Thread by @AGGraziano

Working on revising my jmp, I realized I did not see any thread summarizing the 2017 US Economic Census concentration results, released some months ago. I do so here.
#EconTwitter
1/n

Background: There have been many papers documenting an increase in industrial concentration in the US and other developed countries lately. It does not necessarily mean an increase in the market power of large firms, but the trend worried academics and policymakers.
2/n

Here I use publicly available US Economic Census data from 1997-2017. Caveat: I use 5-digit NAICS codes (about 600 sectors), not 6, to get a cleaner comparison over time. Trade-off: a more disaggregated classification might put us closer to concentration in product markets.
3/n

Main question: do we see an increase in industrial concentration over 1997-2017? Yes, we do when looking at the average share of top 4 and 20 firms—previous papers documented such an increase until 2012 using this data. The trend continued.
4/n

What sectors are responsible for such an increase? There does not seem to be caused by manufactures: the share of top firms did increase sharply at the turn of the century but then stalled. If we check the Herfindahl-Hirschman Index, concentration even decline.
5/n

The number of firms did decrease over this period, though, and those surviving were/became larger, or/and entrants relative to exiters were bigger.
6/n

The wholesale trade sector paints a different picture. The share of top firms did stall in the 2000s, but it continued increasing after the Great Recession.
7/n

The number of wholesale trade establishments continued decreasing, and their average size increased in 2017.
8/n

The retail trade sector gives a clear-cut message: the concentration of sales in large firms increased steadily over the last 20 years.
9/n

The number of retail establishments suffered the recession but then recovered strongly. Given the increase in concentration, this might be capturing an increase in the number of establishments of large firms.
10/n

Service sectors became more concentrated up until 2012 on average. However, the 2017 Census data seems to indicate that the relative size of top firms stabilized.
11/n

12.Interestingly, the number of establishments increased and became larger over the entire period. Very different from what we saw in other sectors.
12/n

In conclusion:
i. Concentration kept on increasing in 2017, but at a lower rate.
ii. Different sector trends can explain this decline in the rate of concentrating industries.
iii. Retail and wholesale trade sectors became more concentrated in 2017.
13/n

iv. The share of top firms in the manufacturing and service sectors has been stable during the last decade.
v. Firms/establishments became larger in all industries.
n/n

Technical notes: confidence intervals are not included for clarity, but all statements on the thread are sustained if I include them. All averages are weighted by 5d sector sales/revenue. Results are similar when I take simple averages.

@ThomasPHI2
@jan_eeckhout
@joseazar
@gimpullitti
@JohnHCochrane
@JHaltiwanger_UM
@tylercowen
@HansbergRossi
@mattyglesias
@Noahpinion
@Mathieu_Parenti

@TheEconomist @FT @fteconomics @sharatganapati @pol_antras @chrisedmond @_gabrielunger @TimTkehoe @DBaqaee @pdacosta @JustinWolfers @paulkrugman @BaldwinRE