Rethinking income inequality, with Chris Giles

Episode Summary

Executive Summary: The episode examines a major dispute over U.S. income inequality: Thomas Piketty, Emmanuel Saez, and Gabriel Zucman argue inequality rose sharply, while Treasury/JCT economists Gerald Auten and David Splinter say much of that rise disappears after methodological adjustments. The discussion breaks down six contested choices and concludes the truth likely lies between the extremes.

Main Topics: The Piketty inequality narrative vs. the Auten-Splinter challenge (Priority: 5/5): The conversation centers on whether U.S. income inequality has soared since 1960 or whether prior estimates overstated the rise due to measurement choices and tax-data adjustments. How tax-return data are turned into inequality measures (Priority: 5/5): Both sides use tax filings, but they differ on how to convert raw returns into a measure of society-wide income distribution and what should count as income. Three largely accepted methodological adjustments (Priority: 4/5): Marriage/family filing patterns, the 1986 tax reform, and the treatment of capital gains are discussed as key adjustments; the first two mostly favor the officials’ view, while capital gains is more disputed. Tax evasion, hidden income, and audit evidence (Priority: 5/5): A major unresolved dispute is how to allocate income missing from tax returns: should it be assigned mostly to high earners (academics) or more broadly/bottom-weighted using audit evidence (officials)? Retirement accounts and misreported income (Priority: 3/5): Some apparent income spikes reflect rollovers between retirement accounts rather than true income, which the officials say should be removed or reduced in inequality estimates. Government transfers and public services (Priority: 4/5): The groups disagree on how to value benefits from health spending, defense, and other public services: per-person allocation versus income-proportional allocation changes inequality results substantially. Policy and broader implications for inequality debates (Priority: 4/5): The episode argues that even if measured income inequality rose less than thought, the U.S. still faces serious distributional problems in health, mortality, and social satisfaction; the result changes the policy questions, not the need for them.

Key Arguments: Raw tax-return data do show the top 1% rising from about 10% to 20% of total income between 1960 and 2020, but that does not by itself prove a true rise in social inequality. Marriage patterns matter because richer Americans stayed married much more than poorer Americans, changing tax-unit composition and inflating top-income shares. The 1986 tax reform altered incentives for how income was reported, causing a temporary spike in top-income shares that is not a real inequality shift. Including capital gains is a conceptual judgment: the academics include them as part of economic well-being, while the officials exclude them to align with national income conventions. The biggest unresolved issue is how to allocate unreported income; both sides know some income is missing, but they disagree on whether it belongs mostly to high earners or more evenly across the distribution. Retirement-account rollovers can create artificial income spikes, making some top-income observations less meaningful than they first appear. Assigning government benefits is philosophically difficult; the officials’ per-capita approach tends to lower measured inequality more than the academics’ income-based allocation. The most defensible conclusion is not that inequality did not rise at all, but that it likely rose less than the stark narratives suggested. Even if income inequality rose less, the U.S. still has major inequality problems in health, mortality, and social outcomes, so the broader societal diagnosis remains troubled.

Data Points: Top 1% share of total income in 1960: about 10% - Raw tax-return data discussed as the starting point for both research camps. Top 1% share of total income in 2020: about 20% - Raw tax-return data discussed as the starting point for both research camps. Change in top 1% share, raw data: about +10 percentage points - The broad, unadjusted increase in income share from 1960 to 2020. Marriage rate of very rich people: from 90% to 85% - Used to show richer people remained married more than others, affecting tax-unit measures. Marriage rate of middle/poorer people: from 67% to 37% - Used to show a much larger decline in marriage among non-rich Americans. Estimated effect of marriage adjustment: about 1.5 percentage points (15% of the apparent rise) - How much of the measured inequality increase is attributed to changing marital filing patterns. Post-1986 tax reform effect: large temporary impact in 1984-1987 - The tax reform changed reporting incentives and inflated inequality measures during that period. Contribution of capital gains adjustment: about 1.5 percentage points - The episode says including capital gains raises measured inequality by roughly this amount. Time span analyzed: 1960 to 2020 - The core period over which the paper disputes the trajectory of U.S. inequality.

Pivotal Quotes: "from about 1960 to 2020, it hasn't really risen at all" — Chris Giles: Summarizing the Auten-Splinter conclusion about long-run U.S. income inequality. "what we're trying to do is go from what people fill in on their tax returns to something that means something for society" — Chris Giles: Explaining the central methodological problem in inequality measurement. "I don't think you can legitimately say that inequality hasn't risen, but I think you can say it hasn't risen as much as some of the big and stark doubling or whatever we had in the debate over the last 10 years or so" — Chris Giles: His bottom-line assessment after weighing the six contested methodological choices.

Implications: The episode suggests inequality debates hinge on measurement choices, not just ideology. Policymakers should focus on the specific channels driving harm—health, mortality, and public-service design—rather than assuming one income statistic captures the whole story.