The textbook I use for a Cost/Benefit Analysis course I teach is Public Finance, 8th edition, by Harvey S. Rosen and Ted Gayer. There's much good in that book. I particularly like their exposition about why most of health care and of education are private goods rather than public goods or goods with large positive externalities.
One part, though, that is off is their exposition of market failure on page 46, specifically, the idea that markets fail by failing to exist. Here's what they write:
After all, if a market for a commodity does not exist, then we can hardly expect the market to allocate it efficiently. In reality, markets for certain commodities fail to emerge. Consider, for instance, insurance, a very important commodity in a world of uncertainty. Despite the existence of firms such as Aetna and Allstate, there are certain events for which insurance simply cannot be purchased on the private market. For example, suppose you wanted to purchase insurance against the possibility of becoming poor. Would a firm in a competitive market ever find it profitable to supply "poverty insurance"? The answer is no, because if you purchased such insurance, you might decide not to work very hard. To discourage such behavior, the insurance firm would have to monitor your behavior to determine whether your low income was due to bad luck or to goofing off. However, to perform such monitoring would be very difficult or impossible. Hence, there is no market for poverty insurance--it simply cannot be purchased.
Basically, the problem here is asymmetric information--one party in a transaction has information that is not available to another. One rationalization for governmental income support programs is that they provide poverty insurance that is unavailable privately. The premium on this "insurance policy" is the taxes you pay when you are able to earn income. In the event of poverty, your benefit comes in the form of welfare payments. (bold in original)
But if monitoring is difficult (i.e., high cost) or impossible, isn't that a relevant cost to take account of in setting up a market? There's no market for people to repair $10 calculators. Does that mean that there's market failure? No. It means that the cost of repair is high relative to the cost of a new calculator. The absence of repairmen is not evidence of market failure. On the contrary, it's evidence of market success. The market has weeded out services that are not worth as much as they cost. Similarly, the fact that someone cannot monitor people at low cost suggests that the absence of poverty insurance is evidence of market success, not market failure.
Notice also how Rosen and Gayer jump from their claim of market failure to an argument for government to provide poverty insurance. They don't ever say that government can monitor at a lower cost. I doubt they believe that it can. And suddenly "poverty insurance" isn't insurance at all, but, rather, is a tax-and-subsidy scheme. Maybe this is why they use the word "rationalization" rather than "rationale." It surely is a rationalization. Also, note that they assume away the problem of monitoring when government is involved, writing, "The premium on this 'insurance policy' is the taxes you pay when you are able to earn income." By using the word "able," they make it sound as if when you're not earning income, it's because you're not able to. But, of course, that's a problem of monitoring. As they wrote in their first paragraph, when discussing the private insurance company, "the insurance firm would have to monitor your behavior to determine whether your low income was due to bad luck or to goofing off. " So, just like the private insurance company, the government would have to figure out whether you were able to earn income or unable to earn income.