The Unassuming Economist

From The Economist magazine:

“They perform far better when forecasting growth years than downturns

SOME INVESTORS worry that America will face a recession in the next few years, after one of the longest expansions on record. Stock indices have fallen by 10% since early October. Yields on short-term government bonds exceed those of some longer-dated ones, often a harbinger of a downturn. Despite this, economic forecasters project GDP growth of about 2% in 2020.

How much confidence should one have in these predictions? For the past 20 years The Economist has kept a database of projections by banks and consultancies for annual GDP growth. It now contains 100,000 forecasts across 15 rich countries. In general, they fared well over brief time periods, but got worse the further analysts peered into the future—a trend unsurprising in direction but humbling in magnitude. If a recession lurks beyond 2019, economists are unlikely to foresee it this far in advance.

Economies are fiendishly complex, but forecasters usually predict short-term trajectories with reasonable accuracy. Projections made in early September for the year ending four months later missed the actual figure by an average of just 0.4 percentage points. Errors rose to 0.8 points when predicting one year out. But over longer horizons forecasts performed far worse. With 22 months of lead time, they misfired by 1.3 points on average—no better than repeating the previous year’s growth rate.

 

The biggest errors occurred ahead of GDP contractions. The average projection 22 months before the end of a downturn year missed by 3.7 points, four times more than in other years. In part, this is because growth figures are “skewed”: economies usually expand slowly and steadily, but sometimes contract sharply. As a result, forecasters seeking to predict the most likely outcome expect growth. However, they adjust too slowly even once bad news arrives, says Prakash Loungani of the IMF. That suggests they are prone to “anchoring”—over-weighting previous expectations—or to “herding” (keeping their predictions near the consensus).”

Read the full article here.

A separate article from Bloomberg says that:

“Recession Signs Hard to Miss If Stock Message Is Taken Seriously

Prices bounce around, emotion obscures logic, signals appear and vanish. The reasons for treating equities as a poor barometer for the economy are many. Right now, that might be for the best.

Pools of gloom await anyone looking for a message in stocks. There’s the $3 trillion in value erased, the bloodbath in banks and the trouncing in transports. In wonkier circles, shrinking valuations and negative rolling returns have started to ring the recession bell. A relatively calm week in the Dow Jones Industrial Average just ended with a 495-point thud.

It’s a pastime on Wall Street these days to look at the carnage, add it all up, and announce that the market is wrong. But what if it’s not? After all, even if equities have predicted “nine of the last five recessions,” as the economist Paul Samuelson famously said, that’s a better record than a lot of humans.

(…)

Sure, markets overshoot, and sentiment gets carried away. Corrections like this one have occurred six other times since the bull market began in 2009. They all sparked growth scares. But none of them a recession.

“The market is wrong,” said Anik Sen, global head of equities at PineBridge Investments. “Clearly it has been a slowdown, but the slowdown can be very transitory in our view. At the end of day, there is enormous pent-up demand, whether it’s capex or technology spending. None of that has changed.”

Still, anyone heeding strategist calls shouldn’t forget Wall Street’s propensity to lean bullish. Over the past two decades when stocks suffered two bear markets, professional forecasters have never once predicted a down year. Economists don’t see one now, either. Eighty-nine surveyed by Bloomberg generate an average prediction of 2.6 percent growth in gross domestic product next year.

Meanwhile, a 2014 study by Prakash Loungani of the International Monetary Fund found that not one of 49 recessions suffered around the world in 2009 had been predicted by the consensus of economists a year earlier. Loungani previously reported that only two of the 60 recessions of the 1990s had been anticipated a year in advance.”

Read the full article here.

Also, see a related note on “There will be growth in the spring”: How well do economists predict turning points?

From a new IMF working paper:

“I regress real GDP growth rates on the IMF’s growth forecasts and find that IMF forecasts behave similarly to those generated by overfitted models, placing too much weight on observable predictors and underestimating the forces of mean reversion. I identify several such variables that explain forecasts well but are not predictors of actual growth. I show that, at long horizons, IMF forecasts are little better than a forecasting rule that uses no information other than the historical global sample average growth rate (i.e., a constant). Given the large noise component in forecasts, particularly at longer horizons, the paper calls into question the usefulness of judgment-based medium and long-run forecasts for policy analysis, including for debt sustainability assessments, and points to statistical methods to improve forecast accuracy by taking into account the risk of overfitting.”

From a new working paper:

“Economic forecasts are often disseminated via a survey of professionals (i.e. “Consensus”). In this paper we compare and contrast the Consensus with a crowdsourced alternative wherein anyone may submit a forecast. We focus on U.S. Nonfarm Payrolls and find that, on average, Consensus is more accurate, but the best crowdsourced forecasters are superior to the best Consensus forecasters. We also find that information plays a key role. When the Consensus is uncertain and herds together, the crowdsourced forecasts appear to be more. Our findings provide evidence that crowdsourcing might provide a valuable supplement to traditional macroeconomic forecasts.”

A new Bloomberg post cites my study:

“Nine turbulent weeks and a correction in U.S. stocks have left analysts with a thorny question. What’s the market saying about the economy? And while few see incontrovertible signs investors are bracing for a recession, it’s a word that’s been coming up more as they seek a signal in the chaos.

From the ascent of defensive industries to the sudden craze for companies that resist volatility, stocks are acting in ways that have presaged slowing growth in the past. That makes sense: gains in the economy and corporate earnings are forecast to ease in 2019 from this year’s torrid pace.

Befitting that, most of the charts that follow reflect observations by analysts who don’t see a recession as the most obvious conclusion. Many view the sell-off as healthy after a 10-year run of gains. But with a trade war flaring and the Federal Reserve set to boost interest rates again, the number of stock researchers who are at least willing to mention the possibility is rising.”

“Economists haven’t always done a great job predicting contractions. A 2014 study by the International Monetary Fund’s Prakash Loungani found that not one of 49 recessions suffered around the world in 2009 had been predicted by the consensus of economists a year earlier. Loungani previously reported that only two of the 60 recessions of the 1990s had been anticipated a year in advance.”

“[…] the economic indicators that often precede recession — yield curve inversion and rising unemployment — are not flashing warning signs. The yield curve is flat but not inverted and the unemployment rate keeps falling, as opposed to rising when a recession approaches.”

From a new post by Francis Diebold:

“I just returned from a great BFI Conference at U Chicago, Developing and Using Business Expectations Data, organized by Nick Bloom and Steve Davis.

Wonderfully, density as opposed to point survey forecasts were featured throughout. There was the latest on central bank surveys (e.g., Binder et al.), but most informative (to me) was the emphasis on surveys that I’m less familiar with, typically soliciting density expectations from hundreds or thousands of C-suite types at major firms. Examples include Germany’s important IFO survey (e.g.,Bachman et al.), the U.S. Census Management and Organizational Practices Survey (e.g., Bloom et al.)., and fascinating work in progress at FRB Atlanta.

The Census survey is especially interesting due to its innovative structuring of histogram bins. There are no fixed bins. Instead users give 5 bins of their own choice, and five corresponding probabilities (which add to 1). This solves the problem in fixed-bin surveys of  (lazy? behaviorally-biased?) respondents routinely and repeatedly assigning 0 probability to subsequently-realized events.”