Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
1
“The Supraview of U.S. Equity Return Predictive Signals –
The Last 40 Years and Changes in the Last 5 Years”
Prof. John Hand UNC Chapel Hill
SQA Nov. 15th, 2013
GOALS OF MY TALK
2
● I will give you my SUPRAVIEW of what academics have discovered since 1970 about the firm-specific characteristics that historically have predicted the cross-section of expected U.S. stock returns.
● My coauthors Jeremiah Green (Penn State) and Frank Zhang (Yale) and I term these firm-specific characteristics as “return predictive signals” or RPS.
● Disclaimer: Certain parts of this talk represent my personal opinions, and are not necessarily shared by my coauthors.
2
RPS SUPRAVIEW
● Almost all prior academic RPS research has centered on discovering the identify and returns to one new RPS per paper – the “one new tree” approach.
● What Green, Hand and Zhang (RAST, 2013a) do is take the supraview = looking at the whole forest.
● Taking the supraview and “getting high” is worth the effort and risk because the view from above the trees provides insights that cannot be seen from within the trees on the forest floor.
3
POPULATION of RPS (GHZ, RAST 2013a)
● We approximated the population of RPS by:
● Searching top-tier U.S. accounting, finance and practitioner journals for RPS papers published Jan. 1970 – Dec. 2010.
● Following up references in these papers
● Searching SSRN for working papers in these Subject Matter eJournals:
● FEN Capital Markets: Market Efficiency ● FEN Capital Markets: Asset Pricing / Valuation ● ARN Financial Accounting
4
3
DATABASE of RPS (cont’d)
● We categorized each RPS as:
● Accounting-based > In financial statements ● Finance-based > Dependent on stock price ● Other-based > Neither of these
● Minority of ‘mixed’ RPS coded as Finance-based.
● We coded key attributes: Author(s), publication date, first WP date, databases used, RPS name and definition, period of analysis, EW or VW returns, firm characteristics or factors orthogonalized against, and RPS mean return and Sharpe ratio.
5
0
20
40
60
80
100
120
140
160
Cumulative # of ABTS vs. PBTS vs. OBTS, 1970 - 2010
Accounting-based RPS Finance-based RPS Other-based RPS
RESULT # 1 There are 330+ RPS
6
4
RESULT # 2a RPS EW_return stats
TABLE 4 Panel A: All types of RPS Panel B: Accounting-based RPS
Mean Std. dev. Sharpe Mean Std. dev. Sharpe Min. 0.8% 4.3% 0.08 3.3% 4.6% 0.37 25th percentile 6.6% 8.6% 0.68 8.5% 9.2% 0.72 Median 10.8% 11.0% 0.87 12.0% 11.5% 0.91 Mean 12.2% 12.1% 1.04 13.2% 12.4% 1.10 75th percentile 16.1% 14.3% 1.29 17.4% 16.7% 1.38 Max. 35.0% 31.2% 2.98 32.4% 31.2% 2.50 Std. dev. 7.1% 5.2% 0.55 6.9% 4.9% 0.97 Skewness 0.9 1.3 1.0 0.9 1.2 0.9 Number of RPS 237 208 208 115 97 97
7
RESULT # 2b RPS EW_return stats
TABLE 4 Panel A: All types of RPS Panel B: Accounting-based RPS
Mean Std. dev. Sharpe Mean Std. dev. Sharpe Min. 0.8% 4.3% 0.08 3.3% 4.6% 0.37 25th percentile 6.6% 8.6% 0.68 8.5% 9.2% 0.72 Median 10.8% 11.0% 0.87 12.0% 11.5% 0.91 Mean 12.2% 12.1% 1.04 13.2% 12.4% 1.10 75th percentile 16.1% 14.3% 1.29 17.4% 16.7% 1.38 Max. 35.0% 31.2% 2.98 32.4% 31.2% 2.50 Std. dev. 7.1% 5.2% 0.55 6.9% 4.9% 0.97 Skewness 0.9 1.3 1.0 0.9 1.2 0.9 Number of RPS 237 208 208 115 97 97
Sloan (1996)
10.4% 12.8% 0.81
TABLE 4 Panel A: All types of RPS Panel B: Accounting-based RPS
Mean Std. dev. Sharpe Mean Std. dev. Sharpe Min. 0.8% 4.3% 0.08 3.3% 4.6% 0.37 25th percentile 6.6% 8.6% 0.68 8.5% 9.2% 0.72 Median 10.8% 11.0% 0.87 12.0% 11.5% 0.91 Mean 12.2% 12.1% 1.04 13.2% 12.4% 1.10 75th percentile 16.1% 14.3% 1.29 17.4% 16.7% 1.38 Max. 35.0% 31.2% 2.98 32.4% 31.2% 2.50 Std. dev. 7.1% 5.2% 0.55 6.9% 4.9% 0.97 Skewness 0.9 1.3 1.0 0.9 1.2 0.9 Number of RPS 237 208 208 115 97 97
8
TABLE 4 Panel C: Finance-based RPS Panel D: Other-based RPS
Mean Std. dev. Sharpe Mean Std. dev. Sharpe Min. 3.7% 4.8% 0.35 0.8% 4.8% 0.35 25th percentile 6.6% 9.4% 0.61 4.3% 6.8% 0.52 Median 11.9% 11.8% 0.98 6.4% 9.0% 0.75 Mean 13.5% 12.6% 1.11 8.1% 10.6% 0.81 75th percentile 16.9% 14.4% 1.30 10.6% 11.5% 1.05 Max. 35.0% 30.9% 2.98 22.8% 30.8% 1.97 Std. dev. 7.6% 5.0% 0.63 5.4% 6.2% 0.44 Skewness 0.8 1.3 1.1 1.2 1.9 0.7 Number of RPS 72 68 68 50 43 43
Jegadeesh & Titman (1993)
11.4% 18.7% 0.61
5
RESULT # 2c RPS VW_return stats
TABLE 5 Panel A: All types of RPS Panel B: Accounting-based RPS
Mean Std. dev. Sharpe Mean Std. dev. Sharpe Min. -2.4% 3.9% -0.21 -1.4% 4.6% -0.11 25th percentile 3.5% 8.6% 0.34 4.2% 8.8% 0.41 Median 7.1% 11.0% 0.61 7.1% 11.0% 0.67 Mean 8.1% 12.2% 0.70 8.0% 12.2% 0.63 75th percentile 10.9% 15.2% 0.84 10.6% 13.8% 0.83 Max. 35.0% 27.9% 3.09 30.1% 27.3% 1.47 Std. dev. 6.3% 5.0% 0.58 5.3% 4.9% 0.31 Skewness 1.4 1.0 2.1 2.0 1.5 0.0 Number of RPS 99 87 87 35 27 27
TABLE 4 Panel A: All types of RPS Panel B: Accounting-based RPS
Mean Std. dev. Sharpe Mean Std. dev. Sharpe Min. 0.8% 4.3% 0.08 3.3% 4.6% 0.37 25th percentile 6.6% 8.6% 0.68 8.5% 9.2% 0.72 Median 10.8% 11.0% 0.87 12.0% 11.5% 0.91 Mean 12.2% 12.1% 1.04 13.2% 12.4% 1.10 75th percentile 16.1% 14.3% 1.29 17.4% 16.7% 1.38 Max. 35.0% 31.2% 2.98 32.4% 31.2% 2.50 Std. dev. 7.1% 5.2% 0.55 6.9% 4.9% 0.97 Skewness 0.9 1.3 1.0 0.9 1.2 0.9 Number of RPS 237 208 208 115 97 97
VW EW
9
RESULT # 2d RPS , Sharpe stability
Figure 2, Panel A: Mean RPS returns Figure 2, Panel B: Mean RPS returns All types of RPS combined together Accounting-based RPS only t-stat on [EW RPS] slope = 0.7 t-stat on [EW RPS] slope = 1.6 t-stat on [VW RPS] slope = 0.2 t-stat on [VW RPS] slope = –0.3
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
1970 1980 1990 2000 2010
Hedge returns over time
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
1970 1980 1990 2000 2010
ABTS: HRet over time
10
Figure 3, Panel A: Sharpe ratios of RPS returns Figure 3, Panel B: Sharpe ratios of RPS returns All types of RPS combined together Accounting-based RPS only t-stat [EW RPS] slope = 0.7 t-stat [EW RPS] slope = –0.1 t-stat [VW RPS] slope = 0.3 t-stat [VW RPS] slope = 1.1
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
1970 1975 1980 1985 1990 1995 2000 2005 2010
Sharpe ratios over time
-0.50
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
1970 1980 1990 2000 2010
ABTS: Sharpe ratios over time
RED = VW returns
6
RESULT # 2e RPS vs. , vs. Sharpe
Figure 4, Panel A: vs. Figure 4, Panel B: vs. All types of RPS combined together Accounting-based RPS only
[RPS_EW] = 4.7% + 0.61 [RPS_EW] [RPS_EW] = 4.0% + 0.74 [RPS_EW] (4.2) (7.3) (2.4) (6.1)
[RPS_VW] = 2.1% + 0.49 [RPS_VW] [RPS_VW] = 4.2% + 0.24 [RPS_VW] (1.3) (4.1) (2.6) (2.0)
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
0% 5% 10% 15% 20% 25% 30% 35%
ABTS: Ann_HRet vs. Ann_Stdev
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
0% 5% 10% 15% 20% 25% 30% 35%
ABTS: Ann_HRet vs. Ann_Stdev
11
Figure 4, Panel C: vs. Sharpe ratio Figure 4, Panel D: vs. Sharpe ratio All types of RPS combined together Accounting-based RPS only
[RPS_EW] = 2.0% + 0.10 SR[RPS_EW] [RPS_EW] = 2.8% + 0.09 SR[RPS_EW] (2.8) (16.1) (2.3) (9.3)
[RPS_VW] = 2.7% + 0.08 SR[RPS_VW] [RPS_VW] = 2.7% + 0.07 SR[RPS_VW] (3.9) (9.8) (2.5) (4.9)
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
ABTS: Ann_HRet vs. Ann_SR
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
ABTS: Ann_SR vs. Ann_HRet
RED = VW returns
RED = VW returns
Figure 5, Panel A: RPS discovered and publicly reported by academics, 1970-2010
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
ABTS: Ann_HRet vs. Ann_SR
Panel B: RPS commonly used by sophisticated investors (as proxied by J.P. Morgan, 1/27/11)
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
JPM factors -- Russell 3000 -- Sharpes vs. Returns
Largest 1000 Smallest 1000 Smallest 2000 Largest 200
Panel B: RPS commonly used by sophisticated investors (as proxied by J.P. Morgan, 1/27/11)
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
JPM factors -- Russell 3000 -- Sharpes vs. Returns
Largest 1000 Smallest 1000 Smallest 2000 Largest 200
Panel B: RPS commonly used by sophisticated investors (as proxied by J.P. Morgan, 1/27/11)
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
JPM factors -- Russell 3000 -- Sharpes vs. Returns
Largest 1000 Smallest 1000 Smallest 2000 Largest 200
Panel B: RPS commonly used by sophisticated investors (as proxied by J.P. Morgan, 1/27/11)
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
JPM factors -- Russell 3000 -- Sharpes vs. Returns
Largest 1000 Smallest 1000 Smallest 2000 Largest 200
RESULT # 2f vs. Sharpe Academic RPS vs. Practitioner RPS
12
7
What has changed in last 5 years?
13
● Some academics are still discovering new RPS …
● But other academics are adopting the supraview:
● Harvey, Liu & Zhu (2013) identify 311 RPS and/or factors to study effects of data snooping.
● McLean & Pontiff (2013) document attenuation of 35% in mean hedge returns to 82 RPS after academic discovery.
● Frank, Jeremiah and I have a working paper in which we take the supraview on the cross-section of expected returns. More on that in a minute …
CONCLUSIONS, part 1
1. Adopting the Supraview on RPS uncovers many new facts about RPS that are informative to both academics and practitioners.
2. The time has come for academics to STOP discovering more RPS. Rather, academics should seek to understand what the population of RPS tells us about risk, return, market efficiency, etc.
3. For example, does the presence of N = 330 RPS indicate that the U.S. stock market is massively inefficient, or that there are far more rationally priced sources of risk to be understood?
14
8
MY TALK, PART 2 – GHZ (2013b, WP)
15
● In this new working paper we take the supraview on the cross-section of expected U.S. stock returns. [full text of paper is on SSRN.com]
● How? We update Fama & French (1992) in light of the huge population of documented RPS.
● We report what we think are some remarkable results.
● I will give you my opinion as to some of the implications of our results for academics and investment practitioners.
“The Remarkable Multidimensionality in the
Cross-Section of Expected U.S. Stock Returns”
Jeremiah Green John Hand Frank Zhang
Penn State UNC Yale
Chicago Oct. 24, 2013
9
Cumulative number of RPS, 1970-2010
17
0
50
100
150
200
250
300
350
Cumulative # of ABTS vs. PBTS vs. OBTS, 1970 - 2010
Total RPS
“In the beginning, there was chaos. Then came the CAPM …
Then anomalies erupted, and there was chaos again.”
Cochrane (2011, p. 1058)
18
0
50
100
150
200
250
300
350
Cumulative # of ABTS vs. PBTS vs. OBTS, 1970 - 2010
Total RPS
FF92 goal: “To evaluate the joint roles of market b, size, E/P, leverage,
and book-to-market in the cross-section of average returns of U.S.
stocks.” (p. 428)
10
19
0
50
100
150
200
250
300
350
Cumulative # of ABTS vs. PBTS vs. OBTS, 1970 - 2010
Total RPS
FF92 find: “If assets are priced rationally, our results suggest that stock risks are
multidimensional … Size and book-to-market provide a simple and powerful characterization
of the cross-section of average stock returns for the 1963-1990 period.” (pp. 428-9)
20
0
50
100
150
200
250
300
350
Cumulative # of ABTS vs. PBTS vs. OBTS, 1970 - 2010
Total RPS
“Alas, the world is once again descending into chaos … We are going to have to repeat Fama & French’s (1992) anomaly digestion, but with
many more dimensions. We have a lot of questions to answer:
First, which characteristics really provide independent information about average returns? Which are subsumed by others?”
RPS chaos has grown 10X since FF92
11
Our main results
● We project 93 RPS onto U.S. returns 1980-2012. We find that a remarkably large 26 RPS are multidimensionally priced as defined by their mean FM coefficients having an abs{t-stat} 3.0 in the cross-section of 1-month ahead returns.
● This is 10X the dimensionality of the current 3-dimensional model of firm size, book-to-market and 12-month momentum that is widely used by academics and investment practitioners.
● We think that 26 is likely an underestimate.
21
Our main results (cont’d)
● High dimensionality occurs in part because the average cross-correlation between scaled ranked decile RPS is very small — just 0.03.
● Multidimensional pricing of several famous RPS is quite different to their unidimensional pricing.
● Firm size, book-to-market, and 12-month momentum do not have t-stats large enough to put them in the top 10 largest t-stats in 1-month ahead returns multidimensional regressions.
22
12
Our main results (cont’d)
● Hedge returns on multidimensioned RPS are 2/3rds smaller than those on unidimensioned RPS, while t-stats are between 1/3rd and 2/3rds smaller.
● High dimensionality is present in each decade of our 1980-2012 data period.
● Large firms have far fewer multidimensionally priced RPS than do small firms, but the RPS of large firms explain 3X the amount of cross-sectional return variance.
23
24
Implications that we see
1. Dimensionality of U.S. stock returns is likely higher than we estimate it to be.
2. Controlling for only firm size, book-to-market and 12-month mom is low powered, maybe also biased.
3. NPV of investing in understanding why returns are so multidimensional >> NPV of finding another RPS.
4. How much the estimation of dimensionality has changed over 20+ years means that it is reasonable to take the inferences/conclusions of our study, and future studies, more cautiously than ever.
13
Literature on return dimensioning
● Relatively sparse.
● Papers have either proposed a small competing set of priced RPS to replace firm size, book-to-market and 12-month momentum, or have put forward a larger set of RPS beyond firm size, book-to-market and 12-month momentum.
● Hou, Xue & Zhang (2012) exemplify the former approach, while Haugen & Baker (1996), Fama & French (2008) and Lewellen (2013) illustrate the latter method.
25
Data & empirical methods: RPS supraview database 100 RPS
● Since FF92, almost all RPS research has centered on discovering and reporting the hedge returns to one new RPS per paper.
● In GHZ (RAST, 2013a) we take the RPS supraview. We find that getting high above the RPS forest yields new insights—and puzzles— that cannot be seen from perspective of a single RPS tree.
● We select 100 RPS primarily from GHZ database.
● RPS based only on CRSP, Compustat or I/B/E/S.
26
14
Data
● Individual RPS listed and defined in Tables 1 and 2.
● Not all RPS can be computed per originating paper for every firm at every point in calendar time.
● But we really want to keep as many RPS as possible, and as many observations per RPS as possible.
● So we:
● Set missing Compustat data to zero (e.g., R&D). ● Set all RPS with post-creation missing values to
mean of the non-missing RPS for that month.
27
Table 1 (excerpt)
28
# RPS Acronym Author(s) Date, Journal
1 Beta beta Fama & MacBeth 1973, JPE
2 Beta squared betasq Fama & MacBeth 1973, JPE
3 Earnings-to-price ep Basu 1977, JF
4 Firm size (market cap) mve Banz 1981, JFE
5 Dividends-to-price dy Litzenberger & Ramaswamy 1982, JF
6 Unexpected quarterly earnings sue Rendelman, Jones & Latane 1982, JFE
7 Change in forecasted annual EPS chfeps Hawkins, Chamberlin & Daniel 1984, FAJ
8 Book-to-market bm Rosenberg, Reid & Lanstein 1985, JPM
9 36-month momentum mom36m De Bondt & Thaler 1985, JF
10 Forecasted growth in 5-year EPS fgr5yr Bauman & Dowen 1988, FAJ
11 Leverage lev Bhandari 1988, JF
12 Current ratio currat Ou & Penman 1989, JAE
13 % change in current ratio pchcurrat Ou & Penman 1989, JAE
14 Quick ratio quick Ou & Penman 1989, JAE
15 % change in quick ratio pchquick Ou & Penman 1989, JAE
16 Sales-to-cash salecash Ou & Penman 1989, JAE
17 Sales-to-receivables salerec Ou & Penman 1989, JAE
18 Sales-to-inventory saleinv Ou & Penman 1989, JAE
19 % change in sales-to-inventory pchsaleinv Ou & Penman 1989, JAE
20 Cash flow-to-debt cashdebt Ou & Penman 1989, JAE
21 Illiquidity (bid-ask spread) baspread Amihud & Mendelson 1989, JF
22 1-month momentum mom1m Jegadeesh 1990, JF
23 6-month momentum mom6m Jegadeesh & Titman 1990, JF
24 12-month momentum mom12m Jegadeesh 1990, JF
25 Depreciation-to-gross PP&E depr Holthausen & Larcker 1992, JAE
26 % change in depreciation-to-gross PP&E pchdepr Holthausen & Larcker 1992, JAE
15
Table 1 (excerpt)
29
# RPS Acronym Author(s) Date, Journal
75 Change in # analysts chnanalyst Scherbina 2007, WP
76 Asset growth agr Cooper, Gulen & Schill 2008, JF
77 Change in shares outstanding chcsho Pontiff & Woodgate 2008, JF
78 Industry-adjusted change in profit margin chpmia Soliman 2008, TAR
79 Industry-adjusted change in asset turnover chatoia Soliman 2008, TAR
80 3-day return around earnings
announcement
ear Kishore, Brandt, Santa-Clara & Venkatachalam 2008, WP
81 Revenue surprise rsup Kama 2009, JBFA
82 Cash flow volatility stdcf Huang 2009, JEF
83 Debt capacity-to-firm tangibility tang Hahn & Lee 2009, JF
84 Sin stock sin Hong & Kacperczyk 2009, JFE
85 Employee growth rate hire Bazdresch, Belo & Lin 2009, WP
86 Cash productivity cashpr Chandrashekar & Rao 2009, WP
87 ROA roaq Balakrishnan, Bartov & Faurel 2010, JAE
88 CAPEX and inventory invest Chen & Zhang 2010, JF
89 Real estate holdings realestate Tuzel 2010, RFS
90 Absolute accruals absacc Bandyopadhyay, Huang & Wirjanto 2010, WP
91 Accrual volatility stdacc Bandyopadhyay, Huang & Wirjanto 2010, WP
92 Change in tax expense chtx Thomas & Zhang 2010, WP
93 Maximum daily return in prior month maxret Bali, Cakici & Whitelaw 2011, JFE
94 Percent accruals pctacc Hafzalla, Lundholm & Van Winkle 2011, TAR
95 Cash holdings cash Palazzo 2012, JFE
96 Gross profitability gma Novy-Marx 2012, WP
97 Organizational capital orgcap Eisfeldt & Papanikolaou 2013, JF
98 Secured debt-to-total debt secured Valta 2013, WP
99 Secured debt indicator securedind Valta 2013, WP
100 Convertible debt indicator convind Valta 2013, WP
Table 2 (excerpt)
30
# Acronym RPS definition (annual figures are for most recent fiscal year prior to signal date)
1 beta Beta estimated from 3 years of weekly firm and EW market returns ending month t-1 (with at least 52 weeks of returns
available).
2 betasq Beta squared.
3 ep Annual income before extraordinary items (ib) divided by market cap at end of prior fiscal year.
4 mve Natural log of market cap at month-end immediately prior to signal date (prc*shrout).
5 dy Total dividends (dvt) divided by market cap at fiscal year end.
6 sue Unexpected quarterly earnings divided by market cap at end of most recent fiscal quarter. Unexpected earnings is IBES
actual earnings minus median forecasted earnings if available; else unexpected earnings is the seasonally differenced
quarterly earnings before extraordinary items from Compustat quarterly file.
7 chfeps Mean analyst forecast of annual EPS in month prior to fiscal period end date from IBES summary file minus same mean
forecast for prior fiscal period.
8 bm Book value of equity (ceq) divided by end of fiscal year end market cap.
9 mom36m 24-month cumulative return ending month t-13.
10 fgr5yr Most recently available analyst forecasted 5-year growth.
11 lev Total liabilities (lt) divided by fiscal year end market cap.
12 currat Current assets (act) divided by current liabilities (lct).
13 pchcurrat % change in currat.
14 quick Current assets (act) minus inventory (invt), divided by current liabilities (lct).
15 pchquick % change in quick.
16 salecash Annual sales (sale) divided by cash and cash equivalents (che).
17 salerec Annual sales (sale) divided by accounts receivable (rect).
18 saleinv Annual sales (sale) divided by total inventory (invt).
19 pchsaleinv % change in saleinv.
20 cashdebt Earnings before depreciation and extraordinary items (ib+dp) divided by avg total liabilities (lt).
21 baspread Monthly avg of daily bid-ask spread divided by avg of daily bid-ask spread.
22 mom1m Return in month t-1.
23 mom6m 5-month cumulative return ending month t-2.
24 mom12m 11-month cumulative returns ending month t-2.
16
Table 4
31
Panel A: Distribution of variance inflation factors (VIFs) from pooled time-series
cross-sectional regressions of 1-month ahead stock returns on 100 or 93 RPS
Before
removing
RPS with
VIF > 5
After
removing
RPS with
VIF > 5
# RPS 100 93
Min. 1.0 1.0
Median 1.2 1.1
Mean 3.4 1.3
max. 58.0 4.0
# VIFs > 5 14 0
• betasq • chempia
• currat • stdacc
• pchcurrat • maxret
• mom6m
RPS removed in reducing the
set of 100 RPS to 93 RPS
Panel B: Percentiles of absolute values of cross-correlations
Min.
1st
pctile
25th
pctile Median Mean
75th
pctile
99th
pctile Max.
0.00 0.00 0.00 0.01 0.03 0.03 0.29 0.99
0.00 0.00 0.00 0.01 0.03 0.03 0.27 0.68
N = 100 RPS
N = 93 RPS
Panel B: Percentiles of absolute values of cross-correlations
● Our main design choices are robust, in that in untabulated results, we observe that our general inferences about the multidimensionality of returns are unchanged if we instead use:
● Raw or standardized/normalized RPS ● Winsorized RPS ● Ranking based on all firms combined rather
than separately by size ● WLS vs. OLS ● Not infilling missing observations
32
Main design choices (cont’d)
17
● Conventional time-series of cross-sectional regns.
● We use three future return horizons that differ in how long they cumulate returns over, and how far ahead of the RPS measurement date they are when they begin to cumulate their returns.
33
Fama-MacBeth (1973) regressions
1-month
ahead
returns
13-36 month
ahead returns
Date RPS is
measured
2-12 months
ahead returns
● We use scaled decile ranks approach to defining RPS where each monthly re-measured RPS [0, 1]:
● SDR reduces the effects of outliers.
● SDR coefficients are returns to linear, optimal, pre-transactions costs, dollar-neutral hedge portfolios that are orthogonal to all other independent variables in the regression (Fama, 1976; Abarbanell & Bushee, 1998).
● We term the mean annualized long/short coefficient return estimates as ‘MALSRets’.
34
Scaled decile ranked RPS
18
● We use 3.0 as cutoff for inferring stat. significance.
● Harvey, Liu & Zhu (2013) criticize 1.96 RPS t-stat cutoff because it fails to take data snooping & data mining biases into account.
● “Whether suggested by theory or empirical work, a t-ratio of 2.0 is too low” (HLZ p. 2).
● “We argue that a newly discovered factor today should have a t-ratio that exceeds 3.0”
● Our use of 3.0 also means we don’t really have to adjust for expected number of significant coefs.
35
abs{t-stat} 3.0 cutoff
Table 5. FF92 + mom12m (1980-2012)
36
Panel A: 1 month ahead returns (396 non-overlapping monthly regressions)
RPS Pred. sign MALSRet t-stat.
mve - -3.8% -0.9
bm + 13.4% 3.5
mom12m + 6.1% 5.6
All firms
5,018
1.5%
Mean # obs. per regression Mean adjusted R2
RPS Pred. sign MALSRet t-stat. MALSRet t-stat. MALSRet t-stat.
mve - -1.1% -0.7 2.5% 1.5 -13.3% -3.9
bm + 4.2% 1.1 11.9% 2.8 14.1% 3.5
mom12m + 6.3% 4.4 6.8% 5.9 4.5% 3.3
Large-Cap Mid-Cap Small-Cap
996 1,997 2,019
0.7%2.3% 1.5%
Mean # obs. per regression Mean adjusted R2
19
Table 6 unidim vs. multidim (partial view)
RPS Pred. sign MALSRet t-stat. MALSRet t-stat. MALSRet t-stat. MALSRet t-stat.
1 beta + -3.9% -0.8 0.1% 0.1 -6.8% -1.7 -1.2% -0.6
2 betasq + -4.1% -0.8 -7.0% -1.8
3 ep + 6.6% 1.2 6.4% 4.3 9.8% 2.8 2.6% 1.2
4 mve - -6.8% -1.8 12.1% 3.8 1.5% 0.5 6.8% 7.3
5 dy + 0.9% 0.2 -2.6% -2.6 5.3% 1.2 -1.5% -1.8
6 sue + 18.4% 12.1 10.2% 13.2 1.1% 0.6 2.4% 3.2
7 chfeps + 10.8% 3.6 2.4% 1.2 0.2% 0.1 2.4% 0.6
8 bm + 15.1% 4.6 10.8% 5.4 11.3% 2.9 9.9% 4.7
9 mom36m - 8.7% 5.2 0.8% 1.9 -2.3% -1.8 -0.6% -1.2
10 fgr5yr - -0.9% -0.2 -4.2% -4.1 -1.0% -0.5 0.2% 0.9
11 lev + 6.6% 1.6 7.1% 3.3 7.1% 1.7 7.3% 4.9
All firms, 13-36 month ahead returns
Unidimensional MultidimensionalUnidimensional Multidimensional
All firms, 1 month ahead returns
# abs{t-stat} 1.96 48 47
37 26# abs{t-stat} 3.0 9 8
27 21
RPS Pred. sign MALSRet t-stat. MALSRet t-stat. MALSRet t-stat. MALSRet t-stat.
90 absacc - -0.6% -0.2 -2.1% -2.5 -4.4% -2.8 -0.8% -0.5
91 stdacc - -8.1% -3.0 -7.8% -4.2
92 chtx + 12.5% 9.7 3.3% 4.0 1.4% 1.0 1.0% 2.1
93 maxret - -8.8% -1.8 -6.4% -2.2
94 pctacc - -5.8% -3.3 -2.2% -1.5 -5.3% -5.5 -1.4% -1.9
95 cash + 5.1% 1.3 3.8% 3.5 -0.8% -0.3 1.8% 1.1
96 gma + 4.3% 2.5 2.4% 1.2 3.4% 2.6 -0.7% -0.4
97 orgcap + 8.7% 3.0 3.4% 2.2 3.2% 1.3 3.0% 1.5
98 secured + 1.6% 0.5 0.1% 0.0 0.9% 0.5 3.2% 1.6
99 securedind + 0.6% 0.3 0.9% 0.8 0.8% 1.1 -0.1% -0.2
100 convind + -5.0% -4.1 -3.1% -5.0 -2.7% -2.8 -1.4% -1.6
All firms, 13-36 month ahead returns
Unidimensional MultidimensionalUnidimensional Multidimensional
All firms, 1 month ahead returns
# abs{t-stat} 1.96 48 47
37 26
Mean # obs. per regression
Mean adjusted R2
Sharpe ratio on portfolio of RPS with abs{t-stat} 3
# abs{t-stat} 3.0
2.98 2.03
6,362
7.4%
5,035
0.4%
4,930
6.5%
9 8
5,048
0.4%
27 21
Table 6 unidim vs. multidim (partial view)
20
● Our missing value data resets add noise.
● Our RPS missing value resets add noise.
● We use mixture of pre- and post-publication data.
● Our monthly RPS remeasurement choice often distances us from methods in originating papers.
● Ditto for calendar alignment and return holding period.
● We use scaled decile ranked RPS, not unscaled RPS, or just extreme top/bottom RPS deciles.
39
We think 26 multidimensionally priced RPS is likely an underestimate:
40
Multidim pricing of famous RPS
RPS Pred. sign MALSRet t-stat. MALSRet t-stat.
1 beta + -3.9% -0.8 0.1% 0.1
3 ep + 6.6% 1.2 6.4% 4.3
4 mve - -6.8% -1.8 12.1% 3.8
6 sue + 18.4% 12.1 10.2% 13.2
7 chfeps + 10.8% 3.6 2.4% 1.2
8 bm + 15.1% 4.6 10.8% 5.4
9 mom36m - 8.7% 5.2 0.8% 1.9
11 lev + 6.6% 1.6 7.1% 3.3
22 mom1m - 4.2% 4.5 -1.4% -3.7
23 mom6m + 8.6% 5.4
24 mom12m + 9.8% 6.0 0.7% 1.5
36 acc - -8.0% -4.1 -3.3% -2.8
37 turn + 21.9% 9.1 25.2% 11.1
44 indmom + 25.8% 7.3 7.2% 6.0
45 ps + 5.6% 2.5 -1.3% -1.6
46 dolvol - 1.0% 1.5 -10.0% -10.1
49 sfe + -6.0% -1.4 -16.1% -8.6
68 retvol - -5.8% -1.1 -13.9% -6.7
74 aeavol + 7.7% 6.2 3.3% 6.1
76 agr - -15.8% -5.9 -6.9% -4.2
80 ear + 16.5% 16.6 9.4% 13.8
87 roaq + 13.6% 3.3 10.8% 7.7
100 convind + -5.0% -4.1 -3.1% -5.0
Unidimensional Multidimensional
All firms, 1 month ahead returns
21
41
Top 10 t-stats (1-month ahead returns)
# RPS Pred. sign MALSRet t-stat. MALSRet t-stat.
1 ear + 16.5% 16.6 9.4% 13.8
2 sue + 18.4% 12.1 10.2% 13.2
3 turn + 21.9% 9.1 25.2% 11.1
4 dolvol - 1.0% 1.5 -10.0% -10.1
5 sfe + -6.0% -1.4 -16.1% -8.6
6 roaq + 13.6% 3.3 10.8% 7.7
7 rsup + 7.9% 3.5 7.4% 7.5
8 retvol - -5.8% -1.1 -13.9% -6.7
9 aeavol + 7.7% 6.2 3.3% 6.1
10 saleinv + 3.1% 2.2 5.4% 5.1
11 bm + 15.1% 4.6 10.8% 5.4
20 mve - -6.8% -1.8 12.1% 3.8
56 mom12m + 9.8% 6.0 0.8% 1.9
Unidimensional Multidimensional
All firms, 1 month ahead returns10 largest multidimensional t-stats
42
Table 7 – Additional multidim results
RET horizon 1980-2012 1980-1989 1990-1999 2000-2012
1 month 47, 26 24, 15 32, 19 36, 22
Estimation decade
Panel A: By decade: Number of multidimensional RPS abs{t-stat} 3.0, 1.96
Panel D: By firm size: Mean FM multidimensional RPS regression adj. R 2
RET horizon All firms Large-Cap Mid-Cap Small-Cap
1 month 6.5% 16.8% 8.7% 4.4%
2-12 months 9.0% 21.0% 12.6% 7.1%
13-36 months 7.4% 16.5% 10.4% 4.2%
Panel C: By firm size: Number of multidimensional RPS abs{t-stat} 3.0, 1.96
RET horizon All firms Large-Cap Mid-Cap Small-Cap
1 month 47, 26 16, 5 35, 17 34, 21
2-12 months 30, 13 13, 5 23, 6 22, 4
13-36 months 21, 8 15, 4 23, 11 18, 5
22
Fig. 2A – MALSRet attenuation Multidim = 0.34 x unidim
43
Unidimensional MALSRet
Multidimensional MALSRet
44
-15
-10
-5
0
5
10
15
-10 -5 0 5 10 15 20
1-month ahead returns 2-12 month ahead returns 13-36 months ahead returns
Fig. 2A – t-stat{MALSRet} attenuation Multidim = [0.59, 0.38] x unidim
Multidimensional t-stat{MALSRet}
Unidimensional t-stat{MALSRet}
23
45
Fig. 3 – MALSRet attenuation by pub Post-pub = 0.72 x pre-pub Post-pub = 0.63 x pre-pub
● A hard nut to crack – we offer one perspective.
● We hypothesize that mispricing is more likely to manifest in short-length near-term returns, while the rational pricing of economic risks is more likely to be seen in longer-length further-away returns.
46
What might multidimensionality say about rational pricing vs. mispricing?
1-month
ahead
returns
13-36 month
ahead returns
Date RPS is
measured
2-12 months
ahead returns
24
Table 9 – Our estimation of the RPS that likely capture economic risks vs.
RPS that likely reflect mispricing
47
Rational pricing of economic risks Mispricing
sue #6 pchcapx_ia #40 dy #5 turn #37 fgr5yr #10 chfeps #7
bm #8 ill #62 ep #3 dolvol #46 mom1m #22 pchsaleinv #19
lev #11 mve #4 saleinv #18 aeavol #74 nincr #43 mve_ia #27
sfe #49 agr #76 indmom #44 nanalyst #50
retvol #68 ear #80 std_dolvol #47 tb #57
rsup #81 std_turn #48 lgr #60
roaq #87 idiovol #53 ms #64
convind #100 zerotrade #71 stdcf #82
chtx #92 absacc #90
cash #95
MispricingRational pricing of economic risks
Next most likely Most likelyMost likely Next most likely Next most likely Next most likely
48
Implications that we see
1. The 26-dimensionality of U.S. stock returns is likely understated, not overstated.
2. Controlling for only firm size, book-to-market and 12-month mom is low powered, maybe also biased.
3. NPV of investing in understanding why returns are so multidimensional >> NPV of finding another RPS.
4. How much the estimation of dimensionality has changed over 20+ years means that it is reasonable to take the inferences/conclusions of our study, and future studies, more cautiously than ever.
25
49
Conclusions & future work
We are the first to state that our paper raises more questions than it answers. For example:
Why are returns so multidimensional?
Exactly which RPS reflect rational economic pricing, and which reflect mispricing, and why?
If there really are close to 20 RPS that each reflects mispricing, where does such diverse mispricing come from in the first place?
Do multidimensioned factors work better or worse than RPS?
Do U.S. multidimensioned RPS price outside U.S. and non-equity?
How well do multidimensioned RPS explain the cross-section of higher moments of returns, and why?
Do our results mean that equity portfolio managers have been mis-evaluated over the past 30 years?
50
This concludes my talk. Thank you for inviting me.
I look forward to hearing your views.