NBER PAPER ANALYSIS: Diegmann, Müller, Schoefer (W35265)

TEXT START

The employer size wage effect—the empirical regularity that workers at larger establishments earn more—has long been interpreted through the lens of monopsony power. This paper kills that narrative dead.

THE DISSECTION

This is a methodological correction with structural implications, but the authors pull their punch in the final paragraph where it matters most. They document something economically significant: within large multi-establishment firms (MEFs), wage differences across establishment sizes vanish entirely once you hold the firm constant. The ESWE—the "canonical fact"—is revealed as a composition artifact. Large firms pay more as firms, not because large establishments exercise monopsony power over local labor supply.

What they're actually finding is efficiency wage behavior at scale:
- Large firms pay above-market-clearing wages
- They then ration employment—leaving excess labor supply that they don't hire
- This is the opposite of monopsony: they're not extracting from scarcity, they're subsidizing above-market wages and being choosy

The meta-analysis is a useful service to the literature. The canonical finding was never robust to the right specification.

THE CORE FALLACY

The paper correctly identifies the monopsony fallacy but then defaults to efficiency wage theory as the alternative—and efficiency wage theory is itself a stabilization narrative dressed as economics.

Efficiency wage theory says firms pay above-market wages to increase productivity, reduce shirking, and attract higher quality workers. Fine. But the paper doesn't ask the obvious follow-up: why can large firms sustain above-market wage premia while smaller firms cannot?

The answer is structural market power—but of a different kind than monopsony. It's oligopolistic rents, scale economies, and barriers to competitive displacement that allow large MEFs to extract surplus and redistribute a portion as wage premia. The paper gestures at this ("accommodating above-market-clearing wage premia") but doesn't name it clearly.

HIDDEN ASSUMPTIONS

1. Stable firm boundaries: The analysis assumes firm identity is meaningful and fixed over the observation period. Under rapid AI-driven restructuring, this is increasingly fiction. Firms don't stay large and wage-premium-paying for long if they can automate the premium-generating workers.

2. Labor as the binding constraint: The efficiency wage story implicitly assumes labor supply elasticity matters. Under DT, the relevant constraint shifts to capital substitution possibility. Large firms don't pay premia because they need to—they pay premia because they've extracted rents that haven't yet been automated away.

3. German labor market as representative: Germany has strong codetermination, concentrated industry, and specific wage-setting institutions. The result may be partially specific to a labor market structure that's already under stress from automation pressure, not a general finding.

4. Zero ESWE within firms = no monopsony anywhere: The paper overstates the negative. The within-firm result doesn't prove small single-establishment firms aren't monopsonists. It just proves large MEFs don't use establishment-level monopsony.

SOCIAL FUNCTION

This paper is partial truth with prestige packaging. It makes a real empirical contribution—the firm-fixed-effects identification is clean and the data is excellent (German population-wide linked data). But its theoretical implications are deliberately defanged.

It should have said:

The persistence of above-market wage premia within large firms, combined with rationed employment, is consistent with rent-sharing from oligopolistic market positions. As AI enables competitive entry and reduces scale economies, these premia are not stable foundations—they are artifacts of a specific competitive regime currently under structural dissolution.

That paper doesn't exist. This one stays safely inside efficiency wage theory.

THE VERDICT

The ESWE literature was built on a misattribution. Large firms pay more not because of monopsony but because of scale rents. This is actually more alarming under the Discontinuity Thesis, not less:

• Large firms paying efficiency wages while rationing employment are precisely the structures most vulnerable to AI-driven displacement. They're already not hiring everyone who wants to work at market wages. When AI makes the remaining workers more productive, the rationing gets worse, not better.
• The 25% of German employment "where the ESWE disappears" represents the most fragile premium—sustained by competitive structures (scale, barriers, market power) that AI directly erodes.
• As AI reduces the value of the cognitive and organizational work those premium workers perform, the rent-sharing that funds the premium evaporates. You don't get monopsony wages falling—you get the premium collapsing entirely as the rent base dissolves.

The monopsony story was wrong. But the efficiency wage story is a transitional narrative. It describes a wage-setting equilibrium that exists between the competitive capitalism of the postwar era and whatever comes after. It is not a stable endpoint.

VIABILITY SCORECARD (DT LENS)

SURVIVAL PLAN

For the workers currently inside the 25% receiving firm-level wage premia:

• You are not in a stable premium. You are in a temporarily protected position within a firm that is itself in a temporarily protected position within a market.
• The protection is not your productivity. It's your firm's market power. When that market power is competed away by AI-native entrants, the premium goes first.
• Sovereign path: Acquire equity stakes in AI-native competitive entrants while the premium income is still flowing. Convert wage rent into capital rent before the premium collapses.
• Servitor path: Become indispensable to the specific human judgment functions that remain un-automated within large firms—but understand these niches shrink faster than institutional loyalty suggests.
• Verification path: The paper's data methodology is excellent. Verification arbitrage in labor market economics—applying firm-fixed-effects logic to other contexts—is a real analytical skill.

The paper is worth reading for the identification strategy and the meta-analysis. The conclusions are institutionalist comfort food.