māori – and that raises questions about whether this data modelling could lead to bouts of negative racial profiling.

The Minister does not shy away from the issue when challenged.“It is a really interesting matter and one we have to be cognisant of, but equally we are very open about the fact that one of our core challenges is that we have a significantly high representation of Māori in the justice system,” Adams says

“We are quite public about it, that Māori are over-represented in the justice system. But the point we always make alongside that is they are over-represented both as offenders and as victims.

“Equally we have to be clear that we can’t simply paint these things as simply being a problem in poorer brown communities.”

Adams says similar data is being used in some countries to actually address what she calls “subconscious racism or other types of stereotyping in the justice system”.

“In America, for example, they are using a number of bail and sentencing matrices like this, or calculations like this, to feed in information to ensure that the characteristics of the defendant are not subconsciously affecting the decision maker’s views. Yes you have to be very careful because you don’t want to weight it so that if they are Māori, they are therefore more likely to be high-risk.

“But if you set it up carefully, they [matrices] can be a way of addressing subconscious bias because you are only entering the factors that are relevant to the actual situation – the risk profile of that particular offender, the criminal history of that particular offender… and ensuring that any stereotyping doesn’t enter into it.”

Asked if she has particular ambitions or targets in terms of reducing the level of crime, the Minister replies: “Fundamentally the goal is for these young people to have much better lives than the current trajectories have them on.

“And when you start thinking of them as three- and four-year-olds and houses where family violence is prevalent, where drugs and alcohol are prevalent, where they are not being encouraged to turn up to school regularly, it is a pretty awful outlook and life for them, quite apart from the fact that they are likely to end up as criminals in the justice system and create victims of their own.”

Steve Lewis is a journalist and senior adviser with Newgate Communications.