A lot of the policy debate and academic debate is around what policies we should use to improve outcomes for society. But a couple of recent books and a recent paper all make the point, in very different contexts, that how things get done (or not get done) matters perhaps as much if not more in many contexts. Karthik Muralidharan’s book Accelerating India’s Development: A State-led roadmap for effective governance argues that weak state capacity is a key factor limiting India’s development. Ezra Klein and Derek Thompson’s Abundance argues that in the U.S., “laws made to ensure that government considers the consequences of its actions have made it too difficult for governments to act consequentially”. And at a much more micro level, Noam Angrist and Rachael Meager show in a recent working paper that most of the heterogeneity in the size of the effects of targeted educational instructional programs across studies in Kenya, Ghana, and India can be accounted for by two implementation factors: who is delivering the content (teachers or volunteers) and how much it is actually getting delivered (take-up and fidelity).
Partial evidence of the problem
Both books do a great job documenting lots of infuriating examples of good-sounding policies that don’t get implemented or don’t get implemented well. In India, Muralidharan notes the scale of the problem in delivering education (“half the fifth-grade children in rural India cannot read at a second-grade level”), nutrition (“nearly 35 percent of India’s children are stunted”), justice (“the Indian court system has accumulated a backlog of nearly 30 million cases”), and other basic public services. In the U.S., Klein and Thompson discuss the lack of housing, rail infrastructure, and clean energy being built. For a close to home example, the Washington Post recently had a story on how a Takoma Park vacant parking lot that was meant to be turned into new housing has not seen work start 25-years after a developer was selected.
However, by only focusing on good things that don’t get done, we perhaps miss the other side of the coin: stopping bad policies from being implemented. I’m reminded of an interview Austin Goolsbee did, about being on the Council of Economic Advisors “I have a friend who was on the Council of Economic Advisers a long time ago. And when I took the job, he said the job of the Council of Economic Advisers, like a good gardener, is 10 percent planting seeds and 90 percent pulling weeds… And so the fear always, as the economists leave - nobody likes them. Nobody wants to talk to them. But they prevent some of the dumbest things from happening.” So the challenge is improving state capacity, while still maintaining the checks and balances that allow that capacity to be used for good and not dumb.
What causes this problem?
Klein and Thompson see a lot of the problem as being to do with what they note Nicholas Bagley calls “the procedure fetish”, where policy becomes “process-focused rather than outcomes-oriented….the state’s legitimacy would be earned through compliance with an endless catalog of rules and restraints rather than through getting things done for the people it claimed to serve”. Muralidharan likewise notes how the Indian system’s “focus on paperwork shifts supervisor and staff attention to compliance and record-keeping rather than service delivery or outcomes”. This is coupled with what Klein and Thompson dub Everything-Bagel Liberalism “they often add too many goals to a single project. A government that tries to accomplish too much all at once often ends up accomplishing nothing at all”. Such arguments may sound familiar to people working in development as well – but focus and deregulate is not all that it takes.
Indeed, the problem is not that we need the government to just get out of the way – there are many good things we want the state to deliver, and we need better management practices that strengthen the state’s capacity to do this. Muralidharan uses an organizational economics lens to highlight six areas which determine effectiveness, that have many similarities to our work with large firms: (i) the use of data and outcome measurement (KPIs) -“the problem of reliable data is so severe that a senior government official once told me that ‘we are mostly flying blind’”; (ii) personnel management – how are public sector workers hired, trained, supervised, and motivated (“studies show that personnel management is the most important component of management that matters for organizational effectiveness”; (iii) how cost-effectively the government spends money; (iv) the quality and quantity of revenue it raises; (v) how responsibilities are allocated within the organization – in India what should be done at the Federal vs decentralized level – in firms the analog being also what should be centralized vs decentralized; and (vi) and optimal allocation of tasks within and outside the organization – here what should be done by the state vs delivered by the market or civil society – and what the state can do to make the market work better.
Both books note that this does not mean that governments can not get anything done – they give examples of emergency responses where things have been done effectively and fast – from the rebuild of the I-95 bridge in Philadelphia that was destroyed by a gasoline tanker truck hitting it in 2023, to what Muralidharan calls “mission mode” in India, such as conducting the world’s largest elections and vaccination campaigns, and relief operations during natural disasters. So the state can act fast, but most of the time does not.
These disasters may be one time where the goal of policy is clear, singular, and almost everyone agrees on what needs to be done, and the state is trying to make this happen. In his review of Muralidharan’s book, Jishnu Das “his recommendations have little to say about the multiple occasions where the state acts mendaciously, or is called up to adjudicate among different interest groups, or indeed, when its core function requires it to actively reduce the welfare of some citizens, such as in policing and the courts”. This is where I think we need “Goldilocks state capacity” – strong enough to deliver good things and adjudicate disputes, but weak enough to not be able to deploy a lot of bad policies without scope for pushback.
Where does Impact Evaluation fit into this?
The Abundance book is much more about diagnosis and a big picture call to action than specific solutions. However, it does have two key messages that resonated with me on how to change. The first is that most major inventions don’t work well at first (they illustrate through discussing electric light bulbs and penicillin) and need to be tweaked and incrementally improved. “These micro inventions are often more important than the original idea”. Second, they note the need for another type of management technique – bottleneck analysis – to try and figure out where policy efforts on change should focus.
This is where impact evaluations and RCTs fit in. Muralidharan’s book and Angrist and Meager’s paper highlight experimental evidence of how state capacity can be improved through interventions that improve state capacity, with Muralidharan arguing that these can be often ten times more cost effective than just spending more on government as usual. These can be relatively simple reforms – filling supervisor positions to reduce teacher absenteeism, using biometric authentication and new payment mechanisms to reduce corruption and inefficiency in public transfers, filling judge vacancies to speed up case resolutions and unlock assets, subgrouping students by learning level more, etc. As Esther Duflo argued in her economist as plumber paper, “tinkering and adjusting will be necessary since our models gives us very little theoretical guidance on what (and how) details will matter”. In the introduction to his forthcoming global edition of his book, Muralidharan notes he hopes this will also act as a bridge between the debate on how much development research should be focused on “big picture” questions not amenable to impact evaluations and smaller questions that may be less important to national development. The effective state capacity to get good things done is definitely a big question, and there is a role for experimentation in figuring out how exactly to do so.
Join the Conversation