In the last five years, there have been numerous interesting developments in the public sector innovation space. In Singapore, public policymaking and services have traditionally been the domain of a small number of elite technocrats. As in other countries, the more strategic the policy, the smaller that number of technocrats involved. However, as digital and social media become increasingly ubiquitous, with an Internet penetration of 88% in Singapore, the nature of information treatment — in particular the consumption, creation, and dissemination of information — inadvertently creates a shift in the social contract as we know (knew) it.
With a small group of like-minded colleagues, I’ve been working in the area of public sector innovation, relating to the urbanisation process in developing countries as well as public sector issues in the urban context. This has taken us from re-imagining municipal public service models in Bangladesh and informing Bhutan’s employment policy, to healthcare and hospital design in Singapore. Along the way, we’ve learnt a lot about public policy, public services and the nature of public sector innovation. This includes how government bodies may not always talk to each other, the tension between politics and administration, conservative cultures that may not support the change (and risk-taking) needed to meet stated objectives, and the tradeoffs between desired outcomes and resources.
As we look to the future, the interesting developments and trends that are emerging lead us to a number of hunches about the future of public sector innovation in Singapore. These hunches are also implicit questions about the future of cities and populations, the potential of technology, and the nature of the human condition.
HUNCH #1: Behavioural insights will be increasingly important in designing policy for citizens.
As populations become more varied and textured, it becomes more difficult to create policy in a number of domains that are premised on generalisations of the various groups they seek to serve.
In a project on urban poverty and neighbourhood transformation, one of the relevant demographic groups was single mothers. Many of them were experiencing financial difficulty and struggled to provide healthy food for their children as whole foods cost more than processed foods. We saw single mothers forgoing jobs and employment income at the cost of better nutrition. They prioritised being at home with their children so that they would not be in contact with what they alluded to be negative influences in the neighbourhood. This told us the importance of understanding the behaviours that prioritise one need over another: that getting low-income mothers to take up regular jobs by offering incentives for successful job placement may not have nearly been as effective as, for example, a programme seeking to enable mums to work independently from laptops or kitchens.
HUNCH #2: Blending disciplines and methods for public policy is powerful, but the blending will increasingly require people who can serve as translators.
In the urban poverty project mentioned above, we applied three different approaches to understanding urban poverty at the neighbourhood level: Ethnography (a method from anthropology) which involved interviews and observation; data analysis of aggregated case file data of social assistance recipients; and a basic form of system dynamics mapping (for example, how family environments may impact educational attainment, and in turn employability, income, health outcomes, etc.). This type of mapping surfaced an insight: that within the factors relating to family environments, it was parental attitudes to a child’s education that had the largest impact on his or her success at school.
It was through ethnography that we gained a deeper understanding of how, on an employment-related problem, many blue-collar workers found it challenging to find employment after an injury, because their existing financial situation made it difficult for them to receive the appropriate treatment and recover fully to be fit for work.
By running the data, we were able to uncover an interesting correlation between employment status and incidence of domestic violence.
These stories have come about from very different research methods. How do we piece them together to form a fuller understanding of the problem area? “Translators” who understand enough from anthropology, for example, could say that a certain finding is a signal to dig deeper in one line of inquiry, or that we should look into a geospatial mapping to uncover spatial clues in the urban space to explore spatial connections to the problem.
HUNCH #3: Complexity means we need to allow for emergence.
Complexity tells us we can’t be sure that A causes B. It may just be happening as a function of the millions of events that take place at any given time. But when we don’t know that one thing causes another, then it’s impossible to create a “model” of how a particular system runs. In the case of the urban (or urbanising) context, cities or emerging cities are perfect examples of a “complex adaptive system”. It’s not a closed system, so we can’t be sure of all the variables we may want to map out. So how can we create a model like so many consultants do?
Well, beyond the fact that we would need to develop many, many assumptions, it also means that the solutions that are generated are evidence-based. Quantitative data is just one form of evidence. But it does mean doing the uncomfortable thing of suspending one’s reliance on frameworks or models — to simply allow multiple data points out there, quantitative and otherwise, to surface a pattern. We say then that such properties — whatever form they may take — are emergent. According to an anecdote, the University of California, Irvine, in its first year of infrastructural planning did not put concrete footpaths in its campus; instead, administrators laid out grass all across the campus, and had students and faculty walk however they wished to get to where they wanted. By the end of the year, trails had been formed by the natural paths used, also known as desire lines. It was over these desire lines — the emergent property from use and desire — that the concrete footpaths were laid.
Photo from Welshgeek
HUNCH #4: Public service design may become more experimental — and risk-taking.
As the world becomes more complex and we start to deal with more diversity in our population on different fronts, there are more competing needs and interests than before. Resources, as always, remain finite. How can one even design policy for massive groups of people anymore?
We can go back to empirical evidence. One form of generating evidence within a project context is to conduct experiments. These can be quick and dirty, such as prototyping in Lean Startup Method (LSM) which has become de rigeur with Silicon Valley startups. LSM takes an approach of quickly developing simple hypotheses — about the customer or user, on what the problem or need is, and what the solution might be. These are then tested in the field with real “potential customers” in small, rapid cycles in succession, helping the solution provider or startup founder(s) understand better what their target customer needs and what solution they should build. On the other end of the spectrum, there are the more rigorous (and typically more resource-intensive) randomised control trials (RCTs), which form the basis of many study findings in psychology and behavioural economics. Once the domain of pharmacology and psychology labs in universities, RCTs are now a powerful way to understand the effect of an intervention with a sample population before a decision to potentially roll them out at a larger scale.
These approaches come from two very different worlds, but we have a(nother) hunch: what if we combined these together? An LSM experiment to get out the biggest kinks, for e.g. the riskiest assumptions, validated or invalidated, and then with a clearer idea on the parameters that are most relevant, conducting an RCT to look at the size of effect that we are testing. On deciding whether or not to conduct an RCT, we would look at resource availability relative to the stakes of the decision, which may be with regard to a policy, programme or service.
HUNCH #5: Problem definition will be increasingly key in not chasing down the wrong problem.
As Einstein memorably said, “If I had an hour to solve a problem I’d spend 55 minutes thinking about the problem and 5 minutes thinking about solutions.” The clarity and sensibility of this quote seems obvious once articulated — which is why it was ironic for us to realise that, as a team of behavioural researchers, data geeks and designers, this wisdom is not as widely applied in practice as it could or indeed should be.
Let’s say you’re a mayor of an emerging city in Sri Lanka, like Jaffna. You’re trying to tackle a slum area that’s expanding in Thirunelveli, which is struggling with crime, poor education, access to healthcare, and a whole slew of other things. And let’s just say, there were five-year (or even 10-year) plans created, at the federal level, that were not realised after multiple cycles of promises. You’re embattled by competing priorities set against very finite resources — and worse, your civil servants are getting disillusioned. Perhaps the first thought you have is, “Let’s set up a job creation programme.” That’s completely valid. But maybe what you really needed to do was to first get a hold of gang crime, in order to create the stability needed for employment. But without investigation, how would you have known that? Not to mention you would’ve taken a completely different solution path.
For problems with such complexity, where does one begin? Perhaps you should look at data and pick out strange correlations to examine. But these are just data points. Maybe you should also see actual realities on the ground, and take a leaf from anthropology and conduct some ethnography on the streets. These may possibly uncover some behavioural insights that might be key to your problem (and solution). But how would you frame your challenge? Maybe the field of design could offer something here, and even guide your innovation efforts on when to “diverge” and when to “converge”. And of course because these issues are inherently at a systems level, maybe you build a system dynamics map (which could look like a very messy spider web).
Problem definition is key. Good problem definition will ensure that public servants do not plan, execute and evaluate the wrong solutions. It will avoid playpumps being possible perpetuators of child labour. It will enable the creation of things that can continue to be built upon by locals, with local materials — and most importantly, to serve the citizens they are meant for in the first place.
Bernise Ang is Principal and Methodology Lead at Zeroth Labs, an experimental lab which brings together behavioural insights and systems analysis, and applies them to social policy issues, creating new forms of services, products, and innovating new business models. An edited version of this essay was published in Singapore: A Democracy of Deeds and Problem Solving [Volume 25 (2016) of Commentary, an annual publication by the National University of Singapore Society (NUSS)], in June 2016.
Bernise is also featured on The Future of Us Ideas Bank Page (Dreamers)
Top photo from thinkstock