"What's this business worth?" is, implicitly, the question every single trader is answering every time they make a trade. If there were an easy answer, it wouldn't be so lucrative to successfully come up with answers.

But one surprisingly useful starting point is also the fairly obvious one: add up the accounting value of all the company's assets, subtract the value of its liabilities, and you’ll get an accountant's-eye-view of what the company is worth: the book value. Despite its shortcomings, book value is a surprisingly good measure of a business—which is why buying low price-to-book-value stocks tends to outperform over time. In other words, if you consistently bet against a sophisticated worldview where Visa, Coca-Cola, and Google have valuable assets that aren't reflected in the balance sheet, and bet on a view that banks, oil companies, and other tangible asset-heavy sectors are the place to be, you will (on average!) make some money over time.

One driver of this is that a company has low returns on capital it will shrink over time, because it can't easily raise more funds and because management's incentive is to shrink the business. Meanwhile, if a company has high returns on capital, it will see those slowly converge with the average over long periods, simply because the more unusually-profitable their first investment category is, the harder it will be for them to get a repeat performance. So, in a sense, book value-based investing is a bet on mean reversion: good industries will attract more capital until they're not so good over time, and there is almost no industry so bad that it can't become good when enough competitors call it quits and leave the market.

(This is part of the general topic of "factor investing," or identifying broad categories of companies that a) move together, and b) produce better risk-adjusted returns over time. It's a topic for a future post.)

Even though betting on book value is simplistic, it's hardly simple. Accounting is meant to create a conservative picture of what a company’s assets are worth, which might feel like a rock-solid science, but there's an art to it. The most simple form of accounting, cash accounting, just keeps track of cash flows that come in and out, but this has some issues we'll get to shortly. A better system is accrual accounting, where entries on a balance sheet represent likely outcomes that are tied to specific events.

On a cash accounting basis, you get revenue when money hits your bank account. Consider The Diff, a business near and dear to me. When someone signs up for an annual subscription, they pay $220, and that money hits my bank account the next day. The accrual accounting way to recognize that revenue is by spreading the payment over the life of its contract, either in a straight line ($220 a year / 365 days a year = $0.603/day) or, to be really sophisticated, by estimating how many issues that subscriber is paying for and recognizing revenue the instant they receive a paid issue. But if I do this instead, I’m presented with a new problem: I have added $220 to my balance sheet, but I'm only recognizing a fraction of that each day. To account for this, I would need to add a corresponding liability, usually labeled "deferred revenue", and then my profits actually consist of decreases in the deferred revenue line item.¹

On a cash accounting basis, buying a long-lived and periodically-replaced asset is an expense; cash leaves the balance sheet. On an accrual basis, the way you'd look at this is that you changed the form of your assets, from cash to something else, but didn't change the amount. However, you did spend money, and that needs to be recognized somehow. The answer is to depreciate the asset over time: if it lasts for four years before it has to be replaced, and it has a scrap value of 20% of its initial cost, then you'd recognize 20% of the cost each year as the replacement date approaches. (And then, if you sold for more or less than the scrap value you'd estimated, you would need to recognize a profit or a loss on that transaction.)

This creates lots of opportunities for clever behavior, both because the nature of some expenses is debatable, and because the depreciation schedules can be flexible. If two companies own the same assets, but one depreciates them over six years and the other over eight, the one with the longer depreciation schedule will report higher accounting profits, but will have the same cash flows. And some expenses can get capitalized instead of depreciated: AOL looked more profitable than it was in the 90s because it was capitalizing some marketing expenses. The SEC eventually decided this was not acceptable.

When companies have assets on the balance sheet and something changes that affects the value of those assets, they’re required to adjust by writing the value down. This happens in many different cases: during an acquisition, for example, there's generally a difference between the acquired company's book value and the amount paid for it. The acquirer's balance sheet won't balance if that gets ignored—if cash dropped by $100m, but assets only went up by $10m, something's wrong. So they book an asset called "goodwill," representing the intangible value of the business they bought. If that business doesn't work, they have to write down the goodwill. I hate to pick on AOL excessively, but they set an all-time record for annual losses when they wrote down the AOL goodwill from the AOL/Time Warner merger.

Accounting at a bank is particularly tricky. In Q1 of 2020, for example, JPMorgan Chase's consumer banking unit reported a 95% reduction in net income compared to the prior quarter, a $4.0bn drop. But in terms of actual realized credit losses compared to interest income, profits had actually gone up slightly—what changed was that in Q4 '19, they added $1.2bn to their provision for credit losses, i.e. the accounting line for loans that they hadn't tried to collect on yet but did not expect to collect in full. In Q1 '20, that provision was $5.8bn. Ignoring the provision for credit losses, though, i.e. focusing on the money coming in rather than hypothetical changes in future income, profits were actually up $500m from the previous quarter.

In one sense, their reduction in profits was entirely fictional: they were reporting losses based on their expectations for the future. In another sense, though, it's exactly right. Suppose you lend your friend some money for a year, collateralized by your friend's car. (In this scenario, I guess you're not very close friends.) The next day, your friend loses their job and totals their car. You're absolutely sure you will never see a dime of that loan any time soon. When did the loss happen? In a cash accounting sense, it hasn't happened yet, and won't happen until the loan comes due and doesn't get paid. In an accrual accounting sense, the loss happens the instant you have strong evidence of what will happen to that loan.

Banks have another accounting wrinkle: there are some losses they don't have to recognize. A bank will generally put some of its assets into very safe bonds, which are definitely going to pay off at 100 cents on the dollar but which may fluctuate in price in the meantime. These bond price fluctuations are driven by interest rates, but in the end if you buy and hold a 30-year US treasury bond, you're getting 100 cents on the dollar once it matures. However, this can lead to lively fluctuations, and banks have had cases where rates went up enough that the loss on their bond portfolios exceeded their equity. The basic argument for why they shouldn't have to recognize this as a loss is:

1. They intend to hold to maturity, and they will recognize a loss if they sell.

2. They'll only have to sell if depositors pull out so much money that they run out of cash.

3. Federal deposit insurance makes individual depositors unlikely to do that. Companies with larger deposits might, though, and some banks raise "wholesale" funding, making short-term borrowings at large scale. (This, too, is a future Capital Gains topic.) These deposits are skittish whenever there's a whiff of a problem, so a bank that's too reliant on wholesale funding can suddenly collapse. On the other hand, bank treasury departments are well aware of this, and their one job is to stay boring.

The utility of book value varies over the life of companies. Sometimes it's pretty worthless, especially when a company a) has some kind of locked-in intangible value that's hard to disrupt, like a government monopoly, and b) borrows money to buy back stock. That can give a company a steeply negative book value, even though the business is clearly worth something. Moody's, for example, had a negative book value for a few years, and its tangible book value remains negative.

In the earliest days of a business, book value has the least relevance. When a pre-revenue company is raising money, it's valued based on what it could turn into, not the assets on the books. When you’re negotiating a pre-money valuation, it’s a bad sign to say things like “Well, we have some very nice chairs.” But companies are fated to age, and to eventually reach a point where what you see on the balance sheet is an increasingly good approximation of what you get.

Read More in The Diff

The Diff has covered matters related to price-to-book a few times.

• A recent pair ($) of posts ($) looked at Silicon Valley Bank, whose mark-to-market (but unrealized!) losses on an interest rates bet briefly exceeded their equity.

• Another piece looks at why the returns from systematic value investing were weaker in the last decade ($) or so, until recently.

• This piece ($) briefly looks at the intersection of statistically cheap stocks and accounting games.

1. One way to look at this is that businesses that charge upfront are essentially raising startup capital from their customers. This was absolutely the case in the early days of The Diff, when it wasn't yet viable as a standalone business, but was generating cash flow from annual subscriptions. Like any kind of borrowing, this is only as good as what's collateralizing it, and some paid newsletters end up in a situation where their "liability," the promise to produce another twelve months of content, doesn't justify the effort. The graceful thing to do is to shut down and offer a pro-rata refund, but if the money's been spent on food and rent that's not an option.

Help us learn more about you!

We'd like your feedback so we know what to cover, so if you haven't already, please fill out this one-minute form.