Some services are most cost-effectively provided by governments, others by the private sector, and it's not at all predictable which will fall into which category.For instance, if you left a publicly-traded airport and on the way out you paid a fee for using a toilet, you're probably in socialist Europe rather than the capitalist United States. On average, of course, the more active governments are more likely to control more institutions, but industry-by-industry and country-by-country, there's a lot of variance. And some of these differences result from historical contingencies; the US used to privatize the business of packaging and insuring mortgages via Fannie and Freddie, but at least in Fannie's case that started out as an accounting gimmick to keep their debt off the government's balance sheet.

And, as it turns out, accounting and gimmickry are a common feature of debates about privatization. Specifically, there's an intuition that goes like this: a government-provided service can run at breakeven, and produce some kind of public benefit. A private company trying to produce exactly the same benefit will still need to produce returns for shareholders, and those returns will have to come from some combination of higher prices for consumers and lower wages for workers.¹ For users, and for workers, the need to turn a profit means that a privatized postal service would be strictly worse.

But is that need really restricted to the private sector? If it were true, it would mean that government-operated entities had a cost of capital advantage, meaning that if they competed with the private sector, they'd tend to gain share over time. UPS and FedEx have to return capital to shareholders, USPS doesn't, so if they operate with remotely comparable efficiency, USPS would win.

But this gets at the distinction between accounting profits and economic profits. The US postal service loses money in GAAP terms (their report, much like their for-profit peers, provides a non-GAAP reconciliation that flatters the numbers.).² But the USPS balance sheet also has assets, $46bn worth, and that's a book value, not a present value—if they were to auction off their network to competing companies, some of those longstanding locations in dense cities would go for quite a premium. Generally, midtown office buildings have appreciated in value over the last century, albeit with some recent wobbles.

Determining how much capital they have is tricky; both UPS and Fedex trade premiums to book value, but some of that is because a company that borrows against appreciating assets and returns some of that money to shareholders will tend to depress its book value per share even if intrinsic value is going up. And there's also the question of what price to put on that capital. The discount rate for USPS might be lower than comparable companies because they're a regulated monopoly. But that discount rate also needs to factor in risks. What if, for example, someone found a way to transmit first class mail by converting it into a stream of bits that could be sent over wires, making it faster, cheaper, and more secure? And what if other companies started offering parcel delivery? Things could get dicey for USPS, and they might find that many of the asset purchases they underwrote based on more optimistic assumptions about long-term demand for their services were impaired! Bumping up their cost of capital is a way to account for this; it's a way to say that capital outlays should be considered in light of the full range of possible outcomes, not just the scenario that justifies spending.

If we don't do this, the main problem we run into is that the benefits of public and private sector capital expenditure are incommensurable. If USPS could deliver mail faster with another $10bn in capital, is that a worthwhile investment or not? For USPS, we have to debate about it; for Fedex and UPS, we can look at how the stock price reacts to an announcement about this to get an instantaneous judgment. Meanwhile, we don't have any sense of how much capital we're tying up in mail delivery compared to the optimal amount. It could be too much, and it could be too little. Given the incentives at work—USPS will find its pensions more defensible if it continues to deliver mail, and can point to expensive rural delivery as a symbol of how indispensable it is even if some of those rural addresses, on the margin, still receive mail only because it's subsidized.

In the end, there is a profit tax, but it's in the opposite direction: not requiring an entity to produce an accounting profit means instructing it to run at an economic loss. It would be a strange coincidence if the positive externality of USPS relative to potential replacements happened to be precisely equal to its cost of capital. Better to estimate both and have numbers to compare against than to make a best guess and refuse to verify whether that guess was correct.

The Diff often looks at cost of capital as a deciding factor in strategy, though usually in the private sector. A few examples:

• Being “capital light” is more about where you are in the cycle than what kind of industry you’re in.

• In the absence of good established metrics, make your own.

• One reason cost of capital is hard to talk about is that the meaning of interest rates has changed over time.

• You can analyze big tech companies from the framework of their willingness to use cheap capital well ($).

Sometimes broad trends in the cost of capital actually make some otherwise-unfundable businesses work.

Join the discussion!

1  An optimistic libertarian view is that these businesses will be more efficient, and on average that's true. But there are two important caveats: first, one of the reasons the private sector is more efficient is that inefficient operators go under, so absolute gaps in efficiency include survivorship bias. But in the case of privatizing a public service, the public usually doesn't want there to be a meaningful risk that it goes away entirely. Second, one element of this efficiency gap is that government agencies tend to continuously expand the scope of needs that they serve. Uber doesn't need some kind of policy for dealing with violent, mentally-ill people who use their service, because a) those people are less likely than average to have smartphones, credit cards, and disposable income, and b) Uber can just ban them. But some subway systems don't ban such people, giving them a wider range of expensive problems to deal with. This ends up being a pricey ad-hoc redistribution system that would be almost universally unpopular if it were described as such, but a more efficient private-sector operator that refused to do this would, in fact, be more efficient in part because it wasn't providing an expensive service.

2  They are a bit sniffier about it than public companies, though. The main thing they want to net out is changes to the actuarial cost of their pensions, "adjusted for the impact of the [Postal Service Reform Act of 2022] and other costs outside of management's control, including workers’ compensation expenses caused by actuarial revaluation and discount rate changes, and the amortization of the [Civil Service Retirement System], and [Federal Employees Retirement System] unfunded liabilities." In other words, like many tech companies in the 90s and early 2000s, they would really love it if they could pay their employees in a way that has equivalent economic costs to salaries, but that gets treated differently. It is entirely within their control to stop offering defined-benefit pensions, and to swap these for higher wages, but if they did that they'd find that the wage they replaced them with led to adjusted losses in line with GAAP losses. The USPS insists that its pension calculations are unfair, but the unfairness is that they're being required to calculate the present value of their pension liability, just like anyone else who offers a pension.