"人们买了保险，的确是会放松警惕，车祸的比例会上升。但为什么会出现这种情况？你不能说是因为“道德风险”，因为“道德风险”的定义就是买了保险故意撞车，用定义去解释定义，不是同义反复是什么？"
(1) My interpretation of (李俊慧's interpretation of) Cheung's arguement is that:
- Moral hazard assumes people cheats when monitoring is imperfect. So if we use moral hazard to explain why people "cheats". It would be tautological.
(2) Firstly, I don't think "“道德风险”的定义就是买了保险故意撞车", moral hozard just constructs a situation with the type of asymmetries of information that develop subsequent to the signing of a contract.
"买了保险故意撞车" is the result of the maximization postulate in this moral hazard setting. So you can say this is an immediate dedeuction result, but not a tautology.
(Note, however, results of logical dedeuction is by definition always consistent with the assumptions. So if you want, you can consider all assumptions, relevant definitions and deduced implication as a proposition as a whole, in this case, it is always a tautology!)
(3) Yes, you're right. It is meaningless to answer why people cheats by assuming it. But as far as I know, moral hazard models are not aimed to answer this question. ( Just like (traditional) monetary economics is not aimed to answer the question why we have money, but just assume its existence. If you use monetary theory to answer this question, it would be tautological.)
Instead, they consider, given a moral hazard setting, how can we explain certain behaviour of agents or contractual arrangements.
For example, in the case of insurance, we consider how the equilibrium insurance contract depends on the information structure (observibility of agents'action) and other exogenuous variables. We want to know when we will have a full/partial compensation contract. Is an answer to this type of question a tautology?
Another example is the moral hazard game of firm owners & managers. We DO NOT use moral hazard to explain why agents shirk. But we assume imperfect monitoring, and see how equilibrium wage contracts are determined.
For example, we want to see in what situation we will have a fixed wage contract, a proportional contract, a linear contract, a short-term contract, a long-term contract, a tournamant (eg. promotion reward). How the equilibrium contractual arrangement are related to the information structure, monitoring cost, number and characteristics of agents, ... Are answers to these questions tautological? Are all these implications not testable? I don't think so.
(4) You may say "hey, we can explain all these very well by using the concept of transaction cost."
In principal, yes. I guess you can re-label the information variable & monitoring cost by the concept of transaction cost. But is that really a better and clearer approach? To be a reasonable candidate, at least we need to know what transaction cost exactly mean. Because if we don't specify what tranaction cost refers to in real life, we can always use transaction cost as a "residual term". Whatever we see, we can say, "oh, because the transaction cost is so high/low that we observe this phenomena ..." If this is the case, the explanation using transaction cost becomes not refutable! That's why I always ask for a precise definition of this concept.
According to Cheung's definition, it seems that the cost of "playing tennis" is also a transaction cost. This is because playing tennis is not conceivable in an one man economy. That's surely not a good definition, no?
SEE：http://bbs.efnchina.com/dispbbs.asp?boardID=57&ID=8572&replyID=27811&skin=1
