Market power over workers (rather than consumers) appears central to the behaviour of large firms - with direct policy implications, since effective interventions may differ greatly depending on which is side of market power is being targeted.
Could some other bias drive these empirical correlations? Possibly - in the end, markups aren’t observable. But it's worth noting that the combination of exceptional data and state-of-the-art methods points to a picture that's quite different from what's typically assumed.
An alternative explanation is based on labour-replacing technology, as larger firms may shift from labour to intermediates. Yet, taking this into account following Raval (2023) doesn't change our key result: cross-sectionally, markups (markdowns) fall (grow) in firm size.
Labor market power could rationalise this inconsistency: under monopsony, the labour-based markup captures both price markups and wage markdowns. What we show is that this term of joint market power grows in firm size because larger firms have higher markdowns, not higher markups.
The choice of the variable input turned out to be key. Markups fall in firm size when they are estimated using intermediates (or just energy), but not using labour - explaining our divergence from previous findings.
To dig deeper, we got access to a special administrative database on German manufacturing companies, with financial statements matched to firm-level product prices. Immune to price bias, the negative correlation held across a range of specifications.
This started while working on the 8th CompNet data collection (@iwh-halle.bsky.social). Running some safety checks, we noticed something odd: across all available European countries, the cross-sectional correlation between markups and firm size was consistently negative.
Glad to see my work with Matthias Mertens (@mitsloan.bsky.social) forthcoming at Econ Letters!
A short thread for anyone studying market power.
TL;DR: In our data, larger firms exhibit *lower* markups - but much higher markdowns.
www.sciencedirect.com/science/arti...
The best cross-country data-building effort is no doubt CompNet at IW Halle. For this reason, there are several interesting cross-country papers coming from the team, including Matthias Mertens' and @bmottironi.bsky.social's paper on cost markups and wage markdowns. 1/2
That's very kind - thank you!
Thank you!
Always very kind, Maria! Thank you!
A big thank you to everyone who contributed to this work with feedback and comments! Especially my fantastic supervisors: @johnvanreenen.bsky.social, @deriddermaarten.bsky.social, and Alan Manning.
Here's the full draft:
t.co/5Q0IMXFY8a
And here's my website, if you'd like to explore more:
sites.google.com/view/bmottir...
Overall, the key message is that labour market power impacts not only wage setting and rent extraction but, crucially, production efficiency—so much so that it can serve as a power explanation for cross-regional productivity differentials.
Lastly, I look at geographical disparities: I estimate higher levels of labour market power in Southern Italy, which account for 40% of the South’s productivity gap with the North.
Calibrating the model, I estimate that moving from a 25% excess MRPL (competitive benchmark) to a 50% excess (typical labour market) results in a 21% drop in aggregate productivity, which combines with lower production labour for a total 35% fall in output.
The observed empirical patterns align with the theory.
At the province-level, labour market power is associated with:
-More misallocation
-More entrepreneurship
-Smaller firm size
-Reduced use of intangibles
-Lower productivity
Here are some empirical findings:
Italian labour markets are far from the competitive equilibrium. On average, the estimated marginal revenue product of labour (MRPL) exceeds wages by ~50% in typical labour markets.
I examine it in three steps:
-Theory: a tractable GE model, proving each channel analytically.
-Validation: new evidence on monopsony & market performance across Italian provinces, with CompNet data.
-Quantification: the model is calibrated to estimate aggregate losses.
As a result: (i) misallocation towards small firms, (ii) excess entry of less talented entrepreneurs, and (iii) limited technology diffusion.
These three key channels combine to diminish aggregate productivity.
That’s the paper’s idea—simple and (hopefully) intuitive.
When wages are lower, entrepreneurship can be more remunerative than labour, even for less talented individuals:
-> Labour market power prompts the entry of too many entrepreneurs, who end up managing underperforming firms that lack the necessary scale for technology adoption.
(2) Average firm size is directly affected by the <labour vs entrepreneurship> occupational choice: more entrepreneurs means more firms, fewer workers.
So, how does *labour market power* factor in?
By lowering wages!
Let’s start with two apparently unrelated observations:
(1) Firm size is crucial in modern economies. Intangible technologies offer major productivity gains but come with high sunk costs—often too high for small businesses.
Alright, I've got zero followers right now—let's see if this thread on my JMP can escape the depths of the internet.
TL;DR: Labour market power leads to large productivity losses.
Keep reading if you’re interested in misallocation, tech diffusion, and regional disparities (you should be!).
