Why do monopoly concentrations distort macro statistics?

The short answer

The standard macro framework treats the corporate sector as a representative firm. Aggregate productivity, aggregate labor share, aggregate markups — each is supposed to summarize what’s happening across the firm distribution. This works well when firms are roughly similar. It fails badly when a small number of dominant firms behave very differently from the median.

Since 1980, the US economy has shifted in exactly that direction. The De Loecker-Eeckhout-Unger 2020 paper, using Compustat firm-level data, shows that the rise in aggregate markups from 21% to 61% above marginal cost was driven entirely by firms in the top decile of the markup distribution. The median firm’s markup did not change.

The angle that distinguishes this finding: the aggregate is true but misleading. “Average” markups have risen, but the typical firm faces the same competitive pressure as in 1980. What has changed is that a small number of firms have escaped competition and capture vastly more value. This compositional shift biases nearly every aggregate macro statistic produced from firm-level data.

→ New to industrial organization? Financial education hub

What the data shows

The most cited firm-level evidence comes from De Loecker-Eeckhout-Unger (2020), with corroboration from Autor-Dorn-Hanson-Katz-Van Reenen on superstar firms.

The empirical context (DEU 2020, ADHKV 2020, Census, BLS, 1980-2024):

• US aggregate markup: 1.21 in 1980 → 1.61 in 2016 (DEU 2020)
• P90 markup: from ~1.5 in 1980 to ~2.5 in 2016 (top decile)
• Median (P50) markup: essentially unchanged 1980-2016
• US corporate profits as % GDP: from ~1% above-normal in 1980 to ~8% in 2016
• Top-4 firm sales share rising in ~75% of US industries 1997-2012 (Autor et al)
• Business dynamism (firm entry rate): from ~14% in 1980 to ~8% in 2018 (Census BDS)

The exception worth noting: the Basu (2019) and Traina (2018) critiques argue that DEU’s markup estimate depends sensitively on whether selling, general and administrative (SG&A) expenses are treated as fixed cost or marginal cost. Under alternative assumptions, the rise in markups is smaller. The empirical magnitude is contested, even if the qualitative trend is widely accepted.

→ Dataset: US Corporate Debt to GDP

Why it happens — the macro mechanism

Concentration distorts macro statistics through three distinct channels.

Channel 1 — Compositional bias in aggregates. When you compute an average across firms weighted by sales (which is how most macro aggregates work), the answer is heavily influenced by a small number of large firms. If those firms have unusually high markups, low labor share, and high productivity, they pull the aggregate in their direction even when the median firm is unchanged. This is why DEU’s aggregate markup rose 33% while the median did not move. See our FAQ on labor share decline.

Channel 2 — Reallocation versus within-firm change. The labor share decline can in principle come from two sources: each firm reducing its labor share (within-firm), or market share shifting from high-labor-share firms to low-labor-share firms (reallocation). Autor-Dorn-Hanson-Katz-Van Reenen 2020 show that reallocation toward superstar firms explains most of the US labor share decline, not within-firm changes. Same point applies to TFP measurement. The angle that distinguishes this view: the standard “average labor share fell” narrative obscures that most firms are unchanged. See our FAQ on TFP slowdown and our FAQ on intangibles.

A third channel concerns measurement of competition itself.

Channel 3 — Concentration measures fail to capture market power. Standard concentration measures (CR4, HHI) are computed at the industry level. But “industries” are statistical categories that may bundle very different markets. Amazon competes with Walmart in some segments, with Apple in others, with no one in cloud computing. Aggregate industry concentration may be falling even as effective market power rises in specific submarkets. Werden-Froeb document that industry-level HHI in the US has actually been roughly stable since 2000, despite clear evidence of rising firm-level dominance.

Synthesis by regime. In the postwar era 1948-1980, US firm distribution was relatively flat — the top decile and the median moved together. Aggregate measures of markups, labor share, productivity and concentration tracked the underlying firm-level reality faithfully. From 1980 to 2000, technology, globalization and antitrust loosening allowed a small number of firms to escape competition; the gap between the top and the median began to widen. Since 2000, the divergence has accelerated dramatically — superstar firms in tech, retail and finance now have markups, profits and productivity vastly above the median, distorting every aggregate measure. The pivot is that macro statistics built for a world of representative firms no longer faithfully represent the modern US economy.

The aggregate told us markups had risen 33%. The median told us nothing had changed. Both were true. Neither was the whole story.

→ Analytical frame: Macro-financial regimes

What it means for different economic actors

Equity investors. The S&P 500 is itself a sales-weighted index, which means it is dominated by superstar firms. Index investors are getting concentrated exposure to firms with very high markups — this is partly why the S&P 500 has outperformed equal-weight indices over the past decade. But the same concentration creates regime risk: a regulatory or technological challenge to superstar firms would have outsized effects on the index.

Policymakers. Antitrust policy is rebuilt around firm-level rather than industry-level analysis. The Khan-Wu-style approach at the FTC since 2021 explicitly targets firm-level dominance even when industry HHI is moderate. The macro statistics that justified loose antitrust 1990-2015 were partly artifacts of measurement error.

Researchers and analysts. Anyone using aggregate corporate data must increasingly distinguish between the median firm experience and the aggregate experience driven by superstars. The two diverge meaningfully on most variables of interest since 2000.

A common error is to treat the median firm and the aggregate firm as interchangeable. They were broadly similar in the postwar era; they have diverged dramatically since 2000. Reading aggregate data as if it represents the typical firm produces systematically misleading conclusions.

Practical observation

Go deeper

Related questions

• Why is labor’s share of GDP declining globally?
• How does the rise of intangible capital change valuations?
• What is total factor productivity and why has it slowed?
• Why do a few stocks dominate index returns?

Frequently asked questions

Why has the median firm’s markup not risen?

Because the markup rise is a story about firm reallocation, not within-firm dynamics. According to De Loecker-Eeckhout-Unger (2020), the typical US firm in 2016 faces roughly the same competitive pressure and earns roughly the same operating margin as in 1980. What has changed is that some firms — those that successfully captured network effects, data advantages, scale economies, or regulatory capture — have escaped competition entirely. They earn very high markups, take growing market share, and pull the aggregate up. The economy is becoming more bifurcated, not uniformly more concentrated.

How much of the apparent decline in business dynamism reflects measurement error?

Some, but not all. The US firm entry rate fell from ~14% in 1980 to ~8% in 2018 according to Census BDS data. Part of this reflects compositional shift — older industries (with low entry) growing relative to younger ones. Part reflects superstar incumbency that deters entry. Part reflects regulatory burden growth. Decker-Haltiwanger-Jarmin-Miranda (2017) attempt to disentangle these — the conclusion is that even after correcting for sectoral mix, business dynamism has fallen meaningfully since 2000.

Are these distortions specific to the US?

The US shows the most extreme concentration trends, but they exist elsewhere. Bajgar-Berlingieri-Calligaris-Criscuolo-Timmis (2019) document that aggregate concentration has risen in 16 European countries since 2000, though more modestly than in the US. Philippon’s “The Great Reversal” (2019) argues that Europe is now more competitive than the US in many sectors — a complete reversal of the 1990s pattern. The empirical work is most developed for the US but the cross-country evidence is consistent with a global, US-led concentration trend.

Last updated — 12 July 2026