Jerry Pournelle's "Iron Law of Bureaucracy" implies that leaders of bureaucratic organizations will seek to maximize the power and influence of the organization at the expense of its stated goals - but is that true in the real world?

Julie Dolan of Macalester College examined surveys of government administrators and found that, surprisingly enough, high-ranking federal bureaucrats tended to prefer less government spending than the general public, even on issues that their own departments are responsible for.

Here is the abstract from her paper, "The budget-minimizing bureaucrat? Empirical evidence from the senior executive service" that was published in the journal Public Administration Review:

In a representative democracy, we assume the populace exerts some control over the actions and outputs of government officials, ensuring they comport with public preferences. However, the growth of the fourth branch of government has created a paradox: Unelected bureaucrats now have the power to affect government decisions (Meier 1993; Rourke 1984; Aberbach, Putnam, and Rockman 1981). In this article, I rely on two competing theories of bureaucratic behavior-representative-bureaucracy theory and Niskanen's budget-maximization theory-to assess how well the top ranks of the federal government represent the demands of the citizenry. Focusing on federal-spending priorities, I assess whether Senior Executive Service (SES) members mirror the attitudes of the populace or are likely to inflate budgets for their own personal gain. Contrary to the popular portrayal of the budget-maximizing bureaucrat (Niskanen 1971), I find these federal administrators prefer less spending than the public on most broad spending categories, even on issues that fall within their own departments' jurisdictions. As such, it may be time to revise our theories about bureaucratic self-interest and spending priorities. [emphasis added]

I was able to read the paper here for free, but I had to register first.

See also: The Case FOR Bureaucracy

New to LessWrong?

New Comment
10 comments, sorted by Click to highlight new comments since: Today at 6:43 PM
[-][anonymous]13y130

Well, without having read the whole article, it seems to me that the reason they're getting this result is that we're comparing bureaucrats to the general public.

The general public doesn't know how much is actually spent. They vastly overestimate spending on foreign aid, for instance, and welfare. It's plausible that bureaucrats prefer levels of spending that are roughly commensurate with actual operating costs, while ordinary citizens are just mis-estimating.

The fact that the public doesn't know the actual budget explains several strange artifacts in polling. For instance, when you poll people, a large percentage think foreign aid should be cut, but when you ask them what percent of the budget it should be, it's still higher than the actual percent! There's a similar phenomenon with taxes: people tend to think they should be cut, but they also overestimate how high taxes are in the first place, so they sound like they're calling for higher taxes.

It might be interesting to have two versions of a poll on government spending, one of which tells the participants the current spending levels before asking the questions, and compare the results.

Stated preference or revealed preference?

(I know, I could find it out by reading the paper, but I think this is important enough that it should have been in the post. Downvote if you think I'm misbehaving.)

Stated preference.

Interesting. How would you explain behavior like NASA management's cooking of shuttle safety numbers before the Challenger explosion, then? Richard Feynman is always a good read. It seems clear that at some level bureaucrats have often tried to optimize the wrong things, in this case "percieved safety," but it seems reasonable that lots of other wrong things will get optimized, including money.

It's also a little problematic because the survey doesn't measure quite what we want to measure. It could be a genuine effect, or it could be something like the difference between believed belief and actions in the real world, with effects not seen in the survey emerging in the workplace? Or maybe, due to the general perception it's simple dishonesty - I looked up the source of the data, and it's a face to face interview survey, which picks up more bias from that kind of thing.

If anything, doesn't the fact that chief bureaucrats see little need for more funding imply that bureaus really are wasteful?

Looks like a witch-hunt to me: if they want more funds it means they're wasteful (and greedy), if they don't want more funds it means they're wasteful (and extravagant).

Not that either can't be true, but you can't take both A and non-A as positive evidence for the same particular thesis. Your post would require you to endorse the statement "if they wanted more funding, that would suggest that they are not wasteful".

First of all, the study doesn't claim that the high-level bureaucrats don't want more funding, just that they want smaller increases than the general public. That fits in with the theory that bureaucrats want to expand their domains. However, I suspect that high-level bureaucrats have better incentives to be efficient than (a.) the public, or (b.) the politicians who actually create budgets. They are more likely to understand how much funding their group already has and more likely to be fired or criticized for inefficiency.

It's pretty widely known that polls usually reflect support for more public funding for everything except foreign aid and the arts. Economically, that makes sense. The public is not sensitive to the cost when it is just an abstract question (should x get more funding?) And of course the politicians who create budgets have little reason to be careful with money (they don't get to keep it if they don't spend it, so they might as well convert it into political capital with some group or another).

So that seems to paint a picture of a system where the public and politicians have an unlimited demand for spending because the marginal cost is near zero, and the bureaucrats get to expand their domains without being seen as unusually greedy or wasteful.

See also: The Case FOR Bureaucracy

That link bothers me immensely. While many of the things they describe are myths (I consider the most important part of politics to be math, and so the "there's always more waste to cut!" guy drew a laugh here), many of the things they describe are strawmen / they handle them wrongly.

For example, consider myth #5. When you compare private industries as a whole to public industries as a whole, the difference is a statistically insignificant .4%. However, what about when you compare apples and apples, not apples and oranges? The U.S. Postal Service gets a rating of 76.1%, private mail carriers get a rating of 84.5%. Public health clinics get a rating of 74.4%, private doctors' offices get a rating of 80.6%. If you average those two actual comparisons together, private industry's edge is 7.3%, almost a full 10% better than the public industry's ratings.

(Also, I would suggest "evidence against" rather than "refuting" when discussing a result, just like "evidence for" is a better description than "proving.")

[-][anonymous]13y00

Interesting. How would you explain behavior like NASA management's cooking of shuttle safety numbers before the Challenger explosion, then? Richard Feynman is always a good read. It seems clear that at some level bureaucrats have often tried to optimize the wrong things, in this case "percieved safety," but it seems reasonable that lots of other wrong things will get optimized, including money.

It's also a little problematic because the survey doesn't measure quite what we want to measure. It could be a genuine effect, or it could be something like the difference between believed belif and actions in the real world, with effects not seen in the survey emerging in the workplace? Or maybe, due to the general perception it's simple dishonesty - I looked up the source of the data, and it's a face to face interview survey, which picks up more bias from that kind of thing.