The New Zealand government’s ambitious commitment to reduce child poverty by half over the next decade is challenging agencies to deeply consider new approaches to empowering front line workers, while dealing with the underlying drivers of poverty – especially housing, education and employment.
These approaches include, a sharp focus on the needs and situation of impoverished children, embracing design thinking to develop new services, using technology to give caseworkers the tools and intelligence they need, and strongly embracing data-driven approaches to measuring success and identifying the broader causes of poverty.
A central delivery focus has already been created through the establishment of Oranga Tamariki – Ministry for Children. But the ambition to cut child poverty by half –with major statutory reportable milestones to be regularly meet– represent a major step up in ambition and demands new approaches and thinking.
At a recent seminar these approaches were considered by a panel of experts and discussed with New Zealand officials from a variety of government agencies.
The panel included Professor Jonathan Boston from the School of Government at Victoria University Wellington, the deputy chief executive for Insights and Investment at the Ministry for Social Development, Nic Blakeley, the lead for the cross agency Service Innovation Lab, Pia Andrews (nee Waugh), deputy chief executive of Safety of Children in Care at Oranga Tamariki, Jen Ritchie-Campbell, and IBM’s managing partner New Zealand, Bettina Baer.
Understanding the policy problem
In a presentation detailing the complexity of the policy challenge, Boston said there is no single correct measure of poverty. There needs to be a variety of measures to highlight different features of this multi-dimensional problem.
He noted developed countries use two main poverty measures: income below a certain agreed threshold and material deprivation – missing out on more than a certain number of essential goods and services. Boston observed these metrics generate different results and only partially overlap.
Boston told the seminar child poverty and material hardship can be reduced, but large reductions will be costly fiscally and will require significant policy changes; material hardship rates will be more difficult to reduce than income-based measures.
Increases in family benefits announced in the recent New Zealand Budget would begin to address the long term decline in income support which had been capped since the early 1990s.
Boston also observed that European evidence showed there needed to be cross political support for the overall strategy to succeed. Sweden had seen strong improvements in child poverty off the back of a whole of government approach supported by the main political parties. The United Kingdom had seen their efforts be less effective, after changes in priorities by different governments.
Boston noted material deprivation is much more multi-faceted and created a powerful reason to redesign services around the material needs of the children and their families and carers – rather than government system needs. The methodologies around this citizen driven approach are being incubated through the New Zealand government, with Internal Affair’s Service Lab providing a rapid and impressively pragmatic approach to relentless problem solving.
Child led policy and service design
The seminar was told the key to this forensic approach is clearly identifying the children in need and precisely understanding the context of their situation so that front line workers can fully respond to their real needs.
Empowering front line case workers was considered key to any reforms. The seminar heard observations that deep engagement with social workers and front line responders had been a powerful lesson for other Australasia jurisdictions – such as NSW – looking to implement large scale reform of child level human services.
It was noted that it was important to bring together all the relevant players – government and non government – to understand these needs. Many of these players, like community service providers, have deep knowledge, experience and wisdom about the challenges impoverished families are facing. Capturing the insights of case workers about the practical needs of children and their carers would also ensure any new approaches are developed from a bottom up perspective.
This was considered a critical paradigm change from previous initiatives around child poverty, where the programs had been typically top down, often designed through a central office policy lense, rather than the detailed needs of end users.
Through the Ministry for Social Development (MSD) there is already interesting work being done to ensure their clients (and by extension their children) are aware of the various benefits and support available, through a new online eligibility guide. Rather than taking the compliance focused approach popular in Australian jurisdictions, the aim is to get the benefits into the hands of those who needed it, to alleviate their disadvantage – the actual outcome of the benefits’ system.
This opened up a discussion around the design of apps, specially tailored to make it easy for carers and families to access government benefits and services. A not dissimilar approach had been adopted in Japan, where IBM and Apple combined together to design iPad apps for Japan Post for the elderly to access government services.
It was observed that a key need of front line workers was to be across the various interactions other agencies and NGOs have had with the client child. This would enable a more holistic approach and ensure key information and intelligence was not locked up in siloed information systems.
There was a need for case workers to be able to easily review the whole case history of children and their carers.
Building intelligence into the system
This prompted a discussion about employing intelligent technology to support social workers and agencies seeking to contribute into the broader poverty reduction program.
We have already seen artificial intelligence (AI) technologies applied in the health care and justice sector to bring together all the relevant information into an unified application.
AI technologies have been trained to not only read thousands of relevant materials but also to make recommendations on what has been learnt from previous cases.
The seminar heard that in Melbourne this work was being done through a large IBM research lab attached to the University of Melbourne, where intelligent applications are able give specialist doctors recommendations as to the best treatment, based on case file and the latest published research. The work is world leading and has further bolstered Melbourne’s reputation as a world leading medical research city.
The same approach has been employed in Florida to used advanced analytics to drive down juvenile recidivism in the Miami correctional system. In the Florida case a highly disciplined approach to tracking key data associated with juveniles had seen incarceration rates fall 26 per cent over three years.
The program found one of the most effective ways to deter future offending is to properly evaluate every child’s risks and needs, and provide the appropriate services and programs to match their unique profile. The parallels with child poverty interventions are obvious.
These self learning technologies have very broad applications. Designed around a typical human services domain, the seminar heard there are applications that will intelligently manage all the case management interactions, underpin professional development programs, while providing data insights both at the front line, but also for policy and predictive purposes.
This large scale tapping and capturing of the front line experience could fundamentally shift the dial from case workers forever trying to keep up, to one where social workers have powerful cognitive tools to support their professional judgments.
Care needed around AI
At the same time the seminar heard it was important to be cautious about the use of predictive analytics and that bias can come from within the data and from how algorithms are constructed. In particular to ensure decision making by AI is testable, traceable and accountable.
As data is integrated and shared the importance of robust information governance was underlined. New Zealand’s relative data maturity is reflected in a Privacy, Human Rights, and Ethics framework that the MSD has developed to take a structured approach to considering all the relevant factors when designing new tools that use people’s personal information.
It was also noted that while AI will be an important part of any future solution, it had to be deployed as part of a wider reworking of how government partners with community and trusted intermediaries to work through wicked problems.
In this regard embracing critical service design thinking, robust co-design, transparent rules, deep engagement with impacted communities, and front line staff were seen as equally, if not more, critical.
Using data to understand complex policy linkages
With these caveats, it is the data that these systems create that arguably provides the largest opportunity to apply modern evidence based policy lies.
Over the last decade New Zealand has built a strong capability around measurement of its public programs and a focus on outcomes and so is ideally placed to benefit from using advanced analytics to track complex multi-factorial policy challenges such as child poverty and well-being.
Many family poverty issues are caused by factors like poor housing, health and disability, so there is a major opportunity to build a data model to track the effectiveness of the various programs. This tracking would begin to reveal the upstream drivers that are causing disadvantage and enable policy makers to focus on the underlying causes of poverty, rather than simply dealing with the consequences of deprivation.
The real power of AI is its ability to recognise and distinguish complex patterns in order to predict specific outcomes and to learn and improve over time. Built on a cognitive platform the whole ten year child poverty reduction program is expected to generate a treasure trove of actionable insights. These insights will only get richer as more data from the various initiatives are fed into policy makers and performance reviews.
The key observation was that whatever technologies and applications are applied, it was important to build these applications off a cognitive self-learning platform, so all the ongoing learnings can be captured and reused.
The seminar heard the great power of AI is to, over time, remove bias and look at what the data patterns are really revealing. This is enormously powerful for understanding and monitoring of the complex public policy challenge that Professor Boston detailed.
At best we have poor understandings of the linkages and relationships, between say housing, education, health and employment and location. Work done by the Victorian Government around social mobility had revealed the sophistication of these linkages and the need to holistically consider and model the whole system.
It was noted Stats NZ has already built global leading data integrations and offers enormous capability, which Australian jurisdictions are now trying to rapidly develop. (This plus a strong heritage of measurement for outcomes over the last decade has positioned New Zealand uniquely among its digital peers).
By doing this at a whole of government level, it would enable the full suite of government programs to be tracked. This will enable the government to determine the overall success or otherwise of the large scale public investment the national government is making into child poverty reduction.
It will also give a far clearer idea of the relative effectiveness of various initiatives.