Economists haven’t thrown out the models yet (but they will)

Danny Crichton

There are many things that are in short supply these days, but one of them is cognitive flexibility.

Humans are used to confronting the extreme complexity of our modern world with some mélange of data, modeling, heuristics, and “gut instinct.” We can’t know everything about everything, and so we have to truncate our perspective to order the chaos before us. In linear times, that process actually (mostly) works — the economy generally trends one way, and while models and instincts can be a bit off, they are fairly robust to reality.

The novel coronavirus presents us with a novel scenario — one we have never experienced before in the modern, globalized economy. No model and no gut is up to the challenge of evaluating exactly what is going on. Every way we see the world — quite literally everything — has to be rebuilt from scratch.

Take two of the most important leading indicators of economic health that were released last week in the United States: the latest unemployment insurance weekly claims which came out on Thursday, and the employment surveys for March which came out on Friday.

Here is Max Reyes at Bloomberg conveying the predictions of economists before the release of those insurance figures:

The median projection from economists surveyed by Bloomberg puts the figure at 3.5 million, slightly above the prior reading, which was already more than quadruple the previous record.

The real result was 6.6 million — off by a factor of two. Think of how many standard deviations away from the median you have to be to get a number like that.

The real kicker for me though came the next day. The unemployment figures (like many economic statistics) are tricky, for while they are announced the first Friday of each month, their survey period ends more than two weeks prior. So the “March” employment data really represents mid-February to mid-March. Normally, that discrepancy would only be discussed by a bunch of econometricians in a grad seminar, but given the gargantuan changes in the economy these days, that artifact of survey design is suddenly front and center.

Here is Eric Morath in the WSJ summing up the views of economists prior to the release of the statistics:

Economists surveyed by The Wall Street Journal through Monday forecast employers cut 10,000 workers from payrolls and the unemployment rate for March ticked up to 3.7% from 3.5%. Those benign numbers miss mass layoffs announced by restaurants, retailers, manufacturers and others because they occurred in the second half of March, after the survey period.

Morath at the end of a previous article gets at the challenge:

Possible outcomes for the March report are unusually wide. Surveyed economists are forecasting a range of employers adding 312,000 jobs to shedding 1.25 million. Some expect the unemployment rate to hold at a 50-year low, others project it will rise above 5%.

“No forecasting models are built to deal with the unique situation we have,” said Brad Hershbein, an economist at the Upjohn Institute for Employment Research. “There’s tremendous uncertainty.”

And indeed, Hershbein was right, because the median was wildly off:

Payrolls decreased by 701,000 jobs in March, the Labor Department said Friday, as efforts to contain the virus disrupted the U.S. economy [and] The unemployment rate for March rose to 4.4% from 3.5% in February, the largest one-month increase in the rate since January 1975.

10,000 versus 701,000.

One of the hardest parts for experts — and the press — is how to convey the sheer level of uncertainty in the market these days. A range of +312,000 to -1.25M is not a range — it’s every possible reality under the sun (that said, the economist(s) who predicted hundreds of thousands of new jobs in the midst of a global pandemic should probably stop going to the dispensary and shelter in place).

In the startup world, the most common refrain I get from VCs is “I don’t know.” That’s actually healthy, as no one knows what the market is going to do next week, let alone three months or two years out. While there are a variety of scenarios that are reasonable, those scenarios are often so incompatible that it creates analysis paralysis.

Will travel pick up in two years due to pent-up demand, or will we see a long-term secular decline in exploring the world? Will retail be radioactive, or will the novel coronavirus clear away a bunch of deadwood shops and reset the retail landscape, making it among the most dynamic sectors of the economy this coming decade?

No one knows. All that data and all those models, heuristics, and instincts can give you some perspective, but they are just as likely to prevent you from seeing the new world as they are to help you understand it.

Dividends versus share buybacks

Cognitive inflexibility isn’t just a problem among experts — everyday consumers and retail investors are just as susceptible.

Top banks around the world are reconsidering their dividends as they think through reserve requirements and their own lending needs. That doesn’t sit well with some investors:

A group of Hong Kong-based retail investors who own 2% of HSBC Holdings’ shares are demanding a scrip dividend [equity dividend] in place of the cash distribution that was canceled by the bank for the first time in 74 years under pressure from U.K. regulators, and are threatening to mobilize to force a special shareholders’ meeting.

Shoring up and maintaining faith in the banking system is absolutely critical for the long-term health of these finance stocks. So a short-term dividend could very well come at the cost of nothing less than the complete collapse in value of the underlying asset.

Yet, the signals of a dividend are so strong that banks are willing to overlook their own financial health in order to not send a bad signal to the market. Richard Henderson at the Financial Times writes this morning about the negative outlook for dividends in the futures markets:

Dividends are a marker of financial stability for listed companies that try to raise payouts year after year, and which are typically penalised by investors when they fail to do so. The steady growth of dividends contrasts with the recent explosive growth of share buybacks, which are seen as a more flexible way to hand money back, as programmes can be paused or scrapped without much fanfare.

“You don’t want to cut dividends because it’s a really bad signal,” said Mr Velis. “Once you’ve established your dividend you only cut in extreme circumstances.”

“You only cut in extreme circumstances.” But apparently a massive health emergency is not an extreme circumstance. Stability is so important in fact, that banking executives are plowing ahead as if nothing in the world has changed:

Meanwhile, Morgan Stanley chief James Gorman told CNBC that suspending dividends would be a “very poor thing to do” and would be “destabilising”, while Citigroup boss Mike Corbat told the same channel the bank’s dividend was “sound and we intend to keep paying it”.

In a linear world, such a heuristic probably makes sense — stable companies should have stable dividends. But given the current situation, you can see why companies might need to take a step back from shoveling cash out the door. And yet they don’t … locked into a model of the past and not thinking about how to navigate the future.

These are novel times, and we need to recognize the limitations of the heuristics, models and all the other cognitive apparatuses that we use to make sense of the world. And then we need to throw them away. This is not a time to double down on what we learned 15 years ago about how the economy functions. This is an opportunity to rebuild our way of thinking for the long haul.