Welcome to Known Unknowns, a newsletter that will tell you how to predict the future—or, to be more precise, how to not predict the future.
What is risk?
My first book, An Economist Walks into a Brothel, has been out for about two weeks now, and I know I’ve been oversharing a bit. I promise I’ll stop soon. But there are a few stories I want to highlight that are getting less attention.
One is about risk measurement, which sounds so dull, but it may be—and I know this sounds overdramatic, but bear with me—the most important issue of our time. Risk is a measurement, an estimate of all of the things that might happen, good or bad, and how probable they are. We never know the future, but we can take a guess and make plans around it. Risk is our guess.
The best and most scientific guesses are based on data. The problem is that data (by definition from the past) is a terrible way to predict the future. Even forecasts are based on parameters derived from historical data.
Some data is better than others. I tell the story of man who came to Hollywood thinking that he could tame it and predict which movies would become blockbusters. It’s a classic Hollywood parable about a young, hopeful upstart who initially finds fame and fortune (with a Monte Carlo simulation!) and who then flies too close to the sun, thus falling back to earth.
But he was destined to fail from the start, because movie data is so terrible as it changes constantly. Do you remember when DVDs were a source of revenue? And it’s hard to get consistent estimates, because if you plot potential profit scenarios, the distribution is extremely skewed. These issues of fast-changing data and skew raise apply to more than just movies. Whenever we make a decision, we all need to determine what data from the past is most useful, and what is now irrelevant. Anyone who models long-term interest rates knows that problem well—do the 1970s tell us anything about interest rates today? Will the next 20 years repeat themselves? What about inflation? Will that ever be a risk again? It depends on what data you choose to use.
We also need to be mindful about what we mean by risk. Do we mean volatility, or the range of things that typically happen, or tail risk? The answer is that it depends both on the risk and the nature of the data.
These questions have always been important in finance, or in any business decision. But now more than ever, in the age of big data, they matter to everyone. Big data promises more accuracy and predictability, but it doesn’t answer the questions that I raise—it actually makes them more pressing.
Now, everyone can get risk estimates about almost everything, all on their phone —movies they’ll like, restaurants they’ll enjoy, approximate travel time to practically any location. But if these estimates are based on irrelevant data (perhaps a rave restaurant review is for a chef who no longer works there) or is skewed (like a divisive art house film), these estimates could be misleading.
How to prepare for an uncertain war
You might also be thinking about the things that aren’t even in our risk estimates; outcomes we never could have imagined. I spoke to General HR McMaster about that. After all, war never goes according to plan, and yet the military keeps on making plans that often fall short.
It has become popular to argue that because risk models sometimes don’t anticipate everything that could possibly happen, we should not use them. But that’s the wrong approach. If a tool works 90% of the time, that’s a lot better than never working at all. It’s still worth using it. But you can’t assume that the tool works 100% of the time, so you need to stay flexible and change your plan if you need to. McMaster argues that balancing planning and flexibility is what makes or breaks empires. Planning is essential, but war never goes as planned.
And this is also true in finance. Measuring risk and managing it makes things less risky in the end, and this enables growth. But if you assume that you’ve covered everything and then lever up, you may be in trouble. Leverage can be dangerous, because it leaves so little room for flexibility if things go wrong in ways that we never could have expected.
Debt at the household, firm, or national level can be a great way to fuel growth and to earn even greater rewards. But if things get nasty and income drops, debt can become a big problem. You can plan for contingencies, insure and hedge, but something can always happen that you don’t expect, whether it’s a hurricane, a bad recession, or even a war.
I am worried about Connecticut
Celebrities are just like us. 50 Cent had a hard time selling his house and ultimately took a loss on it. Granted, there’s not a huge market for a behemoth estate in Farmington, CT. And it’s not decorated in the traditional New England style either. But it seems that many people with big houses in the suburbs are having problems selling them, too.
This just shows how data changes. A history of rising house prices in the suburbs doesn’t mean it will still happen in the future, especially if there’s a structural change in which young, upwardly mobile people want to live in cities instead.
Paying someone a regular income that you can never cut is both risky and expensive. That’s why companies don’t offer DB plans. States and cities do, but that’s mainly because they haven’t realized the cost of doing so.
The few corporate DB plans that are left are trying to unload their obligations by offering beneficiaries a lump sum payout when they retire. And the Treasury has just made it easier for companies to do this. Josh Gotbaum thinks that this is terrible, because it sticks retirees with all of the risk. But Olivia Mitchell argues that sometimes this may not be such a bad thing. Many retirees are in debt, and they can use the lump sum to pay off what they owe, which means that they’ll be less exposed to interest rate risk in the future.
Lump sums can also be a better option if you expect to die soon. But if you expect to live 20 or 30 more years, the odds are that you’re better off with the annuity.
I think the most important quality of a good Fed Governor (in addition to expertise) is humility. Monetary policy is a blunt and limited tool. It cannot do the following:
Prop up the stock market indefinitely
Impact structural unemployment (unemployment caused by changes in skill demand or globalization)
Have a meaningful effect on long term interest rates (even QE)
Increase the wages or improve the job prospects for specific, targeted minority groups
You know what causes a premature recession or harms Fed independence? It’s not a quarter point rate increase, it's when the Fed overestimates what it can do and it over reaches. Both the left and right appear ignorant of the long, messy, Pre-Volcker history of monetary policy. Bad Fed picks are a symptom of a much bigger problem.
Here’s a case where there is no structural break, history is a good guide here on what not to do.
In other news
Japan loses a mid-size city worth of people per year
Is it honestly surprising that firms don’t go public any more? Why would you, when it means that everyone finds out that you will actually never earn a profit? Better to stay private.
Until next time, Pension Geeks!