Literature Review: Idiosyncratic Volatility Puzzle (1990-2024 Update)

Project: Revisiting Ang, Hodrick, Xing, Zhang (2006) in the modern US equity universe. Window: 1990-2024. Method: CAPM-based daily-residual idiovol, quintile sorts, Fama-MacBeth.

This bibliography organizes the literature into four buckets: (i) Does the puzzle exist? (ii) Why does the puzzle exist? — including a sub-bucket of benchmark factor models used to compute alphas — (iii) Does the puzzle still exist? and (iv) Methods literature for the Fama-MacBeth specification. Entries within each bucket appear in approximate chronological order. Each annotation closes with a proximity label: foundational, direct precedent, methods, or context. The closing sections present a frontier map and a gap statement.

Bucket (i) — Does the Puzzle Exist? Original, Replications, Measurement

ang2006cross

Ang, A., Hodrick, R. J., Xing, Y., and Zhang, X. 2006. “The Cross-Section of Volatility and Expected Returns.” Journal of Finance 61(1): 259-299.

The seminal paper. Sorts NYSE/AMEX/Nasdaq stocks 1963-2000 into quintiles by lagged-month idiosyncratic volatility (FF3 daily residuals) and finds the highest-idiovol quintile underperforms the lowest by about -1.06% per month, with an FF3 alpha of -1.31%. The pattern is robust to size, book-to-market, leverage, liquidity, volume, turnover, bid-ask spread, momentum, and dispersion in analyst forecasts. Our paper: directly replicates and extends this design to 1990-2024 using CAPM residuals. foundational

ang2009high

Ang, A., Hodrick, R. J., Xing, Y., and Zhang, X. 2009. “High Idiosyncratic Volatility and Low Returns: International and Further U.S. Evidence.” Journal of Financial Economics 91(1): 1-23.

Extends the puzzle to 23 developed markets and finds the negative idiovol-return relation is pervasive internationally, with co-movement across countries suggesting a common source rather than a US-specific anomaly. Re-confirms US results with stricter controls and demonstrates the pattern is not driven by small stocks or short windows. Our paper: establishes that the puzzle is not a fluke; sets up the question of whether it survives in the post-2000 US sample. direct precedent

bali2008idiosyncratic

Bali, T. G., and Cakici, N. 2008. “Idiosyncratic Risk and the Cross Section of Expected Stock Returns.” Journal of Financial and Quantitative Analysis 43(1): 29-58.

Argues the AHXZ result is sensitive to portfolio weighting, breakpoint choice (NYSE vs all), data frequency (daily vs monthly), and screen for low-priced/illiquid stocks. Shows that under value-weighting with NYSE breakpoints and exclusion of small stocks the alpha differential weakens or disappears. Our paper: motivates our robustness specification with both equal- and value-weighted portfolios and explicit price/liquidity screens. direct precedent

campbell2001have

Campbell, J. Y., Lettau, M., Malkiel, B. G., and Xu, Y. 2001. “Have Individual Stocks Become More Volatile? An Empirical Exploration of Idiosyncratic Risk.” Journal of Finance 56(1): 1-43.

Documents a secular rise in firm-level idiosyncratic volatility 1962-1997 without a corresponding rise in market or industry volatility. Provides the time-series backdrop against which any idiovol-pricing study must be read, and shows idiovol is large enough to matter for portfolio diversification. Our paper: the 1990-2024 sample begins at the peak of this trend, so understanding the trend in the sample period is essential. context

goyal2003idiosyncratic

Goyal, A., and Santa-Clara, P. 2003. “Idiosyncratic Risk Matters!” Journal of Finance 58(3): 975-1007.

Argues average stock variance (mostly idiosyncratic) forecasts the market return — a positive aggregate idiovol-return relation in time series. Predates AHXZ and frames idiovol as priced under-diversification at the market level. Our paper: useful contrast — the aggregate positive relation coexists with a cross-sectional negative one, a tension our paper inherits. context

Bucket (ii) — Why Does the Puzzle Exist? Explanations

fu2009idiosyncratic

Fu, F. 2009. “Idiosyncratic Risk and the Cross-Section of Expected Stock Returns.” Journal of Financial Economics 91(1): 24-37.

Argues that expected (not lagged realized) idiovol matters, and estimates it via an EGARCH model on monthly returns. Finds a positive idiovol-return relation under this conditional measure — claiming the AHXZ puzzle reflects mean reversion in the lagged measure, not a true mispricing. Subsequent work (Guo, Kassa, Ferguson 2014, JFQA) shows this result is partly an artifact of look-ahead bias in the EGARCH fit. Our paper: clarifies that we follow AHXZ in using lagged realized idiovol and discuss Fu’s expected-volatility critique. direct precedent

boyer2010expected

Boyer, B., Mitton, T., and Vorkink, K. 2010. “Expected Idiosyncratic Skewness.” Review of Financial Studies 23(1): 169-202.

Builds a cross-sectional model of expected idiosyncratic skewness and shows it negatively predicts returns — lottery-like stocks earn lower average returns. Because high-idiovol stocks tend to be high-skewness, this offers a partial explanation for the AHXZ result rooted in preference for skewness. Our paper: skewness is one of the leading explanations we discuss but do not test directly. context

bali2011maxing

Bali, T. G., Cakici, N., and Whitelaw, R. F. 2011. “Maxing Out: Stocks as Lotteries and the Cross-Section of Expected Returns.” Journal of Financial Economics 99(2): 427-446.

Introduces MAX — the average of the five highest daily returns in the prior month — as a direct proxy for lottery demand. MAX largely subsumes idiovol in cross-sectional regressions, suggesting the AHXZ puzzle is at heart a lottery-preference phenomenon. Our paper: MAX is the leading competing characteristic; we will report idiovol sorts both unconditional and conditional on MAX. direct precedent

han2011investor

Han, B., and Lesmond, D. A. 2011. “Investor Sentiment and Liquidity-Biased Idiosyncratic Volatility.” Review of Financial Studies 24(5): 1590-1629.

Shows that bid-ask bounce and zero-return days contaminate daily-residual idiovol estimates, particularly for small/illiquid stocks. After correcting for microstructure noise, the idiovol-return relation attenuates substantially. Our paper: motivates our minimum-price and minimum-volume screens and the use of NYSE breakpoints in robustness. direct precedent

huang2010return

Huang, W., Liu, Q., Rhee, S. G., and Zhang, L. 2010. “Return Reversals, Idiosyncratic Risk, and Expected Returns.” Review of Financial Studies 23(1): 147-168.

Argues that the negative idiovol-return relation is largely driven by short-term return reversal — high-idiovol stocks just experienced positive returns, and the next month they reverse. Controlling for last month’s return dramatically weakens the puzzle. Our paper: return-reversal control is part of our robustness battery. direct precedent

stambaugh2015arbitrage

Stambaugh, R. F., Yu, J., and Yuan, Y. 2015. “Arbitrage Asymmetry and the Idiosyncratic Volatility Puzzle.” Journal of Finance 70(5): 1903-1948.

Explains the puzzle via the interaction of mispricing and arbitrage asymmetry: idiovol deters arbitrage of overpriced stocks more than of underpriced ones because short-selling is costlier. Thus high-idiovol overpriced stocks remain overpriced (low returns) and high-idiovol underpriced stocks correct slowly (high returns); the average effect is negative. Our paper: perhaps the most cited modern explanation; we summarize it in the literature section and note we do not test it. direct precedent

hou2016have

Hou, K., and Loh, R. K. 2016. “Have We Solved the Idiosyncratic Volatility Puzzle?” Journal of Financial Economics 121(1): 167-194.

Conducts a horse race among more than a dozen proposed explanations using a standardized decomposition method. Finds that lottery-preference proxies (MAX, expected skewness) and one-month return reversal jointly explain roughly 60-80% of the puzzle, while other channels (illiquidity, dispersion, short-sale constraints) explain little incrementally. Our paper: the benchmark survey we cite when describing the state of explanations; positions our update against this 2016 stocktake. direct precedent

chen2012does

Chen, Z., and Petkova, R. 2012. “Does Idiosyncratic Volatility Proxy for Risk Exposure?” Review of Financial Studies 25(9): 2745-2787.

Decomposes the FF3-residual idiovol into innovations to aggregate idiosyncratic volatility and idiosyncratic shocks. The aggregate component carries a negative risk premium, accounting for part of the AHXZ alpha. Our paper: an alternative framing — idiovol is partly a systematic risk factor in disguise. We cite as a competing interpretation. context

liu2018absolving

Liu, J., Stambaugh, R. F., and Yuan, Y. 2018. “Absolving Beta of Volatility’s Effects.” Journal of Financial Economics 128(1): 1-15.

Shows the beta anomaly is driven by the positive correlation between beta and idiovol combined with the negative idiovol-alpha relation among overpriced stocks. Excluding overpriced high-idiovol stocks renders the beta anomaly insignificant. Tightly connects the idiovol puzzle to the beta anomaly through the mispricing/arbitrage-asymmetry channel of Stambaugh, Yu, and Yuan (2015). Our paper: reinforces that the idiovol puzzle does not stand alone — its survival is informative about a family of related low-risk anomalies. direct precedent

asness2020betting

Asness, C. S., Frazzini, A., Gormsen, N. J., and Pedersen, L. H. 2020. “Betting Against Correlation: Testing Theories of the Low-Risk Effect.” Journal of Financial Economics 135(3): 629-652.

Decomposes the low-risk effect into a leverage-constraints channel (priced via beta and correlation) and a behavioral lottery-demand channel (priced via idiosyncratic risk and SMAX). Both channels operate; idiovol-based factors load on sentiment while correlation-based factors load on margin debt. Our paper: the most recent JFE statement of the lottery/leverage taxonomy; clarifies that an updated idiovol estimate also speaks to the lottery side of this decomposition. direct precedent

Sub-bucket: Benchmark Factor Models

These papers are not explanations of the idiovol puzzle. They define the factor-model benchmarks used to compute the alphas reported in the AHXZ literature, including ours.

fama1993common

Fama, E. F., and French, K. R. 1993. “Common Risk Factors in the Returns on Stocks and Bonds.” Journal of Financial Economics 33(1): 3-56.

The FF3 factor model — MKT, SMB, HML — that AHXZ use to extract idiosyncratic residuals. Establishes that size and value capture systematic variation absent from the CAPM. Our paper: we use CAPM residuals (one-factor) as the headline definition of idiovol but report FF3 residuals as robustness, following AHXZ. methods

carhart1997persistence

Carhart, M. M. 1997. “On Persistence in Mutual Fund Performance.” Journal of Finance 52(1): 57-82.

Introduces the momentum factor (UMD/WML). The four-factor model (FF3+MOM) is the standard benchmark for alpha decompositions in the idiovol literature. Our paper: used in alpha decomposition of the long-short idiovol portfolio. methods

Bucket (iii) — Does the Puzzle Still Exist? Recent Updates and Robustness

chu2020idiosyncratic

Chu, Y., Hirshleifer, D., and Ma, L. 2020. “The Causal Effect of Limits to Arbitrage on Asset Pricing Anomalies.” Journal of Finance 75(5): 2631-2672.

Uses Regulation SHO’s randomized pilot to show that relaxing short-sale constraints attenuates the idiovol puzzle (along with other anomalies). Provides causal evidence consistent with Stambaugh-Yu-Yuan’s arbitrage-asymmetry mechanism. Our paper: confirms that the puzzle is sensitive to the short-sale environment — and the post-2007 universal SHO regime is part of our sample. direct precedent

mclean2016does

McLean, R. D., and Pontiff, J. 2016. “Does Academic Research Destroy Stock Return Predictability?” Journal of Finance 71(1): 5-32.

Documents that, on average, anomalies decay by about 26% after in-sample publication and 58% after publication, consistent with arbitrageur learning. Lists idiovol among the affected anomalies. Our paper: directly motivates the question — has the post-AHXZ decay continued through 2024? direct precedent

hou2020replicating

Hou, K., Xue, C., and Zhang, L. 2020. “Replicating Anomalies.” Review of Financial Studies 33(5): 2019-2133.

Replicates 452 published anomalies with NYSE breakpoints and value-weighting and finds about half fail at the 5% level. Idiovol survives under value-weighting in their q-factor framework but with a smaller magnitude than the equal-weighted result. Our paper: the methodological touchstone for our robustness section — we report both VW and EW alphas with NYSE breakpoints. direct precedent

bali2017lottery

Bali, T. G., Brown, S. J., Murray, S., and Tang, Y. 2017. “A Lottery-Demand-Based Explanation of the Beta Anomaly.” Journal of Financial and Quantitative Analysis 52(6): 2369-2397.

Generalizes the lottery-demand interpretation to the beta anomaly and reinforces evidence that demand for lottery-like payoffs is the unifying channel behind several volatility/beta puzzles. Our paper: situates the AHXZ puzzle inside a broader family of “lottery anomalies.” context

barillas2018comparing

Barillas, F., and Shanken, J. 2018. “Comparing Asset Pricing Models.” Journal of Finance 73(2): 715-754.

Bayesian model-comparison framework for ranking factor models. Relevant for the choice of alpha benchmark (CAPM vs FF3 vs FF5 vs q-factor) in the long-short idiovol portfolio. Our paper: cited when justifying our use of multiple benchmark models for alpha computation. methods

Bucket (iv) — Methods Literature for the Fama-MacBeth Specification

famamacbeth1973

Fama, E. F., and MacBeth, J. D. 1973. “Risk, Return, and Equilibrium: Empirical Tests.” Journal of Political Economy 81(3): 607-636.

The two-pass procedure: estimate factor loadings in a first-pass time-series regression, then run cross-sectional regressions of returns on loadings each period and take the time-series mean of the slopes with Newey-West-style standard errors of the period-by-period coefficients. The estimator of choice for nearly every paper in this bibliography that reports cross-sectional risk premia. Our paper: the FM regression of monthly excess returns on lagged idiovol plus controls is one of our two headline specifications, alongside quintile sorts. methods

shanken1992estimation

Shanken, J. 1992. “On the Estimation of Beta-Pricing Models.” Review of Financial Studies 5(1): 1-33.

Derives the errors-in-variables correction to FM standard errors when first-pass betas are estimated rather than known, and integrates ML and two-pass approaches. Shows the uncorrected FM procedure overstates the precision of price-of-risk estimates. Our paper: the canonical reference when reporting Shanken-corrected standard errors alongside the unadjusted FM standard errors. methods

petersen2009estimating

Petersen, M. A. 2009. “Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches.” Review of Financial Studies 22(1): 435-480.

Compares OLS, White, clustered, Fama-MacBeth, and Rogers standard errors in finance panel data sets. Shows that when residuals exhibit a firm-effect, clustering by firm is essential; when residuals exhibit a time-effect, FM or clustering by time is essential; double clustering is appropriate when both are present. Our paper: governs our standard-error choice in the FM regressions — we cluster by month, with Newey-West adjustment for the time-series of FM coefficients. methods

jegadeesh2019empirical

Jegadeesh, N., Noh, J., Pukthuanthong, K., Roll, R., and Wang, J. 2019. “Empirical Tests of Asset Pricing Models with Individual Assets: Resolving the Errors-in-Variables Bias in Risk Premium Estimation.” Journal of Financial Economics 133(2): 273-298.

Proposes an instrumental-variables approach that allows individual stocks (rather than portfolios) as test assets while delivering consistent ex-post risk-premium estimates. Demonstrates that under the corrected procedure, market and factor risk premia become insignificant once asset characteristics are controlled — sharpening the case for characteristic-based interpretations. Our paper: methods reference for the FM specification at the individual-stock level; we note the EIV concern when discussing the magnitude (not just the sign) of the idiovol slope. methods

Frontier Map

Bucket	Papers
(i) Does the puzzle exist?	ang2006cross, ang2009high, bali2008idiosyncratic, campbell2001have, goyal2003idiosyncratic
(ii) Why does the puzzle exist? — explanations	fu2009idiosyncratic, boyer2010expected, bali2011maxing, han2011investor, huang2010return, stambaugh2015arbitrage, hou2016have, chen2012does, liu2018absolving, asness2020betting
(ii) Benchmark factor models (sub-bucket)	fama1993common, carhart1997persistence
(iii) Does the puzzle still exist?	chu2020idiosyncratic, mclean2016does, hou2020replicating, bali2017lottery, barillas2018comparing
(iv) Methods	famamacbeth1973, shanken1992estimation, petersen2009estimating, jegadeesh2019empirical

The literature is mature on bucket (i) for the original sample window (1963-2000) and well-developed on bucket (ii) through 2020. Bucket (iii) is unsettled. McLean and Pontiff (2016) predicts attenuation, Hou, Xue, and Zhang (2020) finds survival under VW with q-factor benchmarks, and the post-Reg-SHO era plus the rise of indexing and high-frequency arbitrage capital create new reasons to expect decay. The benchmark factor models (FF3, FF4) and methods literature (FM, Shanken, Petersen, JNPRW) are settled tools.

Empty cells the literature has not addressed:

No transparent decade-by-decade AHXZ replication on the 1990-2024 sample. Hou, Xue, and Zhang (2020) cover idiovol within a 452-anomaly survey; no paper takes the AHXZ design end-to-end on this window.
No update on whether the post-publication anomaly decay documented by McLean and Pontiff (2016) has continued through 2024, specifically for idiovol — i.e., whether the spread is now indistinguishable from zero in the most recent decade.
No clean comparison of CAPM-residual vs FF3-residual idiovol in the modern sample, with both equal-weighted and value-weighted portfolios and NYSE breakpoints, holding the rest of the design fixed. Most papers either swap one design lever at a time or test one specification.
No sample-period sensitivity analysis of how the puzzle behaves across the regime shifts within 1990-2024 — pre/post Reg SHO 2007, pre/post the 2003 decimalization, and the indexing/passive-flows era (2010+).

Our paper fills cell 1 directly and provides material relevant to cells 2 and 4 as part of the same exercise. Cell 3 is partially addressed by our robustness battery (CAPM headline, FF3 robustness; EW headline, VW robustness; NYSE breakpoints robustness).

Gap Statement

The literature contains a foundational result (Ang, Hodrick, Xing, Zhang 2006), a body of competing explanations through 2020, and a methodological literature on replication (Hou, Xue, Zhang 2020) — but no recent paper takes the original AHXZ research design and runs it on the 1990-2024 US universe with the explicit goal of measuring whether the negative idiovol-return relation has survived. We fill this gap by replicating the exact AHXZ exercise — CAPM-based daily-residual idiosyncratic volatility, equal-weighted quintile sorts, and Fama-MacBeth cross-sectional regressions — on common stocks (CRSP shrcd 10/11) ex-financials and ex-utilities over 1990-2024. Our contribution is descriptive and methodological: a transparent update that shows what the AHXZ-style estimate looks like today, decade-by-decade, with NYSE-breakpoint and value-weighted robustness in the spirit of Hou, Xue, and Zhang (2020). We make no causal claim and do not adjudicate among the explanations in bucket (ii); rather, we provide the updated stylized fact against which future explanations must be tested.