Ambulances: cui bono?

Public services and risk-taking

Who benefits from public services? Or, to give a particular example: who benefits from a well-functioning ambulance service?

There is one obvious and entirely correct answer: the patients. People having a stroke, kids with sepsis, little old ladies who fall down the stairs—those in medical emergencies who are helped (whose lives are often saved) by quick responses and emergency treatment. Ambulances help these people in the first instance.

You can also say that the friends and family of the patient benefit from this—their loved ones remain alive and well—and maybe also that society at large benefits via the future health, happiness, and productivity of the patients.

All of these answers are correct, but also boring. Are there any other groups of people who benefit, in more indirect but perhaps more interesting ways, from a well-functioning ambulance service? I think the answer is yes: a well-functioning ambulance service effectively subsidises certain types of risk-taking.

Here’s a high-level analysis of one way that might work:

1. Suppose that there’s some endeavour (e.g., rock-climbing) that people take pleasure in, but which comes along with a small risk of injury or even (in the worst-case scenario) death. You’d enjoy the activity if you did it, but you’re worried about the risks, so you avoid it.

   But if you know that, in case of an emergency (e.g., a serious fall off the rocks), there are ambulance workers who are willing and able to give you emergency care and take you to the hospital, this reduces the possible downsides of the activity, making you more likely to give it a go. So in a society where the ambulance service is functioning well, people will be more willing to take part in these risky endeavours.

If an activity is risky to your health—in particular, if it carries a small risk of medical emergency—then you will ipso facto be less likely to perform it. But the ambulance service offsets that risk somewhat, and so makes the activity more attractive. People who are predisposed to enjoy risky activities are thus benefitted by the ambulance service more than those who are predisposed to enjoy less-risky ones: potential rock-climbers get the benefit, potential knitters don’t.

Is this a good thing? After all, there’s nothing inherently wrong with avoiding risky behaviour; should we really be encouraging people to take more risks, especially when it means subsidising certain activities over others? Well, here’s one argument that we should. Evidence from decision theory and psychology suggests that (in general) people are too unwilling to take risky actions, avoiding them even when the risks should be outweighed by the possible reward.¹ This means that they will tend to avoid risky activities more often than they should, paying an overly high premium for safety and certainty.² A well-functioning ambulance service, by reducing the risks associated with possible medical emergencies, makes it easier for risk-averse people to do the things that make them happy.

The above is a variation on a quite standard template for behavioural-economics arguments, and (like those arguments) it is implicitly reliant on some quite controversial background assumptions.³ So let’s see if we can’t try to give a different account of how subsidising risk-taking through an ambulance service might be a good thing. Here’s a possible story:

2. You, like me, might not own a car—especially if you live in a large or particularly dense city. And you might also agree with me that owning a car is a bad thing on the margin: cars contribute to pollution and congestion, and widespread car use incentivises the government to build infrastructure that is bad for the life of the city.

   Yet living without a car is risky. To be sure, it reduces your risk of being in an accident (although if you cycle, perhaps not by that much); but it comes along with a whole host of other costs. In particular, it’s harder to get around in emergencies: if you’re having a stroke and can’t get someone to drive you to the hospital, you certainly can’t just cycle there, and the bus will hardly be quick enough in an emergency where every second counts.

   But, if you can rely on the ambulance service, this risk gets offset: there are a group of people whose job it is to drive you to the hospital in the event of an emergency, so you don’t have to worry. This effectively subsidises car-free living; and if you want to encourage more car-free living, this is a good thing.

I feel like this story is pretty plausible. Certainly, if my partner were to get pregnant and I was not confident in the UK’s ambulance service, I would definitely buy a car so I could take her to the hospital when necessary. This would cost me money, yes, but it would also come at a (vaguer, more subjective) cost to my lifestyle. A functioning ambulance service subsidises my behaviour, benefitting me more than those who already have a car or who wouldn’t mind getting one.

But if owning a car is a bad thing, then subsidising my behaviour is exactly what we should want. This argument doesn’t assume anything about people’s risk-aversion or irrationality. It just says that people are acting in a risk-averse manner that, even if it is rational, has bad effects on society; thus, a service that reduces the risks (causing people to change their behaviour) is good for society. There are many other examples that fit this pattern, not just car use.

Of course, my point is so general that it can go both ways. Risk-taking behaviour, while sometimes positive, can be negative; we need to judge each case on its own merits. And perhaps the most negative risk-taking behaviour is violent crime. Here’s another possible story:

3. For almost every violent crime, there’s a chance that it turns into a homicide, even if that was not the criminal’s intention. If this seems a little weird to you, it’s worth looking at the surprising number of people killed by a single punch—never mind when criminals start using weapons, even just as a threat. Things go wrong, situations escalate, people have emergencies, etc.

   Now, in most jurisdictions, the police take homicide very seriously, taking much more action over homicide than almost any other category of crime. So when someone dies as a result of violent crime, it leads to a police crackdown; and the risk of this crackdown acts as a deterrent to violent crime, over and above the deterrent of normal police action against violent crime.

   But when there’s a well-functioning ambulance service, more victims of violent crime can be saved from death. This is of course a good thing in itself; but it reduces the risk that a violent crime will turn into a homicide, which in turn reduces the risk of a police crackdown. And through this pathway, the ambulance service effectively subsidises violent crime.

Sounds bad!

How strong are these effects? I imagine 3 is pretty weak, at least on our current margins—probably too weak to be picked up in the data, if the effect even exists. 2 is probably also quite weak. (Indeed, since driving a car comes along with its own risks, the effect of an ambulance service could in principle go the other way, making car-driving more popular rather than less.) Effects like the rock-climbing example in 1 seem to me like they would be much stronger, but it would be harder to disentangle these effects from others in an empirical study—e.g., you couldn’t just compare ‘effectiveness of ambulance service’ and ‘prevalence of risky activities’ across a wide range of countries, because whether or not a country has an effective ambulance service depends on a whole range of factors that might also affect its citizens’ propensity for risk-taking.

But I think that the general point probably holds (the examples given are just illustrative, not meant to be exhaustive): a well-functioning ambulance service, on top of benefitting people in medical emergencies, has a more general and diffuse effect of subsidising certain types of risk-taking behaviour. It thus benefits those who are predisposed towards these types of behaviour—e.g., potential rock-climbers, cyclists, and violent criminals.

And my judgment is this is probably net-positive. A well-functioning ambulance service gives people the subjective freedom to do the things that make them happy without being overly worried about risks; even if some people use this freedom to do bad things, it’s normally a good general approach to endorse the freedom.

---

I think the above analysis extends beyond just the ambulance service. More broadly, well-functioning health service also lets people take more risks with their health, knowing that they can offset some of them. Outside of healthcare, benefits for unemployed people can encourage more risk-taking from those in the job market: people can afford to look around for better employment rather than just sticking with a sure source of income. In general, public services and programmes in general tend to subsidise risky behaviour, because they offset some of the risk: different programmes subsidise different types of behaviour, but the structure is similar across all of them. This is sometimes recognised in the case certain public services and programmes (especially the dole), but it is not often realised how general it is; the reflections on the ambulance service above are intended to bring that generality out.

I think this offers an interesting way to think about the ‘safety net’ metaphor for public services and programmes. Normally, the metaphor is cashed out like this: a safety net is something that catches you when you fall, just as the state effectively protects those who suffer bad outcomes like unemployment or illness. But another (not incompatible) way of thinking about it is this. A safety net is something that reduces risk, making it more attractive for you to partake in otherwise-risky behaviours.⁴

Economists call this kind of situation, where you become more willing to take risks because you know someone else will bear part of the cost, a ‘moral hazard’—a terrible name because it is supposed to just be a bit of technical terminology, but it has obvious negative connotations. To be sure, if people would otherwise behave entirely optimally, then the introduction of ‘moral hazard’ is unquestionably a bad thing because it causes them to change their behaviour away from the ideal. But that’s a crazy assumption!

In real life, we have to use our judgment to decide whether or not the introduction of a ‘moral hazard’ will be a good or bad thing. And it is a reasonable judgment to make that people should often be enabled to take more risks; the previous section offers some reasons to think this way. If you, like me, think that increased risk-taking is often net-valuable (on the margin), then this gives you a new way to think about the importance of public services.

I leave with a final prediction: as the state begins to creak and public services begin to fail, then we should expect a general movement towards small-c conservatism in citizens’ decision-making—less risk-taking and less boldness. This, once again, seems like a bad thing to me for a time of crisis; your judgment may differ.

1. Most of this literature begins from the famous Allais paradox, introduced in Allais (1953) [nota bene: written in French], which is an example that suggests most people are risk-averse in a manner that is inconsistent with orthodox decision theory. A classic empirical paper which tries to show that most people have risk-averse ‘Allais preferences’ is Slovic and Tversky (1974), and Oliver (2003) who discusses the issue in a healthcare setting. Weber (1998) is a compelling philosophical argument against attempts to ‘explain away’ the Allais paradox and risk-aversion.

Note, however, that the argument of the main text only works on the assumptions that having Allais preferences are a bad thing, that it makes you overly risk-averse (not just reasonably risk-averse). This is essentially the assumption that orthodox decision theory prescribes the right way to think about risky choices. And that is a can of worms that I am not opening in this post.

2. In addition (this one’s just for the nerds), I think we should expect insurance to be underprovided if people are risk-averse in this manner, at least in cases where people can avoid the risky behaviour entirely: too few people would be willing to pay insurance premiums (because they have a less-risky alternative—simply avoiding the risky behaviour), so insurance provision becomes less profitable. But this is obviously sensitive to the assumptions that are used.

3. As mentioned at the end of footnote 1; see Dietrich et al. (2020) for a technical analysis of some of the relevant issues, drawing on Broome (2013).

4. This insight is partially captured through the analogy with insurance, but only partly—not least because something like an ambulance service doesn’t function much like insurance in practice. There are also quasi-Jamesian considerations: the presence of a literal safety net can make you less likely to fall, by reducing your fear and making you more confident; by analogy, the presence of a social safety net can increase your confidence and make good outcomes more likely.