The study, published in the Atlantic Economic Journal, examines the reduced-form relationship between the statutory top marginal federal income tax rate and subsequent growth in real gross domestic product per capita using annual US data from 1929 through 2024. Output figures come from the Bureau of Economic Analysis National Income and Product Accounts, while the tax series comes from the Tax Policy Center’s historical record of the highest marginal income tax rates. The authors, Mike Kimel and Brandon Parsons, deliberately kept the empirical approach simple, estimating ordinary least squares regressions with Newey-West standard errors that correct for heteroskedasticity and serial correlation.

The headline result is a concave relationship. In a linear specification, the lagged log top marginal tax rate carries a positive and statistically significant coefficient, associated with higher subsequent growth even after conditioning on the lagged growth rate. In a nonlinear specification that adds the squared, mean-centered tax rate, the linear term is positive and significant while the squared term is negative and significant. The fitted quadratic reaches its peak near a 64 percent top rate, a level the authors stress is a feature of the fitted curve rather than a structural optimum or policy threshold, since its location is sensitive to specification and sampling variation. Because the sample mean tax rate sits near 59 percent, most historical observations fall on the upward-sloping portion of the relationship.

The timing structure matters for how the result should be read. Tax rates enter the regressions with a one-year lag: the rate in year t-1 is paired with growth from year t to t+1, so the tax environment is fixed before the outcome window begins. The authors note that this lagged design reflects the temporal ordering of their proposed mechanism and helps mitigate simultaneity concerns, but they are explicit that it does not resolve policy endogeneity or omitted-variable bias. Tax rates change in response to wars, recessions, inflation, budget pressures and shifting political coalitions, and the paper does not claim to identify exogenous tax shocks or estimate a structural causal parameter. The specifications also intentionally omit standard growth controls such as demographics, human capital and structural investment measures, a choice the authors describe as a transparent benchmark that risks over-conditioning if loaded with endogenous regressors.

A key robustness question is whether the pattern is merely a relic of the mid-century tax environment, when top rates were far higher than today. The authors test this in two ways. First, they add a post-1982 indicator and its interactions with both tax terms to the quadratic specification; the interactions are small, individually insignificant, and jointly insignificant, with an F-test failing to reject that all three post-1982 terms equal zero at a p-value of 0.42. Second, they estimate the linear specification separately for 1930 to 1982 and 1983 to 2024. The earlier subsample coefficient is close to the full-sample estimate but imprecise, while the post-1982 coefficient is positive and significant at the 1 percent level, even though top rates ranged only from 28 to 50 percent in that era, entirely below the estimated turning point. The authors caution that the larger post-1982 point estimate should not be read as evidence of a stronger relationship, since the log specification’s coefficient depends partly on the range of tax variation and the later subsample contains relatively few observations. The appropriate conclusion, they argue, is continuity: no evidence of a structural break, and a positive association that persists into the modern buyback era.

To explain the pattern, the paper develops what the authors call the Aston Martin mechanism, a deliberately narrow framework built around the cost of income extraction for owner-managers of closely held firms. For such individuals, the marginal tax rate is not an abstraction but the price of taking money out of the firm. At a 10 percent tax rate, financing $500,000 of personal consumption, roughly the price of a late-model Aston Martin Vanquish, requires about $556,000 of extractable firm income. At a 90 percent rate, the same consumption requires $5 million. Higher marginal rates therefore raise the pre-tax cost of withdrawal, shifting the margin toward retention inside the firm, where retained funds are not taxed at the point of retention by the personal income tax.

The framework then links retention to investment through a financing constraint. If outside finance were frictionless and costless, retention would matter little, since firms with profitable projects could simply raise external capital. But when financing constraints bind, actual investment equals retained funds, so higher retention translates into higher investment and, eventually, higher output. This assumption draws on the investment-cash-flow sensitivity literature beginning with Fazzari, Hubbard and Petersen in 1988, and on more recent dynamic panel evidence that higher retention ratios improve firm performance and valuation. The mechanism also predicts its own limits: capital faces diminishing returns, the behavioral response to taxation may weaken at very high rates, income effects can push owners to extract more pre-tax income to preserve consumption, and financing constraints stop binding once retained funds exceed the first-best investment level. Together these forces generate the concave shape seen in the data.

The authors are careful about what the evidence does and does not show. The aggregate time series cannot distinguish owner-managed firms from other firms, and no firm-level or size-segmented variables exist over the full 1929 to 2024 period, so the analysis asks only whether the historical record is consistent with the mechanism’s implications, not whether the mechanism operates within any particular class of firms. The paper does not test the mechanism directly; the comparative static that higher rates reduce extraction holds only when the substitution effect dominates any offsetting income effect. The authors also emphasize that the estimated relationship should be read as a historical average across different institutional regimes rather than a single stable structural parameter. The mechanism’s strength depends on three conditions: binding financing constraints, a large wedge between the cost of extraction and retention, and limited alternative extraction channels such as capital gains treatment, shelters or income shifting. Where those conditions fail, the standard supply-side prediction is more likely to prevail.

The institutional history frames that dependence. The Tax Reform Act of 1986 sharply reduced the top statutory rate and changed the relative treatment of corporate and personal income, likely compressing the extraction wedge. The Securities and Exchange Commission’s adoption of Rule 10b-18 in November 1982 created a safe harbor for open-market share repurchases, launching the modern buyback era in which large corporations increasingly returned earnings to shareholders or held them as cash rather than reinvesting them, behavior that could work against the retention-to-investment channel. Yet the estimated relationship shows no statistically significant break across that divide. The paper situates itself against prior work in both directions: narrative-identification studies such as Romer and Romer and Mertens and Olea found that tax cuts raise output in the short to medium run, while Piketty, Saez and Stantcheva found that top rate cuts since the 1970s across 18 OECD countries raised top income shares but not economic growth.

The policy implication the authors draw is deliberately restrained. The results caution against the assumption that cutting top marginal rates reliably raises growth, and they suggest that in settings where financing constraints bind and extraction is costly, as in many emerging economies today where firms depend heavily on internal funds and closely held firms account for a larger share of activity, top rates also shape whether income stays inside firms to finance investment. The authors do not offer a simple policy rule and note that the next step is direct testing in firm-level administrative microdata, where responses in retention and investment to tax changes should concentrate among closely held firms facing binding constraints and be absent among firms with ready access to external capital. For now, the contribution is the documented pattern itself: over nearly a century of American history, the relationship between top tax rates and subsequent growth was not uniformly negative, and simple monotonic claims about taxation and growth, the authors argue, miss important features of the data.

Subject of Research: The reduced-form relationship between US top marginal income tax rates and subsequent economic growth from 1929 to 2024

Article Title: An Unexpected Relationship Between Taxes and Economic Growth

Article References: Kimel, M., & Parsons, B. (2026). An Unexpected Relationship Between Taxes and Economic Growth. Atlantic Economic Journal. https://doi.org/10.1007/s11293-026-09865-3

Image Credits: AI Generated

DOI: 10.1007/s11293-026-09865-3

Keywords: tax policy, economic growth, marginal tax rates, GDP per capita, owner-managed firms, income retention, investment financing constraints, supply-side economics, share buybacks, reduced-form regression, Tax Reform Act of 1986, closely held firms