Incentives drive outcomes: markets, socialism, natural monopolies and charities

Nathan Chadwick · October 27, 2026 · 15 min read

TLDRRead the short version

People respond to incentives, and most of the argument about how to run a society is really an argument about which incentives you are building, not which values you hold.
Where the incentive is clean and the target is simple, it works fast. The 5p bag charge cut bags from the big supermarkets by more than 95%. The sugar levy cut sugar in taxed drinks by about 47%.
Socialism fails the incentive test. It tries to replace the signal that tells people what to make and how hard to work. China's Xiaogang farmers got it back in 1978 and their output jumped. E. O. Wilson's line, quoted by Gad Saad, is 'Great idea. Wrong species.'
A natural monopoly breaks the market's main discipline, which is that you can leave. You cannot switch your railway track or your water main. That is where public ownership, or hard regulation, has a real case: rail track, water pipes, power wires. Not the shops that sit on top of them.
Charities have the opposite problem. The person who pays is not the person who benefits, so nothing punishes a feel-good programme that does not work. Saad calls the extreme version suicidal empathy.
The test for any policy, public or private, is the same: who gets rewarded, for what, and what happens when it fails?

Show me the incentive#

Adam Smith put it better than I can: we do not expect our dinner from the benevolence of the butcher, the brewer or the baker, but from their regard to their own interest. That line gets quoted as a defence of greed. I read it as a design principle. If you want a particular outcome, do not hope people will want it. Arrange things so that wanting it pays.

The other half is less comfortable. Incentives do not care about your intentions. A system built by kind people, to do a kind thing, will still produce whatever it rewards. If it rewards looking busy, you get busy. If it rewards the story, you get the story.

I want to run that one idea across four places: ordinary markets, socialism, the bits of the economy that are natural monopolies, and charities. In each, ask the same three questions. Who gets rewarded? For what? And what happens when it goes wrong?

Where a clean incentive just works#

The best evidence is boring and British.

In October 2015 England started charging 5p for a single-use carrier bag in big shops. The main supermarkets had handed out about 7.6 billion bags in 2014. By 2019/20 they sold 226 million, a fall of more than 95%. Nobody was lectured. Nobody needed to feel differently about plastic. A small price appeared at the till, and the behaviour moved.

The Soft Drinks Industry Levy is the cleverer version, because it did not tax the customer first. It taxed the manufacturer on sugar content, with a threshold. Between 2015 and 2024 the sales-weighted sugar in drinks in scope fell from 3.7g to 1.9g per 100ml, about 47%. The government's own review says roughly 65% of drinks that had been above the 5g threshold reformulated to get below it. Companies did the work themselves, because they were the ones holding the incentive. No health campaign did that.

What these have in common is a clear target, a price signal, and a party who can actually respond. Hold onto that, because the rest of this post is about places where one of those three is missing.

Incentives go wrong in markets too#

I am not going to pretend this is a story where private business has all the answers. Markets are exactly as good as the incentives inside them.

Wells Fargo is the standard example. Sales targets and compensation incentives led employees to open roughly 1.5 million deposit accounts and apply for roughly 565,000 credit cards that customers had not authorised, according to the bank's own analysis. Later legal filings put the number of bogus accounts as high as 3.5 million. The company fired 5,300 staff, but the staff were not the root cause. The scorecard asked for accounts opened. They gave it accounts opened.

That is the same lesson as the bag charge, pointed the other way. A powerful incentive aimed at a proxy (accounts) instead of the real goal (customers who are better off) will hit the proxy every time.

Why socialism keeps failing the incentive test#

First, a definition, because this word gets slippery. By socialism I mean the state owning and planning production, so that prices, profits and losses stop doing the job of telling people what to make and how hard to try. I do not mean the NHS, a state pension, or a generous welfare system. The Nordic countries are market economies with high taxes and large safety nets. That is a different thing from state ownership of the means of production.

My problem with the planned version is not that its goal is unkind. The goal is usually kind. The problem is mechanical, and there are two parts to it.

The first is information. Hayek's 1945 argument is that the knowledge needed to run an economy is scattered across millions of people, and the price system is how it gets passed around. Someone somewhere finds tin is scarcer, the price rises, and everyone who uses tin economises without needing to know why. Take the price away and a planner has to guess.

The second is motivation. If you will be fed whether or not you work hard, the sensible response is to work less hard. The cleanest natural experiment I know is Xiaogang, in Anhui province, China. In 1978 the village's 18 households secretly divided up the commune's land. They promised to hand over the state quota and keep whatever was left. Grain output went from around 15,000kg to something like four to six times that the following year, depending on which account you read, and the practice spread nationwide. Across China between 1978 and 1984, total grain output rose 33.6% and real agricultural value added rose 52.6%. One estimate attributes nearly half of that increase to the household responsibility system alone. Same land, same people, same weather. Different reward.

This is where Gad Saad's evolutionary framing gets useful. In The Parasitic Mind he quotes E. O. Wilson on socialism: "Great idea. Wrong species." His own gloss is that humans are both cooperative and competitive, that we form hierarchies, and that we are not indistinguishable worker ants, so any system built on a false picture of human nature is doomed. You can disagree with a lot of what Saad says elsewhere and still see that the core point holds up. A system that needs people to behave differently from how they actually behave is a system that will not work.

But some things are not markets#

If I stopped there, this would be a standard free-market post. It is not, because there is a class of thing where the market's main discipline does not exist.

The main discipline of a competitive market is exit. If the baker is bad, I go to another baker. The baker knows this, and so behaves. That whole mechanism assumes there is another baker.

A natural monopoly is the case where there cannot sensibly be one. It is cheaper for a single provider to serve the whole market than for several to compete, because the fixed cost of the network is enormous and duplicating it is waste. You do not want three rival railway tracks to your town, three water mains under your street, or three sets of power lines on the same road. And once you have one, the customer cannot walk.

So the incentive that normally keeps firms honest is switched off. Two things can replace it. You can regulate hard, which tries to imitate competition from the outside. Or you can own it publicly, which tries to align the operator with the people it serves. Neither is automatic. Here is how it has actually gone.

Rail#

The British railway privatisation of the 1990s split the track from the trains. The track company would hold a monopoly, would be remote from customers, and might struggle to attract investment. That was pointed out at a policy conference in 1988, before the model was adopted. After the Hatfield crash in October 2000, Railtrack's weaknesses in maintenance and asset management were exposed. The IPPR's verdict at the time is the cleanest statement of the incentive problem I have found: Railtrack "appeared unable to reconcile its public interest objectives with the interests of its shareholders to maximise profits". It went into administration in 2001 and was replaced by Network Rail, a not-for-profit company limited by guarantee.

To be fair, the same IPPR paper notes that passenger numbers grew strongly until Hatfield and that privatisation introduced some things worth keeping. It is not a story with only one villain.

The train operators are now moving the same way. Under the Passenger Railway Services (Public Ownership) Act 2024, 10 of the 14 contracted operators are in public ownership, starting with South Western Railway in May 2025. Great Western is due on 13 December 2026, Avanti on 7 March 2027, and the programme is meant to finish by the end of 2027. A Railways Bill to create Great British Railways is going through Parliament.

I think this is the right call for the network and probably for the trains, for the reason above: there is nowhere else to go. But I would not claim it is proven. It is too early to judge outcomes. Public ownership removes the dividend. It does not remove the need for an incentive. It swaps a shareholder for a minister, and a minister's incentive is the next election, not the 07:42 to Marylebone.

Water#

Water is the case I find hardest to defend as it stands. A water company has a captive customer base, a regulator, and a licence. That looks like a safe, boring, infrastructure business. Thames Water has a debt pile of about £20bn, its shareholders have walked away, and it is under the effective control of its creditors. On 15 October 2026 it asked them for another £2.4bn because it expected to run out of cash by the end of 2026 or the start of 2027. It has also been given a record £122.7m fine for sewage spills and unlawful shareholder payouts.

I read that as an incentive failure. If you can load a monopoly with debt, pay out dividends, and rely on the fact that nobody can let a water company fail, the gains go to the owners and the risk goes to the customer and the environment. It is not that the people involved are uniquely bad. It is that the rules reward it.

The government's own white paper is blunt that four regulators with overlapping and conflicting remits have not delivered, and proposes abolishing Ofwat and creating one new regulator. That reform has been slow, and what happens to Thames in particular is still undecided as I write. I would be wary of claiming I know the right answer for Thames specifically. The general point stands: a monopoly with a regulator is not a market, and pretending otherwise has been expensive.

Energy, where you have to split it#

Energy is the case that shows why you have to be precise. The wires and pipes are a natural monopoly. Selling electricity and gas to households is not obviously one.

Britain tried competition in the selling part. Between 2010 and May 2022 at least 73 new suppliers entered the market and at least 65 left. Between July 2021 and May 2022, 29 failed. When wholesale gas prices rose about six-fold in 2021 and the price cap stopped suppliers passing that on, those that had not hedged properly went under. The National Audit Office estimated that this would cost customers around £2.7bn, about £94 each on every bill, and found that Ofgem's licensing approach had increased the risk and cost of failure. The NAO also noted that new entrants had probably made some bills cheaper in earlier years. So it is not a clean verdict either way.

My conclusion is that the case for public ownership is strongest at the network layer: the track, the pipes, the wires. It is much weaker for everything that sits on top, where there can be real competition and where failing firms can fail. Mixing the two up is how you end up either defending the indefensible or nationalising a lot more than you need to.

What good public ownership would need#

This is opinion, not research, but it follows from the rest of the argument. If you take a natural monopoly into public hands, you have to rebuild the incentives deliberately, because the market will not do it for you:

One clear objective with published measures that the public can understand. Punctuality, cost per mile, leaks per day, not a list of fifteen aims.
Benchmarking against comparable operators, at home and abroad, so that bad performance has somewhere to be seen.
Pay and promotion that depend on outcomes managers can actually influence.
A hard budget with real consequences, rather than an assumption that the Treasury will always cover the gap.
Competition wherever competition is possible: building trains, maintenance contracts, retail supply.
An independent watchdog that is not run by the owner.

Public ownership of a natural monopoly is not the opposite of incentive thinking. It is incentive thinking applied to a place where the market's own incentive has gone missing.

Charities: where the feedback loop breaks#

Now the opposite problem. In an ordinary market the person who pays is also the person who benefits, so they check whether it works. In a charity they are different people. The donor pays. A stranger benefits. The donor usually never sees the outcome, and the beneficiary cannot take their business elsewhere.

That breaks the loop that tells everyone whether the thing is working. What gets rewarded instead is whatever makes the donor feel good at the moment of giving.

Suicidal empathy#

This is where Gad Saad's idea is useful. In his 2026 book Suicidal Empathy: Dying to Be Kind, and in earlier writing, he describes an empathy system "tricked into an orgiastic hyperactive form of empathy, deployed on the wrong targets". Another way he puts it: the inability to implement optimal decisions when you are conditioned to prioritise empathy, or displays of empathy, over a rational course of action. He contrasts it with reciprocal altruism, which is an evolved mechanism, and he says suicidal empathy is not.

A few of the remedies he proposes, as summarised in published reviews, are the ones I want to borrow: resist the immediate hit of moral satisfaction, avoid first-order thinking that ignores second and third-order consequences, and reject the single sympathetic example that overrides the statistics.

I should be straight about one thing. The term is Saad's own, it is contested, and it is not an established category in psychology, so I am using it as a lens and not a diagnosis. The book applies it to some highly charged political subjects. I am not relying on those examples here. The mechanism stands without them: care aimed by how vivid something is, not by what it achieves.

The experiment#

The sharpest evidence for that mechanism is a small study that is not about politics at all. Small, Loewenstein and Slovic gave people $5 and a Save the Children appeal. People shown statistics about millions of hungry children gave an average of $1.14. People shown one named seven-year-old girl, Rokia, gave $2.38.

The less comfortable part is what happened next. When people were taught about the bias, or put in an analytical frame of mind, they gave less to Rokia but did not give more to the statistical victims. Total giving fell. So the answer is not simply to tell people to be more rational. Willpower and exhortation do not reliably fix it. The fix has to be built into the system: who evaluates, what gets funded, and what counts as success.

The charity incentive#

There is a second incentive problem inside the charity itself. This part is my reasoning, not a finding. A charity's revenue depends on donors' feelings, so the cheapest way to raise money is a vivid story, not a rigorous result. And a charity that completely solved its problem would have no reason to exist. Nobody is acting in bad faith. The structure just rewards visibility over effect.

The clearest example I know is the PlayPump. It was a children's roundabout wired to a water pump, so that kids playing in a village would pull up clean water. It is hard to imagine a story that is easier to love. In September 2006 the then US First Lady announced a $16.4 million investment from government agencies and foundations to roll it out across Africa, and it collected celebrity and political backing along the way.

Then UNICEF evaluated it in 2007. It called the technology itself innovative and robust, so this was not a case of a stupid idea. The trouble was everything around it. Each unit cost about $14,000, up from $6,500 two years earlier without explanation, which would have paid for several conventional handpump boreholes. In Zambia, users at 63% of the sites visited said they had not been properly consulted and had been given no choice of technology. In one Zambian community adults were paying children to play on it. UNICEF concluded the rollout strategy needed serious and urgent revision, and the aggressive rollout ended in 2009.

Look at who was rewarded. The donors got a lovely photograph and a story. The people with the actual problem, who needed water, had no vote and no exit. Nobody in that chain was paid to find out whether the thing delivered more water per dollar than a boring handpump. That is the missing feedback loop, in one project.

What I take from all of this#

I came out of this with a handful of rules.

Ask who gets rewarded for what. Not who means well. The scorecard decides the outcome, whatever the mission statement says.

Aim the reward at the real goal, not a proxy. Wells Fargo rewarded accounts. The sugar levy rewarded sugar reduction, which is what it was for.

Be honest about which kind of market you have. Where customers can leave and competitors can enter, let that do the work. Where they cannot, either regulate as though you mean it or own it, and rebuild the incentives on purpose.

Split the monopoly from the market. The pipes, wires and track are one thing. The retail shop on top is another.

Do not fund by feeling. If the person paying cannot see the result, someone has to measure it for them. Randomised trials, independent evaluators, and boring outcome numbers are how you protect donors from their own kindness.

Measure outcomes, and be prepared to be wrong. The PlayPump made a lovely story and a poor water supply. Finding that out needed someone independent to go and look.

None of that is cold. It is the opposite. If you care about what happens to people, you have to care about what actually happens to them, not just about how it feels to try.