• Capital Gains
  • Posts
  • Why Macro Policy is Such a Bad Predictor of Macro Data

Why Macro Policy is Such a Bad Predictor of Macro Data

The Thermostat Problem and the Limits of Empiricism

In partnership with

Know someone who might like Capital Gains? Use the referral program to gain access to my database of book reviews (1), an invite to the Capital Gains Discord (2), stickers (10), and a mug (25). Scroll to the bottom of the email version of this edition or subscribe to get your referral link!

An old economics joke, first coined by Milton Friedman, goes like this: a macroeconomist hears about the marvelous new technologies of air conditioners, heaters, and thermostats. He's told that when they work in combination, they can alter the temperature of a home, protecting people from the discomfort of seasonal weather fluctuations. So he resolves to investigate: in winter, the heater is working hard, and the house is a comfortable 72 degrees. In the summer, the AC is cranking away, and the house is... still 72. So he concludes that temperature is a constant, and the electric bill would be lower without all these superfluous machines.

It's a silly parable, but consider the following much more real-world examples:

  • If lowering interest rates causes higher growth, why was growth so slow during the zero-rates era after the 2008 financial crisis? Why was growth slower in zero-rate Japan than the rest of the world over a longer period than it was post-2008?

  • If new housing reduces housing prices, why is it that higher housing supply correlates with higher, rather than lower, rents?

  • If stock prices are tied to economic growth, why did the market crash in 2008 with unemployment at 6%, and then stage a massive rally in 2009 with unemployment at 9%?

All of these are examples of two related phenomena:

  1. No economic model can explain 100% of what happens in the economy.1

  2. In many cases, a variable that has a causal impact on one part of the economy is also reacting, either to that same part of the economy or to something else.

So before we smash the thermostat, let’s dig into each of those real world examples, and dismantle the Thermostat Critique instead:

The reason low rates and low growth persisted into the 2010s throughout most of the developed world, and in Japan before that, was that there was a widespread shortfall in demand because of contracting credit, and fiscal policy wasn't aggressive enough to offset it. Which is a mouthful. A more folksy explanation is that if nobody has a job, nobody's spending money, and if nobody's spending money, then nobody's hiring. In Japan, this took the form of indebted companies that didn't restructure their debt, and kept servicing it instead; they were paying off loans at 100 cents on the dollar that might have been worth half that, and because they'd borrowed so much, they didn't have spare funds to invest in growth.2 In the US and Western Europe, there was a similar overhang as banks worked off the bad debts on their balance sheets and built up equity so they'd be robust ahead of the next crisis. The low interest rates at that time were great for asset values—hate equities all you want, but if you're earning 2% on the ten year it's hard not to buy them at more than double the earnings yield. But rich people have a low marginal propensity to spend, especially when they're getting capital appreciation instead of salaries, bonuses, dividends, and distributions. So stocks were up, but demand was low, and unemployment took a very long time to recover. An investor might say that the financial crisis was well and truly over in early 2011, when the S&P recovered its decline from the US's sovereign credit downgrade. Or, at the latest, in 2013, when it hit an all-time high. But for workers who weren't in the market, a better date for the end of the crisis was March of 2017, when unemployment finally got back to 2007 levels.

The housing story is simpler. Real estate developers want to make money, and they're implicitly making a very simple calculation: unlevered return equals rental yield plus long-term rent growth. When wages in a given city are rising, they'll build what they can. Even if there are restrictions on growth, a change in profit expectations will still produce a change in how much they try to build. If a real estate developer says "You'd have to pay me literally a million dollars to deal with SF's bureaucracy," sometimes the supply/demand imbalance says "How about two million?"

For the last phenomenon, of stocks and employment saying contradictory things about growth, it doesn't always shape up that way. The 60s and the 90s had low unemployment and a booming equity market, for example. But the stock market is discounting future changes—stocks decline, not just because of bad headlines today, but because of the expectation of worse headlines in the future. The equity turnaround in 2009 didn't happen when the macro data reached their lows—it happened when housing prices fell a bit more slowly than they had in the prior few months, and when the unemployment rate started going up 0.4% each month instead of 0.5%. Markets react to where the average participant thinks things are going, which is, by definition, not where they are right now.

Does this make such signals useless? No, but it does make them tricky. Turning back to the thermostat example, the obvious reason the analogy doesn't work is that we know how a thermostat does work, and we know that stable temperatures throughout the year are not a fact of life and do demand an explanation. Economics is trickier than that, and what we usually have are partial, situational explanations, all of which work on a (variable) lag. But the marginal impact is still there, even if it's swamped in the short term by whatever it's a response to.

Rise and grind, dollar bills!

Put your money to work in a high-yield cash account with up to $2M in FDIC† insurance through program banks.

Get started today, with as little as $10.

Read More in The Diff

The Diff has implicitly talked about this issue a few times:

1. With the edge case that sufficiently extreme models, over long enough time periods, do explain everything. For example, if an asteroid wipes out all of humanity, you can predict that GDP will drop regardless of what the Fed does in response—to astrology's credit, there are times when the movement of heavenly bodies is in fact the only thing that determines our fates. In the other direction, if we get energy or intelligence that's too cheap to meter (and either one will probably lead to the other in short order), then that, too, is all that matters.

2. Elsewhere, I've described Japan in that period as the world's most depressing credit bubble. If a bubble involves irrationally overpaying for assets, and if we don't treat the borrower as a special category when it comes to buying debt, then Japanese corporate borrowers were wildly overpaying for the specific bonds and loans they had issued, instead of restructuring, and giving creditors a smaller amount of debt but a big slug of their companies' equity in return.

Share Capital Gains

Subscribed readers can participate in our referral program! If you're not already subscribed, click the button below and we'll email you your link; if you are already subscribed, you can find your referral link in the email version of this edition.

Join the discussion!


or to participate.