I've suggested here several times before in various contexts that corruption is more complex (and more interesting as a problem) than the usual linear focus on the "bad" foreign government leader(s) or, as Milanovic says, a "corrupt state." But how does "organized crime and its supporters become the largest employers in the country"? One focus may be on traditional, informal economic transactions in a given country (baksheesh, etc.). When these are elevated to the status of official ways of getting things done in a country on a larger scale, we've got what international agencies generally think of as corruption. But these transactions may remain - at least on the small scale - important parts of social life and the daily distribution of goods and services. In tackling corruption defined in these terms, an international agency necessarily - though perhaps unconsciously - is taking on important elements of the culture which may have deep historical roots. Indirectly disrupting these processes in the name of stemming large-scale corruption by replacing them with "proper" forms of market transaction often requires deeper transformations of a society that the international development agencies, complex as they are, are ill-equipped to handle. When they fail, they blame the corrupt leader.
Once organized crime and its supporters become the largest employers in the country, they play the same role that a more conventional business plays in other countries. They try to influence the political process. Moreover, they need to control the political arena - election of presidents and parliaments - even more tightly than "normal" business people because their very existence depends on having a government willing to tolerate violation of international rules as the country's main activity.
The government structure that emerges is "endogenous": It reflects domestic social and economic structure, which in turn is the outcome of greater international trade and economic incentives, much like other countries, except that the governance structure is, almost inevitably, more corrupt. The recent World Bank and International Monetary Fund's insistence on reforming governance in these countries is bound to fail because the cause is misdiagnosed.Governance is viewed by the international organizations as something "exogenous," something that a country just happens to have and which - through a better electoral process, more transparent laws and more honest lawmakers - can be improved. Thus the international organizations are in a permanent, and fruitless, search of an "honest" lawmaker, an Eliot Ness who will bust corruption and illegality. They fail to notice that governance structures respond to underlying incentives, and to expect an honest person to rise to power in a corrupt state is akin to expecting a person with no financial backing from big business to be elected president of the US. In both cases, the outcome of a political process reflects the country's underlying economic conditions.
A different approach is necessary: legalize the currently illegal activities like prostitution and drug use and modify the often draconian US and European immigration laws that stimulate human trafficking. If prostitution and drugs indeed became like haircuts and candies, their production would obey the same rules: Countries that export beauty services and confectionary products are not notably more corrupt than others. Some of the current entrepreneurs would remain in these activities, others would move to others. In either case, there would be a general "normalization" akin to what was observed after prohibition on alcohol sales was lifted in the US. Thousands of "bootleggers" became normal producers of alcohol, alcohol-linked criminality decreased, and only a minority of those with preference for high risk and crime moved to other illegal activities.
Corruption is sustained through at least three further systematic elements: various structural conditions of the global system with diverse concrete effects (such as the international borrowing privilege); structural adjustments that put market assumptions first at the expense of goods and services desperately needed on the ground in the short term, whatever the particular political system or government's effectiveness; and incentive packages structured around the age-old economic assumption of individual self-interest, especially in countries with strong communitarian values. There's much more to say here than I can right now. But when we take into account these and various other system-focused elements of corruption (conceptualizing corruption as misdirected distribution of material assistance and undemocratic distortion of ongoing processes of institutional reconstruction), anti-corruption policies shake off their linear assumptions and become self-critical as well as at least potentially more experimental when policies fail (which they do, over and over).
The key is that meaningful reforms do not begin in the corrupt states themselves, but in the rich world that is the main consumer of illegal goods and services. This requires a total overhaul in our thinking about the root cause of a corrupt state. Many of the most corrupt states are "corrupt" because they specialize in goods and services that are deemed illegal. But what is illegal today is not necessarily illegal tomorrow. "Illegality" is a historical category, as the long history of accepted prostitution and drug use shows. Thus if illegality is the main cause of corrupt governments, then the best way to root out corruption is to remove illegality.Milanovic suggests that a little linguistic legerdemain can get rid of the problem. While I think there's actually a profound point embedded in here regarding our own institutionalized conceptions of propriety, it's lost through his cavalier framing of the solution, reminiscent of older libertarian proposals to legalize drugs and thus subject them to the market and minimal regulation (assuming that illegality is maximal regulation). But corruption is not merely about a blackmarket trade in "illegal" goods and services. Theft from public funds, for example, can hardly be legalized and subject to market mechanisms. In the developed nations, public funds used for dubious purposes may seem like stealing but these processes aren't usually subject to public market forces either. Corruption is, however, a product of economic and noneconomic structural conditions outlined above, and incentives based in undemocratic features of governance (such as lack of transparency and accountability, rather obviously).
Corruption often takes place in the space provided by severely weakened institutions of governance. Focusing on economic incentives may very well contribute to our understanding of corruption, and the political problem is not merely endogenous. Milanovic rightly points this out in his criticism of the IMF and World Bank. But in the end the problem is more political than economic, and so both the development agencies and Milanovic have only partially correct views: the IMF/World Bank because of the linear view that corruption is endogenous, and Milanovic because of the view that corruption is a matter of stunted market forces. A clearer understanding of the exogeneity of corruption would lead us to understand how various economic and political assumptions shared by the international development agencies recreate structural conditions that allow corruption to flourish at cross-purposes with their own stated objectives.