Expecting the unexpected

Both of the following might be true:

1. A recession became much more likely after 2021.

2. During any given year, a recession is unlikely.

I’ll use a numerical example to illustrate the way I think about recession risk.

Assume that during normal times, the Fed aims for steady NGDP growth, say 4%/year. Because the Fed is a bit “clumsy”, actual NGDP growth fluctuates over time. Let’s suppose that 30% of the time the growth rate is at least 2% too high, and there’s also a 30% chance that it’s at least 2% too low. There’s a 10% chance that NGDP growth is at least 4% too high, and a 10% chance that growth is at least 4% too low.

Now assume that a more than 4% drop in NGDP growth leads to a recession. In that case, there’s a 10% chance of recession in a given year during normal times.

Now assume that inflation has become a big problem. The Fed knows that a 4% drop in NGDP growth leads to recession, so they aim for a “soft landing”. They hope to reduce NGDP growth by 2% each year. Now there is a much higher risk of recession. A mere 2% error to the downside causes NGDP growth to fall by 4%, triggering a recession. The risk of recession rises from 10% to 30%. That’s still not particularly high, but if the anti-inflation policy is done two years in a row, then the recession risk rises to 51%. Over three years it’s over 65%.

To summarize, it’s possible to enter a period where recession clearly becomes increasingly likely, even as it remains exceedingly difficult to predict the timing of a recession with any confidence.

The term “soft landing” can be defined in several different ways. An extremely soft landing would allow inflation to fall while growth continues and unemployment remains below 4%.

The Fed is forecasting a rise in unemployment to roughly 4.5%. That sort of moderately soft landing might be called a “mini-recession”. It might seem odd to forecast a mini-recession, as (AFAIK) the US has never experienced that phenomenon. Unemployment always rises by less than 1% (percentage point) or more than 2%. But that seems to be roughly what the Fed is aiming for, and lots of other countries have had mini-recessions. There’s no obvious reason why we could not have that outcome.

The recession scare of 1966 offers another lesson. In that case, the Fed failed so badly in its attempt to bring down NGDP growth that they ended up overshooting to the upside in 1967-68 (after a brief dip in 1966.) This illustrates an important point—policy failure is much more likely under an anti-inflation regime than under a normal regime. A bit too tight and you end up in recession. A bit too easy and the high inflation continues. There’s far more room for error under a normal regime. Thus NGDP growth slowed a couple percentage points in 2016 without triggering a recession (but it did cost Hillary the election.)

All macro models are wrong, including the one I provided here. In reality, NGDP misses are not normally distributed. If my model were exactly true, then the US would have more mini-recesisons than outright recessions. But we haven’t had any mini-recessions. Errors on the contractionary side can easily snowball into a sharp downturn, due to the Fed’s flawed policy regime (lack of level targeting.)

Nonetheless, this numerical example illustrates a few points that recession forecasters in the media don’t seem to grasp. Just as it’s really hard to predict when a bus driver will accidentally steer a bus off the road, it’s difficult to predict when the Fed will accidentally steer NGDP growth too slow (or too fast). On the other hand, just as buses are more at risk on narrow mountain roads, the economy is more at risk when he Fed is engaged in an anti-inflation policy.

Fed policy actually seems roughly on track at the moment, even though the policy regime remains highly flawed. In other words, the central point in the distribution of possible outcomes is a roughly appropriate slowdown in NGDP growth, but the distribution is much too wide due to the Fed’s flawed policy regime. I’m worried about both recession and high inflation.

How could the Fed make accidents less likely? How can they reduce the “fat tails” in the NGDP growth distribution? How can they become less “clumsy”? Check out my license plate:

PS. Soon the US will announce a big drop in 12-month CPI inflation, to the low 3s. Don’t believe the hype—inflation remains a big problem. Base effects.



25 Responses to “Expecting the unexpected”

  1. Gravatar of Ray Lopez Ray Lopez
    9. July 2023 at 19:00

    Based. I’m in the high-inflation camp too. Nice vanity plate.

  2. Gravatar of Michael Sandifer Michael Sandifer
    9. July 2023 at 19:16


    Off-topic, if you haven’t tried the new GPT-4 baesd tool, called Code Interpreter, you may want to consider doing so. It was just widely released Friday, and it does require a $20/month subscription, but it’s well worth it for those doing any sort of data analysis/data science.

    I want to emphasize that this does not require coding. You can upload a raw Excel file that isn’t even cleaned, tell Code Intepreter what you want done with the data, and it will do it. It is an automated data analyst.

    I copied and pasted the Feds Funds Futures data from the CME website into an Excel file, uploaded it raw, and simply used this prompt:

    “Can you clean up this data and produce a chart of the implied expected Fed Funds rate based on this data displaying the Fed Funds futures curve?”

    The chart it produced was very impressive, and took only seconds. I needed only to further prompt it to bold the font and add data labels, to my preference.

    I stress, this doesn’t require coding, but it obviously can employ Python libraries to create data visualizations.

    With regard to coding, as an aside, it is much better than ChatGPT-4. It’s greatly speeding up my website development now.

    This is truly the dawn of a new age. What will happen to data analysts?

  3. Gravatar of Rajat Rajat
    9. July 2023 at 22:34

    Thanks Scott, a very interesting post. I was wondering about your numerical example until I realised for the first time something that I probably should have realised much earlier, but had perhaps never been particularly important in the decade pre-Covid: That recessions occur when NGDP growth is significantly lower than recent rates, even though NGDP growth may remain well above zero and even if it remains above the long term (implied) target. That’s presumably because (or to the extent that) wages growth adjusts slowly to changes in NGDP growth. In other words, if NGDP growth slows – even to target – rapidly, the music stops and some workers are left without a chair. Is that right? If so, it means that the task for the Fed is not just to return NGDP to target (say 4-5%), but to do so slowly enough that real wages do not rise fast enough to cause unemployment.

    Also, by ‘normal’ regime, do you mean a ‘NGDPLT’ regime?

  4. Gravatar of Student Student
    10. July 2023 at 06:36

    The Musk post got me thinking… you should challenge Jerome Powell to a cage fight over the NGDPLT regime.

  5. Gravatar of ssumner ssumner
    10. July 2023 at 09:25

    Michael, Interesting.

    Rajat, I should have clarified that by “normal” I meant an economy roughly at equilibrium, with no need for austerity or stimulus.

    Student, It’s interesting that someone as brilliant as Musk feels a need to continually act like an immature 8th grader. Didn’t he tweet “Zuck is a cuck”?

  6. Gravatar of Student Student
    10. July 2023 at 11:24

    It is amazing a guy that rich and powerful acts like that… and it’s also amazing… in an odd way, I like it. I would pay 50 bucks to watch that fight on pay per view lol.

  7. Gravatar of Student Student
    10. July 2023 at 11:24

    I would be rooting for a simultaneous knockout lol.

  8. Gravatar of Sara Sara
    10. July 2023 at 12:38

    All good, but if you take this logic to its natural conclusion, then there is no Federal Reserve at all.

    It’s better to have nobody at your party around the fine China who is “clumsy” then to have a few who are “less clumsy.”

    We should have learned this in the 30’s, after the Fed reduced rates and fueled the 20’s boom, and when Keynes admitted he critiqued Mises book without ever reading it because — get this — he couldn’t read German. That’s how arrogant Monetary economists are. A guy critiques a book he cannot read, then publishes a manifesto in 1936 which does nothing but glorify spending, and lo and behold the profession actually takes it seriously. It’s pseudoscience!!!!

  9. Gravatar of Kevin Erdmann Kevin Erdmann
    10. July 2023 at 12:53

    On your ps, shouldn’t we say “don’t believe the hype, MoM inflation (without the lagging and otherwise problematic CPI shelter component) has already approximated the 2% target rate for 11 months.”?

  10. Gravatar of Garrett Garrett
    10. July 2023 at 14:21

    Kevin, I think 2% annualized MoM inflation would still be a problem due to compounding as the level has shifted up significantly from the past decade trend up to 2020. 2024 inflation is still going to be much higher than it was expected to be in 2019-2021, which I would think still has some distortionary effects due to long term planning. And current long term planning out to the second half of this decade is probably being distorted due to increased uncertainty regarding the path of NGDP.

    Hope Scott corrects me anywhere I’m wrong!

  11. Gravatar of ssumner ssumner
    10. July 2023 at 14:27

    Kevin, And without the problematic food and energy component, inflation remains a big problem.

    It’s best to focus on wage inflation or NGDP, and just ignore the CPI entirely. But to the extent that the CPI matters, then it’s likely to be much too high going forward.

    Garrett, I see that as a separate problem—the Fed’s abandonment of average inflation targeting, which was a major error. But even ignoring past mistakes, inflation remains hot going forward.

  12. Gravatar of Solon of the East Solon of the East
    10. July 2023 at 15:22

    The Reserve Bank of Australia targets a 2% to 3% rate of inflation, while the People’s Bank of China targets a 3% rate of inflation.

    A bit of an oddball is the Reserve Bank of India that targets a 4% rate of inflation, +/- 2%.

    US macro economists have made the rate of inflation a bit of a fetish.

    It might be interesting to do a Google search and find out if macro economists talk more about “prosperity” or “inflation.”

  13. Gravatar of Bobster Bobster
    10. July 2023 at 20:55


    What about the argument that the longer we wait to tame inflation the worse a recession will be?

    Because of higher inflation expectations

  14. Gravatar of ssumner ssumner
    10. July 2023 at 21:43

    Bobster, I agree.

  15. Gravatar of Kevin Erdmann Kevin Erdmann
    10. July 2023 at 23:49

    What tips spreads today would make you comfortable?

  16. Gravatar of spencer spencer
    11. July 2023 at 06:03

    Before Powell eliminated reserve requirements, you could always determine whether a recession was or not imminent. Monetarism has never been tried.

  17. Gravatar of spencer spencer
    11. July 2023 at 06:06

    The 10-month rate-of-change in our means-of-payment money bottomed in June. Without additional QT, there will be no recession in 2023.

  18. Gravatar of spencer spencer
    11. July 2023 at 06:34

    re: “the Fed failed so badly in its attempt to bring down NGDP growth that they ended up overshooting to the upside in 1967-68 (after a brief dip in 1966.)”

    These two articles by Dr. Leland J. Pritchard, Ph.D. Economics, Chicago 1933, M.S. Statistics, Syracuse (Phi Beta Kappa).



    This article sums up the error:


    Savings flowing THROUGH the nonbanks never leaves the payment’s system.

  19. Gravatar of spencer spencer
    11. July 2023 at 06:45

    The “demand for money” has fallen by 18% since C-19. This has increased the transaction’s velocity of money (not income velocity).

  20. Gravatar of ssumner ssumner
    11. July 2023 at 08:38

    Kevin, I’m not all that uncomfortable with the tips spreads, which are just one of many factors that I consider. (Although a bit lower would be better.)

    My point is that policy has been a fiasco over the past couple of years and the doves at the Fed are 100% to blame. The decision to abandon average inflation targeting was a disaster.

  21. Gravatar of Ray Lopez Ray Lopez
    11. July 2023 at 20:12

    The aging Boomer Scott Sumner is, like aging generals, fighting the last war. While I too am an ‘inflationista’ as I lived through the 1970s (nostalgia), the truth is, as a paper by the Bank of England showed a few years ago, inflation has been dropping by and large for the last 500 years (even the New World Spanish silver mines were not that inflationary, after adjusting for population growth). One reason is technology always makes things cheaper for the same unit of money. On occasion there’s a spike in inflation but it’s always temporary and the long run tends to be deflationary. Not that it matters (money is everywhere and always neutral).

  22. Gravatar of spencer spencer
    12. July 2023 at 05:04

    “According to Dr. Milton Friedman, the main reason for the non-neutrality of money in the short-run is the variability in the time lag between money and the economy.”

    The neutrality of money has been denigrated (changes in the money supply only affect nominal variables and not real variables).

    See: “Inflation continues to outpace Americans’ rising wages – for the 26th straight month”.

  23. Gravatar of spencer spencer
    12. July 2023 at 05:17


  24. Gravatar of spencer spencer
    12. July 2023 at 05:20

    As Batra Ravi explains: “What matters is the size of the wage-gap”.
    Batra, pg. 142, “Greenspan’s Fraud”:

    “If demand and supply are to be balanced over time, then either wages rise in sync with productivity, or productivity growth must be matched by the growth of wages plus debt.”

  25. Gravatar of TheMoneyIllusion » A wise owl TheMoneyIllusion » A wise owl
    14. July 2023 at 08:15

    […] would be getting back to 2% with only a small increase in slack. I discuss this in a recent post on the difficulty of achieving a soft […]

Leave a Reply