Would Labour have handled the A level fiasco better than the Tories? There's a reason to think so, and it has nothing to do with the parties' relative competence.

Instead, it's because Labour ministers would have had different sensibilities, or sympathies. They would have been more awake than the Tories to the situation of 18-year-olds from poor schools and hence more alert to developments which might hurt them.

The point here is a Smithian one. Smith thought that sympathy "arise[s] from an imaginary change of situations with the person principally concerned." But, as Smith saw, our imagination is limited. It's easier to put yourself in someone's shoes if you have actually been in them, or if you have friends or family in those shoes, or if you expect to be in them later. Although there are many exceptions on either side, Labour MPs therefore sympathize more with people from poor backgrounds whilst Tories are more likely to sympathize with landlords, bankers or bosses. For this reason, some sections of capital would have disproportionate influence over a Tory government even if not a single pound changed hands.

Not that capital wants to leave it at that. It's the principle of sympathy that justifies JP Morgan hiring Sajid Javid. Politicians around the world will look at that and think "that could be me". This feat of imagination will increase their sympathies with bankers.

Obvious as it seems, this Smithian idea explains a lot. It explains why some people sympathize more with migrants than others. If you feel an outsider in the UK - say because of your race, class, sexuality or just idiosyncratic temperament - you'll find it easier to put yourselves in the shoes of those who feel compelled to leave their country. Logically, rightist lovers of freedom should support freedom of movement as much as leftists. But differences in sympathies mean that in fact things are otherwise.

It also explains why so many on the left had misgivings about New Labour, even though it did much to help working people such as introducing the minimum wage and tax credits. There was a feeling that its sympathies lay more with bosses and bankers than with working people.

It also explains the long-running electoral success of the Tories. Their sympathies make them instinctively more likely to placate enough of the rich and powerful to earn a reputation for competence. Nobody gives a damn if your policies hurt single parents in Wigan.

Which is why it matters that top jobs are dominated by people from private schools. Because we all find it easier to sympathize with people like us, the dominance of the privately educated therefore skews decision-making (not least in the media) towards the interests of the rich and against those of the poor. And this would be the case even if the decision-makers were honest and of good will.

All this both justifies representative democracy and helps explain why it goes wrong.

Voters aren't qualified to judge the technical competence of the parties, nor even their detailed policies - not even in aggregate, given that the conditions required for the wisdom of crowds are absent. What they can do, though, is say which set of sympathies they prefer. And that's reasonable. In politics there is an agency problem - the danger that the government won't act in our interests. Knowing that its sympathies are in line with our own, however, reduces this risk.

Differences in sympathies, however, can also help explain some of the bigger policy failures of our times. Thatcher's introduction of the poll tax arose from not sympathizing with people who didn't own property. The occupations of Iraq and Afghanistan might have been less disastrous if New Labour had better sympathy with military men, and wouldn't have occurred at all if Blair had more sympathy for working people and less for George Bush. The introduction of Universal Credit would have gone better had there been sympathy for benefit recipients. And so on up to the A level mess.

Which brings us to a problem. The greater are divisions within a society, the harder it is for one side to sympathize with the other. Which means that policy errors are more likely. As long ago as 1995 Patrick Dunleavy argued (pdf) that the UK was "unusually prone to make large-scale, avoidable policy mistakes" - a point confirmed by Ivor Crewe and Anthony King in The Blunders of our Governments and of course corroborated by subsequent events. Which poses the question. Might it be that this unusual proneness to error arises in part from our deep class divisions*? If so, then good technocratic government actually requires the abolition of those divisions.

* Dunleavy, Crewe and King offer more proximate explanations for bad government. But at least one of these - overconfidence - is most certainly a class issue.