English: Social Investment Analytics Layer launch
28 September, 2016
Social Investment Analytics Layer launch
I’m pleased to see such a good turnout for what is, I believe, a significant step towards better services for people, and better government for the country.
Over the last five years, we’ve talked about social investment and about doing what works in social services.
Throughout that time agencies have been developing and assembling the pieces of a system that will deliver on this promise.
That promise requires understanding the people we in government are working with, and to measure what impact services are having on outcomes, and for whom. That’s what social investment is about.
Today is a major step in the process to build this system of measurement, evaluation and feedback that means governments today, and in the future, can rely less on anecdote, and more on actual services effectiveness.
That’s a fundamental shift for a public service, and I want to take a moment to acknowledge the considerable progress that’s already been made.
We’ve built a world class repository of data in the Integrated Data Infrastructure (IDI) under Statistics New Zealand.
We’ve established data and analytics shops around the public sector.
We’ve adopted the investment approach in Welfare in 2012, and in social housing in 2015, and the concepts are being increasingly applied in the Health, Education and Justice sectors.
We established the Social Investment Unit (SIU) just over a year ago.
And, most fundamentally, we’ve shifted the conversation within government from policy to outcomes.
That’s the good news.
But we’re still at the start of understanding the consequences of working with information.
Because information is different. It’s different from bricks and mortar, and the policy making process many of us are used to. It’s different from the array of unique bilateral contracts that characterise our accumulated social spend.
Information is costly to create, but it’s almost costless to share and replicate. It’s infinitely scalable once it’s created. It means anyone, anywhere can know anything for nothing.
Information is much more valuable when it’s shared.
As you’ll see in the SIU presentation, it’s hard to overstate the value of what’s being revealed to us for the first time through the use of the IDI and other data stores in government.
That value is so high that we can count the cost of not sharing data, of not releasing code, of not documenting your output in terms of all the New Zealanders who have to wait longer for assistance – or not receive it at all.
Time after time – almost daily, in fact – I see examples of valuable information and IP created at the taxpayer’s expense and for the benefit of citizens being withheld by agencies.
So today’s announcement is not only about the fact that the SIU has assembled a new analytical layer that builds on the code that many of you do when you work in the IDI, and which is going to make it faster and easier for users of the IDI to unlock new insights in the social sector.
That’s an important development in its own right.
But perhaps even more important than this is the lessons we can take for how the SIU has delivered this result.
I’ll come back to that shortly.
Today we are launching the Social Investment Analytical Layer.
I’m told it is a cleaned and quality-assured data layer that covers nearly all of the social sector.
The code that produces this data layer will be published on code-sharing web site GitHub and the Statistics New Zealand web site, and it will be fully documented.
The layer maps around 66% of the $53 billion of social sector spending back to anonymised individuals.
This mapping means that, for the first time, we can see cross-sector impacts on individual people and their services use across time.
For instance, it means we can understand how changes in policy or funding in one part of the social sector affects spending and outcomes in other parts of the social sector months or years downstream based on real events.
That’s important when billions of future dollars across sectors can hang on apparently innocuous policy decisions today. Understanding and managing those pools of liability and risk is a real issue for governments.
This layer means we can shine a light on what drives our costs.
That’s the first major benefit this new layer delivers.
The second is that by creating a standardised layer of cleaned data, it means users coming in to the IDI don’t have to first spend months cleaning data themselves before they can start on the research. That’s done.
I want to focus on this second point for a moment.
The idea of creating a standardised layer of cleaned data ready for analysis that means each user isn’t repeating work already done by others isn’t particularly new.
In fact, it’s obvious.
So we can increase the return on the Government’s $24 million investment into the IDI since 2013.
The drive to build that analytics layer didn’t come from mainstream agencies who have few incentives to consider the impact of their decisions and actions outside their own budgets.
Those agencies should now be building data sets so they can be published and used by others.
Data, and code used to produce that data, are not a private good. They are a public good that can drive value for our customers.
And this sort of tool will become a requirement for agencies to use to analyse existing spend and bid for new money.
When I look across agencies, what I see at times is a reluctance to share data. A reluctance to publish reusable code. A reluctance to release some findings that some might consider unhelpful.
Agencies that see risks but not rewards from finding out the truth.
Finding out the truth is not a risk.
Revealing uncertainty about what we thought was the truth is not a risk.
Now that the technology will be available, the bigger risk to agencies will be not using it, and not being geared for other others to look into their business.
The SIU has provided a model process for this exciting new capacity.
Part of the benefit of this launch is that agencies can see how to go about publishing their work.
Another aspect of this is that the SIU has worked out what skillsets it needs to enable this information to be more accessible.
The lesson, for me at least, is that the small amounts of the right skillsets can generate enormous leverage and opportunity across the public sector.
There’s also no doubt that the insights generated at this scale will create challenges for implementation.
Translating these insights into organisational change, into new ways of commissioning, and new ways of working with suppliers and with our customers, will be genuinely challenging.
It will be possible for our customers to know as much as us.
Public agencies need to be preparing for the inevitable – where anyone can know anything, anywhere, now, for nothing.
This will create a rich, but contestable, information and policy environment.
That means the SIU and other agencies are going to face increasing competition for analytical insight and real world service solutions.
My sense is that the public service has moved a long way in the last six months.
There’s a growing understanding of how measurement can help tell us what we need to be doing.
And I’m pleased to report this new analytical layer is being complemented by improvements in the IDI access process run by Statistics New Zealand.
You can see some of progress by visiting the Stats NZ web site.
As these products become more common, and this will be one, it will become clear that accessing and using information is no longer a specialised activity.
A further explanation of the analytics layer can be found here: https://youtu.be/0GVCpBrxF2o