top of page
  • Brendan Stec

How did we not see this coming?

The questionable value of predicting recessions. What does this mean for your portfolio?

Fears are high among American investors today. Scary memes like trade war! and rate hikes! and recession! cling for relevancy through articles in the Wall Street Journal and discussions on CNBC. And that last one - recession! - seems particularly potent. Only 10 years ago was The Great Recession of 2008, a shocking financial collapse that erased over $11.5 trillion of household net worth, tossed millions of Americans out of work, and instilled a "how did we not see this coming" sense of frustration among investors that held poorly-performing investments. Not all recessions are as severe as The Great Recession, but investors today are particularly afraid of getting blindsided by the next one that is "due." So, how do we see this next recession coming?

The short answer: we can't. Or it is at least extremely difficult to predict when a recession will occur. Economists in the United States technically define a recession to be 2 consecutive quarters of negative GDP growth. The problem is that economists aren't particularly good at forecasting GDP growth, let alone when GDP will shrink and lead to a recession. Let's observe the following facts:

1) Each November from 1993 to 2010, economists from the Survey of Professional Forecasters gave 90% confidence intervals of the following year's GDP growth. These forecasts, which should only be wrong 10% of the time, were wrong more than 30% of the time. From 1968 to 2010, they were wrong almost 50% of the time. (Silver, p. 150)

2) These forecasters were especially wrong during times of recession. In November 2000, the average guess put calendar year 2001 GDP growth at around 3.0%. The actual figure was merely 0.15%. In November 2007, the average guess put calendar year 2008 growth at a shade over 2.0%. The actual figure was negative at -3.3%. (Silver, p. 151)

Figure 1: GDP Forecasts: 90 Percent Prediction Intervals Against Actual Results (Silver, p. 151)

Figure 1: GDP Forecasts: 90 Percent Prediction Intervals Against Actual Results

Source: Nate Silver from his book The Signal and the Noise

3) In the 1990s, economists predicted only 2 of the 60 global recessions a year in advance according to the International Monetary Fund. Even worse, about 25% of these forecasts still predicted positive growth for a country just 2 months before a recession began! (Loungani)

4) In 2018, the IMF extended this analysis to include over 63 countries from 1992 to 2014. The conclusion? Forecasters miss the magnitude of a recession by a wide margin until the year is almost over:

Figure 2: Average GDP Forecasts vs. Actual GDP, Consensus and IMF (An et al, p. 3)

Source: Zidong An, João Tovar Jalles, and Prakash Loungani from their paper "How Well Do Economists Forecast Recessions?"

The above chart shows that in a given year t in which a recession occurs, the average GDP declines 2.8% (represented by the black bar) in a given country. In April of the year before the recession, both consensus and IMF forecasts overestimate GDP growth by over 5 percentage points. In April of the following year, when the recession is already happening, forecasters still project GDP growth for that year 2 percentage points higher than reality! These economists - both in the U.S. and abroad - consistently provide inaccurate estimates of future GDP growth.

They simply cannot predict recessions.

Why do they have so much trouble? To be fair, economists have a lot of challenges stacked up against them. For one, the economic data they rely upon to make their forecasts is noisy and is constantly being revised, including GDP itself. For example, in the fourth quarter of 1977, the government initially reported strong GDP growth of 4.2% only revise this estimate down to a borderline recessionary -0.1% at a later date. This means the very data economists are trying to forecast is of marginal quality and can bounce around for reasons not related to the economy at all.

Second, since there have only been 11 recessions since World War II, economists are working with an extremely small sample size in which to draw inferences about what may cause a recession. Even worse, there are thousands upon thousands of "economic indicators" economists use as potential predictor variables. With thousands of dependent variables and only 11 independent samples, it's clear any model that appears predictive on paper is likely overfitting the data.

Finally, the economy is a complex, constantly-changing system that is challenging enough to understand in the present, let alone in the future. For example, few economists in the 1950's probably predicted that millions of women would enter the labor force in the following decades, effectively doubling the economy's productivity and leading to an unprecedented "Long Boom" of GDP growth through the 20th century. Few economists in 2007 predicted that arcane mortgage-backed securities would enable a nationwide financial crisis, leading to job losses, foreclosures, consumer spending cuts, and The Great Recession. In the economy, small, unpredictable events can trigger massively unforeseen phenomena. Just as a butterfly flapping its wings in Brazil could cause a hurricane in Florida, President Trump's decision to raise tariffs could cause an all-out trade war and consequent recession. Or it could not. But can economists - or anyone for that matter - predict Trump's behavior?

So the data sucks, there's not enough of it, and it comes from a system that's complex and suffers from unpredictable feedback loops. It's no wonder economists have trouble predicting recessions. And it probably means we shouldn't try to predict them either.


If you have a portfolio of stocks, bonds, ETF's, real estate or other investments that would likely decline in value in the instance of a recession, you probably care a lot about when the next recession is happening so you know when to sell. Since timing is clearly difficult, is there anything else you can do to fortify your investments?

Here are 2 important things that matter for an investment portfolio, regardless of whether a recession is looming or not:

1) Diversification: Since different asset classes appreciate and depreciate in value at different times, holding a mix of these different assets can reduce a portfolio's overall variance (i.e. how much it bounces around in value over time). Not to get too technical here, but the below formula for overall portfolio variance summarizes my idea well:

Figure 3: Overall Variance of a Portfolio That Holds Asset A and Asset B

Source: Rusty Guinn's article "You Still Have Made a Choice: Things that Matter #2"

When a recession hammers a risky asset like "Asset A", maybe "Asset B" doesn't do so bad or maybe it even increases in value. Maybe because these two things don't always move together, your overall portfolio doesn't get hammered when a recession happens. And maybe this means you don't care as much when a recession happens, not because you don't know when it will, but because when it does happen you know your portfolio will take less of a hit than if you were not diversified.

This may sound like Investing 101, but how frequently do you waste hours trying to pick stocks, time the market, or predict recessions instead of focusing on a proven strategy of reducing risk? Diversification is always helpful, whether a recession is coming or not.

2) Costs: Doing research on investments, talking markets, and staying on top of the economy's direction is fun. It's sexy. So it's tempting to spend all of your time on these things. But costs - boring old costs - are a drag on your portfolio's return and are important too. What kind of costs am I referring to? There are several:

- Expense ratios: Mutual funds and ETF's charge an annual fee called an expense ratio. A low-cost large cap ETF expense ratio may only be 0.1%, that is, $10 on an investment of $10,000. But some mutual funds charge over 1% annually, or $100 on an investment of $10,000. Are these higher fees worth it for investors? Research from Dimensional Fund Advisors (in their 2017 "Mutual Fund Landscape") argues that mutual funds with the highest expense ratios tend to have the lowest rates of outperformance. If you're a retail investor looking to buy an ETF or mutual fund for your portfolio, it's important to ask yourself: is that $90 of extra fees each year worth it? Maybe it is - but you should at least ask yourself the question and avoid the extra costs if you can.

- Taxes: Investment taxes must be paid on taxable assets that generate income or capital gains. However, these taxes can be managed efficiently in order to minimize tax costs and maximize after-tax returns for your portfolio. For example, for stocks, if you buy a stock and sell it for a capital gain less than 1 year later, you pay a tax rate of 39.6% on the gain. If you buy and hold the stock for longer than 1 year, you only pay 20%. This is a huge difference and is another reason why you should hold stocks for the long run. Of course, if you buy a stock and sell it for a loss, you pay no taxes. By selling losers that cancel out the winners and avoiding short-term gains, you can effectively minimize tax costs from the equities in your portfolio. For a more complete explanation of tax cost management, here is an interesting article that also discusses taxes on bonds and mutual funds.

- Trading costs: Each time you buy and sell a security, you incur a bid-ask spread and (although they're small) commissions. So the more you buy and sell, the more direct trading costs you incur. Pretty straightforward, right? But there is another insidious cost that's hiding in there: the "market timing cost." Suppose you own Tesla, *cough*, Asset A. Asset A is up one day and down big the next. As the anxious, biased, hormone-rich retail investor that you are, you try to buy shares of Asset A when its low and sell shares when its high. Better than doing nothing, right? In fact, due to the fact most investors are horrible at timing markets, you're really buying high and selling low. By trading your portfolio too frequently, you are incurring execution and timing costs that can quickly eat away your returns.

Unlike economic forecasting or investment research, spending time minimizing costs from fees, taxes, and trading costs is guaranteed to be worthwhile for your portfolio.

Given economists' poor track record of forecasting GDP and predicting recessions, any article, report, or economist on TV that claims a recession is looming is not providing you any helpful information for managing your savings. Your first priority should not be macroeconomic forecasting or investment research, but ensuring your portfolio is robustly diversified and efficiently minimizing costs. These are much more straightforward goals that even a novice individual investor can attain.

Download a PDF of this article by clicking here


Disclosures: This commentary is being provided to you as general information only and should not be taken as investment advice. No one can accurately predict the future of the market with certainty or guarantee future investment performance. Past performance is not a guarantee of future results.



An, Zidong, et al. “How Well Do Economists Forecast Recessions?” IMF, International Monetary Fund, 5 Mar. 2018,

Guinn, Rusty. “Wall Street's Merry Pranks: Things That Matter #4.” Epsilon Theory, 12 Dec. 2017,

Guinn, Rusty. “You Still Have Made a Choice: Things That Matter #2.” Epsilon Theory, 10 Aug. 2017,

Loungani, Prakash. “The Arcane Art of Predicting Recessions.” IMF, International Monetary Fund, 18 Dec. 2010,

Silver, Nate. The Signal and the Noise: Why Most Predictions Fail But Some Don't. Penguin Press, 2012.

bottom of page