I guess if we removed Hitler, Stalin, Franco, Pinochet etc... as these may be described as individual leadership failures .... government failures may seem rarely as severe, or serious and frequent as market failures. by market I simply refer to financial (banking and currency),
[WS:] Hmm... I thought of "government failures" in more economic than political terms i.e. as government involvement in propping up faltering demand, investment, etc. From that point of view, Stalin and Hitler were great success stories, even though their politics sucked big time.
There is a certain train of thought in the neoclassical econ (cf. Burton Weisbrod) arguing that "government failure" i.e. inability to finance public goods at the level adequate to meet the existing demand is a result of diversity and democratic process (i.e. lack of consensus which public goods should be publicly funded.) If you place that argument in the context of Arrow's impossibility theorem (http://en.wikipedia.org/wiki/Arrow's_impossibility_theorem) you can make a case for relaxing the non-dictatorship constraint i.e. argue that one needs to remove the non-dictatorship constraint (or "median voter" influence on government in Weisbrod's parlance) to create a social welfare system that meets the multitude of voters' preferences. Weisbrod does not go that way, however, but instead adds a deux ex machina - an extra market force (altruism or nonprofit institutions) to bail the social system from such government failure. The problem with his solution, however, is that he does not have a good behavioral model of his deus ex machina actor, other than claiming his altruistic motivation - which in my view undermines his rationality assumption (see my piece "The Death Knell of Utilitarianism" http://www.springerlink.com/content/w52027l3h21205w7/fulltext.pdf)
The above suggests that from a theoretical perspective, government failure is a much thornier issue than market failure. It is so, because in the neoclassical econ government is an institutional solution to the recognized market failures (i.e. the agent producing public goods, which the markets cannot do efficiently). The system works in a cascading fashion, so to speak: first there are the markets that distribute most of goods most efficiently, and what cannot be efficiently distributed by the markets (i.e. public goods) - is cascaded to government. Now if the government fails, there are only two solutions possible: (i) either find another actor to which that failure is cascaded for a solution (and Weisbrod finds it in philanthropic/nonprofit institutions) or (ii) conclude that the market system can fail altogether to deliver certain kinds of goods.
Since the second conclusion is politically unacceptable, after all we live in the best of possible worlds, no? ;) the first conclusions offers a way out - private philanthropy is pronounced a savior from government failures. Never mind that it undermined the core theoretical assumption of the whole school of thought - namely the sufficiency of the rationality assumption to explain economic behavior. Since philanthropists are not motivated by a desire to maximize their gain, they are not rational in the economic sense - so that non-rationality seems to be the final solution from the failures of "rational" systems (markets and governments).
Of course, moving away from the neoclassical theory, government failure is more of a practical than theoretical problem, and much more serious than a market failure. History has demonstrated that societies can function, and even moderately prosper without efficient markets (cf. x-USSR), but they utterly fail without efficient government that provides sufficient level of investment and distribution of goods (cf. most African states). That is to say, we can live with market failures, but it is much harder to function in case of government failures.
Wojtek