Economics after Neoliberalism: Introducing the EfIP Project

Why it matters, in detail. A multi-cause phenomenon (the post-1980 rise in US wage and top-income inequality) is attributed exclusively to a policy bundle. Decompositions assign a large share of wage-inequality growth to the race between education and technology: in AGK (2020), rising educational wage differentials account for 75% of the 1980-2000 rise and 38% of the 2000-2017 rise. Automation is another large share (Acemoglu-Restrepo 2022: 50-70%, published after the P&P). Institutional channels (DFL 1996; Farber et al. 2021) and top marginal-rate cuts (for top shares) explain other parts. These shares overlap, are not additive and are partly disputed. The direction of the bias is to overstate the policy share; its size cannot be quantified because the essay states no number. Section III's point that the direction of technology is itself policy-shaped (Acemoglu, Korinek briefs) partly reconciles the two views, but the essay does not draw that link.

Evidence. Text: p. 366 ('agreed that the prevailing neoliberal policy framework had failed society, resulting in monumental and growing inequality'; 'we fully embraced these aims'); p. 367 (culprits sentence, unhedged appositive); reference list p. 371 (two items). Versions (scripts/version_phrases.py): the culprits sentence is in V1 (EfIP Policy Brief 1, Jan/Feb 2019, p. 2), V2 (Boston Review, 15 Feb 2019) and the P&P. V1 p. 1 also said 'Organizational and technological changes and globalization have fueled great wealth accumulation among those able to take advantage of them, but have left large segments of the population behind'; this was dropped from V2 and the P&P. Literature (abstracts verified via OpenAlex): Autor, Goldin & Katz (2020), AEA P&P 110: 347-351 (same volume), 'Increased educational wage differentials explain 75 percent of the rise of U.S. wage inequality from 1980 to 2000 as compared to 38 percent for 2000 to 2017'. Deming, EfIP Policy Brief 14 (July 2019): 'An important cause of current levels of economic inequality is growing demand for college-level skills that began in the 1980s, combined with slow growth in the number of young people receiving degrees.' Korinek, EfIP Policy Brief 6 (Feb 2019): technological change 'has been to reduce the market wages of regular (unskilled) workers'. Supporting channels: DiNardo, Fortin & Lemieux (1996, Econometrica 64(5)), deunionization and the falling real minimum wage were important for 1979-88 wage inequality; Farber, Herbst, Kuziemko & Naidu (2021, QJE 136(3)), unions reduce inequality; Autor, Dorn & Hanson (2013, AER 103(6)). Later work: Acemoglu & Restrepo (2022, Econometrica 90(5)) attribute 50-70% of 1980-2016 US wage-structure change to automation-driven task displacement, robust to controls for deunionization and market power (this post-dates the P&P). Tax strand: IRS SOI Historical Table 23, re-read: top bracket rate 70% (1980), 28% (1988), 39.6% (1993), 37% (2018). Splinter (2020, NTJ 73(4)) finds the CBO-based Kakwani index of federal progressivity rose 46% over 1979-2016 (corpus us-income-14, us-wealth-taxes-16).

Why this severity. The diagnosis is the essay's motivating premise and the place where it leans on the inequality literature, but the programmatic thesis (C1, C2) does not depend on which drivers dominate. The listed policy channels have real empirical support, so the claim is overstated and incomplete rather than false.

The authors' response. No direct response to the omission. The P&P's Section III notes that the Acemoglu, Kasy and Korinek briefs treat technological change as endogenous to policy (p. 370). On progressivity, Saez & Zucman maintain that top-end effective rates fell (corpus us-wealth-taxes-16, citing NBER w31912).

Why it matters, in detail. The well-identified US evidence (bunching and event-study designs over 138 state increases, 1979-2016) estimates an employment elasticity with respect to the minimum wage of 0.024 (s.e. 0.025). That rules out large losses for minimum-to-median ratios up to roughly one half (Dube brief) or somewhat higher. It does not identify effects beyond that range, which a $15 federal floor would exceed in many local labour markets. CBO's July 2019 median for $15 was -1.3m jobs (-0.8%), with a right-skewed two-thirds interval from about 0 to -3.7m, and virtually zero for $10. The unqualified phrasing therefore overstates the external validity of a range-specific null. The cost to the essay's argument is small: the null within range already contradicts the competitive-model prediction that any binding minimum reduces employment.

Evidence. P&P p. 368: 'Recent empirical findings, for example, have found that international trade produces large adverse effects on some local communities, that minimum wages do not reduce employment, and that financial liberalization produces crises rather than faster economic growth.' Versions (scripts/version_phrases.py): the clause appears in V1 p. 3 and V2. V1 p. 5 adds 'while the minimum wage debate continues, there is a consensus that it is not an effective tool for intervening in labor markets with wages higher than say, the 30th percentile'; V2 has 'few would claim ... higher than a certain level'. Both are absent from the P&P. The authors' Boston Review 'Final Response' (26 Mar 2019) described Card, Katz and Krueger's findings as showing 'that minimum wage increases may not curtail employment'. Dube, EfIP Policy Brief 4 (Feb 2019): 'little evidence of either overall job loss or losses for lower skilled groups for minimum wages that are up to half the median wage for full time workers'; bolder policies 'will exceed the range of minimum-to-median wage ratios we have seen to date. Therefore, empirical evidence will need to assess whether these policies go too far.' Cengiz, Dube, Lindner & Zipperer (2019, QJE 134(3), abstract) find low-wage jobs 'remained essentially unchanged', 'no evidence of disemployment when we consider higher levels of minimum wages', and 'some evidence of reduced employment in tradeable sectors'. Their NBER WP 25434 Table 1 col. 1 (re-read) gives affected employment +2.8% (s.e. 2.9%) and an employment elasticity with respect to the minimum wage of 0.024 (s.e. 0.025). CBO (July 2019), 'The Effects on Employment and Family Income of Increasing the Federal Minimum Wage' (re-read): for $15 by 2025, a median of 1.3 million fewer employed (0.8%), with a two-thirds range from about zero to 3.7 million; for $12, 0.3 million (range 0-0.8 million); for $10, 'virtually no effect'. Jardim et al., NBER WP 23532 (June 2017, rev. May 2018): Seattle's increase to $13 'reduced hours worked in low-wage jobs by 6-7 percent'. Neumark & Shirley (2022, Industrial Relations 61(4); after the P&P) document summaries ranging from 'higher minimum wages do not reduce employment' to 'most evidence points to adverse employment effects', and conclude there is 'a clear preponderance of negative estimates'.

Why this severity. This is an important undisclosed scope condition on the essay's most policy-relevant empirical claim, and it contradicts a bound stated in the essay's own recommended source. A bounded version (moderate increases within observed ranges show no detectable job loss) would support C3's point equally well, so the thesis survives.

The authors' response. None identified. The authors' own Boston Review Final Response (26 Mar 2019) used the hedged formulation 'may not curtail employment'.

Why it matters, in detail. Estimated growth effects depend on the liberalization measure (equity-market, capital-account or domestic financial), the horizon (temporary level effects versus permanent growth effects) and the identification strategy. BHL find about +1 pp annual growth over five years after equity-market liberalization. RTW find a positive net effect after netting out crisis costs. Macro cross-country correlations for financial openness are weak (Rodrik-Subramanian 2009). 'Crises rather than faster growth' therefore overstates one side. The defensible statement is that liberalization raises crisis risk and that its growth dividend is contested.

Evidence. P&P p. 368 (no citation). Kaminsky & Reinhart (1999, AER 89(3): 473-500): 'financial liberalization often precedes banking crises'. Bekaert, Harvey & Lundblad (2005, JFE 77(1): 3-55; NBER WP 8245): 'equity market liberalizations, on average, lead to a one percent increase in annual real economic growth over a five-year period'. Ranciere, Tornell & Westermann (2006, NBER WP 12806): 'the direct effect of financial liberalization on growth by far outweighs the indirect effect via a higher propensity to crisis'. Ranciere, Tornell & Westermann (2008, QJE 123(1)): in financially liberalized economies systemic risk-taking 'leads to higher mean growth but also to greater incidence of crises'. Henry (2007, JEL 45(4)): studies that properly test the theory 'show that liberalization has significant effects on the cost of capital, investment, and economic growth'. Skeptical side: Rodrik & Subramanian (2009, IMF Staff Papers 56(1): 112-138) start from 'the stylized fact that there is no correlation between long-run economic growth and financial globalization' and argue that foreign finance can have 'adverse long-run growth consequences'. Kalemli-Ozcan, EfIP Policy Brief 18 (Oct 2019): the model prediction of 'an increase in productivity growth' after liberalization 'does not match the experience of countries in South Europe'.

Why this severity. This is an illustrative example. Its crisis half is robust, and its no-growth-dividend half matches one side of a genuinely split literature rather than extrapolating beyond the evidence. Correcting the wording leaves C3 and the thesis intact.

The authors' response. Rodrik & Subramanian (2009) had already reviewed the pro-liberalization evidence (Henry 2007; Kose, Prasad, Rogoff & Wei 2006) and judged it 'unconvincing'. That defends a no-correlation reading, not 'crises rather than growth'.

Why it matters, in detail. Spending as a share of GDP measures effort, not generosity, and is pushed up by ageing and health-cost growth, so it overstates benefit levels. Replacement-rate series show some retrenchment in specific programmes. Still, in PSZ's accounting the post-1980 rise in inequality is in pre-tax income, and taxes and transfers raised bottom-50% growth from +1.4% to +21.0% over 1980-2014 (PSZ attribute all of the post-tax bottom-50% gain to the elderly). The aggregate data do not show a dismantled welfare state.

Evidence. P&P p. 367. Public social expenditure, % of GDP, 1980 to 2019, from OECD SOCX via Our World in Data (scripts/social_spending_check.py; identical to Auditor A's SDMX extract): USA 12.9 to 18.2; GBR 16.9 to 22.3; FRA 20.3 to 30.8; DEU 21.8 to 26.4; ITA 17.3 to 27.6; JPN 9.8 to 22.8; SWE 24.5 to 26.3; CAN 13.4 to 19.8; AUS 10.2 to 20.4; NZL 16.3 to 24.2; DNK 20.4 to 28.7; ESP 15.0 to 24.3; BEL 23.5 to 28.1; AUT 22.0 to 27.8; FIN 17.8 to 29.6; NOR 16.1 to 25.1. NLD fell from 24.4 to 17.5, and Sweden is below its 1993 peak of 34.4. PSZ, NBER WP 22945, p. 3: 'The aggregate flow of individualized government transfers has increased, but these transfers are largely targeted to the elderly and the middle-class'. The age section adds that government redistribution has not enhanced working-age income growth 'despite the rise in means-tested benefits' (Medicaid and the EITC). PSZ Table II (re-read): bottom-50% growth 1980-2014 was +1.4% pre-tax and +21.0% post-tax. Cass, 'Economics After Partisanship', Boston Review forum, 16 Aug 2019: 'it is not as if a bunch of market fundamentalists have actually cut back government provisions ... the United States has undertaken massive expansions of its safety net, entitlement programs, and public education systems'.

Why this severity. This is one item in a framing list. The thesis is unaffected, and a narrower claim about benefit generosity for working-age households would be defensible.

The authors' response. None on this point. The P&P's new Section II paragraph on 'new flavors of conservatism' (p. 370) answers Cass's argument about values, not his welfare-state point, and the 'dismantling' wording was kept.

Why it matters, in detail. Real growth for the bottom half or middle ranges from about +1% (PSZ pre-tax per adult, 1980-2014), through about +21% (CBO market income, Q2-Q3, 1979-2019), to +52-60% (CBO after transfers and taxes). The range depends on the unit (equal-split adults versus size-adjusted households), the deflator, the period, and the treatment of in-kind health benefits (CBO values them at cost, which is contested). The gap is driven mainly by transfers, especially health and old-age benefits, and by household-size adjustment.

Evidence. PSZ Table II (PSZ2018MainData.xlsx, sheet T2, re-read), growth per adult 1980-2014: bottom 50% +1.4% pre-tax and +21.0% post-tax; P0-20 -25.4% and +4.4%; middle 40% +41.8% and +49.4%. PSZ text (NBER WP 22945, p. 3): 'Even after taxes and transfers, there has been close to zero growth for working-age adults in the bottom 50% of the distribution since 1980.' CBO (Nov 2022), The Distribution of Household Income, 2019, supplemental Table 3 (re-read; households ranked by size-adjusted income, 2019 dollars), growth 1979-2019: market income Q2 +21.5% and Q3 +21.4% (about 0.5% a year); income after transfers and taxes Q1 +93.5%, Q2 +60.1%, Q3 +52.3%. Burkhauser, Larrimore & Simon (2012, NTJ 65(1); corpus us-income-03), 1979-2007: median tax-unit pre-tax, pre-transfer income +3.2%, against +36.7% for size-adjusted post-tax, post-transfer household income including health insurance.

Why this severity. A passing descriptive premise. On the concepts closest to 'working class' income (market income, or working-age post-tax income in PSZ) the claim has support, so only the unqualified framing is at issue.

The authors' response. PSZ (2018) moved to equal-split adults, published post-tax series, and report near-zero post-tax growth for working-age bottom-50% adults. Saez & Zucman (2020) reject equivalence-scale adjustments as a 'hybrid notion' (corpus us-income-03).

Why it matters, in detail. Summary and labelling inaccuracy. 'Streamlining' misstates the emphasis of a brief whose central proposals expand enforcement (lower HSR thresholds, burden-shifting). The Kuziemko summary drops the incidence assumption, near-full pass-through of employer premium savings to wages, on which the sign of the middle-class cash-income effect depends.

Evidence. P&P p. 371: 'Jonathan Baker and Fiona Scott-Morton present ideas on streamlining antitrust.' EfIP Policy Brief 11 (May 2019): 'Our program for confronting rising market power has four themes: strengthening antitrust enforcement under current rules, reforming the rules to enhance deterrence of anticompetitive conduct, adopting institutional changes to make antitrust enforcement more effective, and employing regulatory tools to foster competition'; 'Strengthening antitrust enforcement is the most obvious way to enhance competition'. Its proposals include lowering the pre-merger notification threshold and shifting the burden of proof to defendants, and the 'EZ-merge' form is a simplified filing for small acquisitions. P&P p. 370: 'Kuziemko argues that extending Medicare to the whole population would raise middle-class incomes while substantially shrinking the health insurance sector and shrinking bloated payments to providers.' EfIP Policy Brief 15 (Sept 2019): 'Assuming that most of the savings would get passed back to workers via higher wages (and this point deserves more discussion in my view) ... middle-wage workers' take-home pay should proportionately rise'; 'an important and open question is how Medicare for All would guarantee that the resulting savings in national health expenditure get passed on to workers'.

Why this severity. These are wording problems in one-line summaries of documents available to readers. Nothing in the essay's argument turns on them.