Why Does No-one Understand Social Investment? Not Even Bill English

Jess Berentson-ShawTax and Welfare, Uncategorized

Its confusing, this Social Investment business, yet sometimes it sounds so very simple. For example Bill English made it sound very simple in the leaders debate when he said

“…we will go to the social investment ball here [for dealing with poverty]… …right around the country we are targeting right in on the families that need support…we will deal with them one by one if that is what it takes”

– Bill English, Newshub Leaders Debate, September 2017.

Funny thing is though he got it wrong. Wrong that what he was describing was a new idea, and wrong that the tools developed within the context of social investment allowed teachers and social workers to target in on individual families. So why does the man who essentially invented social investment “kiwi-style” not understand what is and is not possible with social investment in New Zealand? And what even is it?

You keep using that word. I do not think it means what you think it means

-Inigo Montoya, The Princess Bride

To be honest it is not clear what social investment actually means in New Zealand, because as an approach it has not really ever been defined. I will get to the problem with not having a clear definition for the overall approach in a bit. For now what we can say about social investment kiwi styles is that it is a collection of tools and ideas that have loosely come together under this broad heading “social investment”, which gets bandied about a fair bit. So if we can’t describe the overall approach, we can at least describe the tools.

Listen to speeches, read the documents produced by the Social Investment Agency and Treasury, talk to a few people, and three main aspects of social investment emerge: 1) The Integrated Data Infrastructure (IDI), 2) The actuarial liability approach (aka return on investment analysis) and 3) Targeting risk factors.

The Integrated Data Infrastructure – a good tool being used inappropriately?

The IDI is an excellent data tool, and is used to inform the latter two components of social investment. The IDI was built by Statistics New Zealand to join up multiple, separate government databases. The type of data in the IDI is administrative; tax, justice, health, and social development data is collected by agencies, as we, the people, interact with them. That means the data collection has not always been designed with a particular research question in mind, and suffers all the limitations of linked administrative data (there are many).

Importantly any analysis undertaken by researchers using the IDI cannot identify individuals in the system.

The IDI is a great source of data for researchers. However, while it has been termed ‘a supermarket’ of data, with the inference that this is the only information about people in New Zealand we need, it is more like a warehouse filled with just bananas; it won’t provide a week of balanced meals. When it comes to finding out about people’s lives and what works for people – multiple research inputs matter.

‘Actuarial Release’ using Return on Investment Analysis

The second component of SI is basically a type of return on investment analysis. The idea is that analysts try and calculate, using the data in the IDI, the return on government money spent (e.g. social housing) in terms of life outcomes (e.g. health, education, use of benefits). The idea being that, like cost-effectiveness analysis, people in government can identify best spend on services.

It is made to sound pretty inhuman when terms like “actuarial release” and “future liability” are used to refer to people using government services and their lives. For example, in reporting on the first better public service target last year the Ministry of Social Development describes the process of moving people off benefits (one assumes the intention is into work, but that is unclear) as such:

“The social investment approach identified that the previous Result 1 target only incorporated a small part of the future liability, because it only referred to the number of clients dependent on welfare. The target for Better Public Services Result 1 now combines both a 25% reduction in working-age client numbers on benefit and an accumulated actuarial release of $13 billion by June 2018.”[PDF]

Words matter here, and these are not good words to use about people’s lives to be honest.

As a concept it is not terrible (despite the language used), people in government should be able to tell what impact the things government do has on peoples lives, however there are so many limitations on the IDI: the data that is selected to represent “positive life outcomes” is riddled with individual values, the government interventions that are selected for analysis are driven by politics (i.e. why social housing and not Superannuation). As a tool it will take a long time to get right, and even then, like any tool, it is only as good as its users. The question is, are their better ways to figure out the impact of government interventions? For example proper randomised control trials of new government policies are used to much greater degree in other settings – kind of like testing the impact of social investment itself as an approach to social policy.

The final component of social investment kiwi style involves targeting.

Targeting of services to ‘at risk’ people– a mostly dead or all dead idea

There are three main types of targeting that get thrown about – and it gets pretty confusing. The first targeting is the targeting of children who are ‘at risk’ of becoming adults who draw heavily on certain types of government funds- e.g.; criminal justice, unemployment, health (superannuation gets excluded as a universal service- that is telling itself).

Using the IDI, some ‘predictive risk modelling’ has been done to see if it is possible to identify what characteristics present in children’s lives, put them at risk of poor outcomes as adults.

To be very clear this is not a process to identify individuals, it can never be used to do so and any indication by the Prime Minister or others that it can shows a real misunderstanding of how this type of analysis works.

Risk modelling data cannot locate or know individual children. The Treasury, who did some initial risk modelling work, has a lot to say about the dangers of using predictive risk modelling and linked data sets like the IDI to forecast costs and outcomes, or try and identify individuals and target services. This quote sums up their position nicely:
“Measuring risk is inexact”

Treasury 2016

Time and again I have heard terms like “zero in on the most at risk”, “targeting individuals at the most risk of poor outcomes” used in the context of social investment and the IDI analysis. The implication for people listening is that we can locate individuals or families using this risk data, but to try and do so would be very unethical given how imprecise the data is. For example, the risk modelling misses many children – about 35% of children identified as being at risk will go on to do just fine as adults, and over 50% of children who have poor outcomes as adults will have no or only 1 risk factor presents as children according to the Treasury [PDF]

“Inigo Montoya: Who are you?

The Man in Black: No-one of consequence

Inigo Montoya: I must know!

The Man in Black: Get used to disappointment”

  • The Princess Bride, William Goldman

The lack of precision with the risk modelling is also a problem for targeting specific issues – the second type of targeting we see. For example people in government have decided to target mental heath as an issue under the banner of ‘social investment’. It is very unclear how these areas of concern are selected- in this case politics probably has played a significant role – but we need a system and process that consistently identifies real need early and we need confidence that the data being used is representative of the entire population and the big problems.

The third type of targeting relates to the type of targeted service selected. For example if child maltreatment is the target area, which targeted programme to use, violence prevention, parenting skills or perhaps parenting employment? There is a narrow focus here – universal services are not even considered, even though they are very effective for addressing these types of issues. Rather, targeted programmes, notably with regard to families, are favoured.

Again this idea of ‘zeroing in’, or reaching into individual families and providing “wrap around services” pops up, and yet again the IDI cannot be used to do so. What can be used is the same system we have been using for years (not in any way unique to social investment) – caseworkers, police, social workers, etc. identify families ‘at risk’ from their work, and interventions or wrap around services to overcome their ‘situation’ are delivered to them.

Yet, time and time again, experience from in NZ and internationally has shown “wrap around services” for problem families are far more popular than powerful, the UK’s latest troubled families programme just reported another “predicted and predicable” failure for the families it was delivered to. This type of targeting is a mostly dead or all dead idea.

“Mostly dead is slightly alive. With all dead, well, with all dead there’s usually only one thing you can do.  Go through his clothes and look for loose change”

– Miracle Max, The Princess Bride

Unpeeling the onion

That is not to say that targeting never works- but it has to be done in the context of what we know already. Scientists have shown that it is the conditions that we live our lives in that have the biggest impact on families’ choices, and ultimately on children’s wellbeing. We must unpeel the top layers of the onion, working properly on the wider social and economic conditions first, with targeting being a cherry on top, when we know we have got everything else right. Most importantly this targeting must be led from people and communities themselves. More top down “you are broken, you need to be fixed” attitude will lead to more misery in social service provision.

The Determinants of Wellbeing & Health

Source: Dahlgren & Whitehead 1991.

Social Investment Kiwi style has no vision for people and that is a big problem

Despite what individual agents operating within the social investment area think and say and do about caring for people, ultimately, as an overall approach, social investment ‘kiwi-style’ lacks a vision for people’s lives. There is little clarity about the outcomes that matter to people, and little precision about what this policy is setting out to achieve in the wider government social policy context. So, into that unoccupied space, have crept the values that have informed our social policy for thirty or so years: individuals as a costly problem who need to be fixed.

“The welfare system in New Zealand is increasingly oriented around the need to restructure individuals, rather than systems” Kingfisher (1999)

We can see in the social welfare working group papers from 2011 the language of cost reduction and people as liabilities that need to be dealt with.

“The delivery agency needs to be accountable for reducing the forward liability and the associated reduction in long-term welfare dependency”

– Social Welfare Working Group, 2011

Professor Ian Shirley, who has worked in social policy for over 30 years, argues that this language – of individuals as the problem that need to fixing- and the dismissal of the importance of addressing the conditions of people’s lives, pre-dates even the Royal Commission on Social Security in 1972.

If we do not make the values clear at the outset of a policy approach and seek to achieve outcomes in line with those values, then inevitably people in government will do what has always done, for the same results, all the while genuinely thinking what they are doing is something new.

Can social investment be something good for people and government?

This week at the social investment summit I will discuss an alternative model of investing in people, one that includes a positive vision for people’s lives, with the values of care, manaaki and trust at its heart. It is a model that does not narrowly focus on an old idea of ‘what works’, but can be used to move us to a truly innovative culture of experimentation in government.

It is a model that gives communities the tools to really find and prove their own solutions work, while similarly giving people in government a very clear and precise idea of how government interventions are impacting on people’s wellbeing.

With people centred vision and values, “what works?” to achieve a cost target, can become instead, what can we try, what can we support, what can we experiment with in an inclusive way, to enable a thriving life for all? Call it social investment if you like, but we can probably come up with a better name.

Dr Jess Berentson-Shaw will be talking at the CPAG Social Investment Summit, with Associate Professor Susan St John, Dr Bill Rosenberg, Dr Simon Chappel, and others at Auckland University on Friday the 7th of September.

Why Does No-one Understand Social Investment? Not Even Bill English was last modified: March 1st, 2022 by Jess Berentson-Shaw
About the Author

Jess Berentson-Shaw

Dr Jess Berentson-Shaw is a science researcher working for the Morgan Foundation. Jess holds a PhD in Health Psychology from Victoria University. Jess has over 10 years’ experience working on applying science and evidence to public policy. She worked on improving the use of science in public health practice in NZ, before working as a Research Fellow at University College in London, where she researched how doctors and clinicians translate scientific evidence into their clinical practice. While in the UK she also developed a national data collection system, which was used to determine what factors contribute to poor outcomes for women and babies during pregnancy and birth. On her return to New Zealand she directed a research group that specialised in the independent evaluation and application of research and science to health policy and practice. Jess loves science and what it can do to make the world a fairer place.