Wednesday, November 01, 2006

Why Bureaucrats don't Maximize their Budgets

William Niskanen, in a book published many years ago, proposed a simple model of government bureaucracy. The more money a bureaucrat controls, the more important he is, so bureaucrats want to maximize their budgets. The bureaucracy knows, and the legislature does not know, what a government bureau can do at what cost. The legislature knows how much any level of output from a bureau is worth to it. So the rational bureau misrepresents its production function in a way designed to trick the legislature into giving it the largest possible budget. It does so by finding the largest level of output that it can produce at a cost just barely below the value of that level of output to the legislature and then exaggerating the cost of any lower level of output by enough to make it higher than its value.

When I first read the argument it struck me that it contained a fundamental mistake. So far as I can remember, I never published that conclusion. I recently came across a version of Niskanen's argument in an online discussion, so thought I might as well take the opportunity to explain why it is wrong.

Consider two bureaucrats. Abe has a ten million dollar budget and is required to purchase $9,900,000 worth of paper to be sent to the IRS for printing tax forms on, leaving him $100,000 for himself, his secretary, and rent for his office. Bernie has a one million dollar budget and is required to do nothing at all. Which would you rather be?

Generalizing the example, I suggest that the size of the budget in Niskanen's model ought to be replaced with the surplus, defined as the difference between the size of the budget and the lowest cost at which the output the bureau has agreed to can be produced. That difference represents resources that bureaucrats can divert to their own purposes.

I'm curious. Niskanen's book was published a long time ago. Have other people proposed the same modification?


back40 said...

"the size of the budget in Niskanen's model ought to be replaced with the surplus"

I don't think so. The surplus isn't the issue, it's the budget since that is the source of power and status. More employees, more purchases, more facilities etc. means more power, status and importance. The surplus is just WAM.

Anonymous said...

Yes, this subject has been throughly studied by economist John Kornai, in the process of understanding the vices of planned economies. He has a very neat theory on "plan bargain" and "soft budget constraint".

Anonymous said...

I agree with back40. The issue is not direct corruption into the bureaucrat's pockets. It's about how many employees he has, and how much buying power.

And it's also a fallacy to consider the "real" cost of providing the service. Through incompetence and graft the cost of services can go up without limit. The bureaucrat doesn't have to hide his true costs -- he simply says that's the best anyone can do. Only free competition could prove him wrong. And that, of course, is the problem with bureaucracy.

Anonymous said...

Thinking like a bureaucrat, I would rather be Abe with 'only' the $100,000 surplus. Bureaucrats are generally risk adverse, and the risk of having $1,000,000 with nothing to do is that it will last a very short time. It is much easier to hide the $100K in the $10M budget for a long time. Second, there are SO many opportunities in $10M to raise the take to $200K, and up. Bureaucrats would rather suck at the teat than bite it off.

Anonymous said...

The bureaucrat doesn't embezzle directly, but (1) uses the large budget as justification for a large salary (this is also a motivation for public stock companies to merge) and (2) uses the large number of inefficient jobs to hide the fact that he hires his friends.
Not very different from what other commenters said.

Also, I don't think bureaucrats are very rational. I think selection effects are probably stronger.

Mike Huben said...

Yes, business bureaucracies are wasteful and inefficient. Which of course is why there are so few large corporations.

Oh -- you're talking about government? What makes you think they're different that way?

cuthhyra said...

From experience with the British Civil Service I have to agree with Back40 as well. In fact I would say that the budget in general is more of a consequence of the drive for more staff, the number of staff defining the size and therefore the importance of your department. The big salary and invites to dine with the PM or President are then consequences of the importance of your department. People in the civil service openly admit that they continue to employ staff whom they have no work for, precisely because they maintain the appearance of an important department.

Douglas Knight is also correct in saying it is not direct embezzlement, it is essentially empire building.

Anonymous said...

See Breton and Wintrobe, JPE v83(1)

William Newman said...

Mike Huben, I find your question about overlooked parallels between private actors and the state to be ironic. This kettle is not black: in general free market economists think about parallels between private and public organizations a lot, enough that they have jargon and specialties for it. This particular case happens to be an illuminating example. Google "david friedman agency problems" goes astray because there are too many webbed documents about Friedman's ideas for protection agencies. But Google "david friedman size of the firm" leads quickly to; search for "Why is the capitalist beach made up of socialist grains of sand?"

How black is the pot? Google for "Mike Huben" leads quickly to Libertarianism in One Lesson. To pick one particular point there, "The FDA is solely responsible for any death or sickness where it might have prevented treatment by the latest unproven fad." Somewhere behind this pithy paraphrase of your opponents' ideas is a critique of the FDA that concedes that private customers in an unregulated medical market make many mistakes, but wonders why voters should do better, and asks for evidence that they do. I don't know how to work backward from your page to a serious attempt to address that particular question. In fact, in general I have trouble finding leftist analyses which address imperfections of governmental and voting mechanisms on the same footing as imperfections of market and voluntary organizations. Any hints, Mike?

Anonymous said...

Didn't Niskanen augment the theory in his 1975 paper to allow for discretionary budget maximization?

All you need is that whatever bureaucrats really maximise (power, prestige, perquistes) correlates with budgets; the rest of the conclusions then follow.

Anonymous said...

Some convex bundle of "budget/influence/status" and "personal surplus" seems to be the best answer.

Having a budget of $10M and surplus of $100,000 is probably worse than budget of $5M and surplus of 101K but better than budget of $11M and surplus of $1,000.

Just like people don't maximize only leisure OR only consumption.