Some Characteristics Are Risk Exposures, and the Rest Are Irrelevant

Kelly, Bryan T.; Pruitt, Seth; Su, Yinan

doi:10.2139/ssrn.3032013

Cited by 27 publications

(21 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Third, using a larger sample helps avoid overfitting by increasing the ratio of observation count to parameter count. That said, our results are qualitatively identical and quantitively unchanged if we filter out these firms.29 We cross-sectionally rank all stock characteristics period-by-period and map these ranks into the [-1,1] interval followingKelly et al (2019) andFreyberger et al (2019).30 The 94 predictive characteristics are based on Green et al(2017), and we adapt the SAS code available from Jeremiah Green's website and extend the sample period back to 1957. Our data construction differs by adhering more closely to variable definitions in original papers.…”

mentioning

confidence: 75%

See 1 more Smart Citation

Empirical Asset Pricing via Machine Learning

2018

View full text Add to dashboard Cite

show abstract

mentioning

confidence: 75%

“…study the multiple comparisons problem using a bootstrap procedure. Giglio and Xiu (2016) and Kelly et al (2019) use dimension reduction methods to estimate and test factor pricing models. Moritz and Zimmermann (2016) apply tree-based models to portfolio sorting.…”

Section: Literaturementioning

confidence: 99%

Empirical Asset Pricing via Machine Learning

2018

View full text Add to dashboard Cite

show abstract

“…One approach is based upon rational asset pricing theory in an efficient market French, 2015, 2018;Hou, Xue, and Zhang, 2015). A second approach is to extract factors using a statistical analysis of the returns to assets or characteristic-sorted portfolios, while potentially imposing structural constraints implied by rational asset pricing theory (Kelly, Pruitt, and Su, 2017;Freyberger, Neuhierl, and Weber, 2017;Kozak, Nagel, and Santosh, 2017;Lettau and Pelger, 2018).…”

mentioning

confidence: 99%

Short- and Long-Horizon Behavioral Factors

Daniel¹,

Hirshleifer

Sun

2017

View full text Add to dashboard Cite

We propose a theoretically-motivated factor model based on investor psychology and assess its ability to explain the cross-section of U.S. equity returns. Our factor model augments the market factor with two factors which capture long-and short-horizon mispricing. The long-horizon factor exploits the information in managers' decisions to issue or repurchase equity in response to persistent mispricing. The short-horizon earnings surprise factor, which is motivated by investor inattention and evidence of short-horizon underreaction, captures short-horizon anomalies. This three-factor risk-and-behavioral model outperforms other proposed models in explaining a broad range of return anomalies. (Barberis and Shleifer, 2003), as well as commonality in investor errors in interpreting signals about fundamental economic factors (Daniel, Hirshleifer, and Subrahmanyam, 2001). Since mispricing predicts future returns owing to subsequent correction, this implies that behavioral factors can be used to construct a factor model that better describes the cross-section of expected returns. 2 Just as firms that are exposed to systematic risk factors earn an associated risk premium, firms that are heavily exposed to behavioral factors earn a conditional return premium (see, e.g., the model of Hirshleifer and Jiang (2010)). French (1993, 2015) construct risk factors based on firm characteristics that they argue capture risk exposures; we instead supplement the market factor with two behaviorally-motivated factors. Here, we argue that the mispricing effects of numerous behavioral biases, occurring at both long-and short-horizons, can be captured by two behavioral factors: a financing factor FIN that captures long-horizon mispricing, and an inattention factor PEAD that captures short-horizon mispricing.Consistent with the behavioral theories discussed above, our long-horizon factor is based on the intuition from the model of Stein (1996), who argues that when a firm becomes over-or underpriced, the optimal response for the firm is to issue or repurchase its own stock, while not necessarily to change its level of investment. Managers are well-positioned to lead their firms to act as arbitrageurs of their own stock prices, since managers have superior information about the intrinsic value of their firms. Even so, if investors were fully rational, they would fully impound the information contained in a firm's decision to issue or repurchase equity (Myers and Majluf, 1984), so that the financing decision would not be a proxy for mispricing. However, in models of investor overconfidence, the market does not fully impound this information. Empirically, there are on average persistent and strong negative abnormal returns following issuance activity, and positive abnormal returns following repurchases. 3Precisely because the market underreacts to issuance/repurchase activity, it is in firms' interests to 2 Several other studies also suggest that behavioral biases systematically affect asset prices. For example, Goetzmann and Massa (2008) constr...

show abstract

“…Recent explicit factor models based on q-theory, the present value relation, and mispricing are given by Hou, Xue, and Zhang (2015), Fama and French (2015), and Stambaugh and Yuan (2016), respectively. Rigorous statistical explanations for crosssectional predictability are proposed by Kozak, Nagel, and Santosh (2017), Kelly, Pruitt, and Su (2017), and Lettau and Pelger (2018).…”

Section: Introductionmentioning

confidence: 99%

The Limits of p-Hacking: A Thought Experiment

Chen

2019

FEDS

View full text Add to dashboard Cite

Suppose that asset pricing factors are just p-hacked noise. How much phacking is required to produce the 300 factors documented by academics? I show that, if 10,000 academics generate 1 factor every minute, it takes 15 million years of p-hacking. This absurd conclusion comes from applying the p-hacking theory to published data. To fit the fat right tail of published t-stats, the p-hacking theory requires that the probability of publishing tstats < 6.0 is infinitesimal. Thus it takes a ridiculous amount of p-hacking to publish a single t-stat. These results show that p-hacking alone cannot explain the factor zoo.

show abstract

Some Characteristics Are Risk Exposures, and the Rest Are Irrelevant

Cited by 27 publications

References 36 publications

Empirical Asset Pricing via Machine Learning

Empirical Asset Pricing via Machine Learning

Short- and Long-Horizon Behavioral Factors

The Limits of p-Hacking: A Thought Experiment

Contact Info

Product

Resources

About