Tests of the permanent income hypothesis based on the consumption Euler equation have produced wildly diverging estimates of excess sensitivity — the response of consumption growth to anticipated income growth — traditionally interpreted as the share of rule-of-thumb consumers who simply spend their current income. A meta-analysis of roughly 3,000 estimates from 144 studies (Havranek and Sokolova, 2020) corrects the mean reported estimate of 0.4 down to about 0.13, attributes the remainder to liquidity constraints, and concludes that pure rule-of-thumb behavior is 'probably' absent — explicitly conditional on the assumption that the entire literature does not share a misspecification that prevents identification of the parameter. The open problem is to settle this question outside the possibly contaminated specification: whether, in primary household data and under identification designs that do not inherit the shared Euler-equation framework, the consumption of households with substantial liquid wealth responds to anticipated income. The causal interpretation of any surviving response — limited liquid wealth, persistent household characteristics, or optimizing-but-non-standard behavior such as near-rationality (small, welfare-cheap departures from full optimization) or present bias (overweighting immediate payoffs relative to all later ones) — remains an open question in the wider literature.

Background

The permanent income hypothesis (PIH) holds that consumers choose consumption based on the expected present value of lifetime resources rather than current income; in its modern form (Hall, 1978), optimizing behavior implies that consumption follows a random walk, so that consumption growth should be unpredictable given information available to the household. The workhorse empirical device is the consumption Euler equation, the first-order condition linking consumption growth to the expected real interest rate and subjective discount rate. A large empirical literature since the 1980s has repeatedly violated this implication: consumption growth appears to respond to anticipated income growth, a phenomenon called excess sensitivity. Campbell and Mankiw (1990) formalized the canonical test, in which a fraction of consumers follow a rule of thumb — they simply consume their current income, without optimizing — and the estimated excess-sensitivity coefficient measures that fraction, while the remaining consumers behave as the PIH predicts. The magnitude of this coefficient matters far beyond consumption economics: quantitative models used for fiscal and monetary policy analysis routinely calibrate the share of rule-of-thumb consumers at 0.2–0.5, and this share is a key determinant of fiscal multipliers. A long-standing alternative explanation is liquidity constraints: households that cannot borrow may display excess-sensitivity-like behavior while still optimizing. The hand-to-mouth literature refined this distinction — Kaplan, Violante, and Weidner (2014) showed that many US households hold substantial illiquid assets (housing, retirement accounts) but almost no liquid wealth ('wealthy hand-to-mouth') and therefore still respond strongly to income; Olafsson and Pagel (2018) confirmed the liquidity gradient with transaction-level data from personal finance software. In this literature the meta-analysis of Havranek and Sokolova (2020) is the most comprehensive synthesis to date. The authors collect roughly 3,000 estimates of excess sensitivity from 144 published studies, encode about 45 variables describing each study's data and method, and use meta-regression and Bayesian model averaging (a technique that averages coefficient estimates across many candidate model specifications in proportion to their posterior support, guarding against arbitrary specification choices) to explain why estimates differ. They find the mean reported estimate, 0.4, falls to about 0.13 once they correct for omitted demographic controls and publication bias — the tendency of journals and authors to preferentially report positive, statistically significant results, which the authors detect as asymmetry in the relationship between estimates and their precision. The corrected residual of about 0.13 can be fully explained by liquidity constraints, implying no evidence of pure rule-of-thumb behavior. Crucially, the authors flag the limits of this inference: controlling for 45 study characteristics cannot rule out a misspecification shared by the entire Euler-equation literature — for example, pervasive measurement error in consumption, non-separability between consumption and leisure (the marginal utility of consumption depending on hours worked, which a consumption-only Euler equation ignores), or particular functional-form, approximation-order, and time-aggregation conventions — that would bias all estimates in the same direction; hence the word 'probably' in their title. The authors also acknowledge that their coding of liquidity constraints across studies is crude. Adjacent evidence sharpens rather than settles the question. Kueng (2018) documented excess sensitivity even among high-income, high-liquid-wealth consumers responding to large, anticipated Alaska Permanent Fund dividend payments, proposing near-rationality (small, welfare-cheap departures from full optimization that become worthwhile when income changes are large relative to adjustment costs) as the mechanism — an observation the meta-analysis can accommodate only through its shock-size moderator. Administrative-data studies of income shocks (Ganong et al., 2020; Baker et al., 2023) consistently find consumption responses concentrated among low-liquidity households, corroborating the liquidity structure, but they mostly measure spending responses to realized or windfall income rather than the Euler-equation parameter for anticipated income growth. A 2024 review of the income-shock transmission literature (Crawley and Theloudis, 2024) concludes that the range of estimates is wide, that there is still no consensus in the profession, and that response heterogeneity cannot be fully explained by observable factors. The mechanism decomposition — liquid wealth versus persistent household characteristics such as risk aversion, time preferences, or financial sophistication — remains actively studied (Ueda, 2026; Jappelli, Savoia, and Sciacchetano, 2026) and unresolved.

Problem Statement

Determine whether the excess sensitivity of consumption to anticipated income growth — the response of consumption growth to income changes known in advance, interpreted since Campbell and Mankiw as the share of rule-of-thumb consumers — is genuinely near zero for households with substantial liquid wealth, or whether the corrected near-zero estimates arise from a misspecification shared across the published Euler-equation literature. Causally interpreting any consumption response to predictable income that survives — distinguishing limited liquid wealth, persistent household characteristics such as risk aversion, time preferences, or financial sophistication, and optimizing-but-non-standard behavior such as near-rationality, quasi-hyperbolic discounting (a present-biased pattern in which the near future is discounted far more steeply than all later periods), or temptation preferences (self-control costs that lower utility whenever wealth could be consumed immediately) — is an actively studied question in the wider literature that a decisive answer would inform. The question is settled if the corrected near-zero estimate of excess sensitivity for liquid households is either corroborated or contradicted by evidence from primary household data under an identification strategy that does not inherit the shared Euler-equation specification, and is shown to survive the candidate shared misspecifications: measurement error in consumption, non-separability between consumption and leisure, and functional-form, approximation-order, and time-aggregation choices. The causal-mechanism decomposition is interpretive context for any surviving response rather than the target of the question.

References

  1. Havranek, T., & Sokolova, A. (2020). Do consumers really follow a rule of thumb? Three thousand estimates from 144 studies say 'probably not'. Review of Economic Dynamics 35, 97–122. DOI: 10.1016/j.red.2019.05.004, https://doi.org/10.1016/j.red.2019.05.004
  2. Hall, R. E. (1978). Stochastic implications of the life cycle-permanent income hypothesis: Theory and evidence. Journal of Political Economy 86(6), 971–987. DOI: 10.1086/260724, https://doi.org/10.1086/260724
  3. Campbell, J. Y., & Mankiw, N. G. (1990). Permanent income, current income, and consumption. Journal of Business & Economic Statistics 8(3), 265–279. DOI: 10.1080/07350015.1990.10509798, https://doi.org/10.1080/07350015.1990.10509798
  4. Kaplan, G., Violante, G. L., & Weidner, J. (2014). The wealthy hand-to-mouth. Brookings Papers on Economic Activity 2014(1), 77–138. DOI: 10.1353/eca.2014.0002, https://doi.org/10.1353/eca.2014.0002
  5. Kueng, L. (2018). Excess sensitivity of high-income consumers. The Quarterly Journal of Economics 133(4), 1693–1751. DOI: 10.1093/qje/qjy014, https://doi.org/10.1093/qje/qjy014
  6. Olafsson, A., & Pagel, M. (2018). The liquid hand-to-mouth: Evidence from personal finance management software. The Review of Financial Studies 31(11), 4398–4446. DOI: 10.1093/rfs/hhy055, https://doi.org/10.1093/rfs/hhy055
  7. Ganong, P., Jones, D., Noel, P. J., Greig, F. E., Farrell, D., & Wheat, C. (2020). Wealth, race, and consumption smoothing of typical income shocks. NBER Working Paper 27552. DOI: 10.3386/w27552, https://doi.org/10.3386/w27552
  8. Baker, S. R., Farrokhnia, R. A., Meyer, S., Pagel, M., & Yannelis, C. (2023). Income, liquidity, and the consumption response to the 2020 economic stimulus payments. Review of Finance 27(6), 2271–2304. DOI: 10.1093/rof/rfad010, https://doi.org/10.1093/rof/rfad010
  9. Brodeur, A., Cook, N., & Heyes, A. (2020). Methods matter: p-hacking and publication bias in causal analysis in economics. American Economic Review 110(11), 3634–3660. DOI: 10.1257/aer.20190687, https://doi.org/10.1257/aer.20190687
  10. Sokolova, A. (2023). Marginal propensity to consume and unemployment: A meta-analysis. Review of Economic Dynamics 51, 813–846. DOI: 10.1016/j.red.2023.08.005, https://doi.org/10.1016/j.red.2023.08.005
  11. Crawley, E., & Theloudis, A. (2024). Income shocks and their transmission into consumption. Finance and Economics Discussion Series 2024-038, Board of Governors of the Federal Reserve System. DOI: 10.17016/feds.2024.038, https://doi.org/10.17016/feds.2024.038
  12. Gomes, O. (2021). Hand-to-mouth consumers, rule-of-thumb savers, and optimal control. Journal of Economic Interaction and Coordination 16(2), 229–263. DOI: 10.1007/s11403-020-00292-4, https://doi.org/10.1007/s11403-020-00292-4
  13. Ueda, K. (2026). Marginal propensity to consume and personal characteristics: Evidence from bank transaction data and survey. Journal of Money, Credit and Banking. DOI: 10.1111/jmcb.70041, https://doi.org/10.1111/jmcb.70041
  14. Jappelli, T., Savoia, E., & Sciacchetano, A. (2026). Intertemporal MPC and shock size. European Economic Review 186, 105303. DOI: 10.1016/j.euroecorev.2026.105303, https://doi.org/10.1016/j.euroecorev.2026.105303
  15. Tran, H. T. (2025). Determinants of hand-to-mouth behavior in Asian developing countries: Does financial access matter? Journal of Social and Economic Development. DOI: 10.1007/s40847-025-00461-w, https://doi.org/10.1007/s40847-025-00461-w