Our Strange Recession

The way we study things in generally experimental sciences is to make multiple observations of the same thing under controlled conditions.

The way we study things in generally non-experimental social sciences is to make multiple observations of things we presume might be the same, and then make sure we have all the other variables that might change that included in our analysis. 

The problem with doing this with recessions is ... they're often very different from the preceding one, and often from the whole set of about 40 recessions that we know of, going back in the U.S. to 1820.

Take the example of the current pandemic/lockdown recession. It's an even 100 years since we had a pandemic during a recessions. And, no country on Earth ever locked down anything before they all pretty much decided to just that late last winter.

Anyway, let's take a look at the labor market to get a sense of how different it is. The two graphs come from the Bureau of Labor Statistics website, which has nice tools for finding and charting data.

First up, here's the unemployment rate. This only goes back to 1948, so it missed the Great Depression, which would be not just off the chart, but probably 2 additional blocks up at 25%.

Even so, we can see how the ongoing recession is different: the peak rate is higher, while always steeper on the LHS this one is even steeper, and while it may not continue this way all the way down, it looks like the descent is faster than in earlier recessions. 

Even stranger is the labor force participation rate:

 

Generally, it's fairly hard to spot recessions in the labor force participation rate. 

Digression: this data is often interpreted incorrectly, and this is as good a time as any to explain. Notice that the hump is roughly symmetrical from about halfway up on the left (around 1980), to around 2015 on the right. Casual observers are inclined to say that the labor force participation rate went up, almost exclusively, because of women entering the workforce. That isn't the case: that was more of a smooth rise up until about 2000 — it does little to explain the hump. Instead the hump is likely from baby boomers being in the labor force during their prime years.

Anyway, it's fair to say from this chart that labor force participation has also behaved unusually in the pandemic/lockdown recession.

Looking at the raw numbers may help.

 

Year and month	Civilian noninsti- tutional population	Civilian labor force						Not in labor force
		Number	Percent of population	Employed		Unemployed
				Number	Percent of population	Number	Percent of labor force
1986	180,587	117,834	65.3	109,597	60.7	8,237	7.0	62,752
1987	182,753	119,865	65.6	112,440	61.5	7,425	6.2	62,888
1988	184,613	121,669	65.9	114,968	62.3	6,701	5.5	62,944
1989	186,393	123,869	66.5	117,342	63.0	6,528	5.3	62,523
1990	189,164	125,840	66.5	118,793	62.8	7,047	5.6	63,324
1991	190,925	126,346	66.2	117,718	61.7	8,628	6.8	64,578
1992	192,805	128,105	66.4	118,492	61.5	9,613	7.5	64,700
1993	194,838	129,200	66.3	120,259	61.7	8,940	6.9	65,638
1994	196,814	131,056	66.6	123,060	62.5	7,996	6.1	65,758
1995	198,584	132,304	66.6	124,900	62.9	7,404	5.6	66,280
1996	200,591	133,943	66.8	126,708	63.2	7,236	5.4	66,647
1997	203,133	136,297	67.1	129,558	63.8	6,739	4.9	66,837
1998	205,220	137,673	67.1	131,463	64.1	6,210	4.5	67,547
1999	207,753	139,368	67.1	133,488	64.3	5,880	4.2	68,385
2000	212,577	142,583	67.1	136,891	64.4	5,692	4.0	69,994
2001	215,092	143,734	66.8	136,933	63.7	6,801	4.7	71,359
2002	217,570	144,863	66.6	136,485	62.7	8,378	5.8	72,707
2003	221,168	146,510	66.2	137,736	62.3	8,774	6.0	74,658
2004	223,357	147,401	66.0	139,252	62.3	8,149	5.5	75,956
2005	226,082	149,320	66.0	141,730	62.7	7,591	5.1	76,762
2006	228,815	151,428	66.2	144,427	63.1	7,001	4.6	77,387
2007	231,867	153,124	66.0	146,047	63.0	7,078	4.6	78,743
2008	233,788	154,287	66.0	145,362	62.2	8,924	5.8	79,501
2009	235,801	154,142	65.4	139,877	59.3	14,265	9.3	81,659
2010	237,830	153,889	64.7	139,064	58.5	14,825	9.6	83,941
2011	239,618	153,617	64.1	139,869	58.4	13,747	8.9	86,001
2012	243,284	154,975	63.7	142,469	58.6	12,506	8.1	88,310
2013	245,679	155,389	63.2	143,929	58.6	11,460	7.4	90,290
2014	247,947	155,922	62.9	146,305	59.0	9,617	6.2	92,025
2015	250,801	157,130	62.7	148,834	59.3	8,296	5.3	93,671
2016	253,538	159,187	62.8	151,436	59.7	7,751	4.9	94,351
2017	255,079	160,320	62.9	153,337	60.1	6,982	4.4	94,759
2018	257,791	162,075	62.9	155,761	60.4	6,314	3.9	95,716
2019	259,175	163,539	63.1	157,538	60.8	6,001	3.7	95,636
2020	260,329	160,742	61.7	147,795	56.8	12,947	8.1	99,587
Monthly data, seasonally adjusted(1)
2020
January	259,502	164,455	63.4	158,659	61.1	5,796	3.5	95,047
February	259,628	164,448	63.3	158,732	61.1	5,717	3.5	95,180
March	259,758	162,721	62.6	155,536	59.9	7,185	4.4	97,037
April	259,896	156,478	60.2	133,370	51.3	23,109	14.8	103,418
May	260,047	158,200	60.8	137,224	52.8	20,975	13.3	101,847
June	260,204	159,797	61.4	142,100	54.6	17,697	11.1	100,407
July	260,373	160,085	61.5	143,777	55.2	16,308	10.2	100,288
August	260,558	160,818	61.7	147,276	56.5	13,542	8.4	99,740
September	260,742	160,078	61.4	147,543	56.6	12,535	7.8	100,664
October	260,925	160,718	61.6	149,669	57.4	11,049	6.9	100,207
November	261,085	160,536	61.5	149,809	57.4	10,728	6.7	100,548
December	261,230	160,567	61.5	149,830	57.4	10,736	6.7	100,663
2021
January	260,851	160,161	61.4	150,031	57.5	10,130	6.3	100,690
Footnotes (1) The population figures are not adjusted for seasonal variation.
NOTE: Revisions to population controls and other changes can affect the comparability of labor force levels over time. In recent years, updated population controls have been introduced annually with the release of January data. Additional information is online at https://www.bls.gov/cps/documentation.htm#pop.

I think it's fair to assume that we were at full employment in January 2019. Using that as a baseline, it's fair to say that unemployment is up by about 4.3 million people.

Given the scale of the numbers, population growth contributes a little to that, but can probably be safely ignored. For example, if population grows by about 0.5%/year (as shown in the leftmost column), each column in the table should go up by about that rate, and for unemployment that might contribute 0.1 (rounded up) to the 4.3 million increase.

For the employed, we are off about 8.6 million. But, from above, only half of those are unemployed. Where did the rest go? They are picked up in the decline in the labor force by 4.3 million. 

Note that for the employed and the labor force that the population increase of 1.3 million will make more of a difference. Doing a rough distribution, I'd say that 1.2 million should have gone into employment (meaning the announced number is down more than it looks), 0.1 into unemployment, and 1.3 into the labor force. So, adjusting a little gets the labor force down by 5.6M, employment down by 9.8M, and unemployment up by 4.2M.

Recall how people can be outside the labor force, but still in the civilian non-institutional population: they do not have a job and they are not looking. Further, if they all were looking, unemployment would be up to about 15.7M or a rate of 9.5%. That should put things in perspective that we are in pretty bad shape: it's comparable to the two highest peaks in the unemployment rate in the top chart.

From my longer perspective, I'll tell you that this is something the Democrats have always worried about in recessions, but which never really showed up in the data very seriously: people not interested in being employed. Now we have that, so it's important not to judge this recession by just its unemployment rate. 

It also means that as far as policy goes, we also shouldn't be excessively focused on the unemployed. Yes, we have a lot of them, and many deserve help from policy. But, since past recessions had far fewer people dropping out of the labor force, it's not clear that existing policies or off-the-shelf proposals are going to be focused on the right people.