How does policy affect elections?

Five models

I’ve been working in and around politics and policy for almost a decade now. So it’s a bit embarrassing to have to admit that I don’t have a good model I trust for how policies affect elections.

What I mean is this: working for a think tank and in advocacy, I’ve spent a lot of time recommending policies. And I’m well aware that a major part of persuading politicians to adopt those policies is convincing them that the policies will help them get (re)elected, or at least not harm their chances. But if I’m honest with myself, I can’t predict the electoral consequences of my suggestions with any great confidence. What I can do, though, is set out the different models I have encountered for thinking through this problem.

Popularism

The simplest model, in some ways the most common and most tempting, is ‘popularism’. Essentially, this holds that elections are won by the parties and politicians with the most popular policies (or at least those that are able to talk more about their popular policies and less about their unpopular policies than their opponents). For a think tanker or advocate, the implication is simple - do some polling, and find out whether your policy is popular or not. See if you can find the best framing to get the public to endorse it. If it is net positive, especially if it is net positive with marginal or swing voters, adopting your policy will make election success more likely.

The trouble with popularism is that voters are ignorant - perhaps rationally so - of most policy. For example, one poll before the 2017 election found that two-thirds of people admitted they don’t read manifestos, and around a tenth didn’t even know what a manifesto is. In that context, is it plausible that more than a handful of policies ‘cut through’ and make a difference to election outcomes? Most policies are likely to be so low salience that they don’t really matter, no?

The minefield model

Popularism entails trying to find popular policies and talk about them. Its negative inverse involves trying to avoid unpopular policies that blow up in your face - what I call the ‘minefield’ model. Popularism reflects greater confidence about being able to predict voter responses. The minefield model says “that policy might seem low salience and irrelevant, but you never know, there’s a chance it could explode and cost us the election”. Every policy is viewed with suspicion as a potential poll tax, pasty tax, gilets jaune debacle. This sort of thing leads to ultra caution - scraping barnacles off the boat, creating a small target, protecting the Ming vase.

The question raised by the minefield model is whether disastrous policies really are so unpredictable. Is it completely random, or can we use the other models to anticipate more or less problematic proposals?

Narrativism

A third model of voters is that they don’t notice individual policies so much as they pick up a general vibe of a party or individual, often presented as a narrative. The precise measures are mere details, but the question is whether they cohere into a picture of pro-business or pro-worker; nativist or multicultural; traditionalist or progressive, or any number of other possible frames.

The implications of narrativism for policy are actually a little ambiguous. For the most part, it implies that politicians should pursue policies that fit with the broader narrative they are trying to present, and advocates should emphasise the congruence of their proposals with politicians’ favoured narrative. We see this a lot - given the current government’s focus on growth, everybody wants to frame their ideas as good for the economy, where a couple of years ago it was “real levelling up entails following my proposal”. But equally, politicians sometimes have more room for manoeuvre if they are being counterintuitive - the old ‘Nixon goes to China’ effect. So, for example, it seems plausible that George Osborne could be bolder in raising the minimum wage, or Boris Johnson in anti-obesity regulation, because those policies went against their images.

Deliverism

Narrativism is attractive to those who see politics about PR and spin, but it risks being a bit untethered from the real world. Deliverism is the materialist counterpart to the idealism of narrativism. It says is that voters don’t notice policies, but they do notice outcomes. Deliverism suggests it doesn’t especially matter about how voters see individual policies - what matters is whether they work to achieve the changes they want to see. On this view, lower business taxes are fine if they boost growth, cruel immigration policies necessary if they reduce numbers, and private involvement in the NHS is worth it to reduce waiting lists (I’m not taking a view on the empirics of any of these examples).

Objectors to deliverism will say that voters don’t have a great handle on societal outcomes either. Consider Tom Calver’s recent article noting the gap between the issues people worry about personally, and their political priorities. If concerns on topics like crime or immigration are abstract rather than mediated by direct experience, politicians can achieve their objectives without getting any credit. Things can get better or worse without voters recognising they’ve got better or worse. Indeed, in the US, perceptions of the economy seem to depend more on which party is in charge than any change in the actual economy itself.

Fatalism

Moreover, deliverism presupposes that politicians can control outcomes. Yet, as in much of life, luck and circumstance play a huge role too. Politicians can promise economic growth, but how much they actually achieve depends on technological breakthroughs, international conflict and decisions made by trading partners as much as their own policies. More generally, there is a huge amount of contingency around electoral success. Incumbents governments around the world have been losing power because of a pandemic and inflation that had very little to do with them.

In that context, it is tempting to encourage politicians not to overthink it. Whether they win or lose the election probably won’t depend on policy, and quite possibly won’t depend on their action. So they should be liberated to pursue whatever they think is best.

I think there’s a grain of truth here, but it almost certainly overstates things.

Electoral politics is a land of contrasts

It’s a cop-out, I know, but I genuinely think that each of these theories has a models of truth. Fatalism is a useful corrective to the myth of the all-powerful politician, and its tempting to encourage policymakers to ignore politics. But it is too strong: implausible to think nothing makes any difference. On the other extreme, popularism shouldn’t be taken literally, and naive polling is a scourge. But popular policies surely help. I think the takeaway is to be alert to salience, to avoid taking polls at face value, but consider how they fit with how politicians are perceived and trying to present itself, and the electoral significance of the potential consequences of the policies. That won’t give us a determinate answer, but hopefully it at least gives us a few competing frames to approach the question.