Bryan Caplan  

When Does Changing Household Size Resurrect American Economic Performance?

Game of Thrones and the... What's 'freedom' (regular huma...
Today Tyler chides anti-stagnationists who point to falling household size:

Persons per family household:

1990: 3.22

2000: 3.24

2010: 3.24

Not so much change, and if you look you will see there is also not so much change for non-family households.  The Census pdf is here.  I have covered this ground before, but the myth of "changing household size means economic progress has been just fine" dies hard.

Unfortunately, Tyler continues his recent habit of truncating the data to make his case look as strong as possible.  (He also strangely reports data for family households rather than all households, but that doesn't sway the results).  What if he'd shown all U.S. households for the whole history of the Current Population Survey?


Persons Per Household

% Change






















Now we get a nuanced story.  Changing household size was a huge deal between 1970 and 1990, and no deal at all between 1990 and 2010.  Since stagnationists like Tyler usually start their tale of woe in 1973, critics can and should complain that household income is a biased measure of individual income.  Adjusted for family size, income rose about 20% more between 1970 and 1990 than stagnationists like to claim.

At the same time, of course, stagnationists are right to assert that changing household size has been irrelevant since 1990.  But making this concession ruins their narrative.  Changing family size was a big deal in the 70s and 80s.  Growth was pretty good in the 90s.  So instead of four decades of woe, the stagnationists are down to one.  If they were upfront about this - if they framed stagnation as a current event rather than Big Think economic history - I'd take them more seriously.

Update: Tyler acknowledged some role for declining family size before 1989 in both The Great Stagnation and this post.  But I stand by my complaints about the post I'm discussing here.

Comments and Sharing

COMMENTS (6 to date)
Tyler Cowen writes:

There is plenty of data on productivity and also individual-level earnings which support a longer-term stagnation story, you are simply ignoring that I have presented it. It is correct to believe, though, that stagnation became a worse problem after 1999.

Rick writes:

Here's labor (1st column) and multifactor productivity (2nd column) growth (compound annual) in the U.S., according to the BLS. I broke growth rates down by decade instead of cyclically-neutral periods just because it was more practical.

1950-2010 2.4% 1.3%
1950-1960 2.8% 1.7%
1960-1970 3.1% 2.1%
1970-1980 1.8% 1.0%
1980-1990 1.8% 0.8%
1990-2000 2.3% 1.0%
2000-2010 2.7% 1.1%

The data for labor productivity growth is for the business sector, while for MFP growth is private business sector (i.e. excludes government enterprises).

While growth rates in the 50s and 60s were definitely quite high, I really wouldn't call posting almost 2% per year in terms of labor productivity growth and 1% per year in terms of MFP growth in the last 40 years stagnation...

I'm not that familiar with U.S. household income data, but we should definitely look at median equivalised household income (i.e. adjusting for changes in household size) instead of just median household income.

Bob writes:

What about looking at earners per household as opposed to persons per household?

What we know from the static income distribution data by quintile is that the household sizes aren't all that very different across quintiles but the number of earners is. The top quintile has something like an 4x the earners per household than the bottom quintile.

Why so many fewer people work (or work less) than in the past is the problem. Wages are up per worker, but we have fewer workers. Income looks stagnant as a result.

Some evidence:

Thomas DeMeo writes:

The Persons Per Household numbers for the years 1990, 2000, and 2010 are different in the quote from Tyler and subsequent chart described as the whole history of the Current Population Survey. Which is correct?

dave smith writes:

Adding to Bob's logic, you'd also need to look at person per household per income quintile.

I think Tyler is dismissing demographic factors too quickly.

Brian writes:

Thomas DeMeo,

Tyler gives persons per FAMILY household while Bryan gives persons per household for ALL households. (See Bryan's parenthetic remark after presenting the Cowen data.) The per family household number will naturally be larger since it excludes single-person households.

Comments for this entry have been closed
Return to top