I’ve posted before (see here, here, and here) about some of my concerns regarding the accuracy of the estimates people sometimes throw around about the total amount of bribes paid each year (sometimes given in absolute terms, sometimes as a percentage of global GDP, but in all cases based on dubious extrapolations from suspect data). But for the moment I want to put those concerns aside to make another point: Even if we knew the total amount of bribes paid, that would not necessarily tell us much of anything about how much bribery costs society. (And that’s true even if we limited attention to economic costs, narrowly construed.) This is not an original point – lots of people have made it, and indeed it’s fairly obvious when you stop to think about it. Yet I keep seeing references to estimates of the amount of bribery that treat these figures as if they were measures of the cost of bribery. (For examples, see here, here, here, here, and here.) But that’s just not right.
The key to understanding why it’s not right is to recognize that a bribe payment is, by itself, just a transfer of resources from one person or entity to another. That transfer by itself does not decrease (or increase) the overall resources in the economy; it just moves resources around. (A qualification: From the perspective of utility, as opposed to wealth, transfers can make a difference if we make the conventional assumption that poorer people value additional wealth more than richer people. But most discussions of the aggregate costs of bribery don’t focus on distributive effects, and they express costs in terms of dollar amounts, not utility, so we can put this qualification to one side for now.)
Does the fact that bribery is simply a transfer of wealth mean bribery is harmless? Of course not! But the economic costs of bribery are not the bribes themselves, but rather the economic distortions and inefficiencies associated with the practice of bribery. These would include, for example: the costs associated with lower domestic and foreign investment in otherwise productive projects (if bribery demands deter investment), excessive red tape (if officials impose excessive bureaucratic requirements in order to extract more bribes), inefficient provision of public goods (if contractors are selected on the basis of their ability to pay bribes rather than their qualifications), distorted public policy priorities (if governments over-invest scarce public resources in inefficient projects that maximize the ability to extract bribes), and misallocation of talent (if the prospect of lucrative bribes lures would-be entrepreneurs into less productive public sector jobs, or induces them to devote their talents to cultivating connections and identifying corrupt rent-seeking opportunities rather than productive ventures).
Once we recognize this, it’s possible to imagine a transaction in which a very large bribe imposes a relatively small cost on the economy, or one in which a relatively small bribe entails very large costs. As an example of the former, suppose that two firms are competing for a lucrative government contract, and that one firm—the one that is slightly less efficient—wins the contract by paying a huge bribe. In this case, the social cost of the bribery is much smaller than the size of the bribe. For an example of the latter, think about the forest ranger who takes a small payment in return for allowing a logger to clear-cut an old-growth forest. or the building inspector who allows the construction of scores of unsafe dwellings in an earthquake-prone area in exchange for petty bribes. Here, relatively small bribes have huge costs.
In the aggregate, I guess it might be reasonable to supposed that the total amount of bribe transfers are correlated with the total costs of bribery. If bribery is, on average, costly to society (which I think it almost certainly is), then more bribery probably signifies more social costs. But the relationship is sufficiently loose that we cannot and should not treat estimates of the aggregate amount of bribery as if they were estimates of the aggregate cost of bribery. The latter might be much smaller or much larger than the former.
I acknowledge this isn’t a hugely consequential point, especially since these numbers are mostly made up anyway (and are really just used for the rhetorical purpose of emphasizing that corruption is a big problem). And as I acknowledged at the outset, this is hardly an original observation; indeed, some readers may find it obvious to the point of banality (and point out that journalistic references that exhibit confusion on this point are at worst sloppy, and mostly harmless overall). But I still think (1) we should strive, to the best of our abilities, to be careful and precise in how we talk about corruption, and (2) the distinction between amounts and costs is itself an important conceptual point to keep in mind when thinking about bribery’s effects, even putting aside whether we care all that much about aggregate quantitative estimates.
Thanks for this really fascinating post! I am not sure, if your point is as trivial as you make it out to be. I am going out on a limb here and have not thought this through at at all but this is what came to mind, while reading your post: I am not sure if we can reduce the societal costs of corruption to efficiency issues. The societal costs (all efficiency and externalities aside) of paying for a service that is meant to be for free would then be equivalent to being robbed or equivalent to a kleptocrat stealing all oil resource of a country. The latter two would also just be private-to-private or public-to-private transfers and according to your definition not count as societal costs. But how about if the kleptocrat takes the money abroad? Just a private to foreign transfer – no societal costs? I think the argumentation then becomes progressively strange which would then lead me back to a conclusion that any wealth transfer that is simply unduly monetized entrusted power and deviates from the policy-intended plus market shaped distribution of wealth should be counted as societal costs of corruption. And then of course your argumentation on the efficiency side applies and might produce smaller, negative or larger net effects so the central argument holds that amounts paid do not equal costs. So I would think the calculation should read cost of corruption = volume of bribes + or minus efficiency costs/benefits, rather than setting the cost of volume of bribes to zero. Again, I have no idea if I am right with this since I have not thought this through in any detail. But it somehow feels reasonable at first blush.