Epistemic Status: Metrics look good
There is a long-standing hypothesis that corporate America focuses too much on short-term results. Larry Summers looks at a new study by McKinsey (McKinsey’s summary here), and concludes in his words that ‘the case is still out on whether corporate short-termism is a problem.’
the short-term hypothesis is ‘corporations systematically take too short a view and do not invest enough for the long term.’
Summers considers the short-term hypothesis reasonable, but does not think McKinsey has provided strong additional evidence. Summers’ objection to the study is that what is being labeled ‘short-termism’ is largely a proxy for quality corporations with good prospects. Taking a long-term approach is a luxury, requiring both a stable short term situation and strong long term prospects, so it is not surprising that such firms provide better long term returns. Causality is reversed: If your long-term returns are strong, you then invest heavily. This can then cause systematic mispricing if investors do not update sufficiently on this information.
Reverse causality is certainly plausible, and there are models where it results in long term focused companies being undervalued without short-termism being a problem.
What I found more interesting was the details of what McKinsey was actually measuring. They use five (see page 3):
- On the theory that long-term firms will invest more, they measure the ratio of capital expenditures to depreciation.
- Earnings Quality. On the theory that long-term firms will generate earnings ‘that reflect cash flow, not accounting decisions’ they measure accruals as a share of revenue.
- Margin Growth. On the theory that long-term firms will avoid driving margins unsustainably high, they measure the difference between earnings growth and revenue growth.
- Quarterly Management. On the theory that short-term firms will care about hitting earnings per share (EPS) targets but long-term firms will not, they measure the incidence of beating EPS by less than 2 cents, versus missing by less than 2 cents.
- Earnings-per-share Growth. On the theory that long-term firms won’t spend money on share buybacks (since they take up money with better uses, and are used to inflate EPS growth), they measure the difference between EPS growth and ‘true earnings growth,’ which seems like it is just the change in shares outstanding.
Three of these requirements – #2, #4 and #5 – are labeled short-term but, once you have controlled for #1, would be better categorized as optimizing metrics. There is a proxy G* that people are using to measure G, where G is the correct stock price, so the corporation is maximizing G*. They are signaling. Capital expenditures and margin growth are business decisions reflecting degree of investment and trade-off between the near term and the future. Making sure you hit an EPS target via creative accounting? That isn’t a trade-off between near term profits and future profits! That’s a trade-off between what is right for the business and what is right for your earnings call and the headline on MSNBC.
Margin growth versus revenue growth, #3, certainly could be short-termism, but it can also reflect good past long-term investments that give you market power, a superior value proposition or a superior cost basis for your product. Google has very high margins for the right reasons. Hasbro obsesses over them like a cancer. My guess is margin growth does correlate with short-termism, but not especially well.
Only capital expenditures, #1, seems clearly to be a proxy for long-term focus, up until Goodheart’s Law gets involved.
What is being called short-termism can be thought of as two distinct things.
First, there is the decision to focus on near-term profits over long-term profits, such as by raising prices in a way that makes money now but sacrifices market share. For this to be a problem, the firm would need to be making a mistake absent changes in stock prices. This is clearly not the whole story. As Summers points out, we have a lot of examples of “unicorn” companies with no revenue being valued very highly. Silicon Valley and Wall Street are perfectly capable of calculating the net present value of expected future revenue streams, and in trading on the potential for future greatness.
That does not mean that future prospects and investments are being priced ‘correctly’ compared to current profits, but I fail to see convincing evidence that this effect is even in one direction or the other. Companies might even be too focused on the long run, in many cases.
It could even make sense for corporations to be more short-term focused now than in the past.
If things like AI, self-driving cars, robots and mobile apps are going to disrupt your business in unpredictable ways, or your technology simply goes through an update cycle every few years, or there is too much political uncertainty, it would be wise not to invest too much in the long-term until you have a better picture. Factories with thirty-year amortization plans are the answer to a particular problem that we may simply not have as much of as we used to. Long term investments are inflexible.
Then again, we also have survey results like this:
Early in 2013 McKinsey and the Canada Pension Plan Investment Board (CPPIB) conducted a McKinsey Quarterly survey of more than 1,000 board members and C-suite executives around the world to assess their progress in taking a longer-term approach to running their companies. The results are stark:
- 63% of respondents said the pressure to generate strong short-term results had increased over the previous five years.
- 79% felt especially pressured to demonstrate strong financial performance over a period of just two years or less.
- 44% said they use a time horizon of less than three years in setting strategy.
- 73% said they should use a time horizon of more than three years.
- 86% declared that using a longer time horizon to make business decisions would positively affect corporate performance in a number of ways, including strengthening financial returns and increasing innovation.
What explains this persistent gap between knowing the right thing to do and actually doing it? In our survey, 46% of respondents said that the pressure to deliver strong short-term financial performance stemmed from their boards—they expected their companies to generate greater earnings in the near term. As for those board members, they made it clear that they were often just channeling increased short-term pressures from investors, including institutional shareholders.
The quoted article’s suggestions for fixing the problem is to make investors think long term and optimize on that basis. This is at least very hard, and their proposed details don’t really make practical sense.
Board members think their companies would be better off if they could ‘take a longer term focus.’ Investors disagree. That does not mean that the board members are correct. They are at least in part self-interested lying liars. If you have a long-term focus, your life as CEO or board member is a lot more relaxed and your chance of being fired is much lower. You and your friends can keep hiring each other and paying each other huge bonuses based on this ephemeral future performance.
Investors don’t like that. Investors want good metrics, forcing companies and boards to focus on those metrics, to make them at least partly be making the company money. The investors will do their best to combine short-term and long-term metrics, but good long term metrics are even harder than good short term metrics, and often need to be unique to a given circumstance. Since investors can buy and sell at market prices at any time, and the market’s overall view sets the price, to a large extent only metrics the entire market can parse can count.
The company is forced to maximize against the metrics of the market, and hill climb on that basis. If the long term can’t be found via a hill climb of widely accepted metrics, the long term won’t be found.
I have talked before about my experience with a particularly nasty form of this. Overpowered Metrics Eat Underspecified Goals. We suffer from a flood of overpowered metrics. The metrics work, in the sense that the market responds to them, and the market price reflects your ability to hit those metrics and the predicted ability of the company to hit those metrics.
When you have overpowered metrics, you maximize those metrics at the expense of other things, and the tighter the feedback loop those metrics give you, the more ‘short-term’ and hill climb-like your maximizations become even when the metric was chosen to target a long-term outcome. Maximizing ‘long-term’ growth in the sufficiently short-term is largely a short-term outcome. Humans and organizations behave more like (narrow) AIs, and they suffer more from the weaknesses of those AIs, unable to plan, unable to break out of local maxima, discarding key aspects of their problems (often without even seeing that those aspects exist) and acting like Literal Genies.
The question, which I will consider beyond the scope of this post, then becomes how to break out of that trap.