Is the "Perfect Indicator" Useful, Once Again?

On employing unemployment data to assess the likelihood of a recession. TL;DR at the end of this post.

Recessions are bull-market killers.

Recessions are very often accompanied by rising rates of unemployment.

More precisely, when the trend of unemployment changes — when the number of people out of work is no longer sinking, but is instead rising — then a recession is probably in store.

As demonstrated by this often-referenced FED UNRATE (Federal Reserve unemployment rate) chart:

You see: no recession (grey zone) without rising unemployment (blue line).

The strategy

A 2016 publication by Philosophical Economics describes a very simple approach which from 1930 to 2016 outperformed Buy & Hold in both absolute, and relative terms.

It employs a very basic Buy signal: when unemployment is sinking, you stay invested in the stock market.

Since economic recoveries tend to go on for years and years, this entails many of the advantages of a buy-and-hold strategy, beginning with low transaction costs. But above all, your performance can’t get ruined by a whipsawing market. That’s a major difference from market-timing strategies — timing the market most often means losing money in a trendless environment.

The Sell signal has two layers: When unemployment is rising, and the stock market is no longer bullish: that is when you sell stocks.

(Unemployment rising: when the UNRATE level is above its 12-month moving average. Stock market bullish: when the S&P 500 is above its 10-month moving average).

How did it perform?

From 1930 to 2016: CAGR 10.8%; maximum drawdown -51.3%; volatility 14%; Sortino 0.933.

In comparison, Buy & Hold during the same decades: CAGR 9.3%; maximum drawdown -78.8%; volatility 18.9%; Sortino 0.659.

(How about a purely momentum-based strategy? Buy or sell stocks at their 10-month moving average? CAGR 9.3%; maximum drawdown -49.6%; volatility 12.4%; Sortino 0.825).

Looking at this chart of the last 45 years, we see what appears to be a wonderfully relaxed and yet profitable reaction to the stock market’s short-term up and downs. The strategy went risk-on from 1983 to 1990, from October 1992 to October 2000, from October 2003 to June 2007, and from 2010 to 2020. In other words, it led you to be invested during most of the bull markets of the past decades.

You sold in due course, too. The stock market troubles of 1981, of 1991, the Dotcom bust of 2001 and 2002, and the GFC of 2008 — in all of these cases, you received a warning, which culminated in these

Risk-off periods (when unemployment was growing, and the stock market was below its 10-month moving average):

August 1981 to August 1982

August 1990 to January 1991

January 2001 to April 2003

November 2007 to June 2009

January and November 2023

April and May 2025.

That’s it!

You could call it, “Buy & Hold, but omitting the worst 4 bear-market episodes in the last 40 years, except for the crash of 1987, the Covid Crash of 2020, and the inflationary bear market of 2022”, but that would be ungainly phrasing. More about the latter inflationary bear market later…

Beware the lag, however

Please note the current unemployment data (as of today, December 15, 2025) is from September 2, hence over three months old. UNRATE always lags.

What looked timely in 2020 — in early February, a primary risk-off signal caused by rising unemployment — was old hat by the time the Federal Reserve actually notified us of strongly growing unemployment, in May 2020. (Luckily, the market was also recovering by then).

In most cases, the lag was no problem. A sell signal for June 2007, effective September 2007? Quite early enough.

October 2000 however, effective January 2001? No cigar, but on the other hand, by late 2000, the end of the Dotcom bubble was no big surprise to anybody who had been watching.

How about the current situation? Unemployment rates have actually been rising since around May of 2023, so we are on notice to sell if and when S&P500 sinks below its 10-month moving average. (If the current unemployment trend stays negative!)

Obviously, we would profit considerably from more timely job market data. Does anybody out there know of a better source? If so, please leave a message in the comments, or feel free to get in touch directly.

So, timing the market based on unemployment data worked quite well… until it didn’t

All was well in love and the job markets until Covid 19 screwed everything up. The government and Federal Reserve were flooding the economy with money from 2020 onwards, which led to a highly positive job market situation. In January 2023, unemployment sank to a generational low of 3.4%.

In the midst of this going on however, in 2022, the stock market reacted very negatively to high inflation and rising interest rates, leading to quite heavy drawdowns of -18.17% for those who were invested in the S&P 500.

A bear market in absence of a recession? Quite unusual, for the U.S. at least.

If in 2022, you were waiting for a signal from the the job market to sell stocks, then you were waiting for something that just didn’t happen. With resulting heavy losses. Bummer!

Nonetheless, the strategy generated with the S&P 500 a CAGR of +12.45% for the 10 years of 2016-2025.

(2016: fully invested = +12%

2017: fully invested = +21.7%

2018: fully invested = -4.56%

2019: fully invested = +31.22%

2020: fully invested = +18.37%

2021: fully invested = +28.75%

2022: fully invested = -18.17%

2023: invested February to October +4.47%; December +4,57%, sum = +9.04%

2024: fully invested = +24.89%

2025: invested January to March -4.15%, June to November +15.62%, sum +11.47%)

So, not terribly bad, I’d say. In fact, it’s a nice improvement from the long-term CAGR of 10.8% from 1930 to January, 2016, despite the poor performance of 2022.

No more recessions? Or just tons of A.I. optimism, with trillions of added liquidity?

But what has been going on since 2022? Why hasn’t the natural sequences happened, where a rising rate of unemployment would be accompanied by a recession? And what does this say about the usefulness of job market data for timing the stock market?

Sometimes, bear markets really do happen without a simultaneous recession. For example, Australia didn’t suffer a recession for 29 years, from 1991 to 2020. Meanwhile, the country suffered several episodes of stock-market contraction.

There is a big difference in duration and depth, though. Non-recessionary bear markets typically have seen an average recovery in about five months for the ASX 200.

Recessionary bear markets have lasted significantly longer, often dragging on for 18-20 months.

Declines in non-recessionary bear markets have also been generally less severe, with the ASX 200 experiencing drawdowns of approximately 20% compared to 30-50% in recessionary periods.

Getting back to the U.S., my base line assumption is that Biden’s and Trump’s deficit-spending policies created a lot of liquidity, bolstering both the stock markets and economy in general.

Add to this a capex boom caused by A.I. mania.

Loads of liquidity can feed the economy for quite some time, but this won’t continue forever. A Trump-appointed Fed doesn’t have a magic formula to prevent recessions, and A.I. won’t feed itself.

I think the 2022 example will continue to be the exception. There will always be some strong corrections, such as those of 1994, 2015 and 2018. Not to mention, crashes like in 1987. But generally, the stock market should continue to go hand in hand with the economic cycle.

What do you think? Please share your thoughts in the comments.

In any case, I think it’s good to know that this model sees our current situation, at the end of 2025, as “tentatively risk-off”. Remember, all it needs is for the stock market to sink below its ten-month moving average to generate a real and proper sell signal. Complacency kills returns!

TL;DR

• The unemployment rate is a historically powerful business-cycle indicator

• A change in the trend of the unemployment rate can tell us whether we need to worry about a recession

• Sell stocks when both the unemployment rate and the stock market are in a negative trend

• Buy stocks when unemployment is improving

• Using these rules to buy and sell stocks, a very simple strategy generated a CAGR of 10.8% from 1930 to 2015, and 12.45% from 2016 to November 2025

• This performance is better than buy & hold, and better than simple momentum-based trading

• This is an un-optimized strategy — no tweaks, no options, no dual convexity à la Harry Long. Neither have we added any potentially quite effective out-of-market assets such as treasuries or gold. Much room for further improvement, I think! What do you say?

(Update December 16: I corrected some typos, and changed one yearly number: from “2003” to “2023”).

One more update Dec 16: looking at EU data, which seem to be more timely than UNRATE, we see no recession in Europe coming up. No wonder the stock market there is doing just as well as the U.S.’s.

https://ec.europa.eu/eurostat/de/web/products-euro-indicators/w/3-30102025-bp