Houses are built and sold on a different model. If we equate authors with a construction company, respectively all the different tradesmen involved in building construction, the one big difference is that from the construction company's point of view houses aren't build speculatively.
A construction company builds a house because a clearly defined entity commissions it to do so, so setting up a one-and-done-style transaction (I pay you for your whole effort in constructing the house and that's that) is only natural, and while it's not cheap, people are also willing to pay that kind of price because once built a house will last decades and also because you have to have some kind of shelter. (Though the existence of renting shows that other economical models for financing housing are possible, too).
Books on the other hand are a different matter. Writing a book usually takes on the order of months to years, but few people are willing to individually pay a writer just in order to be able to read a single book. Paying a writer's cost of living for a year or so might be only an order of magnitude or two cheaper than building a house, but compared to a house the value you get out of a typical book is definitively several orders of magnitude lower than the value you get out of a house.
In order for books to be viable, their cost of production needs to be split up among many people, and what's worse, that set of people isn't really known in advance. So the model we've arrived at is that somebody (either a publishing house, but often actually the author him- or herself) fronts that cost and then expects to make it back bit by bit through individual sales, which means that people wanting to read that book are definitively expected to pay their fair share.
And because the whole thing still is somewhat of a speculative endeavour (a book might be genuinely bad or otherwise unpopular, or just have bad luck/timing and therefore not become popular), it also means that some books will still not make back their cost of production, which conversely also means that, other, more popular books will earn healthy profits – but if you just cut out the profits from the popular works, it might mean the averages on which the whole model is based might no longer work out.
(Yes, for popular authors with a reputation to uphold things like preorders, Kickstarter, etc. can work out, too, but you somehow need to be able to get to that point of popularity in the first place)
As for roads – the relationship between road construction companies and whoever commissions that road is the same as with building houses – a clearly defined buyer, who can afford paying the price of a once-and-done transaction because the road will deliver enough value.
If we're looking at the relationship between whoever commissioned that road and the actual end users on the other hand, there are actually some similarities to the models used for books there: The set of expected users of that road isn't clearly defined, yet the costs need to be split up among them. Most of the time that issue is side-stepped by simply paying for roads out of general (or occasionally special) taxation, but other times you get things like toll roads, were road users are very well expected to pay for them for a long time, possibly even after the original construction costs have long been paid off, and maybe more than would just be required for ongoing maintenance.