Here's the problem: the federal budget makes no distinction between money the government spends and money it invests.
But there is a difference. A dollar spent on Social Security is a dollar that the federal government no longer has. That's different from a dollar spent building, say, a federal courthouse, where the government trades a dollar worth of currency for a dollar worth of courthouse. These kinds of spending are not the same, and the government shouldn't treat them as if they are.
Treating investment and spending as identical can lead to some really dumb decisions. If the government cuts spending on Social Security by $10 billion then it really has saved $10 billion.
But if the government cuts spending on infrastructure repairs by $10 billion then it looks like it's saved $10 billion, but really it's just pushed $10 billion worth of infrastructure repairs into the future — and because the bridges may crumble and the water pipes might break, deferring that $10 billion in spending might mean the government has to spend $20 billion instead.
Enter a capital budget: the idea that the federal budget should treat capital investments differently than other kinds of spending.
This isn't some crazed concept dreamed up by FDR nostalgics. It's how most businesses run their budgets, too. They keep an eye on annual cash flow, which is basically what the federal budget tracks now, but they're interested in more than that: they're interested in future profits and losses, and so they try, as best they can, to break out the investments that are meant to generate those future profits and stem those future losses.
The hard part there is they also estimate how much value their capital assets are losing each year, and they treat that, correctly, like a hit to the company's value. You don't lose money when you initially buy a machine for your company — after all, that was simply an exchange of money for machine, and the machine was presumably worth the money you paid for it. You lose money when the machine stops working, because then you don't have the money or the machine.
So, to use the government as an example, capital budgeting would make clear that if our roads are degrading to the tune of $20 billion a year, then we need to be spending $20 billion a year on repairs just to stay even. Spending nothing means losing $20 billion in value, not holding even.
Should the federal government use capital budgeting?
Well, yes, it probably should. But it also probably can't.
A helpful 2008 Congressional Budget Office report lays out the case for capital budgeting, but this giant paragraph detailing the problems is pretty close to a knockout punch:
Moving to a budget that is more reliant on accrual-based accounting could increase complexity, diminish transparency, and make the federal budget process more sensitive to small changes in assumed parameters, such as depreciation rates. (Indeed, other nations have considered adopting capital budgets, but generally decided against it for those same reasons.) Adopting an accrual approach to only one aspect of the budget could raise concerns as to whether the budgeting system would provide a fair basis for allocating the government’s resources among competing priorities. In addition, providing special treatment to certain areas of the budget, such as capital spending, could make the process more prone to manipulation. Furthermore, simply arriving at a definition of capital for budgeting purposes could be a significant challenge. Concerns about such issues largely explain why previous groups charged with exploring budgetary concept issues—including the 1967 President’s Commission on Budget Concepts and the 1999 President’s Commission to Study Capital Budgeting—have rejected the idea of a separate capital budget for the federal government.
I want to focus, particularly, on the issue of "simply arriving at a definition of capital." The closer you get to this question the faster it breaks down. A building is a capital asset, sure. And so is one of those machines the US Mint uses to manufacture nickels.
But then it gets harder. How about a bridge that the federal government pays for but a state government actually builds and manages? How about investments in education and basic scientific research? Why should investments in physical capital be privileged over investments in intellectual capital? Don't you know we live in a knowledge-based economy now?
Since the point of capital budgeting is to make it easier to spend money on the items defined as capital investments, the pressure to define more and more items as capital would be overwhelming.
Of course, that's true in corporations, too. But corporations tend to have more defined aims, more internal unanimity around mission, and, at any rate, they're ultimately disciplined by whether their investments make enough money for the company to stay afloat. If they're fooling themselves about the likely return on their investments or the pace of depreciation they're going to go bust. Not so with the federal government, which is dizzyingly vast in its aims, deeply divided over its mission, not primarily looking to make money off its investments, and not in danger of going bust the way corporations are.
Capital budgeting is trying to solve a problem that it can't actually solve
I spent some time poking around the internet to see if anyone had a good proposal for solving these problems, and, happily, I came across this idea from Miles Kimball and Noah Smith. Kimball and Smith are about as smart as they come, so the fact that their proposal mostly just punts the problem makes me think there's probably just not a good solution:
1. If experts agree that an expenditure will raise future tax revenue by increasing GDP, then it belongs in the capital budget. If it can pay for itself entirely out of extra tax revenue in the future then it should be 100% on the capital budget. If it can pay for half of its cost out of extra tax revenue in the future, than it should be 50% on the capital budget. The provision "experts agree" requires some sort of independent commission doing an economic analysis with appointees from both parties, and with, say, two-thirds of the commissioners needing to agree that the value of future tax revenue is likely to be above a given level.
2. Even if an expenditure will not raise future tax revenue, it can count as a capital expenditure if it is a one-time expenditure—that is, if it makes sense to have a surge in spending followed by a much lower maintenance level of spending in that area. This will only be true if it pushes the existing stock of infrastructure, other government capital, or knowledge to a higher level than before, not if it just keeps things even. Crucially, by this logic, anything that lets the stock of infrastructure or other government capital decline would count as a negative capital expenditure. This principle enables the capital budget accounting to sound a warning when the nation is letting its infrastructure crumble away, and also allows sensible decisions about shifting funds from older forms of infrastructure toward modern forms of infrastructure needed by a fast-moving economy.
The first condition, in particular, gets to the real issue here. Capital budgeting is trying to solve a problem that it can't actually solve: Congress makes a lot of dumb spending decisions and has a lot of trouble agreeing about what kinds of spending is likely to generate positive returns. But if Congress could agree on smarter ways to spend money, then there would be no problem to solve in the first place. So insofar as capital budgeting needs Congress, or representatives chosen by Congress, to make a lot of hard, forward-thinking judgments about how the government should spend its money, it's really just restating the problem rather than solving it.
By the same token, a Congress that works badly now will probably also implement capital budgeting badly, and so adding it as a new layer onto the budget process might simply create a new space for corruption and confusion. It's easy to imagine the highly funded lobbying pushes that will try to get various subsidies reframed as capital investments, for instance. And while I've really focused here on the problems Congress would have deciding what a capital investment is, the next step for them would be to decide on the correct depreciation rates for those investments, and since that process is fuzzy and error-ridden even in the private sector, it's hard to imagine Congress doing a particularly good job of it.
So while I'm sympathetic to capital budgeting in theory and would like to see it used, it's hard for me to believe Congress, as currently composed, would do a halfway decent job at implementing it. But maybe I'm wrong! In fact, I'd like to be wrong.
Update: In the (now closed) comments, Dnairn points to this report from the National Association of State Budget Officers that outlines how various states implement capital budgets. Worth reading if you want to take the red pill on this topic.