On the Conjoint Nature of Value and Profitability *

Sunil Wahal
WP Carey School of Business
Arizona State University
Tempe, AZ 85287
[email protected]

Eduardo Repetto
Avantis Investors
[email protected]

June 2020

Abstract

Novy-Marx (2013, 2014) argues that profitability and value are philosophically and economically
related: buying highly productive firms at average prices is similar to buying average productivity
firms at low prices. We investigate the risk and return of portfolios that hold the entire market but
tilt towards the joint distribution of stocks that rank highly on both value and profitability. Over
1940-2019, such “tilted market portfolios” generate substantially higher returns than the pure
market portfolio. Even in periods where value has delivered weak returns (2000-2019), tilted
market portfolios offer attractive risk-reward ratios. For investors with long horizons,
bootstrapped simulations of up to 30-year holding periods indicate that the entire distribution shifts
further to the right, generating better outcomes for investors. We conclude that benefits to long-
only investors come from targeting value and profitability jointly, rather than running them side-
by-side or sprinkling one with the other.

*
Wahal thanks the Center for Investment Engineering at ASU for financial support. Wahal is a consultant to Avantis
Investors. Avantis did not provide data or funding for this research. We thank Hank Bessembinder for helpful
comments.

Electronic copy available at: https://ssrn.com/abstract=3635714

1. Introduction
Novy-Marx (2013) argues that profitability is the other side of value in the sense that
purchasing highly profitable assets at average prices is akin to purchasing average profitability
assets at low prices. This perspective matches the Miller and Modigliani (1961) present value
identity because expected profitability, book value, and prices are joint drivers of expected returns.
Novy-Marx (2014) pushes the idea further, showing that selecting stocks based on value and
profitability together yields higher returns than investing in these strategy side-by-side. The
importance of the joint distribution is also evident in Fama and French (2015) who build factors
that account for the correlation structure between value and profitability.1
In this paper, we take the above as an empirical regularity and turn to a more practical
problem: how does an investor use the above economic intuition and results? There are a myriad
of ways in which one could solve what is essentially a portfolio engineering problem. The optimal
solution would vary with the constraints faced by investors. For example, the portfolio solution
for investors not subject to leverage constraints and with the ability to short securities is likely to
be very different than for investors who are bound by these considerations. To focus the analysis,
we consider the perspective of an unlevered long-only investor who wants to use the above science
to improve portfolio outcomes.
We construct portfolios which hold the entire market but tilt systematically in favor of
securities that rank highly on both value and profitability metrics. Using the joint distribution is
conceptually quite different from investing in a value portfolio with a profitability screen, or vice
versa. It is also different from investing in a value portfolio and a high relative profitability
portfolio simultaneously.2 This is because value stocks need not be high profitability stocks and
vice-versa, and/or because the weights ascribed to stocks in a value-only or profitability-only
portfolio are not the same as weights in a joint value and profitability portfolio. The average
monthly returns to tilted market portfolios are higher than a pure market portfolio over the 1940-
2019 period with a small reduction in volatility. The performance improvements are as much as

1
The Miller and Modigliani (1961) model also requires controlling for expected future investment. However, holding
investment constant is difficult for two reasons. First, because investment is a lot less persistent than profitability, it
is difficult to obtain good measures of expected investment. Second, as a practical matter, portfolios formed at the
intersection of value, profitability and investment quickly become ill-diversified.
2
This is sometimes referred to as fund-of-funds solution, which often layers fees on fees.

Electronic copy available at: https://ssrn.com/abstract=3635714

10 basis points per month with only modest tilts and a correlation with the market of 0.99. Even
in periods where value has underperformed growth (2000-2019), tilted portfolios outperform
market portfolios by similar magnitudes. They do so by providing exposure to value and
profitability, as evidenced by factor loadings on three- and five-factor models.
We perform bootstrap simulations of market and tilted market portfolios over a variety of
longer holding periods (1, 3, 5, 10, 20, and 30 years). This exercise helps clarify the distribution
of payoffs to investors over realistic investment horizons. Two facts stand out. First, the
percentage of holding periods with negative payoffs is lower for tilted portfolios than for the
market. Second, the distribution of payoffs shifts slightly further to the right. The bottom line is
that the distribution of outcomes for long-only investors with longer horizons are significantly
improved.
To the asset pricing cognoscenti, there may be nothing surprising here – the outcomes we
observe are precisely what one would expect given the value and profitability premiums. The
interesting part comes from recognizing that the joint distribution matters. From a portfolio
perspective, the devil is in the engineering details: recognizing that value and profitability are
linked allows one to target the premiums more efficiently while being cognizant of trading costs,
portfolio capacity, and other such issues of practical import. The results suggest that long-only
investors can benefit from properly designed portfolios, especially if their horizons are long.
The remainder of the paper is as follows. Section 2 outlines the data and portfolio
construction process. Section 3 describes the tilted portfolios, as well as their returns and
exposures. It also contains results from the bootstrap simulations. Section 4 concludes.

2. Data and Portfolio Construction

There are two approaches to understanding the conjoint nature of value and profitability.
One could target the joint exposure at the individual security level with z-scores that embed
measures of value and profitability, while recognizing the correlation structure between the two
(see, for example, Appendix B in Novy-Marx (2014)). Alternatively, one could use portfolios
formed at the intersection of value and profitability as the building blocks. Since it is somewhat
harder to see tradeoffs between value and profitability at the individual security level, we use the

Electronic copy available at: https://ssrn.com/abstract=3635714

portfolio approach instead. An added advantage is that the weighting matrix of tilted portfolios
constructed from other portfolios is easy to illustrate.
The building block consists of 4x4 grids of value and profitability portfolios.3 As in Fama
and French (2015), we use book-to-market ratios and operating profitability to construct portfolios.
We use independent sorts on value and profitability, forming portfolios at the end of each June and
rebalancing annually. All portfolio returns are value-weighted. Since measurement of expected
returns requires the longest possible time series, we use data from Wahal (2019) to generate a time
series of portfolio returns dating back to 1940.
Table 1 shows the grid, with each cell containing the average market capitalization weight
over the entire time series. There are two important observations to be made from the grid. First,
the diagonal of the matrix (colored in yellow), contains 46.2 percent of the aggregate market
capitalization. We regard these portfolios as “neutral” in the sense of Fama and French (2015)
factors. Second, growth and low profitability portfolios corresponding to the northwest quadrant
(left of the diagonal and colored in blue) represent 35.5 percent of the aggregate market
capitalization. Under the joint metric, these are lower expected return portfolios. In contrast, value
and higher profitability portfolios in the southeast quadrant (right of the diagonal and colored in
green) contain 18.3 percent of the aggregate market capitalization.4 The highest expected return
portfolio (high value and high profitability) contains only 0.8 percent of the aggregate market
capitalization. The implication is obvious and not surprising – it is difficult to find securities that
allow an investor to purchase highly profitable assets at bargain prices. In colloquial terms, there
are not a ton of steals out there.

3. Tilted Portfolios
3.1 Portfolio Construction
We employ a simple approach to constructing market portfolios that tilt towards the joint
distribution of value and profitability. Since the majority of risk that long-only investors bear

3
We use 4x4 portfolio grids rather than finer gradations (e.g. 5x5) to ensure that portfolios remain well-diversified,
particularly early in the time series.
4
In contrast, pure value portfolios (the bottom row of Table 1) only contain about 10 percent of the aggregate market
capitalization. From a portfolio engineering perspective, this makes investing in pure value more capacity constrained
than value and profitability.

Electronic copy available at: https://ssrn.com/abstract=3635714

comes from the market, we maintain market capitalization weights in the diagonal portfolios. All
tilts are therefore generated by overweighting or underweighting portfolios in the northwest and
southeast quadrants that correspond to low and high expected return securities respectively.
Holding the diagonal weights constant (at 46.2 percent), it is first useful to see how the remainder
is distributed absent tilts. Roughly speaking, about 35 percent of the remaining market
capitalization resides in high expected return portfolios, while the remaining 65 percent is in low
expected return portfolios.5
Given these statistics, we create a sequence of tilted market portfolios in which weights to
high expected return portfolios increase by increments of 10 percentage points. We start with a
“40-60” tilt in which 40 percent of the weight is allocated to the high expected return portfolios.
In the 50-50 portfolio, therefore, the off diagonal market capitalization weights are equally
distributed between the two quadrants. Once these tilts are put into place, a remaining issue is how
to allocate weights to the building block portfolios within each quadrant. We use rescaled market
capitalization weights to perform this allocation. For example, in a 50-50 portfolio where 50
percent of the off-diagonal weight is allocated to high expected return securities (in green), the
weights for each individual portfolio are based on its (rescaled) market capitalization weight. Such
an approach has important advantages. It ensures that differences in returns are not driven by
enormous allocations to portfolios that would be infeasible. As an extreme example, suppose that
all 50 percent of the allocation to high expected return portfolios was invested in the highest value
and profitability portfolio. Recall that this portfolio only represents 0.8 percent of the aggregate
market capitalization. Such an allocation would be quite expensive to trade because most of these
firms are small or microcap. Moreover, the allocation would have very limited portfolio capacity
because individual security weights would be prohibitively large. Using rescaled market
capitalization weights, on the other hand, causes individual security weights to not be too large
and portfolio capacity to remain quite high.

5
In Table 1, the sum of the market capitalization weights of low expected return portfolios (in blue) is 35.5 percent.
Dividing that by sum of market capitalization weights of the off diagonal (53.8 percent) is 65 percent, implying that
the high expected return portfolios (in green) have a weight of 35 percent. It is also important to note that these are
time-series averages of weights. These percentages are different in any given year or month.

Electronic copy available at: https://ssrn.com/abstract=3635714

3.2 Portfolio Returns
Table 2 contains average monthly returns and standard deviations of the tilted portfolios.
As a benchmark, we also report these statistics for the market portfolio.6 For each tilted portfolio,
we include t-statistics for the return difference between the tilted portfolio and the market, as well
as the correlation with the market return.
Panel A reports the above statistics for the longest possible sample period, 1940-2019. The
average monthly market return over this period is 0.99 percent, with a standard deviation of 4.19.
Tilted portfolios generate slightly higher returns relative to the market, increasing monotonically
as the tilt increases. For instance, in the 60-40 tilted portfolio, average returns are higher by 10
basis points with a similar standard deviation. The t-statistic for the difference in returns is 5.31.
Also notable, and perhaps unsurprising, is the fact that correlations with the market portfolio do
not change by large magnitudes: the 60-40 tilted portfolio still has a correlation of 0.99 with the
market.
Panel B reports equivalent statistics for 2000-2019, a period in which value stocks have
underperformed growth stocks. Allocating jointly to value and profitability continues to generate
higher returns than the market portfolio. Focusing again on the 60-40 tilted portfolio, it earns 0.70
percent per month compared to 0.61 percent for the market portfolio, with a t-statistic of 2.52. The
hedging value of profitability relative to value is also apparent. In periods where value
underperform, the hedging effect of profitability against value lowers the standard deviation of the
tilted portfolio: the standard deviation of the 60-40 tilted portfolio is 4.20 compared to 4.32 to for
the market portfolio.

3.2 Portfolio Exposures

To provide information on exposures and the source of excess returns, we estimate three-
and five-factor regressions for the tilted portfolios over the entire sample period. Panel A of Table
3 reports coefficients from three-factor regressions with t-statistics in parentheses. Three results

6
The market portfolio return will not be precisely equal to the weighted average returns of the portfolios in the 4x4
grids because constructing value and profitability portfolios imposes specific requirements on accounting data
required to construct underlying portfolios (see Fama and French (2015) for details). We also compute a pseudo-
market portfolio using weighted average returns of the constituent 4x4 portfolios. The gist of the results in the paper
are unchanged.

Electronic copy available at: https://ssrn.com/abstract=3635714

stand out. First, market betas are 1.00 and virtually identical across the portfolios. Second, as
expected, factor loadings on HML are reliably positive. Tilted portfolios provide exposure to
value, increasing monotonically with the degree of the tilt. Second, intercepts relative to the three-
factor model are positive and statistically significant. The magnitude ranges from 4 basis points
per month in the 40-60 portfolio to 6 basis points per month in the 90-10 portfolio. Tilting towards
the joint distribution of value and profitability has clear advantages relative to “pure” value
portfolios.
Panel B shows contains results from five-factor regressions. In addition to positive factor
loadings on HML, the loadings on RMW are also positive and increase with the tilt of the portfolio.
Unsurprisingly, introducing RMW on the right hand side of the regression reduces the intercepts,
rendering them statistically indistinguishable from zero. In other words, the tilts provide for more
finely tuned exposure to priced risk and the returns in Table 1 are compensation for that risk.
The story in Tables 1 and 2 is precisely as one would expect. Joint exposure to value and
profitability provides a meaningful improvement to the market portfolio for long-only investors.
Targeting value and profitability jointly in a market portfolio provides exposure to both, but more
importantly, provides a premium relative to targeting only value. The key here is that the two
metrics are integrated, not treated separately and not with one treated as “primary” and the other
“secondary”.

3.3 Long-Horizon Returns

For long-only investors, issues related to investment horizon are important. In particular,
investors are less concerned with monthly returns than with payoffs over their holding period,
which may be years or even decades. Unfortunately, it is common among practitioner “pitch
books” that tout investment products to present long-horizon return statistics to a strategy with
overlapping periods. For example, one might show returns to a strategy over a 10 year period,
moving forward one month at a time. The sales pitch then involves showing that over decades,
the strategy underperforms only X percent of the time. Such an approach builds in dependence
between successive holding periods and misleads investors. Absent this mechanical dependence,

Electronic copy available at: https://ssrn.com/abstract=3635714

the true distribution of outcomes could be much more variable, generating ex post disappointment
in investment results.
To generate a better representation of long-horizon returns to tilted portfolios, we perform
a simple bootstraps exercise. Bootstrapping (including the block bootstrap) has a long history in
academic finance and is used to extract information about true underlying distributions. A recent
incarnation is that of Fama and French (2018) who use bootstraps to examine the distribution of
long horizon market returns. We adapt their approach as follows.
Since long-only investors are concerned with payoffs, we compute payoffs as
𝑇

1 + 𝑅𝑇 = ∏(1 + 𝑟𝑡 )
𝑡=1

where rt is the simple return for month t, and the gross return is compounded for T months to
generate the payoff 1+RT. We use six horizons (T), corresponding to 1 year, 3 year, 5, year, 10
year, 20 year and 30 year payoffs. The simulation samples with replacement from the 954 monthly
returns between July 1940 and December 2019, assuming independent and identically distributed
returns.7 We use 100,000 such samples (“replicates”), drawing the requisite number of returns for
each horizon. For example, to compute annual payoffs to the market portfolio, we construct
100,000 samples of 12 randomly drawn monthly returns for the market portfolio. For 30 year
payoffs, we randomly draw 360 monthly returns to compute payoffs. The process for tilted
portfolios is identical except that the draws are from the sample of monthly returns to the tilted
portfolios.
The results of this exercise appear in Table 4. Each panel corresponds to a particular
horizon with the top row showing payoffs for the market portfolio, followed by the tilted market
portfolios. The table shows average payoffs, the standard deviation, skewness, the percentage of
the payoffs that are negative (% Neg.), and various percentiles of the distribution.
The basic statistics for the market portfolio across holding periods mirror those of Fama
and French (2018, Table 3) with some minor variation because their sample period is only 1963-

7
One might complain about heteroscedasticity and shifts in volatility over this period. An appropriately modified
block bootstrap could accommodate this concern. We elect to use the simple bootstrap because much of the volatility
change in market returns occurs between 1926 and the 2 nd World War, before the start of the sample period, and
because the block bootstrap requires one to explicitly model changes in the variance structure.

Electronic copy available at: https://ssrn.com/abstract=3635714

2016. The more interesting results reside in differences between the pure market portfolio and the
tilted portfolios at longer investment horizons. The reader can obviously inspect detailed results
from the table so for illustrative purposes, we focus on comparing the 60-40 tilted portfolio to the
market for 3-year and 30-year holding periods.
For a 3-year holding period, the average payoff from the 60-40 tilted portfolio is 1.48
compared to 1.43 for the market portfolio. The standard deviation is only a touch higher (0.37
versus 0.36). More interestingly, the percentage of negative payoffs declines from 10.02 percent
for the market to 7.79 percent for the 60-40 portfolio. This shift in the distribution to the right is
also apparent from the percentiles. For example, the 5th and 95th percentiles for the market
portfolio are 0.90 and 2.07, whereas for the 60-40 portfolio they are 0.94 and 2.17 respectively.
At the 30-year horizon, the distribution of the payoffs is even more interesting. The
average payoff of the 60-40 portfolio is 49.48 compared to 34.82 for the market. But given the
variability in the factor premiums, the standard deviation of the tilted portfolios payoffs are quite
a bit higher (45.40 compared to 32.15 for the market). The rightward shift of the distributions is
also apparent. The 5th percentile for the 60-40 portfolio is a payoff of 9.73, compared to 6.77 for
the market, and the equivalent 95th percentiles are 132.40 and 93.26. The upshot is that the joint
distribution of value and profitability has clear benefits over longer horizons.

4. Conclusions
Value and profitability are related. Pure value portfolios do not deliver the profitability
premium and vice versa. Combining independently constructed value and profitability portfolios
is not the same thing as forming portfolios based on the joint distribution of value and profitability.
Targeting the joint distribution has clear benefits for long-only investors, especially those with
long investment horizons.

Electronic copy available at: https://ssrn.com/abstract=3635714

References

Fama, Eugene and Ken French, 2015, A five-factor asset pricing model, Journal of Financial
Economics 16, 1-22
Fama, Eugene and Ken French, 2018, Long-horizon returns, Review of Asset Pricing Studies 8,
232-252.
Miller, M.H. , Modigliani, F. , 1961. Dividend policy, growth, and the valuation of shares. Journal
of Business. 34, 411–433.
Novy-Marx, 2013, The other side of value: The gross profitability premium, Journal of Financial
Economics 108, 1-28.
Novy-Marx, 2014, Quality Investing, working paper, University of Rochester.
Wahal, Sunil, 2019, The profitability and investment premium: Pre-1963 evidence, Journal of
Financial Economics 131, 362-377.

Electronic copy available at: https://ssrn.com/abstract=3635714

Table 1

Average market capitalization weights of value and profitability portfolios, 1940-2019

Value portfolios are formed based on book-to-market ratios in June of each year following the procedures in Fama
and French (2015). Profitability portfolios are similarly formed based on operating profitability. Both portfolio
series are supplemented with data from Wahal (2019). The table shows the average market capitalization weights
in the 4x4 grid of value and profitability. The diagonal elements (in yellow) represent “neutral” portfolios.
Portfolios in the northwest quadrant (in blue) are referred to the text as low expected return portfolios. Portfolios
in the southeast quadrant (in green) are referred to as high expected return portfolios.

Profitability
Low Prof. 2 3 High Prof.
Growth 2.3 5.3 13.7 26.0
Value

2 2.7 7.7 9.4 5.6

3 3.8 6.4 4.7 1.9
Value 4.4 3.7 1.6 0.8

10

Electronic copy available at: https://ssrn.com/abstract=3635714

Table 2

Monthly returns for tilted market-wide portfolios

The table shows average monthly returns and standard deviations of tilted market-wide portfolios. Tilted portfolios
are created using the value-profitability grid in Table 1. Each tilted portfolio holds the diagonal portfolios (in yellow)
in Table 1 at market capitalization weights. Portfolio tilts are applied to remainder of the portfolio: the northwest
quadrant (in blue) and southwest quadrant (in green). For example, in the 40-60 portfolio, 40 percent of the remainder
of the market capitalization is allocated to the high expected return portfolios in blue, and 60 percent is allocated to
the low expected return portfolios in green. A similar procedure applies for other tilts. The market portfolio is the
CRSP value-weighted market index. T-statistics are for the difference in returns between the average tilted portfolio
and the market.
Tilted Market-Wide Portfolios
Market 40-60 50-50 60-40 70-30 80-20 90-10
Panel A: 1940-2019
Avg. Return 0.99 1.05 1.07 1.09 1.11 1.13 1.15
Std. Dev 4.19 4.17 4.17 4.17 4.18 4.20 4.22
T-stat. (Tilt-Mkt) - 4.84 5.25 5.31 5.26 5.19 5.11
Corr. (Tilt, Mkt) - 1.00 0.99 0.99 0.99 0.98 0.97
Panel B: 2000-2019
Avg. Return 0.61 0.72 0.73 0.70 0.75 0.77 0.78
Std. Dev 4.32 4.22 4.22 4.20 4.23 4.25 4.28
T-stat. (Tilt-Mkt) - 3.19 2.81 2.52 2.30 2.13 2.01
Corr. (Tilt, Mkt) - 1.00 0.99 0.99 0.99 0.98 0.97

11

Electronic copy available at: https://ssrn.com/abstract=3635714

Table 3

Three- and Five-Factor regression for tilted market-wide portfolios

Tilted market-wide portfolios are constructed as in Table 2. The table shows factor coefficients and t-statistics (in
parentheses) from three- and five-factor regressions over the 1940-2019 sample period. The factors are from Ken
French’s website, with the exception of RMW and CMA for the 1940-1963 period which are based on Wahal
(2019).
Tilted Market-wide Portfolios
40-60 50-50 60-40 70-30 80-20 90-10
Panel A: Three-Factor Regressions
Intercept 0.04 0.05 0.05 0.05 0.06 0.06
(3.92) (3.76) (3.50) (3.24) (3.01) (2.81)
Rm-Rf 1.00 1.00 1.00 1.01 1.01 1.01
(382.84) (7.61) (290.51) (250.30) (217.48) (191.08)
SMB -0.04 -0.03 -0.03 -0.03 -0.03 -0.03
(9.17) (21.36) (-6.16) (4.96) (4.00) (3.25)
HML 0.05 0.10 0.14 0.18 0.22 0.26
(13.80) (21.36) (26.34) (29.50) (31.53) (32.88)
Panel B: Five-Factor Regressions
Intercept 0.01 0.01 0.01 0.00 0.00 0.00
(1.14) (0.77) ().43) (0.16) (0.05) (0.20)
Rm-Rf 1.01 1.01 1.01 1.01 1.01 1.01
(401.35) (368.27) (324.15) (281.35) (244.62) (214.44)
SMB -0.01 0.00 0.01 0.01 0.02 0.03
(2.75) (0.49) (1.36) (2.73) (3.71) (4.43)
HML 0.06 0.10 0.15 0.19 0.24 0.28
(12.93) (20.60) (25.80) (29.03) (31.00) (32.21)
RMW 0.09 0.11 0.13 0.15 0.17 0.19
(17.18) (19.63) (20.67) (20.88) (20.70) (20.37)
CMA 0.01 0.01 0.00 0.00 -0.01 -0.01
(1.40) (0.77) (0.22) (0.20) (0.51) (0.75)

12

Electronic copy available at: https://ssrn.com/abstract=3635714

Table 4

Distribution of long-horizon payoffs.

The table shows the distribution of long-horizon payoffs generated from a bootstrap procedure in which 100,000
samples are generated with replacement from the time series of monthly portfolio returns from the 1940-2019 period.
%Neg. shows the percentage of simulation outcomes with negative payoffs for each horizon.
Percentiles
Mean Std. Skew % Neg. P1 P5 P10 P50 P90 P95 P99
Dev
Panel A: 1-Year
MKT 1.13 0.16 0.24 21.95 0.77 0.87 0.92 1.12 1.34 1.40 1.53
40-60 1.13 0.16 0.25 20.33 0.78 0.88 0.93 1.13 1.34 1.41 1.54
50-50 1.14 0.16 0.25 19.90 0.79 0.88 0.93 1.13 1.35 1.42 1.55
60-40 1.14 0.16 0.26 19.49 0.79 0.88 0.94 1.13 1.35 1.42 1.55
70-30 1.14 0.16 0.26 19.09 0.79 0.89 0.94 1.14 1.36 1.43 1.56
80-20 1.15 0.16 0.26 18.76 0.79 0.89 0.94 1.14 1.36 1.43 1.56
90-10 1.15 0.17 0.27 18.43 0.79 0.89 0.94 1.14 1.36 1.43 1.57
Panel B: 3-Year
MKT 1.43 0.36 0.65 10.02 0.75 0.90 1.00 1.39 1.90 2.07 2.44
40-60 1.46 0.37 0.65 8.55 0.77 0.93 1.02 1.42 1.94 2.11 2.48
50-50 1.47 0.37 0.65 8.16 0.77 0.94 1.03 1.43 1.95 2.13 2.51
60-40 1.48 0.37 0.65 7.79 0.78 0.94 1.04 1.44 1.97 2.14 2.52
70-30 1.49 0.37 0.66 7.42 0.78 0.95 1.04 1.45 1.98 2.16 2.55
80-20 1.50 0.38 0.66 7.15 0.79 0.95 1.05 1.46 2.00 2.18 2.57
90-10 1.51 0.38 0.67 6.88 0.79 0.96 1.06 1.47 2.02 2.20 2.59
Panel C: 5-Year
MKT 1.81 0.59 0.92 5.14 0.79 1.00 1.13 1.72 2.59 2.91 3.57
40-60 1.87 0.61 0.92 4.06 0.82 1.04 1.17 1.79 2.67 3.00 3.69
50-50 1.89 0.62 0.92 3.75 0.83 1.05 1.19 1.81 2.71 3.04 3.73
60-40 1.92 0.63 0.91 3.51 0.84 1.06 1.20 1.83 2.74 3.08 3.79
70-30 1.94 0.63 0.92 3.26 0.85 1.07 1.22 1.85 2.78 3.12 3.84
80-20 1.96 0.64 0.92 3.03 0.85 1.08 1.23 1.87 2.81 3.16 3.89
90-10 1.99 0.66 0.92 2.86 0.86 1.09 1.24 1.89 2.85 3.20 3.95
Panel D: 10-Year
MKT 3.26 1.55 1.40 1.12 0.98 1.37 1.63 2.95 5.27 6.19 8.33
40-60 3.49 1.65 1.40 0.71 1.06 1.47 1.75 3.16 5.65 6.61 8.92
50-50 3.58 1.69 1.40 0.60 1.09 1.51 1.80 3.24 5.79 6.77 9.15
60-40 3.66 1.73 1.40 0.53 1.11 1.54 1.84 3.32 5.92 6.93 9.38
70-30 3.75 1.78 1.41 0.47 1.14 1.58 1.88 3.40 6.07 7.10 9.62
80-20 3.84 1.83 1.41 0.41 1.16 1.61 1.92 3.48 6.22 7.29 9.88
90-10 3.94 1.88 1.42 0.38 1.18 1.65 1.96 3.56 6.38 7.49 10.17
Panel E: 20-Year
MKT 10.65 7.53 2.35 0.06 1.85 2.95 3.75 8.70 19.75 24.80 37.77
40-60 12.24 8.62 2.37 0.03 2.16 3.42 4.35 10.02 22.63 28.49 43.41
50-50 12.84 9.04 2.37 0.02 2.26 3.58 4.55 10.51 23.74 29.89 45.31
60-40 13.47 9.49 2.38 0.02 2.37 3.74 4.76 11.02 24.90 31.33 47.58
70-30 14.14 9.99 2.38 0.01 2.47 3.92 4.99 11.56 26.18 32.96 49.98
80-20 14.82 10.52 2.40 0.01 2.57 4.09 5.21 12.10 27.51 34.62 52.48
90-10 15.56 11.10 2.41 0.01 2.67 4.26 5.44 12.68 28.92 36.36 55.22
Panel F: 30-Year
MKT 34.82 32.15 3.36 0.00 3.88 6.77 9.10 25.52 70.17 93.26 158.32
40-60 42.90 39.34 3.36 0.00 4.87 8.46 11.30 31.55 86.29 114.58 193.26
50-50 46.08 42.23 3.35 0.00 5.24 9.07 12.16 33.93 92.65 123.20 207.60
60-40 49.48 45.40 3.36 0.00 5.61 9.73 13.05 36.45 99.57 132.40 222.93
70-30 53.22 48.95 3.37 0.00 5.98 10.42 13.98 39.16 107.08 142.55 241.64
80-20 57.14 52.78 3.39 0.00 6.35 11.12 14.95 41.97 114.88 153.19 259.96
90-10 61.43 57.07 3.41 0.00 6.76 11.87 15.97 44.98 123.68 165.39 280.26

13

Electronic copy available at: https://ssrn.com/abstract=3635714