The Humble Policymaker: A Review

Zero to political philosophy in 60 seconds

I have a big list of ideas I want to turn into blog posts. One of the items on the list reads:

“A lot of optimism about improving government (including my own) comes from a perspective that the government’s purpose is to improve society or people’s welfare. But what follows is we consider the purpose of government to be maintaining its own legitimacy?”

Which is why I was annoyed excited to join the discussion when Polly Mackenzie, Chief Executive of the Demos think tank, went and wrote her paper, The Humble Policy Maker, the third instalment of her series Making Democracy Work. This post is a review of that paper. In a sentence: it’s a great paper, but I think it is missing some key parts of the analysis and therefore the solutions required. I then discuss why I’m more sceptical than Mackenzie is about reforms being able to solve the problem she identifies.

For those unfamiliar with the Making Democracy Work series, Mackenzie (former Director of Policy for Deputy Prime Minister Nick Clegg during the coalition) wrote them after becoming frustrated with the way she and most of her peers in policy and politics understand governing and making policy. The first two papers define the problems as she sees them, of a democracy increasingly distant from the people it is intended to serve, facing intractable problems exacerbated by ‘the exponential age’ we live in, where technological change outpaces our ability to enact social change. I would say they do a pretty good job of succinctly covering most of the mainstream critiques of contemporary liberal Western democracy, but don’t bring anything especially new to the table. I’d encourage anyone interested to read them, as I’ve not done them justice with the above summary.

Summary of The Humble Policy Maker

The third paper - The Humble Policy Maker - is where Mackenzie gets into her proposed solution, although there is one more coming with concrete proposals. She describes two primary types of policymaker: the partisan and the technocrat. The partisan believes in the righteousness of their cause and the malevolence of their opponents. The partisan thinks that if only they could get into power, they could enact their policies to make the world a better place. By contrast, the technocrat treats social problems as puzzles to be modeled and cost-benefit-analysed to come up with an ideal evidence-based policy solution. I see these two groupings broadly mapping onto the political and official branches of our government respectively - although there are technocratic MPs and partisan officials (the latter usually on a specific policy issue, rather than being supportive of a political party). Many people have both tendencies and Mackenzie herself identifies as someone who has been both.

The problem with these two views, Mackenzie says, is that they both ignore the voters or at best treat them as obstacles to achieving their grand vision. Voters often don’t like the solutions that elite policymakers come up with, and so hard problems like social care, the rising cost of ageing populations or housing affordability keep getting a series of sticking-plasters rather than meaningful change needed. Voters may even oppose things in their best interest...where their best interest is defined by elite policymakers.

She recognises that the technocratic approach requires stronger pushback than the partisan approach, and presents the following arguments against too much faith in technocracy:

1. The question of “what works” is only relevant once you’ve agreed on what you’re trying to accomplish. 

2. Not all questions have a “what works” answer. 

3. Even when there is a “what works” answer, technocrats often use over-simplified models of people to design policy solutions which mean the “works” part of “what works” fails to deliver.

I think at times she leans too far into a caricature of technocracy as a group of economists and data geeks designing a model and declaring the world must fit into it. She does recognise that the obvious response to her critique is “we’ll make the models better, and fit things like feelings and relationships into them.” In practice, I think this is what gets done. Most policymakers are very aware of the limitations of their models, and factor this into their thinking. The evidence-based policy movement was never about making everything through models, just shifting the dial a bit from “random biases of policymakers” to “clearly articulated evidence-based rules”. This is a critique of tone rather than substance, however. Mackenzie is explicit that she does not want to completely throw out evidence-based policy or technocracy, just that we need to recognise its limits, especially in managing the unsolved problems piling up at the government’s door.

Mackenzie’s solution is that policymakers need to be more humble. The only examples of successful long-standing policies are those with broad popular support. Policymakers therefore need to focus more on co-developing solutions between parties and with citizens, rather than imposing a top-down solution because the majority party thinks it’s right. The opposite of humble policymaking (arrogant policymaking?) often goes wrong, either because the analysis it was based on was wrong, or because it fails to get broad support and is quickly repealed. 

She argues that policymakers need to develop ways of incorporating the relationships, feelings and other complexities ignored by technocratic models. We need to make governing seem like a process of the government representing the people and mediating between them, not a managerial government to be lobbied and appealed to. Mackenzie recognises that this requires a much more active role from citizens beyond voting once every five years or so. It requires them to take more responsibility for dealing with the implications and trade-offs of their decisions and preferences, rather than outsourcing that to the government.

I started this review with my own brief musing on this same subject. It aligns pretty closely with Mackenzie’s analysis, so it’s no surprise that I’m broadly enthusiastic about the paper. I think it’s pretty obviously right that centralised decision-makers are going to be unsuited to making decisions about big problems. The argument presented is pretty similar to Tocqueville’s analysis of why democracy worked so well in early 1830s America because everyone actively participated. It also has aspects of Hayek’s famous line that “The curious task of economics is to demonstrate to men how little they really know about what they imagine they can design.” I’m deliberately using Hayek to show this isn’t a left-wing thing. The idea “you don’t know half as much as you think you do” dates back at least to Socrates, after all.¹ It would be great if all policymakers had the requisite epistemic humility.

I also like that it has a useful stance on preparing for the long-term future. If we think about democracy not as a solution for solving problems now on 5-year terms, but as a centuries-long solution to the problem of human conflict and war, then seeing the job of government as maintaining its legitimacy seems much more reasonable. It’s also helpful to have a way of clearly bridging the divide between civil servants and their political masters. Civil servants are not just there to advise on what’s best based on economic models - they also have a role in advising on what is politically advisable. Seeing policy as a process of creating legitimacy with the public avoids the need for civil servants to be overtly political in their advice.

Who decides what being humble means?

There is a core question with the idea of humble policy makers who involve citizens in policymaking. Which citizens do we involve, and how? Mackenzie’s reframing doesn’t solve the core problem of government, it just relocates it from “what’s the best solution” to “what’s the best way to engage with citizens?”

The problem is that the question of how to engage citizens has all the same problems as the question of figuring out the ideal solution. It too must be based on some normative, or moral, judgement of what the right way to engage is. There are two main approaches to resolving this. Either we can try and derive the right way to engage with citizens a priori, based on moral reasoning. Or we can say that the right way to engage with citizens is whatever leads to the best solution - but this brings us right back to the question of what the right solution is in the first place.

Here’s an example if the above got too philosophical: the housing affordability crisis, i.e. young people struggling to afford houses in major cities, especially in the South East of England. Mackenzie mentions the work of Garrett Jones in 10% Less Democracy, and he uses this as an example (in the US context, but the problem is similar). Housing is an area with arguably too much citizen participation. Wealthy local homeowners can object “not in my backyard” and these NIMBYs can block development that would meet the demand for housing from young people attracted to high-productivity cities. These young people, and the low-income communities that may get displaced from subsequent gentrification, participate less in the democratic urban planning process for three main reasons. The first is that the gains from participation to these people are much more diffuse than the gains to homeowners from restricting housing. The second is less of a feeling of ownership for recently arrived or renting people, compared to long-standing homeowners. The third is that a lot of the people who benefit from more housing don’t even live in the area currently, because they can’t afford to. These people currently don’t get a say even if they wanted to. If you would like to move from Darlington to Kensington, you can’t currently influence Kensington’s housing policies.²

So what is the solution for more humble policymaking with respect to housing? Let’s start with my second solution: decide the ideal outcome and work backwards. Maybe we decide that we want housing costs to come down, and so to do this, we need to reduce the relative voice of local homeowners and increase the voice of prospective homeowners. Ignoring whether this is practically possible, it fails because we’ve just had to assume that lower house prices are the desired outcome. But the whole problem with the current system is that the most influential voters don’t see this as a desirable outcome.

Let’s look at the other solution for the housing situation: deciding a priori on the right way to engage people. In many ways this is what we currently have. We have decided that the moral approach is ‘local people get to decide on planning policy in their area’. There are clear reasons for this. Local people will understand local problems better. Plus there’s a moral case that locals should get to decide what happens in their area. You might consider weighting things differently. Maybe the government maps out all the groups who could be affected - homeowners, tenants, prospective homeowners, housebuilders - and uses some formula to weight their interests. Or maybe representatives from all these groups sit on a panel, and that panel must come to a majority or supermajority decision they agree on. This is a more promising way of thinking.

There are problems with this way of going, however. The main problem is that deciding on the right way of engaging requires political support in the first place. If homeowners won’t support the imposition of new policies that oppose their wishes, why should they support politicians who want to bring in a new process that diminishes their voice? You can solve this by taking a pragmatic approach and saying that the right method of engagement is whatever the electorate will agree to. But this is insufficient on its own, as it implies no direction of travel beyond the status quo, unless there’s a vast grassroots movement to solve it - in the case of housing, there isn’t. We could say that similar to economic analysis of ‘market failures’ we could develop some sort of analysis of ‘democratic failures’ and use this to approach individual problems to come up with new solutions. But this brings us back to being technocrats, just with a different goal.

The exponential society fights back against the humble policy maker

There is also another, possibly more important, problem in the idea of engaging citizens more closely in policymaking, which is that the world is vastly more complex and global than it was when democracy was invented. This means that the gains from participation are likely to be much lower than the opportunity cost of participating. 

In Tocqueville’s time, everyone could join in a healthy participatory democracy because 90% of everything that happened in people’s lives happened within 20 miles of their home. It was easy to personally know everyone who impacted significantly in your life. In the modern world, we are much better off for being globally interconnected. Decisions made locally affect us, but so too do decisions made in county halls 50 miles away, national governments 200 miles away and board rooms from Silicon Valley to Shenzhen. The world has so much in it, of such vast complexity, that nobody can hope to be well-informed in anything except a tiny portion of it. Even the smartest policymakers may struggle to understand all the corners of one policy area, and they are full-time experts. The decisions made by these bodies also touch on far more people than local decisions.

The solution Mackenzie seems to be leaning towards is more localism and devolution. But the problem is that the world just isn’t set up to allow this. This isn’t primarily because we’ve set it up to be less localised. It’s not mainly a policy choice. It’s an actual, physical reality of the world. It’s literally not possible to make the computer/phone you’re reading this on without a globalised supply chain. Similarly, it’s not possible to have world-class healthcare without a globalised supply chain. There is no going back to the world of Tocqueville and still maintaining our material prosperity. Localised decision-making is incompatible with the world of the ‘100x developer’. Amazon is able to deliver amazing products next-day for cheaper than local stores because they have a group of super-smart people continuously optimising for world-best solutions (the people in your local store are probably worse paid than amazon workers, so it's not explained by that). But Amazon is not one-sized fits-all. Being focused on citizens or customers is not incompatible with centralised decision making.

The way Amazon manages the huge complexity of its organisation is by forcing teams to operate on short feedback loops. They do this by enforcing that every team offers its services as an accessible API. The other teams within the company essentially act like customers of each team and coordination happens at the system and process level, not at the interpersonal level. So if something is going wrong with a particular team, this will clearly show up in the data. But each team is not accountable directly to customers - only Amazon at the company level is accountable to customers. Devolution would be like expecting customers to rate every single team that contributed to their package delivery. The modern world works not by increasing the customer touch-points, but by hiding the complexity of organisation that goes into producing a seamless, one-click experience.

In a world like this, the benefits of any individual contributing to decisions is miniscule. The act of participating in policymaking has a huge opportunity cost - it takes time you could be spending working, playing or being with loved ones. It’s the same reason that an individual voting in a general election has a minuscule influence - but we have huge social pressures and encouragement to vote, plus it’s only once every five years. Having to maintain this level of civic duty for no return at scale seems unlikely.

Essentially I am proposing here that the model of democracy adopted depends on transaction costs, similar to Coase’s theory of the firm. Coase explored the question of why markets are made up of firms/companies rather than just individuals freely trading services and products. Why does Coca-Cola bring together manufacturing, advertising, accounting all into one legal body, rather than just individually contracting people to do work required? Aren’t markets efficient such that it should be cheaper to buy services rather than build them all yourself? Coase’s answer is transaction costs - drafting up a contract for each new piece of work, and coordinating between all those contractors, adds additional costs. To avoid these costs, firms organise work within themselves.

The analogy with democracy isn’t exact, but transaction costs are still essential. Every time somebody participates in democracy, it imposes an opportunity cost on them. Therefore, we build democratic structures to avoid these costs. Rather than an individual weighing in on every policy problem, they elect an individual to send to Parliament to go weigh in on those issues. In fact most don’t even bother looking at the individual - parties arise so that you can outsource much of the vetting of candidates to a broader group you respect. Parliamentarians in turn outsource governing to specialist ministers, and scrutiny of those ministers to specialist committees, because even a full-time MP cannot keep on top of everything in a modern economy. People only participate in democracy to the extent that the benefits they gain from participating roughly equate to the opportunity cost of doing so. In the modern world, the benefits of any individual act are so small that all we want to do is vote once every 5 years and have done with it.

Modern technology does offer some scope for reducing the opportunity cost of participating. Crypto fans will tell you that Decentralised Autonomous Organisations, DAOs, are going to replace traditional hierarchical companies by massively reducing the transaction costs of hiring individuals to do fractional amounts of work. I’m sceptical, but even their greatest proponents would probably only predict a relatively modest reduction in transaction costs - an order of magnitude at most. But the world has grown in complexity many orders of magnitude. Tech can slow the problem, but not eliminate the core issue that the benefits to participating in democracy decrease as complexity of society increases.

We’re not going to invent top-down solutions

Despite my rather existential objections to the possibility of introducing more participation to democracy, I still think it seems like a valuable goal. I remember as a teenager (sorry, old folks) arguing that the Big Society was at least asking the right question, even if it didn’t appear to have the right answers. Its critics seemed content to ignore the question. Presumably it was too hard. Participation is a healthy, even crucial, part of democracy and we should seek to encourage it. I see it more as just one of many values to trade-off, along with the other goals policy makers may seek to advance. To some degree, the government has a budget of legitimacy that it can manipulate through policy just as it controls its financial budget. And it has a much shorter default period than fiscal borrowing, so should be considered carefully.

So how do we ensure that citizens participate given that they have incentives only to participate for a very short period of time? Under my description of the problem, there are two generic types of solution: lower the cost of participation, or raise the benefit. I’ve already spoken a bit about how technology might be able to slightly lower the cost of participation. We’re all used to voting or commenting online, and it can be done in our own time rather than having to turn up to a physical place, and either vote or participate in lengthy discussions of policies or values. Any discussion in this area is obliged to mention the work done on digital democracy in Taiwan, but so far this seems to focus on unexpected low-hanging fruit, not solving intractable problems. I stick by my claim that there are limits to the role this can play, without radically rethinking the core of our democratic system (i.e. removing the principle that you vote for an MP).

So how about increasing the benefit of participation? Citizens assemblies are one method of doing this. Rather than being one of 65 million, you are one of 100 citizens responsible for giving your voice. Consequently, your participation has a much larger individual effect. This is a solution only so far as people accept the legitimacy of the citizen’s assembly, however. I’ve yet to see an example of national change on a controversial issue thanks to citizen assemblies. In my head this is “nice idea, show me the proof.” The main problem is that it still relies on the voting population to put their trust in a group of representatives. It just outsources their opinion to random citizens rather than elected officials - a pretty damning indictment of elected officials that they might be worse than random people off the street. The whole point is they are supposed to be better!

The more I try to think of solutions, the more I think they may sound plausible on first look, but all rapidly build up problems that mean you need to add more and more epicycles to retain the solution. We’d end up with some horrible hybrid system not dissimilar to the one we have now. Perhaps I’m being too humble, and should have more faith in my ability to come up with grand solutions to fixing democracy. But it shouldn’t be surprising that I struggle to invent a workable solution. One implicit theme of this essay is that politics and democracy are a lot more endogenous than policymakers might hope. They arise from complex dynamics of power within society. These dynamics are difficult to predict in advance, so it’s going to be hard to know which activities will take us in the direction we want. One area I agree with Polly Mackenzie on, though: the direction we need to go in requires more participation in policy. I look forward to seeing her prove my scepticism wrong.

1. If I’m feeling particularly confrontational, I could even suggest that Mackenzie’s argument isn’t too far off Margaret Thatcher's when she said 

“I think we have been through a period when too many people have been given to understand that when they have a problem it is government’s job to cope with it. ‘I have a problem, I’ll get a grant. I’m homeless, the government must house me.’ They are casting their problems on society. And, you know, there is no such thing as society. There are individual men and women and there are families. And no governments can do anything except through people, and people must look to themselves first. It is our duty to look after ourselves and then, also, to look after our neighbours. People have got their entitlements too much in mind, without the obligations.”

2. Mackenzie’s decision to say that long-standing policies are inherently successful means she says that pension reforms have been successful. Except pensions are a great example of the people who get to vote failing to consider the interests of those who don’t. In this case, the impact is generational. In the late 20th century, pension policies were introduced under the auspices of being paid for by taxes at the time, as a form of insurance. Except public pensions weren’t properly funded by the contributions made. In effect, future generations pay for them. This is fine if future generations continue to grow in wealth and population. But stagnant growth and population means that we now have far more older people as a proportion of the population than was expected. So we end up with increasing tax burdens on younger members of society to support older members. In theory, this money could be borrowed, but then future generations still have to pay - another bet on growth that I’m not confident of making.