This is pretty key.
Imagine a train line that allows commuters to get to work. Trains are expensive to run, so the actual cost to get a commuter to work and back is £100. The commuters are paid (say) £150 a day, after tax. Is this train worth 2/3 of their post-tax income? Probably not, so they won't use it, and the company can't get workers if there's no other practical way to commute. Workers can take less good jobs near home, earn less, but take home more. Or even no job at all. However, a worker generates substantially more than their post-tax salary in value to the economy as a whole, so subsidy of the train fare creates value by getting them to work and generating that value, even though the train cannot actually turn a profit itself by charging the commuters out of their income.
Even if you say "well the company should just pay more if they've made that value", not all that value manifests directly on the company's bottom line, it includes downstream value, as well as intangible things like worker skills that are more of an abstract societal benefit.
In the same way, roads produce massively more value than people would be willing to pay individually: all the food deliveries in a week might be worth, at retail, about £4 billion, say. If those deliveries can't be made, what will be the cost? £4 billion? Or more because the whole country will become a much less effective economy when everyone is starving and looking for food? And the universal-delivery postal system. Healthcare, childcare, energy, etc etc.