First draft

Comments are welcome

Stock Prices, Inflation and Stock Returns

Predictability

Christophe Boucher♦

CEPN
Université Paris-Nord

Version : 16 December 2004

♦
Correspondance : CEPN, Université de Paris Nord, UFR Sciences Economiques et de Gestion, 99 Avenue J. B.
Clément, 93430 Villetaneuse, France. E-mail : [email protected]

1
 Abstract

This paper considers a new perspective on the relationship between stock prices and inflation, by estimating the
common long-term trend in real stock prices, as reflected in the earning-price ratio, and both expected and realized
inflation. We study the role of the transitory deviations from the common trend in the earning-price ratio and realized
inflation for predicting stock market fluctuations. In particular, we find that these deviations exhibit substantial in-
sample and out-of-sample forecasting abilities for both real stock returns and excess returns. Moreover, we find that
this variable provides information about future stock returns at short and intermediate horizons that is not captured
by other popular forecasting variables.

Keywords: Time-Varying Expected Returns, Stock Return Predictability, Stock Return-Inflation puzzle.
JEL Classification : G12, G14, E44, C53.

2
1. Introduction

There now exists a large literature documenting the predictability of stock returns from
past information. Researchers have identified a number of financial variables that are useful in
predicting future stock returns. These include the dividend-price ratio (Rozeff, 1984; Campbell
and Shiller, 1988a; Fama and French, 1988; Hodrick, 1992), the price-earning ratio (Campbell
and Shiller, 1988b, 1998), the book-to-market ratio (Kothari and Shanken, 1997; Pontiff and
Schall, 1998; Lewellen, 2004), the dividend payout ratio (Lamont, 1998), the term and default
spreads on bonds (Keim and Stambaugh, 1986; Campbell, 1987; Fama and French, 1989), recent
changes in short-term interest rates (Campbell, 1987; Hodrick, 1992; Ang and Bekaert, 2004), the
equity share in total new equity and debt issues (Baker and Wurgler, 2000), the level of
consumption relative to income and wealth (Lettau and Ludvigson, 2001) and the aggregate stock
market volatility in conjunction with these consumption-asset-labor deviations (Guo, 2006).
Many of these variables are related to the stage of the business cycle (Fama and French, 1989;
Lettau and Ludvigson, 2001). Typically, expected returns and business conditions move in
opposite directions.
In this paper, we provide new evidence of the in-sample and out-of-sample predictability
of stock returns. We find that the transitory deviations from the common trend in the earning-
price ratio and inflation provide useful information for predicting both real stock returns and
excess returns.
Dividends, earnings (or multiyear backward moving averages of earnings), book value are
traditionally used to normalize stock prices. As noted by Lamont (1998), the important variable is
the level of stock prices which predicts future returns because stock prices are presumed mean-
reverting, even though the persistence of valuation ratios implies that such restorations took
many years to take shape. Indeed, Fama and French (1988), Campbell and Shiller (1988a,b),
Valkanov (2003) and Lewellen (2004), among others, find that valuation ratios are positively
correlated with subsequent returns and that the implied predictability of returns is substantial at
longer horizons. Since dividend yield only weakly predicts dividend growth, the variation of
dividend yields must be due to changing forecasts of expected returns1. Also, Campbell and

1
As noted by Campbell and Thompson (2004), these results are consistent with the view of value-oriented investors
in the tradition of Graham and Dodd (1934) that high valuation ratios are an indication of an undervalued stock
market and should predict high subsequent returns.

3
Shiller (1998) and Rapach and Wohar (2004a) find that these ratios are useful in predicting future
growth in real stock prices at long, but not short-horizons, using annual data spanning 1872-
19972.
Despite the econometric difficulties relating to the overlapping observations, highly
persistent predictor variables, small samples biases in predictive regressions (Mankiw and
Shapiro, 1986; Stambaugh, 1986, 1999; Richardson and Stock, 1989; Nelson and Kim, 1993,
Kirby, 1997, Ferson et al., 2003), the consensus – after thirty years of empirical works – appears
to be that aggregate returns do contain an important predictable component (Cochrane, 1999;
Campbell, 2000).
However, several recent studies have cast doubt on the predictability of stock returns,
especially from the dividend yield at long-horizons. On the one hand, Bossaerts and Hillion
(1999) and Goyal and Welch (2003, 2004) pointed out that the wide range of variables presented
above have some in-sample predictability but exhibit weak or no out-of-sample predictive
power3. On the other hand, Valkanov (2003), Campbell and Yogo (2004), Torous et al. (2005)
reexamine the evidence for predictability using tests that have the correct size even if the
predictor variable is highly persistent4 and find that the predictive power of the dividend yield at
long-horizons is considerably weakened. Moreover, Ang and Bekaert (2004) show, after
accounting for small sample properties of the standard tests, that at long horizons, excess return
predictability by the dividend-price ratio is not statistically significant, not robust across countries
and not robust across different sample periods. They argue that the ability of the dividend yield to
predict excess returns is best visible at short horizons with the short rate as an additional
regressor.
These new results could be explained by the fact that the valuation ratios are not mean
reverting, especially in the recent period, and therefore non-stationary contrary to the Campbell
and Shiller (1998) hypothesis. Indeed, Goyal and Welch (2003), among others, cannot reject that
dividend yield contain an unit-root over the longest sample period available at quarterly

2
For example, Campbell and Shiller (1998) present scatterplots and R² measures that indicate a weak ability for the
price-dividend and price-earning ratios to forecast real stock price growth over the next year, but a strong and
significant ability to forecast real stock price growth over the next ten years.
3
Campbell and Thompson (2004) show that the findings of Goyal and Welch (2004) are no longer valid, once
sensible restrictions are imposed on the signs of coefficients and return forecasts.
4
Stambaugh (1999), among others, has shown that the apparent predictability of stock returns may be spurious when
the predictor variable is persistent and its innovations are highly correlated with returns.

4
frequency (since 1926)5. In the present value model, non-stationary dividend-price ratio or non-
stationary linear combination of stock prices and dividends implies an explosive bubble
(Campbell and Shiller, 1987; Diba and Grossman, 1988). On the other hand, valuation ratios
might exhibit other forms of non-stationarity that do not imply explosive bubble. Indeed,
Timmerman (1995) shows that when the expected rate of return varies over time, the present-
value model does not generally imply the existence of a stationary relationship between stock
prices and dividends. Also, Carlson, Pelz and Wohar (2002) employ breakpoint tests on the
means of the quarterly valuation ratios and find evidence of one downward break in the dividend-
price ratio and the earning-price ratio at the beginning of the 1990’s. Finally, several authors
suggested that the equity premium dropped sharply over the last twenty years (e.g. Jagannathan,
McGrattan, and Scherbina, 2000; Fama and French, 2002). If this drop is permanent, then this
implies a permanent drop in the dividend-price ratio.
In this paper, we assume a time-varying risk premium which can be expressed as a linear
function of the expected inflation. We study the role of the transitory deviations from the
common trend in the earning-price ratio and inflation for predicting stock market fluctuations. In
particular, we find that these “trend deviations” exhibit substantial in-sample and out-of-sample
forecasting abilities for both real stock returns and excess returns. Moreover, we find that the
residual from the cointegrating relation among the earning-price ratio and inflation provides
information about future stock returns at short and intermediate horizons (from 1 to 12 quarters)
that is not captured by other popular forecasting variables.
The use of our forecasting variable is motivated by the vast empirical literature that has
emphasized the significant negative correlation – in post-war data for the US and other
industrialized countries – between inflation and stock returns (e.g. Fama and Schwert, 1977;
Gultekin, 1983; and more recently Barnes et al., 1999) and between inflation and the level of real
stock prices, as reflected in dividend-price ratio and price-earning ratios (Modigliani and Cohn,
1979; Feldstein, 1980; and more recently, Sharpe, 2002; Campbell and Vuolteenaho, 2004).
The paper proceeds as follows: Section 2 reviews previous research on the negative
relationship between stock returns/stock prices and inflation. Section 3 presents results of
estimating the trend relationship among the earning-price ratio and inflation. Section 4 discusses

5
Also, ADF and KPSS tests indicate that valuation ratios contain an unit-root over the longest sample period
available at annual frequency (1871-2003).

5
data used in our forecasting regressions for stock returns and presents some summary statistics.
Section 5 and 6 report respectively the in-sample and out-of-sample predictability test results.
Section 7 shows long-horizon forecasting results. Section 8 concludes.

2. Stock Prices and Inflation

The observed negative relationship between common stock returns and various measures
of expected and unexpected inflation during the post-World War II period is "troublesome"
because it appears to contradict Fisher's (1930) hypothesis, which states that nominal asset
returns move one-for-one with the expected inflation so that real stock returns are determined by
real factors independently of the rate of inflation. According to Fisher (1930), assets which
represent claims to physical or real assets, such as stocks, should offer a hedge against inflation.
The inflation-stock return correlation has been subjected to extensive study at the end of
1970s and the beginning of 1980s (e.g. Lintner, 1975; Bodie, 1976; Fama and Schwert, 1977;
Jaffe and Mandelker, 1976; Nelson, 1976; Fama, 1981; Pyndick, 1984)6 and was confirmed more
recently (Graham, 1996; Siklos and Kwok, 1999; Barnes et al., 1999).
In analyzing the Fisher hypothesis most of these empirical studies have focused on asset
returns over relatively short time horizons (less than a year). However, Boudoukh and Richardson
(1993) investigate the relation between stock returns and inflation at both short (1 year) and long
(5 year) horizons using long-term annual US and UK data, and obtain the quite interesting result
that at the 1-year horizon nominal stock returns and inflation are approximately uncorrelated,
while at the 5-year horizon the Fisher equation holds.
Other early studies focused on the negative relationship between inflation and the level of
real stock prices, as reflected in dividend-price ratio and price-earning ratio (Modigliani and
Cohn, 1979; Feldstein, 1980). More recently, Ritter and Warr (2002), Sharpe (2002) and
Campbell and Vuolteenaho (2004) confirmed this negative relation.
A number of alternative hypotheses have been advanced in the literature to explain the
negative relation between inflation and stock prices and/or stock returns. These alternatives

6
Most of these studies uses US data, but empirical evidence is also provided at the international level (e.g. Firth,
1979; Solnik, 1983; Gultekin, 1983, Boudoukh and Richardson, 1993).

6
include: (i) a correlation between expected inflation and expected real economic growth (the
“proxy hypothesis” suggested by Fama, 1981) ; (ii) the hypothesis that investors may irrationally
discount real cash flows using nominal interest rates (Modigliani and Cohn, 1979); (iii) changes
in the expected return and risk aversion (i.e. the equity risk premium) and (iv), the inflation non-
neutralities tax code which distorts accounting profits (Feldstein, 1980).
The “proxy hypothesis” suggested by Fama (1981) claims that the negative stock return-
inflation relation is spurious. The anomalous stock return-inflation relation is in fact induced by a
negative relation between inflation and real activity. Fama’s hypothesis predicts that rising
inflation rates reduce real economic activity and demand for money7. Geske and Roll (1983)
proposes a “reverse causality” explanation and argue that a reduction in real activity leads to an
increase in fiscal deficits. Since the Federal Reserve bank monetizes a portion of fiscal deficits,
the money supply increases, which in turn increases inflation.
The empirical evidence of the “proxy hypothesis” is mixed and suggests that it is not a
complete explanation. Kaul (1987) find some support for the proxy hypothesis, however the
findings of Cochran and DeFina (1993) and Caporale and Jung (1997) did not support it. Lee
(1992) and Balduzzi (1995) find strong support for the proposition that more than the proxy
hypothesis is at work and particularly that the rate of interest accounts for a substantial share of
the negative correlation between stock returns and inflation. Sharpe (2002) finds that the negative
relation between inflation and P/Es is attributable partly to lower forecasted real earnings growth.
Also, the “reverse causality hypothesis” is supported by James, Koreisha, and Partch (1985) but
rejected by Lee (1992).
Alternatively, Modigliani and Cohn (1979) suggest that investors collectively suffer from
money illusion and commit two errors in valuing equities: they use a nominal rate to discount real
cash flows (and fail to adjust nominal growth rate of dividends) and they fail to recognize the
capital gain that accrues to the equity holders of firms with fixed dollar liabilities in the presence

7
A closely related explanation, the “variability hypothesis” suggested by Hu and Willett (2000) is that the negative
stock return-inflation relation reflects a causal relation between inflation volatility (which is strongly correlated with
the level of inflation) and future real activity. Friedman (1977) argues that increased inflation volatility (uncertainty)
makes it difficult to extract signals about relative prices from absolute prices; therefore creates economic inefficiency
and depressing future economic activity. The “variability hypothesis” is supported by Hu and Willet (2000) but
rejected by Buono (1989).

7
of inflation. Empirical evidence of money illusion is provided by Ritter and Warr (2002)8 and
Campbell and Vuolteenaho (2004)9. Also, in a related literature, Thorbecke (1997), Bomfim
(2003), Bernanke and Kuttner (2005) and Rigobon and Sack (2004) find a significant response of
stock prices to changes in monetary policy. According to Rigobon and Sack (2004), a 25 basis
point increase in the three-month interest rate results in a 1.7% decline in the S&P 500 index and
a 2.4% decline in the Nasdaq index.
Recently, Campbell and Vuolteenaho (2004) decomposed the dividend yield into a term
due to rationally expected long-run dividend growth, a term due to the subjective risk premium
on the market, and a residual term that they attribute to a deviation of subjectively expected
dividend growth from objectively expected growth. They used a VAR system to construct
empirical estimates of these three components and find that high inflation is positively correlated
with rationally expected long-run real dividend growth; thus the negative effect of inflation on
stock prices cannot be explained through this channel. Campbell and Vuolteenaho (2004) find
that inflation is almost uncorrelated with the subjective risk premium and highly correlated with
mispricing10, supporting the Modigliani-Cohn (1979) view that investors form subjective growth
forecasts by extrapolating past nominal growth rates without adjusting for changes in inflation.
However, the authors recognize the possibility that that some part of what they call mispricing is
in fact a second component of the subjective risk premium, one that is common to all stocks and
does not appear in their cross-sectional measure of risk
Thus, the negative stock return-inflation relation can also reflect changes in the expected
return and risk aversion. Blanchard (1993), Jagannathan, McGrattan, and Scherbina (2001) and
Fama and French (2002), among others,11 interpret the bull market beginning in 1982 as partly
due a falling equity risk premium. Sharpe (2002) examines the effect of inflation forecasts on
required (long-run) real stock returns over the period 1983-2001 and finds that this effect is
substantial. In his model, the log earnings-price ratio is expressed as a linear function of expected

8
Ritter and Warr (2000) produce cross-sectional evidence in support of their money-illusion hypothesis. In cross-
sectional regressions, they find that the amount of undervaluation is positively correlated with leverage and expected
inflation.
9
The persistent use of the “Fed model” by Wall Street which relates the yield on stocks to the yield on nominal
Treasury bonds testifies the money illusion of practitioners (see Asness, 2003).
10
The authors use smoothed past inflation as a simple proxy for this expectation in their implementation. Their
empirical estimates suggest that past smoothed inflation explains nearly 80% of the time-series variation in the
aggregate stock market’s mispricing.
11
See e.g. Arnott and Bernstein (2002), Arnott and Ryan (2001), Claus and Thomas (2001), Heaton and Lucas
(1999).

8
inflation, expected future returns, expected earnings growth rates, and the log of the current
dividend/payout ratio. Investors expectations (future earnings growth and inflation) are drawn
from surveys of professional forecasters. The negative relation between equity valuations and
expected inflation is found to be the result of two effects: (i) lower expected real earnings growth
(as cited above) and (ii) higher required real returns. A one percentage point increase in expected
inflation is estimated to raise required real stock returns about one percentage point, which on
average would imply a 20 percent decline in stock prices12. Also, Blanchard (1993) finds that the
expected equity premium has experienced a long decline since the 1950s from unusually high
level in the late 1930s and 1940s. Blanchard examines the importance of inflation expectations
and attributes some of the recent trend to a decline in expected inflation.
Finally, Feldstein (1980) argued that much of inflation’s negative valuation effect could
be explained by basic features of the current US tax laws, particularly historic cost depreciation
and the taxation of nominal capital gains. However, the empirical evidence of the negative stock-
return relation is also provided at international level and as noted by Ritter and Warr (2002), in
1981, partly in response to high inflation, the US tax code was changed to accelerate
depreciation, reducing the distortions.

3. Estimating the long-term relationship between stock prices and inflation

The present value model assumes that prices depend upon the present value of discounted
future dividends, where the discount rate is equivalent to the required rate of return. In our
empirical implementation we use the loglinear version of the present value model proposed by
Campbell and Shiller (1988). In the loglinear dynamic valuation framework of Campbell and
Shiller, the log dividend-price ratio can be written as:

κ ⎡∞ ∞ ⎤
dt − pt = − + Et ⎢ ∑ ρ j rt + j − ∑ ρ j ∆dt + j ⎥ , (1)
1− ρ ⎣ j =0 j =0 ⎦

12
But the inflation factor in expected real stock returns is also in long-term Treasury yields; consequently, expected
inflation has little effect on the long-run equity premium.

9
where Et denotes investors expectations taken at time t, ∆dt + j denotes dividend growth in t+j,

calculated as the change in the log of real dividends per share, and rt + j denotes log stock return

during period t+j. The expected return equals the real risk-free interest rate plus a risk premium.
ρ and κ are parameters of linearization defined by ρ ≡ 1 (1 + exp(d − p ) and
κ ≡ − log( ρ ) − (1 − ρ ) log(1 ρ − 1) . Equation 1 states that expected stock returns and dividend
growth can be predicted by the log dividend-price ratio.
Following Nelson (1999) and Sharpe (2002), we decompose the log dividends per share
into the sum of the log earnings per share and the payout ratio. Then, the Campbell-Shiller
formula can be rewritten as:

κ ⎡∞ ∞ ∞ ⎤
et − pt = − + Et ⎢ ∑ ρ j rt + j − ∑ ρ j ∆et + j − (1 − ρ )∑ ρ j (d t + j − et + j ) ⎥ , (2)
1− ρ ⎣ j =0 j =0 j =0 ⎦

where et − pt denotes the log earning-price ratio, ∆et + j denotes real earning growth in t+j,

calculated as the change in the log of real earnings per share, and dt + j − et + j denotes the log of the

payout ratio (dividends/earnings) in t+j.

This reformulation enable us to focus on earnings which are more closely related to
economic fundamentals than dividends since they can be affected by shifts in corporate financial
policy. Campbell (2000) argues that dividends creates several difficulties for empirical work.
First, many companies pay cash to shareholders partly by repurchasing shares on the open market
(for fiscal reasons) which biased the dividend yield (see Liang and Sharpe, 1999). Second, many
companies seem to be postponing the payment of dividends until much later in their life cycle.
Fama and French (2001) observe that the proportion of listed US companies paying cash
dividends falls from 66.5% in 1978 to 20.8% in 1999.
In Equation (2), if et − pt is non-stationary, the right hand-side is also non-stationary and

possibly reflects the use of a nominal discount rate, bt , by investors or a time-varying risk

premium which can be expressed as a linear function of the expected inflation, π te . It is generally
agreed, see Stock and Watson (1988, 2003), that interest rates and inflation series are I(1)
variables.

10
 Under the preliminary assumptions (verified after), that dt + j − et + j and ∆et + j are

stationary and et − pt , bt and π te are I(1) processes, we investigate the cointegration relationships

between et − pt and bt , and between et − pt and π te . Then, these presumed cointegrating

relationships imply that a deviation from the long-run equilibrium impacts positively or
negatively the (log) earning-price ratio such that the equilibrium is restored. Indeed, these
potential relationships could not be expected to hold exactly and deviations may arise due to
bubbles, noise trading, fads, and omission of other relevant variables.
The first step in our analysis is to document the negative relations between real stock
prices and various measures of inflation. We use five different methods of computing expected
inflation13. First, under the assumption that investors possess perfect foresight, expected inflation
will be equal to realized inflation (πt). The second method uses once-lagged inflation as the
forecast (πt-1). In the third method, expected inflation is derived from an ARIMA model ( π tari ). In

the fourth method, following Lee (1992) and Zhong, Darrat and Anderson (2003), expected
inflation is modeled rationally as π tkal = Et −1 ⎡⎣π t π t −1 , MBt −1 , bt −1 , IPt −1 ⎤⎦ by using a simple Kalman

Filter (updating) method, where MBt is the growth rate of the monetary base, bt is the three-
month treasury bill rate, and IPt is the growth rate of the industrial production average. In the
fifth method, following Cozier and Rahman (1988), expected inflation is based on a forecasting
model that includes lagged values of the variables used in your fourth method ( π tols ).
We use quarterly data over the post-World War II period (1948:1–2004:1). The quarterly
Standard & Poor’s (S&P) nominal stock prices, dividends, and earnings indexes are from
Campbell and Shiller (1998), which begin in 1926 and extend to 200414. We deflate the three
nominal indexes using the consumer price index (all urban consumers) from the Bureau of Labor
Statistics (BLS) in order to obtain series for real stock prices, real dividends, and real earnings.
The monetary base, the T-bill rate and the growth rate of the industrial production average are
available from the FRED II database of the Federal Reserve Bank of St. Louis15.

13
Fama and Schwert (1977), Geske and Roll (1983) and others use the contemporaneous nominal treasury bill rate as
a proxy for expected inflation. We do not use this method because it will be equivalent to test the nominal discount
rate hypothesis
14
The S&P 500 data are available from Robert Shiller’s home page at http://www.econ.yale.edu/~shiller. The
complete documentation for the data sources is also provided here. Data are updated from the standard and poor’s
web site (S&P 500 Earnings and Estimate Report).
15
Available at http://research.stlouisfed.org/fred2/

11
 The non-stationarity is not rejected for the earning-price ratio, the T-bill rate and inflation
in levels, but the hypothesis is rejected if the variables are expressed in first-differences (see
Table A1 in appendix). Thus, it is possible that the earning-price ratio is cointegrated with our
measures of expected inflation and the nominal risk free rate.
Therefore, we test for cointegration using two distinct methodologies, namely the
multivariate trace statistic developed by Phillips and Ouliaris (1990), and the Johansen and
Juselius (1992) approach (Trace Test). Table 1 displays tests results. The only deterministic
components in the models are the intercept in the cointegration space. The appropriate lag-length
is selected in order to accept the assumption that residuals are white noise based on LM(1) and
LM(4) criteria. As table 1 shows, there is sufficient evidence for one non-zero co-integrating
vectors between et − pt and π t , π t −1 or π tari . On the other hand, the hypothesis of no

cointegration between et − pt and bt can not be rejected at conventional significance level. The

cointegration evidence between et − pt and π tkal or π tols is mixed depending on the implemented
test. For all that, in the remainder of the paper, we focus on realized inflation rather than expected
inflation because we intend to provide evidence of the in-sample and out-of sample predictability
of stock returns from past information.
The long-term relationship between et − pt and π te implies that a deviation from the long-
run equilibrium impacts positively or negatively the (log) earning-price ratio such that the
equilibrium is restored. Now, we would like investigate if these deviations have an impact on the
real stock prices, real earnings or both. We estimate a vector error correction model (VECM) for
real stock prices, real earnings and inflation with restrictions on the cointegrating vector. We
constrained the long-run parameters such as the cointegration vector in the VECM is similar to
that obtained previously in the cointegration relationship between the earning-price ratio and
inflation. Table 2 presents results of the constrained VECM16. The table reveals some interesting
points. First, real earning growth is somewhat predictable by their own lags, by lags of real stock
prices growth and by inflation growth with 6 lags17. Second, the error correction term predicts
real stock prices growth but it doesn’t appear at a statistically significant level in the equations for
real earnings. Also, the magnitude of the coefficient on the error correction term in the inflation
growth equation is substantially smaller than in the real stock prices equation. These results
16
The results of the non-constrained VECM are qualitatively the same.
17
In a VECM with more lags, lags of inflation growth longer than six does not appear significant.

12
suggest that the empirical evidence of the “proxy hypothesis” is weak and especially that
deviations from the shared trend in log earning-price ratio and inflation are better described as
transitory movements in real stock prices than as transitory movements in real earnings or
inflation.
The next step in our analysis is to investigate the role of these transitory movements in
real stock prices in forecasting stock returns. Before that, it is necessary to obtain consistent
estimates of the parameters of the shared trend in log earning-price ratio and inflation. Following
Lettau and Ludvigson (2001), we use the dynamic ordinary least squares (DOLS) developed by
Stock and Watson (1993) to estimate the cointegration parameters. Specifically, the DOLS
estimates the long-run relation directly by OLS augmented by the first difference of the
explanatory variables together with their lags and leads (l) to eliminate the effects of regressor
endogeneity on the distribution of the least squares estimator. Formally, DOLS amounts to
running an OLS on the following specification (in the case of the earning-price inflation relation):

l
et − pt = α1 + α 2 π t + ∑ β i ∆π t −i + ε t (3)
i =− l

where the AIC and BIC criteria are used to determine the appropriate lead/lag length, with a
maximum of 8 lags considered. Equations (4) and (5) report the DOLS estimates (ignoring
coefficient estimates on the first differences) respectively for the parameters of the shared trend
among earning-price ratio and inflation and the shared trend among earning-price ratio and
nominal T-bill rate using data from the fourth quarter of 1951 to the second quarter of 200318:

et − pt = − 3.11 + 10.00 π t , (4)

( −35.67) (6.59)

et − pt = − 3.19 + 8.45 bt , (5)

( −26.69) (4.74)

where the corrected t-statistics appear in parentheses below the coefficient estimates. We also
estimated equation (5), even if no cointegration between earning-price ratio and the T-bill rate

18
We used the same sample as Lettau and Ludvigson (2004a,b) in order to compare our results with theirs.

13
can not be rejected, in order to evaluate the predictive power of the deviations from their relation.
The estimated cointegrating coefficients suggest that a one percentage point decrease respectively
in actual inflation and the T-bill rate is associated with a 10 percent decline and a 8.45% percent
decline in the earning-price ratio and thus in real stock prices.
ˆ t and epb
We denote respectively epi ˆ t , the deviation of (log) earning-price ratio from its

predicted value based on the cointegrating regression (4) and the non-cointegrating relation (5).
Before investigate the predictive power of these two variables for the real return on stocks and
the excess of the return on stocks, we describe the data and provide summary statistics.

4. Asset returns data and Summary Statistics

The data set consists of quarterly observations from 1951:Q4 to 2003:Q2. Stock prices,
dividends per share, and quarterly earnings per share all correspond to the Standard & Poor’s
(S&P) Composite Index described above. Real data are deflated by the Consumer Price Index
(All Urban Consumers) published by the BLS. Let rt denote the real return on the S&P index.

The three month T-bill rate is used to construct the real return on the risk free rate, rf , t , and the

log excess return ( rt − rf , t ).

Log price, pt , is the natural logarithm of the real S&P price level in quarter t. Log

dividends, dt , are the natural logarithm of real dividends per share in quarter t. Log earnings, et ,
are the natural logarithm of real earnings per share in quarter t. Following Lamont (1998), the log
dividend payout ratio is dt − et . The stochastically detrended risk-free rate, rrelt , is the T-bill rate
minus its last four-quarter average. This relative bill rate is used by Campbell (1991) and Hodrick
(1992) to forecast stock returns. Following Fama and French (1989) and Campbell (1987), we
used the term spread, TRM t , the difference between the 10-year Treasury bond yield and the 3-

month Treasury bond yield, and the default spread, DEFt , the difference between the BAA and
AAA corporate bond yields19. Following Lettau and Ludvigson (2001, 2004a,b), we use the
measure of short-term deviations from the long-run cointegration relationship among the natural

19
Interest rate data come from the FRED II database.

14
 ˆ t 20. The
logarithm of consumption (c), labor income (y) and aggregate wealth (a), henceforth cay

aggregate stock market volatility, σ t , is the variance of the daily stock market return data adjust
for the 1987 stock market crash21. Guo (2006) finds that a measure of aggregate stock market
volatility in conjunction with the consumption-wealth ratio exhibits substantial out-of-sample
forecasting power for excess stock market returns.
Table 3 shows basic summary statistics for the real stock return, the excess return and
ˆ t , is no surprising highly positively
their forecasting variables. Your estimated trend deviation, epi

correlated with et − pt , dt − pt and epb

ˆ t . Correlations with the real return, the excess return, the

ˆ t and the term spread are positive, and negative with the
relative bill rate, the payout ratio, cay
stock market volatility and the default spread. As reported in the previous literature, many of the
forecasting variables are highly persistent. The log earning-price inflation ratio does not escape
this rule.
Figure 1 plots the log earning-price inflation ratio and the excess return on the S&P
ˆt
Composite Index over the period 1951:Q3-2003:Q2. The figure shows that large swings in epi
precede large swings in excess returns over the entire sample. However, this pattern does not hold
for several episodes as in the mid of the sixties or in the second half of the nineties when the
earning-price ratio fells sharply but excess returns remain positive. This suggests that some non-
linearities or structural break could occur in the underlying parameters governing this relationship
or in the coefficient estimates of the cointegrating relation between the earning-price ratio and

20
The deviation of (log) aggregate consumption from its predicted value based on a cointegrating regression between
(log) consumption, (log) aggregate assets and (log) aggregate labor income. The consumption, net worth, labor
income data and the generated variable cay over the period 1951:Q4 to 2003:Q2 are obtained from Sydney
Ludvigson at New-York University (http://www.econ.nyu.edu/user/ludvigsons/). The theoretical justification that
Lettau and Ludvigson present for the predictive power of this variable is the log-linearized version of the standard
budget constraint relating wealth, consumption, and portfolio returns where the unobservable aggregate wealth
variable is approximated by aggregate assets and labor income. However, Brennan and Xia (2004) argue that the
predictive power of their variable arises from a “look-ahead bias”. Also, Rudd and Whelan (2002) argue that Lettau
and Ludvigson use a set of variables that do not belong together in an aggregate budget constraint, thereby testing a
cointegrating relationship that is not implied by their theory. Rudd and Whelan cannot reject the hypothesis that
cointegration is absent from the data once they employ measures of consumption, assets, and labor income that are
jointly consistent with an underlying budget constraint.
21
The daily Dow Jones index was obtained from www.economagic.com. Following Campbell et al. (2001), Guo
(2006) adjust downward realized stock market variance for 1987:Q4 because the 1987 stock market crash has
confounding effects on it. They replace the 1987:Q4 observation by the second largest realized stock market variance
in the sample. However, our sample is larger than in Guo (2006) and then, the second largest realized stock market
variance differs. So, the predictive power of the stock market variance could be different than in the originally study.

15
inflation. Nevertheless, theses episodes remain specific and transitory, as reflected in the
subsequent continue downturn in excess returns at the end of the 1990’s.

5. Quarterly Forecasting Regressions

We report in this section the in-sample regression results. Inoue and Kilian (2004) argue
that in-sample tests should be preferred because they have greater power than out-of-sample tests
(even adjusted for data mining). The predictive regression model takes the form:

yt +1 = α 0 + α1 zt + ut +1 , (6)

where yt is either real stock returns or excess returns to holding stocks from period t - 1 to period

t, zt is a control variable believed to potentially predict future returns and ut +1 is a disturbance

term.
The predictive ability of zt is typically assessed by examining the t-statistic

corresponding to α̂1 , the OLS estimate of α1 , as well as the goodness-of-fit measure, R 2 . We

estimate the regression (6) by OLS and use Newey-West (1987) adjustment to the standard errors
of the coefficients to correct for serial correlation and heteroscedasticity.
Before examine the predictability of both real stock returns and excess returns, we
ˆ t on future real stock prices growth, ∆pt , real dividend
investigate the predictive power of epi

growth, ∆dt , real earning growth, ∆et , the payout ratio, dt − et , and future real returns on the risk

free rate, rf ,t .

We examine two different vector autoregression models (VAR) where the endogenous
variables are each regressed on their own lags and to the lagged value of your estimated trend
ˆ t . In the first VAR, endogenous variables are the real stock price growth, the real
deviation, epi
dividend growth, the payout ratio and the real return on the risk free rate. In the second VAR, we
substitute the real dividend growth by the real earning growth. Table 4 reports the VAR
estimates. We focus on the relationship between future endogenous variable and the estimated

16
 ˆ t predicts real stock prices growth, real dividend growth
trend deviation. Table 4 shows that epi

ˆ t is not significative at a 5% significance level in the

and the payout ratio. The coefficient on epi
equations for the real risk free rate return and earning growth. This result is similar to that
obtained previously where the long-term deviations among earning-price ratio and inflation (the
error-correction term) does not enter at a statistically significant level in the equation for real
ˆ t predicts dividend growth22 and
earning growth in the VECM framework. On the other hand, epi
the payout ratio. This suggests that the corporate financial policy could be affected by the
transitory deviations from the common trend in earning-price ratio and inflation. These results
ˆ t could forecast real stock returns and excess returns by
also suggest more generally that epi
forecasting both real stock prices and real dividends.
Table 5 reports one-quarter ahead forecasts of both real returns and excess returns of
stocks over the riskfree rate. All models include a constant term. The firsts rows of each panel
show that the one lag of the dependent variable is a weak predictor of future returns. This model
predicts only 2.5% of next quarter’s variation in real stock returns and 1.7% of next quarter’s
ˆ t , is significant and has more
excess returns variation. The log earning-price inflation ratio, epi

ˆ t , the log earning-price T-bill ratio,

explanatory power than the consumption-wealth ratio, cay

ˆ t , the dividend-price ratio and the earning-price ratio23. Regressions of real stock returns and
epb

ˆ t produce adjusted R² of 9.6% and 9.4% respectively. Moreover,

excess returns on one lag of epi

ˆ t indicates that the coefficient estimate is nonzero

the Newey-West corrected t-statistic for epi
with very high probability. These results are not affected by whether the lagged value of the
dependant variable is included in the regression as an additional explanatory variable (rows 5 and
20).
These results are robust to alternative specifications in estimating the log earning-price
inflation ratio. They are not sensitive to the value of l in estimating the DOLS specification, the
choice of estimation method or the measure of expected inflation. In appendix, Table A3 reports

22
Alternative specifications of the VAR indicates that dividend growth is not predictable by the dividend-price ratio
or the earning-price ratio. This is in agreement with a large literature that documents the poor predictability of
dividend growth by the dividend yield (e.g., Campbell, 1991; Cochrane, 1991; Lewellen, 2004).
23
Table A2 in appendix shows that these results are robust to different specifications for the normalized stock prices.
We also show regressions with the “log dividend-price inflation ratio” as the sole predictive variable. The predictive
power of this variable is inferior to that of the log earning-price inflation ratio

17
that the forecasting results for different specifications of the log earning-price inflation ratio are
very similar24.
ˆ t and cay
In order to compare the forecasting power of epi ˆ t , we include both in the same

regression (rows 8 and 23). These two variables are both significant and regressions produce
ˆ t contains information about future
higher R² than in the univariate models. This suggests that epi

ˆ t.
asset returns that is not included in cay
To check the robustness of our results, we augment the precedent regressions by adding a
variety of variables that are useful in predicting future stock returns (row 24 and 9). These
include the payout ratio, the term and default spreads on bonds, the relative bill rate and the stock
market volatility. These regressions have more explanatory power than the precedent models.
However, only the relative bill rate has a significant predictive power among these five
ˆ t and epi
supplementary variables. The two trend deviation terms, cay ˆ t , are still strongly

significant.

6. Out-of-Sample Tests

Some recent studies (e.g., Bossaerts and Hillion, 1999; Goyal and Welch, 2003, 2004)
expressed concern about the apparent predictability of stock returns because while a number of
financial variables display significant in-sample predictive ability, they have negligible out-of-
sample predictive power. Also, our forecasting results presented above could suffer from a
ˆ t are
“look-ahead” bias that arises from the fact that the coefficients used to generate epi
estimated using the full sample.
To address these issues, we examine, in this section, the out-of-sample predictability of
both real stock returns and excess returns by distinguishing two cases. In the first, agents are
ˆ t , epi
assumed to know the cointegration parameters of cay ˆ t , epb
ˆ t , which are estimated using the

full sample. In the second case, the cointegration parameters are estimated recursively using only

24
We experimented with various lead/lag length in estimating the DOLS specification and we used the cointegrating
parameters obtained in the previous section based on Johansen’s (1988) full information maximum likelihood
approach. We also considered the earning-price inflation ratio from the cointegration relationship among the earning-
price ratio and one lag of inflation.

18
information available at the time of forecast. Moreover, we present out-of-sample predictability
ˆ t and epi
results using the two-period lagged value of cay ˆ t because these variables are available

with a one-month delay relative to financial indicators. This scenario gives some idea of how the
model would perform if a practitioner, who must rely on real-time data, uses it. Goyal and Welch
(2003, 2004) indeed recommend that one should adopt “the perspective of a real-world investor”
(who did not have access to ex-post information).
We present two types of comparisons in order to evaluate the out-of-sample predictive
ˆ t : nested comparisons and non-nested comparisons. In the nested comparisons, we
power of epi
compare a benchmark “restricted” model with an unrestricted model which include both the
ˆ t . In the non-nested comparisons, we
explanatory variables of the restricted model and epi
compare competitive models with different explanatory variables (popular forecasting variables
ˆ t versus epi
as the dividend-price ratio or cay ˆ t ).

We use four statistics to compare the out-of-sample performance of our forecasting

models: the mean-squared forecasting error (MSE) ratio, the Clark and McCracken’s (2001)
encompassing test (ENC-NEW), the McCracken’s (2004) equal forecast accuracy test (MSE-F)
and the modified Diebold-Mariano (MDM) encompassing test proposed by Harvey, Leybourne
and Newbold (1998)25. We apply the ENC-NEW and MSE-F tests for the non-nested comparisons
and the MDM test for the nested comparisons. We report the MSE ratio in both nested and non-
nested comparisons.
The ENC-NEW encompassing test, is a modified Harvey, Leybourne, and Newbold
(1998) test statistic adapted to address the fact that the limiting distribution of this test statistic is
nonnormal when the forecasts are nested under the null26. The ENC-NEW statistic provides a test
ˆ t ) incorporates all the relevant
of the null hypothesis that the restricted model (which exclude epi
information about the next quarter’s value of the dependent variable, against the alternative

25
Professor Simon Van Norden is gratefully thanked for providing us the program of the MDM test.
26
Forecast encompassing is based on optimally constructed composite forecasts. Intuitively, if the forecasts from the
restricted regression model encompass the unrestricted model forecasts, the additional variable included in the
unrestricted model provides no useful additional information for predicting returns relative to the restricted model
which excludes this variable; if the restricted model forecasts do not encompass the unrestricted model forecasts,
then the additional variable does contain information useful for predicting returns beyond the information already
contained in a model that excludes this variable. Tests for forecast encompassing are similar to testing whether the
weight attached to the unrestricted model forecast is zero in an optimal composite forecast composed of the restricted
and unrestricted model forecasts.

19
 ˆ t ) provide additional information that
hypothesis that the unrestricted model (which include epi
could be used to significantly improve the restricted model’s forecast.
The MSE-F test is a test of equal MSE. The null hypothesis for this test is that the
restricted model has a MSE that is less than or equal to that of the unrestricted model; the
alternative is that the unrestricted model has lower MSE. Clark and McCracken (2001) show that
these two tests have the best overall power and size properties among a variety of tests proposed
in the literature.
The MDM test, is a modified Diebold and Mariano (1995) test statistic to test for forecast
encompassing between two non-nested models and to account for finite-sample biases. This test
statistic is formed by asking whether the difference in forecast errors between two models is
correlated with the forecast error of the model that is encompassing under the null. The null
ˆ t , encompasses model 1 where the
hypothesis is that the competitor model 2, without epi

ˆ t.
predictive variable is epi
As in Lettau and Ludvigson (2001), we use the first one-third observations for the initial
in-sample estimation and form the out-of-sample forecast recursively in the remaining sample.
The initial estimation period begins with the fourth quarter of 1951 and ends with the first quarter
of 1968. The model is recursively reestimated until the end the of sample.
Figure 2 plots the recursively estimated coefficient on inflation. The point estimates show
large variations until the end of the 1970s because it requires a relatively large number of
observations to consistently estimate the cointegration parameters. After a stabilized phase, the
inflation coefficient increased progressively during the 1990’s, possibly reflecting a higher
inflation aversion of investors in a permanent low inflation environment27. However, the results
ˆ t does not deteriorates if the
presented below indicate that the forecasting ability of epi
cointegration parameters are estimated recursively relative to the fixed parameters using the full
sample.
We report results of the out-of-sample one-quarter-ahead nested forecast comparisons of
real stock returns and excess returns in Tables 6 and 7 respectively. We consider two restricted
(benchmark) models: a model that includes only a constant as a predictor and a model that
27
The Andrews-Quandt and Andrews-Ploberger structural break tests (Andrews and Ploberger, 1994; Hansen, 1997)
indicate a structural break in the parameter stability in 1989:Q3. The Bai and Perron (2003) test detects also one
structural break in 1989:Q3. These results could also reflect the persistent deviation from the common trend in
earning-price ratio and inflation that appears in the 1990’s.

20
includes both a constant and the lagged dependent variable as predictive variables. The nested
comparisons are made by alternately augmenting the benchmark with either the one-period
ˆ t , or the two-period lagged value, denoted epi
lagged value of epi ˆ t −1 . We present results based on

a fixed cointegrating vector where the cointegrating parameters are set equal to their values
estimated in the full sample and a recursive reestimated cointegrating vector.
Consistent with the in-sample regression results, we find that the unrestricted model
ˆ t ) has smaller MSE than the constant restricted model or the autoregressive
(which include epi
restricted model. Tables show that regardless of whether the cointegrating parameters are
ˆ t is used as a predictive
reestimated, or whether the one- or two-period lagged value of epi

ˆ t provides no
variable, both ENC-NEW and MSE-F tests reject the null hypothesis that epi
information about future stock returns at the 1% significance level.
Results of the out-of-sample one-quarter-ahead non-nested forecast comparisons of real
stock returns and excess returns are shown in Tables 8 and 9 respectively. We compare
ˆ t is the sole predictive variable with
alternatively the model 1 in which the lagged value of epi
“competitor models” in which either the lagged dependent variable, lagged dividend-price ratio,
lagged earning-price ratio, lagged dividend payout ratio, lagged detrended bill rate, lagged value
ˆ t (with/without the measure of stock market volatility), lagged value of epb
of cay ˆ t is the sole

predictive variable. A constant is included in each of the forecasting equations.

ˆ t forecasting model produces lower MSE than any of the
The results indicate that the epi
“competitor” model. Moreover, the MDM encompassing test indicates that the model using
ˆ t contains information that provides superior forecasts to those produced by most of
lagged epi
the other models. The findings are statistically significant at better than the two percent level in
almost every case, regardless of whether the cointegrating parameters are reestimated28.
ˆ t has displayed statistically significant
In summary, the results presented indicate that epi
out-of-sample predictive power for both real stock returns and excess returns over the postwar
period, and contains information that is not included in lagged value of the dependent variables or
a model of constant expected returns. The non-nested forecasts comparisons results suggest also

28
Except when the log earning-price inflation ratio is recursively reestimated and the competitor models include the
consumption-wealth ratio or the dividend-price ratio. However, in theses cases, The inverse MDM tests that our
variable encompasses the competitor models are not rejected with greater p-value.

21
 ˆ t would be consistently superior to forecasts using any other popular
that forecasts using epi
forecasting variables.

7. Long-horizon Forecasts

In this section, we investigate the relative predictive power of the log earning-price
inflation ratio for long-horizon stock returns. The relatively modest absolute value of the
coefficient on the error-correction term in the VECM framework presented above and the
graphical evidence of persistent deviations from the common trend in earning-price ratio and
ˆ t should provide useful information for predicting
inflation (see Figure 1) both suggest that epi
stock returns at intermediate horizons.
We use two different methodologies in order to evaluate the long-horizon predictability of
stock returns. The first consists of single-equation regressions as in Lettau and Ludvigson (2001)
that provide a simple way to summarize the marginal predictive power of each forecasting
variable and the overall explanatory power of the forecasting equations. The second consists of
the two out-of-sample tests for nested forecasts models presented above: the encompassing ENC-
NEW test and the equal forecast accuracy MSE-F test. Since these remaining tests have
nonstandard limiting distributions that are usually dependent upon unknown nuisance parameters,
we follow Clark and McCracken (2004) in using a bootstrap procedure similar to that in Kilian
(1999) to estimate asymptotically valid critical values and construct asymptotically valid p-
values29. Following Rapach et Wohar (2004b), we use a restricted (benchmark) model of constant
returns for long-horizon forecasts.
The k-period dependent variable, yk ,t + k , in these long-horizon regressions is measured by
k
yk , t + k = ∑ yt +i . We consider horizons of 1 to 24 quarters and a very long horizon of 48 quarters.
i =1

Before presenting results from long-horizon regressions of real stock returns and excess
returns, we report in Table 10 results of single-equation regressions of stock prices growth,
dividend growth and earning growth at long-horizons. Theses results confirm findings presented

29
The bootstrap procedure is briefly described in appendix and in more details in Clark and McCracken (2004).

22
 ˆ t predicts stock prices growth and dividend growth but has no predicting power
in Table 5 that epi

ˆ t increases with k until it reaches a peak around

for earning growth. The predictive power of epi
2-3 years. Beyond this peak, it decreases progressively until a horizon of 6 years.
On the contrary, the dividend-price ratio and earning-price ratio display no forecasting
power for any dependent variables at short and intermediate horizons except for dividend growth
with the dividend-price ratio at horizons of 12 and 16 quarters (row 5). The valuation ratios
become significant at very long horizons but the predictive power of the dividend-price ratio and
the earning-price ratio, for which we do not reject the non-stationary, probably suffer from a
spurious regression problem. Overlapping observations (when k > 1) are not independent and that
induces serial correlation in the disturbance term (Richardson and Stock, 1989). Even when
robust standard errors are used to compute t-statistics (using the Newey-West procedure), in
finite samples, there is a strong tendency for the t-statistic to increase in absolute value, as the
overlap increases, whether or not there is a relationship between the variables (e.g. Hodrick,
1992; Nelson and Kim, 1993; Goetzmann and Jorion, 1993) 30.
Tables 11 and 12 present results of long-horizon regressions of both real stock returns and
ˆt
excess returns at horizons ranging from 1 to 48 quarters. First rows of these tables show that epi
has statistically significant forecasting power for both real stock returns and excess returns at
ˆ t is in the most of cases superior of any
long horizons. Moreover, the forecasting power of epi
other predictive variable at horizons ranging from 1 to 12 quarters (rows 1 to 6). As in the
ˆ t increases with horizon until a
precedent long-run regressions, the predictive power of epi

ˆ t , cay
horizon of 3 years after that it progressively decreases. When we include epi ˆ t , the payout

ratio, the stochastically detrended short rate, the term spread and the default spread together in
one regression (rows 7), R² statistics are higher at horizons ranging from 1 to 12 quarters than in
ˆ t replaces epi
regressions where the dividend-price ratio or epb ˆ t (rows 8 and 9).

ˆ t as
Figure 3 plots realized and predicted excess returns by the univariate model with epi
the sole predictor (row 2) for different horizons. It provides a simple way to visualize the

30
As noted by Valkanov (2003), overlapping a non-trivial fraction of the sample produces a persistent variable that
behaves very much like a I(1) process. Indeed, at these very long horizons, we can not reject a unit root process at a
5% level in the dependent variables (ADF test). Ferson et. al. (2003) provide simulation evidence that predictors with
large autocorrelation coefficients suffer from a spurious regression problem if the true process for the dependent
variable is also persistent.

23
 ˆ t contains at different horizons. The figure shows
information about future excess returns that epi
the improvement in the ability of the log earning-price inflation ratio to forecast future excess
returns as the horizon increases until k = 12 and then the progressively decrease in the predictive
ˆ t.
power of epi
Table 13 presents MSE-F and ENC-NEW out-of-sample statistics of both real stock
returns and excess returns at horizons ranging from 1 to 48 quarters. The p-values are generated
using the bootstrap procedure described in appendix. As in the precedent section, we present
results based on a fixed cointegrating vector and a recursive reestimated cointegrating vector. The
ˆ t ) has smaller MSE than the constant
table shows that the unrestricted model (which include epi
restricted model at horizons less than 6 years.
Regardless of whether the cointegrating parameters are reestimated, the ENC-NEW and
ˆ t provides no information about future excess
MSE-F tests reject the null hypothesis that epi
returns at the 5% significance level for horizons of 1 to 16 quarters. The ENC-NEW and MSE-F
ˆ t has no predictive power at the 5% significance level for future real
tests reject the null that epi
stock returns at horizons less than 4 years except the ENC-NEW when the cointegrating vector is
reestimated at horizons of 8 and 12 quarters. In these two last cases, we reject the null at the 10%
level.

8. Summary and conclusion

The observed negative relationship between stock prices/stock returns and both expected
and realized inflation during the post-World War II period is “troublesome” because it appears to
contradict the Fisher Hypothesis, which states that expected stock returns move one-for-one with
expected inflation since stocks are claims on “physical” or real assets. The inflation-stock
return/stock prices correlation has been subjected to extensive study since a quarter century.
However, there is less consensus on what drives this negative relation.
In this article, we consider a new perspective on the relationship between stock prices and
inflation, by estimating the common long-term trend in real stock prices, as reflected in the
earning-price ratio, and both expected and realized inflation. We estimated a VECM, based on a

24
variant of the Campbell-Shiller price-dividend model, for real stock prices, real earnings and
realized inflation where we constrained the long-run parameters such as the cointegration vector
in the VECM is similar to that obtained previously in the cointegration relationship between
earning-price ratio and inflation. The results of the constrained VECM suggest that the empirical
evidence of the “proxy hypothesis” is weak and especially that deviations from the shared trend
in real stock prices, real earnings and inflation are better described as transitory movements in
real stock prices than as transitory movements in real earnings or inflation. This implies that a
deviation from the long-run equilibrium impacts positively or negatively stock prices such that
the equilibrium is restored.
We investigate the role of these transitory deviations from the common trend in the
earning-price ratio and inflation for forecasting stock returns. We find that the log earning-price
inflation ratio predicts real stock prices growth, real dividend growth and the payout ratio while it
does not predict – consistent with the VECM results – real earning growth. This suggests that the
corporate financial policy could be affected by the transitory deviations from the common trend
in earning-price ratio and inflation. These results also suggest more generally that the log
earning-price inflation ratio could forecast real stock returns and excess returns by forecasting
both real stock prices and real dividends.
Indeed, we find that the trend deviations from the share trend in the earning-price ratio
and inflation exhibit substantial in-sample and out-of-sample forecasting abilities for both real
stock returns and excess returns. Moreover, we find that these trend deviations provide
information about future stock returns that is not captured by other popular forecasting variables
over short and intermediate horizons (from 1 to 12 quarters) and that the log earning-price
inflation ratio is the best univariate predictor of stock returns over theses horizons.
Also, our results do not support the hypothesis of Modigliani and Cohn's inflation illusion
that states that investors use a nominal rate to discount real cash flows. First, we can not reject the
hypothesis of no cointegration between the earning-price ratio and the nominal risk free rate over
our sample, whereas there is sufficient evidence for one non-zero co-integrating vectors between
the earning-price ratio and expected inflation/realized inflation. Second, the predictive power of
the log earning-price T-bill ratio is always inferior to that of the log earning-price inflation ratio.
In this article, we examined the forecasting ability of the log earning-price ratio through a
linear regression method. However, as shown in Figure 1, there are several episodes, as in the

25
mid of the sixties and in the second half of the nineties, where the earning-price ratio fells sharply
but excess returns remain positive. Also, some recent works (e.g. Coakley and Fuertes, 2003;
Bohl and Siklos, 2004; Ma and Kanas, 2004) documenting non-linearities in the U.S. stock
market valuation ratios. These suggest that an extension of our work would be to investigate
whether a non-linear model can improve forecasts of stock returns.

26
References

Andrews, D.W.K. and W. Ploberger (1994), “Optimal Tests When a Nuisance Parameter is Present Only Under the
Alternative,” Econometrica, 62, 1383-1414.
Ang, A. and G. Bekaert (2004), “Stock return predictability: Is it there?,” Manuscript, Columbia University.
Arnott, R. D and P.L. Bernstein (2002), “What Risk Premium is ‘Normal’?,” Financial Analysts Journal, 58(2), 64-
85.
Arnott, R.D. and R.J. Ryan (2001), “The Death of the Risk Premium,” Journal of Portfolio Management, 27, 61-74.
Asness, C.S. (2003), “Fight the Fed Model,” Journal of Portfolio Management, 30(1), (Fall), 11-24.
Bai J. and Perron P. (2003), “Computation and Analysis of Multiple Structural Change Models,” Journal of Applied
Econometrics, 18(1), 1-22.
Baker, M. and J. Wurgler (2000), The equity share in new issues and aggregate stock returns, Journal of Finance, 55,
2219-2257.
Balduzzi, (1995), “Stock Returns, Inflation and the ‘Proxy Hypothesis’: A New Look at the Data,” Economics
Letters, 48, 47-53.
Barnes, M., Boyd, J. H. and B.D. Smith (1999), “Inflation and asset returns,” European Economic Review, 43, 737-
754.
Bernanke, B.S. and K.N. Kuttner (2005), “What Explains the Stock Market’s Reaction to Federal Reserve Policy,”
Journal of finance, forthcoming, June.
Blanchard, O.J. (1993), “Movements in the equity premium,” Brookings papers on Economic Activity, 2, 75-138.
Bodie, Z. (1976), “Common Stocks As A Hedge Against Inflation,” Journal of Finance, 31, 459-470.
Bohl, M.T. and P.L. Siklos. (2004), “The present value model of U.S. stock prices redux:a new testing strategy and
some evidence,” Quarterly Review of Economics and Finance, 44, 208-223.
Bomfim, A.N. (2003), “Pre-announcement effects, news effects, and volatility: Monetary policy and the stock
market,” Journal of Banking and Finance, 27, 133-151.
Bossaerts, and Hillion (1999), “Implementing Statistical Criteria to Select Return Forecasting Models:What do we
Learn?,” Review of Financial Studies, 12, 405-428.
Boudoukh, J. and M. Richardson (1993), “Stock Returns and Inflation: A Long Horizon Perspective,” American
Economic Review, 83(5), 1346-1355.
Brennan, M. and Y. Xia (2004), “tay’s as good as cay,” unpublished paper, Anderson School Business, University of
California, Los Angeles.
Buono, M.J. (1989), “The Relationship Between The Variability Of Inflation And Stock Returns: An Empirical
Investigation,” Journal of Financial Research, 12, 329-340.
Campbell, J.Y. (2000), “Asset Pricing at the Millennium,” Journal of Finance, 55, 1515-1567
Campbell, J.Y. (1991), “A variance decomposition for stock returns,” Economic Journal, 10(1), 157-179.
Campbell, J.Y. (1987), “Stock returns and the term structure,” Journal of Financial Economics, 18, 373-399.
Campbell, J.Y., Lettau, M., Malkiel B. and Y. Xu (2001), “Have Individual Stocks Become More Volatile? An
Empirical Exploration of Idiosyncratic Risk,” Journal of Finance, 56, 1-43.
Campbell, J.Y., Lo, A. and A.C. MacKinlay (1997), The econometrics of financial markets, Princeton University
Press.
Campbell, J.Y. and R.J. Shiller (1998), “Valuation ratios and the long-run stock market outlook,” Journal of
Portfolio Management, Winter, 11-26.
Campbell, J.Y. and R. J. Shiller (1988a), “Stock Prices, Earnings and Expected Dividends,” Journal of Finance, 43,
661-676.
Campbell, J.Y. and R.J. Shiller (1988b), “The Dividend-Price Ratio and Expectations of Future Dividends and
Discount Factors,” Review of Financial Studies, 1(3), 195-228.
Campbell, J.Y. and R. Shiller (1987), “Cointegration and tests of present value models,” Journal of Political
Economy, 95(9), 1062-1088.
Campbell, J.Y. and S.B. Thompson (2004), “Predicting the Equity Premium Out of Sample: Can Anything Beat the
Historical Average?,” working paper, Harvard University.
Campbell, J.Y. and T. Vuolteenaho (2004), “Inflation Illusion and Stock Prices,” American Economic Review,
Papers and Proceedings, 94, 19-23.

27
Campbell J.Y. and M. Yogo (2004), “Efficient Tests of Stock Return Predictability,” Working Paper, Harvard
University, Department of Economics.
Caporale, T. and C. Jung (1997), “Inflation and Real Stock Prices,” Applied Financial Economics, 7, 265-266.
Carlson J.B., Pelz E.A. and M. Wohar (2002), “Will the Valuation Ratios Revert to Their Historical Means? Some
Evidence From Breakpoint Tests,” Journal of Portfolio Management, Summer, 22-35.
Clark, T.E. and M.W. McCracken (2004), “Evaluating long-horizon forecasts,” Manuscript, University of Missouri
at Columbia.
Clark, Todd E. and M.W. McCracken (2001), “Tests of Equal Forecast Accuracy and Encompassing for Nested
Models,” Journal of Econometrics, 105 (November), 85-110.
Claus, J.and J. Thomas (2001), “Equity Premia as Low as Three Percent? Evidence from Analysts’ Earning
Forecasts for Domestic and International Stock Markets,” Journal of Finance, 56, 1629-1666.
Coakley, J. and A. Fuertes. (2003), “Momentum and reversal dynamics in U.S. valuation ratios,” Working Paper,
University of Essex.
Cochran, S.J. and R.H. DeFina (1993), “Inflation’s Negative Effects on Real Stock Prices: New Evidence and a Test
of the Proxy Hypothesis,” Applied Economics, 25, 263-274.
Cochrane, J.H. (1999), “New Facts in Finance,” Economic Perspectives, Federal Reserve Bank of Chicago, 23, 36-
58
Cochrane, J.H. (1991), “Volatility tests and efficient markets: A review essay,” Journal of Monetary Economics,
127, 463-485.
Cozier, B. and A. Rahman (1988), “Stock Returns, Inflation, and Real Activity in Canada,” Canadian Journal of
Economics, 21, November, 759-74.
Diba B. and H. Grossman (1988), “Explosive Rational Bubbles in Stock Prices?,” American Economic Review, 78,
520-530.
Diebold, F.X. and R.S. Mariano (1995), “Comparing Predictive Accuracy,” Journal of Business and Economic
Statistics, 13, 253-263.
Fama, E.F. (1981), “Stock Returns, Real Activity, Inflation, and Money,” American Economic Review, 71(4), 545-
565.
Fama, E.F. and K.R. French (2002), “The Equity Premium,” Journal of Finance, 57, 637-659.
Fama, E.F. and K.R. French (2001), “Disappearing Dividends: Changing Firm Characteristics or Lower Propensity
to Pay?,” Journal of Financial Economics, 60, 3-43.
Fama, E.F. and K.R. French (1989), “Business conditions and expected returns on stocks and bonds,” Journal of
Financial Economics, 25, 23-49.
Fama, E.F. and K.R. French (1988), “Dividend yields and expected stock returns,” Journal of Financial Economics,
22, 3-25.
Fama, E.F. and G.W. Schwert (1977), “Asset Returns and Inflation,” Journal of Financial Economics, 5, 115-146.
Feldstein, M. (1980), “Inflation and the stock market,” American Economic Review, 70, 839-847.
Ferson, W. E., Sarkissian, S. and T. Simin (2003), “Spurious Regressions in Financial Economics?,” Journal of
Finance, 58, 1393-1413.
Firth, M. (1979), “The Relationship between Stock Market Returns and Rates of Inflation,” Journal of Finance, 34,
743-749.
Fisher, I. (1930), The Theory of Interest, Macmillan, New York.
Friedman, M. (1977), “Nobel Lecture: Inflation and Unemployment,” Journal of Political Economy, 85, 451-472
Geske, R. and R. Roll (1983), “The Fiscal and Monetary Linkage between Stock Returns and Inflation,” Journal of
Finance, 38, 1-34.
Goetzmann, W.N. and Jorion (1993), “Testing the predictive power of dividend yields,” Journal of Finance, 48, 663-
679.
Goyal, A. and I. Welch (2004), “A Comprehensive Look at the Empirical Performance of Equity Premium
Prediction,” Working Paper, April.
Goyal, A. and I. Welch (2003), “Predicting the equity premium with dividend ratios,” Management Science, 49(5),
639-654.
Graham, F. C. (1996), “Inflation, Real Stock Returns and Monetary Policy,” Applied Financial Economics, 6, 29-35.
Graham, B. and D.L. Dodd (1934), Security Analysis, first edition, Mc-Graw Hill, New York, NY.
Gultekin, N.B. (1983), “Stock Market Returns and Inflation: Evidence from Other Countries,” Journal of Finance,
38, 49-65.
Guo, H. (2006), “On the Out-of-Sample Predictability of Stock Market Returns,” Journal of Business, 79(3),
forthcoming.

28
Hamilton, J.D. (1994), Time Series Analysis, Princeton: Princeton University Press.
Hansen, B.E. (1997), “Approximate Asymptotic P-Values for Structural Change Tests,” Journal of Business and
Economic Statistics, 15(1), p 60-67.
Harvey, D.I., Leybourne, S.J. and P. Newbold (1998), “Tests for Forecast Encompassing,” Journal of Business and
Economic Statistics, 16(2), 254-259.
Heaton, J. and D. Lucas (1999), “Stock Prices and Fundamentals,” in B.S. Bernanke and Rotemberg J.J. (eds), NBER
Macroeconomics Annual, 14, London: MIT Press, 213-242.
Hodrick, R.J. (1992) “Dividend yields and expected stock returns: alternative procedures for inference and
measurement,” Review of Financial Studies, 5(3), 357-386.
Hu, X. and T.D. Willett (2000), “The Variability of Inflation and Real Stock Returns,” Applied Financial Economics,
10(6), 655-665.
Inoue, A. and L. Kilian (2004), “In-sample or out-of-sample tests of predictability: which one should we use?,”
Econometric Reviews, 23(4), 1-32.
Jaffe, J. and G. Mandelker (1976), “The ‘Fisher Effect’ for Risky Assets: An Empirical Investigation,” Journal of
Finance, 31, 447-458.
Jagannathan, R., McGrattan E.R. and A. Scherbina (2001), “The Declining U.S. Equity Premium,” NBER Working
paper 8172.
James, C., S., Koreisha and M. Partch (1985), “A VARMA Analysis of the Causal Relations among Stock Returns,
Real Output, and Nominal Interest Rates,” Journal of Finance, 40, 1375-1384.
Johansen, S. (1988), “Statistical analysis of cointegrating vectors,” Journal of Economic Dynamics and Control, 12,
231-254.
Johansen, S. and K. Juselius (1992), “Testing Structural Hypotheses in a Multivariate Cointegration Analysis of the
PPP and the UIP for UK,” Journal of Econometrics, 53, 211-244
Kaul, G. (1990), “Monetary Regimes and the Relation between Stock Returns and Inflationary Expectations,”
Journal of Financial and Quantitative Analysis, 25(3), 307-321.
Keim, D.B. and R.F. Stambaugh (1986), “Predicting Returns in the Stock and Bond Markets,” Journal of Financial
Economics, 17, 357-390.
Kilian, L. (1999), “Exchange rates and monetary fundamentals: What do we learn from long-horizon regressions?,”
Journal of Applied Econometrics, 14, 491-510.
Kirby C. (1997), “Measuring the predictable variation in stock and bond returns,” Review of Financial Studies, 10(3),
579-630.
Kothari, S.P. and J. Shanken (1997), “Book-to-market, dividend yield, and expected market returns: A time series
analysis,” Journal of Financial Economics, 44, 169-203.
Lamont, O. (1998), “Earnings and Expected Returns,” Journal of Finance, 53, 1563-1587.
Lee, BS (1992), “Causal Relations Among Stock Returns, Interest Rates, Real Activity and Inflation,” Journal of
Finance, 47, 1591-1603.
Lettau, M. and S. Ludvigson (2001), “Consumption, aggregate wealth, and expected stock returns,” Journal of
Finance, 56, 815-849.
Lettau, M. and S. Ludvigson (2004a), “Expected Returns and Expected Dividend Growth,” Journal of Financial
Economics, forthcoming.
Lettau, M. and S. Ludvigson (2004b), “Understanding trend and cycle in asset values: Reevaluating the wealth effect
on consumption,” American Economic Review, 94(1), 276-299.
Lewellen, J.W. (2004), “Predicting returns with financial ratios,” Journal of Financial Economics, 74, 209-235.
Liang, J.N. and S.A. Sharpe (1999), “Share repurchases and employee stock options and their implications for S&P
500 share retirements and expected returns,” Working paper, Federal Reserve Board.
Lintner, J. (1975), “Inflation and Security Return,” Journal of Finance, 30, 259-280.
Ma Y. and A. Kanas (2004), “Intrinsic Bubbles Revisited: Evidence from Nonlinear Cointegration and Forecasting,”
Journal of Forecasting, 23, 237–250
Mankiw, N.G. and M.D. Shapiro (1986), “Do we reject too often?,” Economic Letters, 20, 139-145.
McCracken, M.W. (2004), “Asymptotics for out-of-sample tests of Granger causality,” Manuscript, University of
Missouri at Columbia.
Modigliani, F. and R.A. Cohn (1979), “Inflation, Rational Valuation and the Market,” Financial Analysts Journal,
35, 24-44.
Nelson, W.R. (1999), “The Aggregate Change in Shares and the Level of Stock Prices,” Finance and Economic
Discussion Series, no. 1999-08, Federal Reserve Board.
Nelson, C. (1976), “Inflation and Rates of Return on Common Stocks,” Journal of Finance, 31, 471-483.

29
Nelson, C.R. and M.J. Kim (1993), “Predictable stock returns: The role of small sample bias,” Journal of Finance,
48, 641-661.
Newey, W. and K.J. West (1987), “A simple, positive semi-definite, heteroskedasticity and autocorrelation
consistent covariance matrix,” Econometrica, 55, 703-708.
Phillips, P.C.B. and S. Ouliaris (1990), “Asymptoticproperties of residual based tests for cointegration,”
Econometrica, 58, 165-193.
Phillips, P.C.B. and Perron, P. (1988), “Testing for a unit root in timeseries regression”, Biometrika, 75, 335-346.
Pindyck, R.S. (1984), “Risk, inflation, and the stock market,” American Economic Review, 74(3), 335-51.
Pontiff, J. and L.D. Schall (1998), “Book-to-market ratios as predictors of market returns,” Journal of Financial
Economics, 49, 141-160.
Rapach D.E. and M.E. Wohar (2004a), “Valuation Ratios and Long-Horizon Stock Price Predictability,” Journal of
Applied Econometrics, forthcoming.
Rapach, D.E. and M.E. Wohar (2004b), “In-sample vs. out-of-sample tests of stock return predictability in the
context of data mining,” Working Paper, Seattle University.
Richardson, M. and J.H. Stock (1989), “Drawing inferences from statistics based on multiyear asset returns,” Journal
of Financial Economics, 25, 323-348.
Rigobon R. and B. Sack (2004), “The impact of Monetary Policy on Asset Prices,” Journal of Monetary Economics,
forthcoming.
Ritter, J.R. and R.S. Warr (2002), “The Decline of Inflation and the Bull Market of 1982-1999,” Journal of Financial
and Quantitative Analysis, 37(1), 29-61.
Rozeff, M. (1984), “Dividend yields are equity risk premiums,” Journal of Portfolio Management, 11, 68-75.
Rudd J. and K. Whelan (2002), “A Note on the Cointegration of Consumption, Income, and Wealth,” Finance and
Economics Discussion Series, 2002-53, Board of Governors of the Federal Reserve System, November.
Sharpe, S.A. (2002), “Reexamining Stock Valuation and Inflation: The Implications of Analysts’Earnings
Forecasts,” Review of Economics and Statistics, 84, 632-48.
Siklos, and B. Kwok (1999), “Stock Returns and Inflation: A New Test of Competing Hypotheses,” Applied
Financial Economics, 9, 567-581.
Solnik, B. (1983), “The Relation between Stock Prices and Inflationary Expectations: the International Evidence,”
Journal of Finance, 38, 35-48.
Stambaugh, R.F. (1986), “Biases in regressions with lagged stochastic regressors,” Working Paper No. 156,
Graduate School of Business, University of Chicago.
Stambaugh, RF (1999), “Predictive regressions,” Journal of Financial Economics, 54, 375-421.
Stock, J.H. and M.W. Watson (2003), “Forecasting Output and Inflation: The Role of Asset Prices,” Journal of
Economic Literature, 41, 788-829.
Stock, J.H. and M.W. Watson (1993), “A simple estimator of cointegrating vectors in higher order integrated
systems,” Econometrica, 6, 783-820.
Stock, J.H. and M.W. Watson (1988), “Testing for Common Trends,” Journal of the American Statistical
Association, 83, 1097-1107.
Thorbecke, W. (1997), “On Stock Market Returns and Monetary Policy,” Journal of Finance, 52, 635-654.
Torous, W., Valkanov, R. and Yan, S. (2005), “On Predicting Stock Returns with Nearly Integrated Explanatory
Variables,” Journal of Business, 78(1), forthcoming.
Valkanov, R. (2003), “Long-Horizon Regressions: Theoretical Results and Applications,” Journal of Financial
Economics, 68(2), 201-232.
West, K.D. (1996), “Asymptotic inference about predictive ability,” Econometrica, 64, 1067-1084.
Zhong, M., Darrat, A.F. and D.C. Anderson (2003), “Do US stock prices deviate from their fundamental values?
Some new evidence,” Journal of Banking and Finance, 27, 673-697.

30
 Table 1. Phillips-Ouliaris and Johansen Cointegration Tests

P-O Trace Test Johansen Trace Test

Zt # of
Test 90% 95% Test 90% 95%
coint. lags LM(1) LM(4)
Stat. CV CV relation Stat. CV CV

[ ep , π ]
t t
60.05 47.59 55.22
0
1
23.29
3.98
17.79
7.50
19.99
9.13
6 5.19
(p=0.27)
7.78
(p=0.10)

[ ep , π ]
t t −1
61.71 47.59 55.22
0
1
20.84
4.53
17.79
7.50
19.99
9.13
4 4.17
(p=0.38)
9.16
(p=0.06)

⎡⎣ ept , π tkal ⎤⎦ 0 11.58 17.79 19.99 5.69 3.31

55.57 47.59 55.22 4
1 5.59 7.50 9.13 (p=0.22) (p=0.51)

⎡⎣ ept , π tarima ⎤⎦ 0 22.85 17.79 19.99 4.31 7.87

69.36 47.59 55.22 6
1 2.99 7.50 9.13 (p=0.37) (p=0.10)

⎡⎣ ept , π tols ⎤⎦ 0 11.72 17.79 19.99 4.59 5.08

67.77 47.59 55.22 13
1 1.96 7.50 9.13 (p=0.33) (p=0.28)

[ ep , tb ]
t t
33.61 47.59 55.22
0
1
17.20 17.79 19.99
5.29 7.50 9.13
3 2.40 3.27
(p=0.66) (p=0.51)
Note: The table reports tests of the null hypothesis of no cointegrating relationships against the
alternative of one or more cointegrating vectors. “Lags” gives the number of lags in the estimated VAR
model. The appropriate lag-length is selected in order to accept the assumption that residuals are white
noise based on LM(1) and LM(4) criteria. A test statistic greater than the specified critical value suggests
rejection of the null of no cointegration. Significant coefficients at the 5% level are highlighted in bold
face.
 Table 2. Estimated constrained VECM

i ∆pt ∆e t
∆π t
∆p t − i 1 0.11 0.09 -0.01
2 -0.04 0.06 0.00
3 -0.01 -0.01 0.01
4 0.00 0.00 0.00
5 -0.04 0.02 0.01
6 -0.01 -0.10 -0.02
∆et − i 1 -0.10 0.69 0.01
2 0.04 0.07 0.01
3 -0.05 -0.03 0.00
4 0.15 -0.44 0.01
5 0.05 0.43 -0.03
6 -0.09 -0.16 0.03
∆π t − i 1 -0.48 0.35 0.25
2 0.69 0.07 0.11
3 -1.74 -0.15 0.28
4 1.05 -0.52 -0.32
5 0.11 0.29 -0.04
6 0.23 -0.75 -0.02
ˆ t −i
- epi 1 -0.04 0.00 -0.00
ˆ t = pt − et + 18.92π t − 3.52
- epi
Note: The sample period is firth quarter of 1948 to firth quarter
2004. Significant coefficients at the 5% level are highlighted in
bold face.

32
 Table 3. Summary statistics

Correlation Matrix
rt rt − rf ,t et − pt dt − pt dt − et RRELt DEFt TRM t σt ˆ t
cay ˆt
epi ˆ t
epb
rt 1,00 0,15 0,23 0,26 0,05 -0,21 0,10 0,10 0,00 0,31 0,33 0,29
rt − rf ,t 1,00 0,21 0,25 0,07 -0,20 0,11 0,14 0,11 0,31 0,32 0,32
et − pt 1,00 0,90 -0,28 0,08 0,38 -0,19 -0,13 0,25 0,69 0,79
dt − pt 1,00 0,18 0,01 0,34 -0,03 -0,24 0,36 0,73 0,82
dt − et 1,00 -0,16 -0,10 0,37 -0,23 0,23 0,05 0,01
RRELt 1,00 -0,28 -0,23 -0,19 -0,16 0,06 0,09
DEFt 1,00 0,28 0,27 0,06 -0,03 0,03
TRM t 1,00 0,04 0,32 0,00 -0,02
σt 1,00 -0,06 -0,31 -0,23
ˆ t
cay 1,00 0,28 0,30
ˆt
epi 1,00 0,83
ˆ t
epb 1,00

Univariate Summary Statistics

Mean 0.05 0.04 -2.73 -3.43 -0.70 0.00 0.01 0.01 0.00 0.00 0.00 0.01
SD 0.08 0.07 0.39 0.39 0.18 0.01 0.00 0.01 0.00 0.01 0.32 0.34
Max 0.28 0.29 -1.92 -2.78 -0.27 0.05 0.03 0.04 0.00 0.03 0.83 0.82
Min -0.23 -0.23 -3.84 -4.50 -1.19 -0.04 0.00 -0.03 0.00 -0.04 -0.88 -0.81
Autoc. 0.16 0.15 0.97 0.95 0.96 0.51 0.91 0.80 0.51 0.83 0.96 0.95
Note: The sample spans the fourth quarter of 1951 to the second quarter of 2003 except for the term spread,
TRM t , which begin the second quarter of 1953.

Table 4. Estimates of the VAR models

Equation 1 Equation 2

Endogenous variable ∆pt ∆d t d t − et rf ,t ∆pt ∆et d t − et rf ,t

ˆ t −1
epi 0.055 0.008 0.026 0.006 0.053 -0.015 0.026 0.002

(p-value) (0.001) (0.009) (0.004) (0.224) (0.002) (0.064) (0.003) (0.448)

R
2
0.04 0.17 0.95 0.73 0.05 0.48 0.95 0.71

LAG =2 LAG =1

Both AIC and SIC are used to select the lag length. Estimated coefficients of the endogenous
variables are not shown. Significant coefficients at the 5% level are highlighted in bold face.

33
 Table 5. Forecasting Quarterly Excess Returns

Constant lag ˆ t
cay ˆt
epi ˆ t
epb σt dt − pt et − pt dt − et RRELt TRM t DEFt
# R²
(t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat)

Panel A: Excess Returns

0.037 0.147
1 (6.520) (2.171)
0.017
0.044 1.813
2 (8.515) (4.460)
0.086
0.041 0.059 1.702
3 (8.184) (0.939) (4.371)
0.084
0.043 0.072
4 (8.669) (4.418)
0.094
0.041 0.054 0.068
5 (7.595) (0.853) (4.153)
0.092
0.042 0.065
6 (8.163) (3.981)
0.083
0.039 0.060 0.061
7 (6.921) (0.887) (3.670)
0.082
0.044 1.401 0.057
8 (9.526) (3.495) (3.708)
0.139
♦ 0.035 1.181 0.075 87.137 -0.003 -1.170 0.170 -0.116
9 (1.657) (2.939 (4.700 (1.199) (-0.092) (-2.607 (0.159) (-0.267)
0.166
0.034 1.298 0.069 114.880 -0.614
10 (5.265) (3.334) (4.474) (1.609) (-1.561)
0.152
0.130 1.558 0.032
11 (3.127) (3.629) (2.113)
0.107
0.158 1.436 0.034
12 (2.814) (3.312) (2.034)
0.108
0.213 0.050
13 (4.011) (3.181)
0.063
0.164 0.044
14 (3.957) (2.977)
0.049
♦♦ 0.043 0.061
15 (8.389) (3.922)
0.085

Panel B: Real Returns

0.039 0.158
16 (6.633) (2.254)
0.025
0.047 1.868
17 (8.889) (4.469)
0.089
0.044 0.068 1.737
18 (8.440) (1.020) (4.332)
0.089
0.046 0.074
19 (9.022) (4.381)
0.096
0.043 0.063 0.069
20 (7.537) (0.974) (4.164)
0.096
0.046 0.061
21 (8.310) (3.583)
0.071
0.042 0.085 0.055
22 (6.781) (1.188) (3.298)
0.073
0.047 1.446 0.059
23 (9.958) (3.449) (3.650)
0.144
♦ 0.034 1.201 0.076 71.131 -0.006 -1.227 0.501 -0.073
24 (1.536) (2.948) (4.721) (0.968) (-0.184) (-2.752) (0.473) (-0.170)
0.175
0.132 1.619 0.031
26 (2.999) (3.618) (1.964)
0.109
0.157 1.504 0.032
27 (2.676) (3.306) (1.876)
0.109
0.166 0.044
28 (3.907) (2.886)
0.047
0.215 0.049
29 (3.949) (3.078)
0.060
♦♦ 0.046 0.062
30 (8.703) (3.830)
0.085

Note: The table reports estimates from OLS regressions of stock returns on lagged variables named at the head
of a column. Regressions use data from the fourth quarter of 1951 to the second quarter of 2003, except for
regression 13 and 24 (indicated by ♦), which begins in the second quarter of 1953, the largest common sample
for which all the data are available. Rows 15 and 30 (indicated by ♦♦) report estimates with the log dividend-
price inflation ratio as regressor. Newey-West corrected t-statistics appear in parentheses below the coefficient
estimate. Significant coefficients at the 5% level are highlighted in bold face.

34
 Table 6. One-Quarter-Ahead Forecasts of Excess Returns:
Nested Comparisons

ENC-NEW MSE-F
Comparison
Row unrestricted vs. MSEu/MSEr Statistic 99 percent CV Statistic 99 percent CV
restricted
Panel A: Cointegrating Vector Reestimated
1 ˆ t vs. AR
epi 0.9392 9.066** 4.251 9.339** 3.970
2 ˆ t −1 vs. AR
epi 0.9472 7.998** 4.251 8.068** 3.970
3 ˆ t vs. const
epi 0.9264 11.621** 4.251 11.129** 3.970
4 ˆ t −1 vs. const
epi 0.9344 10.562** 4.251 9.793** 3.970
Panel B: Fixed Cointegrating Vector
5 ˆ t vs. AR
epi 0.9328 12.889** 4.251 10.227** 3.970
6 ˆ t −1 vs. AR
epi 0.9472 10.020** 4.251 7.908** 3.970
7 ˆ t vs. const
epi 0.9216 16.266** 4.251 12.004** 3.970
8 ˆ
epit −1
vs. const 0.9376 13.213** 4.251 9.418** 3.970
Note: The MSE-F statistic is used to test the null hypothesis that the MSE for the unrestricted model forecasts is
less than or equal to the MSE for the restricted model forecasts. The ENC-NEW statistic is used to test the null
hypothesis that restricted model forecasts encompass the unrestricted model forecasts. We estimate the
cointegration parameters recursively in panel A and using the full sample in panel B. We consider a restricted
(benchmark) model of autoregressive returns (AR) in rows 1,2,5 and 6. A restricted (benchmark) model of
constant returns (const) is considered in rows 3,4,7 and 8. Each of these model includes a constant. MSEu is the
mean-squared forecasting error from the relevant unrestricted model in each row; MSEr is the mean-squared
error from the relevant restricted model. A number less than one indicates that the unrestricted model has lower
forecasting error than the restricted model. The initial estimation period begins with the fourth quarter of 1953
and ends with the first quarter of 1968. The model is recursively reestimated until the second quarter of 2003. A
* (**) denotes significance at the five (one) percent level.

35
 Table 7. One-Quarter-Ahead Forecasts of Real Returns:
Nested Comparisons

ENC-NEW MSE-F
Comparison
Row unrestricted vs. MSEu/MSEr Statistic 99 percent CV Statistic 99 percent CV
restricted
Panel A: Cointegrating Vector Reestimated
1 ˆ t vs. AR
epi 0.943 8.406** 4.251 8.665** 3.970
2 ˆ t −1 vs. AR
epi 0.950 7.430** 4.251 7.541** 3.970
3 ˆ t vs. const
epi 0.930 11.141** 4.251 10.668** 3.970
4 ˆ t −1 vs. const
epi 0.938 10.192** 4.251 9.500** 3.970
Panel B: Fixed Cointegrating Vector
1 ˆ t vs. AR
epi 0.931 13.027** 4.251 10.706** 3.970
2 ˆ t −1 vs. AR
epi 0.945 10.075** 4.251 8.380** 3.970
3 ˆ t vs. const
epi 0.916 16.938** 4.251 12.888** 3.970
4 ˆ t −1 vs. const
epi 0.933 13.745** 4.251 10.248** 3.970
Note: See Table 6.

36
 Table 8. One-Quarter-Ahead Forecasts of Excess Returns:
Nonnested Comparisons

MDM test
Test
Row Model 1 vs. Model 2 MSE1/MSE2 p value
Statistic
Panel A: Cointegrating Vector Reestimated
1 ˆ t vs. rt − rf ,t
epi 0.963 3.099** 0.002
♦
2 ˆ t vs. dt − pt
epi 0.976 1.972 0.051
3 ˆ t vs. et − pt
epi 0.970 2.740** 0.007
4 ˆ t vs. dt − et
epi 0.948 2.405* 0.018
5 ˆ t vs. RRELt
epi 0.963 3.003** 0.003
♦♦
6 ˆ t vs. cay
epi ˆ t 0.991 1.561 0.121
7 epi ˆ t + σt
ˆ t vs. cay 0.967 2.556* 0.012
8 ˆ t vs. epb
epi ˆ t 0.949 2.699** 0.008
Panel B: Fixed Cointegrating Vector
9 ˆ t vs. rt − rf ,t
epi 0.960 3.435** 0.000
10 ˆ t vs. dt − pt
epi 0.973 2.660** 0.009
11 ˆ t vs. et − pt
epi 0.967 2.374* 0.019
12 ˆ t vs. dt − et
epi 0.946 2.969** 0.003
13 ˆ t vs. RRELt
epi 0.960 3.513** 0.000
14 ˆ t vs. cay
epi ˆ t 0.996 2.982** 0.003
15 epi ˆ t + σt
ˆ t vs. cay 0.976 3.696** 0.000
16 ˆ t vs. epb
epi ˆ t 0.990 2.037* 0.044
Note: The MDM test, is a modified Diebold and Mariano (1995) test statistic to test for forecast encompassing
ˆ t as
between two non-nested models and to account for finite-sample biases. Model 1 always uses just lagged epi
a predictive variable; Model 2 uses one of several alternate variables. All of the models include a constant. The
null hypothesis is that the model 2 encompasses model 1. We estimate the cointegration parameters recursively
in panel A and using the full sample in panel B. The column labeled “MSE1/MSE2” reports the ratio of the root-
mean-squared forecasting error of Model 1 to Model 2. A number less than one indicates that the model 1 has
lower forecasting error than the model 2. The initial estimation period begins with the fourth quarter of 1953 and
ends with the first quarter of 1968. The model is recursively reestimated until the second quarter of 2003. A *
(**) denotes significance at the five (one) percent level. ♦ The inverse encompassing test that under the null
model 1 encompasses model 2 is not rejected (p-value = 0.675). ♦♦ The inverse encompassing test that under the
null model 1 encompasses model 2 is not rejected (p-value = 0.353).

37
 Table 9. One-Quarter-Ahead Forecasts of Real Returns:
Nonnested Comparisons

MDM test
Test
Row Model 1 vs. Model 2 MSE1/MSE2 p value
Statistic
Panel A: Cointegrating Vector Reestimated
1 ˆ t vs. rt − rf ,t
epi 0.963 3.008** 0.003
♦
2 ˆ t vs. dt − pt
epi 0.977 1.902 0.059
3 ˆ t vs. et − pt
epi 0.972 2.803** 0.006
4 ˆ t vs. dt − et
epi 0.950 2.433* 0.016
5 ˆ t vs. RRELt
epi 0.968 2.880** 0.005
6 epi ˆ t ♦♦
ˆ t vs. cay 0.992 1.512 0.133
7 epi ˆ t + σt
ˆ t vs. cay 0.967 2.593* 0.011
8 ˆ t vs. epb
epi ˆ t 0.950 2.698** 0.008
Panel B: Fixed Cointegrating Vector
9 ˆ t vs. rt − rf ,t
epi 0.958 3.572** 0.000
10 ˆ t vs. dt − pt
epi 0.970 2.823** 0.005
11 ˆ t vs. et − pt
epi 0.965 2.363* 0.019
12 ˆ t vs. dt − et
epi 0.943 3.063** 0.003
13 ˆ t vs. RRELt
epi 0.961 3.499** 0.000
14 ˆ t vs. cay
epi ˆ t 0.994 3.118* 0.002
15 epi ˆ t + σt
ˆ t vs. cay 0.974 3.798** 0.000
16 ˆ
epit vs. ˆ
epbt 0.980 2.356* 0.020
Note: See Table 8. ♦ The inverse encompassing test that under the null model 1 encompasses model 2 is not
rejected (p-value = 0.671). ♦♦ The inverse encompassing test that under the null model 1 encompasses model 2 is
not rejected (p-value = 0.376)

38
 Table 10. Long-Horizon Regressions for Stock Prices Growth , Dividend Growth and
Earning Growth

Forecast Horizon k
# Regressors 1 2 3 4 8 12 16 20 24 48

Panel A : Real Stock Prices Growth

1 ˆt
epi 0.100 0.150 0.194 0.235 0.357 0.462 0.473 0.473 0.461 1.084
(3.042) (3.274) (3.464) (3.584) (4.992) (4.005) (3.462) (2.516) (1.904) (4.067)
[0.08] [0.12] [0.14] [0.16] [0.32] [0.21] [0.16] [0.11] [0.09] [0.21]
2 dt − pt 0.039 0.062 0.088 0.112 0.193 0.248 0.238 0.309 0.420 1.644
(1.249) (1.372) (1.525) (1.622) (1.769) (1.885) (1.605) (1.834) (2.043) (5.711)
[0.02] [0.03] [0.04] [0.05] [0.08] [0.08] [0.05] [0.05] [0.06] [0.34]
3 et − pt 0.029 0.047 0.068 0.088 0.136 0.166 0.131 0.129 0.156 1.106
(0.988) (1.127) (1.322) (1.446) (1.393) (1.331) (0.951) (0.769) (0.772) (4.527)
[0.01] [0.01] [0.02] [0.03] [0.03] [0.03] [0.01] [0.01] [0.01] [0.29]

Panel B : Real Dividend Growth

4 ˆt
epi 0.025 0.038 0.052 0.067 0.118 0.153 0.166 0.170 0.188 0.473
(3.130) (3.213) (3.440) (3.707) (4.310) (4.480) (4.174) (3.580) (3.511) (8.168)
[0.10] [0.13] [0.16] [0.18] [0.21] [0.21] [0.17] [0.14] [0.13] [0.48]
5 dt − pt 0.001 0.004 0.007 0.011 0.040 0.067 0.077 0.071 0.090 0.494
(0.209) (0.364) (0.477) (0.662) (1.495) (2.112) (2.022) (1.608) (1.722) (5.420)
[0.00] [0.00] [0.00] [0.01] [0.03] [0.05] [0.04] [0.02] [0.03] [0.36]
6 et − pt 0.001 0.001 0.003 0.005 0.019 0.030 0.022 0.000 0.000 0.297
(0.164) (0.195) (0.198) (0.279) (0.620) (0.744) (0.477) (0.007) (0.016) (3.993)
[0.00] [0.00] [0.00] [0.00] [0.00] [0.01] [0.00] [0.00] [0.00] [0.24]

Panel C : Real Earning Growth

7 ˆt
epi -0.003 0.001 0.013 0.031 0.089 0.056 -0.066 -0.157 -0.205 0.438
(-0.073) (0.015) (0.151) (0.297) (0.622) (0.371) (-0.469) (-1.092) (-1.378) (3.321)
[0.00] [0.00] [0.00] [0.00] [0.01] [0.00] [0.00] [0.02] [0.04] [0.17]
Note: The table reports estimates from OLS long-horizon regressions of real stock prices growth (Panel A), real
dividend growth (panel B) and real earning growth (Panel C) on lagged variables. For each regression, the t-
statistics, listed in parentheses, rely on a Newey-West correction. Adjusted R² statistics appear in square
brackets. Significant coefficients at the five percent level are highlighted in bold. The sample period is fourth
quarter of 1952 to third quarter 1998. Significant coefficients at the 5% level are highlighted in bold face. The
sample period spans from fourth quarter of 1951 to the second quarter of 2003

39
 Table 11. Long-horizon Regressions of Excess Returns

Forecast Horizon k
# Regressors 1 2 3 4 8 12 16 20 24 48
1 ˆ t
cay 3.452 5.003 6.294 7.688 11.465 13.458 13.235 11.972 10.344 3.154
(4.542) (4.787) (5.287) (5.616) (4.992) (5.128) (4.568) (4.053) (2.813) (0.615)
[0.138] [0.188] [0.218] [0.256] [0.323] [0.327] [0.245] [0.153] [0.096] [0.00]
2 ˆt
epi 0.139 0.193 0.246 0.295 0.495 0.648 0.674 0.635 0.583 0.744
(4.535) (4.380) (4.356) (4.395) (4.746) (5.001) (4.808) (3.771) (3.044) (3.747)
[0.155] [0.195] [0.229] [0.256] [0.380] [0.425] [0.379] [0.262] [0.191] [0.185]
3 ˆ t
epb 0.123 0.176 0.220 0.260 0.424 0.561 0.609 0.628 0.635 0.911
(4.013) (4.063) (3.999) (4.002) (4.441) (4.965) (4.999) (4.551) (4.005) (5.468)
[0.135] [0.179] [0.203] [0.224] [0.323] [0.384] [0.372] [0.304] [0.258] [0.316]
5 dt − pt 0.096 0.142 0.188 0.232 0.408 0.549 0.670 0.830 0.969 1.583
(3.333) (3.485) (3.646) (3.787) (4.740) (5.783) (7.084) (8.469) (8.668) (9.030)
[0.107] [0.151] [0.192] [0.227] [0.378] [0.448] [0.492] [0.518] [0.526] [0.584]
6 et − pt 0.086 0.126 0.165 0.200 0.355 0.475 0.524 0.574 0.620 1.167
(3.097) (3.235) (3.344) (3.417) (3.870) (3.975) (4.050) (3.964) (3.970) (8.262)
[0.084] [0.118] [0.147] [0.170] [0.278] [0.326] [0.334] [0.311] [0.304] [0.601]
♦ -1.615
7 ˆ t
cay 2.604 0.115 4.998 6.130 9.176 10.043 9.448 9.571 10.274
(3.553) (2.189) (4.659) (4.943) (5.304) (5.769) (5.124) (4.854) (3.778) (-0.583)
ˆt
epi 0.127 3.927 0.211 0.240 0.346 0.443 0.513 0.545 0.510 0.560
(4.189) (4.138) (3.855) (3.745) (4.170) (5.068) (5.517) (3.861) (3.079) (2.278)
dt − et -0.028 -0.036 -0.017 0.001 0.050 0.062 0.242 0.499 0.514 -1.337
(-0.531) (-0.508) (-0.198) (0.005) (0.401) (0.470) (1.371) (2.077) (1.697) (-3.454)
RRELt -1.785 -2.651 -2.794 -2.742 0.450 3.155 5.301 7.101 6.207 4.953
(-2.010) (-2.168) (-1.748) (-1.581) (0.295) (1.609) (2.138) (2.717) (2.101) (1.131)
DEFt 0.605 0.393 0.953 1.818 3.591 9.920 20.162 30.972 35.475 36.410
(0.286) (0.138) (0.268) (0.450) (0.764) (2.491) (3.557) (4.350) (4.565) (3.623)
TRM t -0.433 -0.856 -1.393 -1.686 -1.658 0.539 2.906 2.295 -0.587 8.095
(-0.507) (-0.776) (-1.034) (-1.033) (-0.766) (0.245) (1.033) (0.682) (-0.161) (1.796)
[0.248] [0.325] [0.365] [0.402] [0.526] [0.593] [0.583] [0.499] [0.392] [0.601]
♦ -2.871
8 ˆ t
cay 2.601 3.886 4.968 6.128 9.170 9.828 8.867 8.905 10.220
(3.639) (4.142) (4.570) (4.774) (4.734) (4.940) (4.359) (4.162) (3.932) (-1.201)
ˆ t
epb 0.106 0.147 0.178 0.201 0.278 0.372 0.486 0.605 0.668 0.816
(2.999) (3.038) (2.984) (2.919) (3.321) (4.365) (5.620) (5.982) (5.985) (3.661)
dt − et -0.016 -0.019 0.002 0.022 0.089 0.115 0.321 0.628 0.736 -0.636
(-0.276) (-0.250) (0.020) (0.202) (0.634) (0.795) (1.856) (3.193) (3.401) (-1.479)
RRELt -1.239 -1.906 -1.879 -1.682 2.086 5.131 7.906 10.333 10.526 13.033
(-1.322) (-1.457) (-1.094) (-0.896) (1.216) (2.319) (3.230) (5.032) (4.391) (3.137)
DEFt 0.424 0.139 0.665 1.496 3.437 9.591 19.874 31.244 37.746 48.775
(0.185) (0.045) (0.169) (0.326) (0.577) (1.581) (2.643) (4.233) (5.116) (5.725)
TRM t -0.115 -0.408 -0.826 -1.006 -0.490 2.140 4.981 4.504 1.762 10.960
(-0.131) (-0.352) (-0.582) (-0.568) (-0.199) (1.023) (1.929) (1.493) (0.590) (3.005)
[0.218] [0.295] [0.328] [0.366] [0.478] [0.555] [0.585] [0.566] [0.499] [0.666]
♦
9 ˆ t
cay 2.469 3.594 4.502 5.517 7.785 8.223 7.053 6.874 8.152 -7.272
(3.151) (3.514) (3.786) (3.900) (3.850) (3.798) (3.017) (2.974) (3.373) (-2.713)
dt − pt 0.094 0.140 0.180 0.209 0.347 0.445 0.575 0.759 0.928 1.246
(2.515) (2.746) (2.851) (2.851) (4.016) (4.481) (5.960) (6.862) (9.284) (4.908)
dt − et -0.080 -0.114 -0.118 -0.117 -0.152 -0.191 -0.052 0.179 0.256 -0.905
(-1.371) (-1.460) (-1.257) (-1.095) (-1.127) (-1.281) (-0.301) (1.006) (1.437) (-3.210)
RRELt -1.958 -2.958 -3.218 -3.248 -0.621 1.853 3.790 5.151 5.153 7.743
(-2.011) (-2.206) (-1.869) (-1.759) (-0.349) (0.896) (1.583) (2.299) (2.345) (2.069)
DEFt -3.375 -5.541 -6.705 -7.094 -11.400 -9.253 -4.146 0.244 1.100 8.609
(-1.213) (-1.563) (-1.529) (-1.386) (-1.680) (-1.418) (-0.567) (0.032) (0.152) (0.924)
TRM t 0.281 0.232 0.007 -0.045 0.985 3.979 7.347 7.420 5.602 15.271
(0.311) (0.195) (0.005) (-0.026) (0.434) (1.958) (2.726) (2.603) (1.966) (4.223)
[0.198] [0.283] [0.326] [0.368] [0.521] [0.586] [0.615] [0.619] [0.601] [0.718]
Note: The table reports estimates from OLS long-horizon regressions of excess returns on lagged variables. For
each regression, the t-statistics, listed in parentheses, rely on a Newey-West correction. Adjusted R² statistics
appear in square brackets. Significant coefficients at the five percent level are highlighted in bold. The sample
period is fourth quarter of 1952 to third quarter 1998. Significant coefficients at the 5% level are highlighted in
bold face. The sample period spans from fourth quarter of 1951 to the second quarter of 2003, except for
regression 7, 8 and 9 (indicated by ♦), which begins in the second quarter of 1953, the largest common sample
for which all the data are available.

40
 Table 12. Long-horizon Regressions of Real Stock Returns

Forecast Horizon k

# Regressors 1 2 3 4 8 12 16 20 24 48
1 ˆ t
cay 0.094 5.193 6.543 8.004 12.094 14.446 14.827 14.429 13.865 7.270
(4.555) (4.822) (5.320) (5.684) (5.329) (5.560) (5.041) (4.569) (3.431) (1.158)
[0.144] [0.195] [0.223] [0.259] [0.322] [0.328] [0.258] [0.182] [0.137] [0.012]
2 ˆt
epi 0.143 0.200 0.254 0.303 0.505 0.659 0.695 0.671 0.637 0.829
(4.554) (4.456) (4.451) (4.483) (4.711) (4.818) (4.616) (3.710) (3.051) (3.594)
[0.159] [0.199] [0.230] [0.253] [0.353] [0.383] [0.337] [0.238] [0.179] [0.158]
3 ˆ t
epb 0.117 0.167 0.208 0.246 0.400 0.527 0.575 0.599 0.626 0.013
(3.615) (3.662) (3.605) (3.600) (3.890) (4.263) (4.326) (4.046) (3.699) (5.119)
[0.116] [0.153] [0.171] [0.186] [0.256] [0.294] [0.276] [0.223] [0.195] [0.268]
5 dt − pt 0.096 0.144 0.190 0.235 0.417 0.564 0.699 0.880 1.058 1.911
(3.251) (3.432) (3.598) (3.750) (4.743) (5.764) (6.953) (8.232) (8.526) (9.525)
[0.103] [0.147] [0.185] [0.217] [0.353] [0.413] [0.447] [0.473] [0.490] [0.587]
6 et − pt 0.086 0.128 0.168 0.205 0.367 0.496 0.560 0.362 0.711 1.449
(3.027) (3.203) (3.346) (3.450) (3.965) (4.088) (4.188) (4.067) (4.056) (9.528)
[0.082] [0.117] [0.145] [0.166] [0.266] [0.309] [0.320] [0.306] [0.313] [0.640]
♦ 1.912
7 ˆ t
cay 0.067 4.071 5.203 6.415 9.868 11.153 11.126 12.091 13.905
(3.574) (4.140) (4.559) (4.779) (5.243) (5.614) (5.193) (5.315) (4.807) (0.565)
ˆt
epi 2.680 0.177 0.218 0.247 0.348 0.444 0.528 0.583 0.577 0.637
(4.249) (4.008) (3.921) (3.783) (4.006) (4.735) (5.400) (4.115) (3.442) (2.415)
dt − et -0.031 -0.040 -0.024 -0.008 0.047 0.062 0.234 0.490 0.509 -1.668
(-0.552) (-0.527) (-0.257) (-0.071) (0.336) (0.402) (1.237) (2.061) (1.705) (-3.850)
RRELt -1.839 -2.723 -2.859 -2.811 0.690 3.709 6.283 8.595 8.138 7.571
(-2.124) (-2.204) (-1.743) (-1.555) (0.388) (1.686) (2.376) (3.181) (2.677) (1.569)
DEFt 1.303 1.534 2.490 3.795 7.583 15.245 26.266 38.193 44.450 50.101
(0.632) (0.546) (0.700) (0.903) (1.325) (2.729) (3.567) (4.566) (5.028) (4.482)
TRM t -0.421 -0.907 -1.481 -1.821 -2.085 -0.046 2.517 2.150 -0.475 9.837
(-0.492) (-0.807) (-1.066) (-1.051) (-0.844) (-0.018) (0.781) (0.556) (-0.115) (1.873)
[0.263] [0.340] [0.377] [0.411] [0.513] [0.576] [0.576] [0.526] [0.455] [0.648]
♦ 0.540
8 ˆ t
cay 2.798 4.208 5.397 6.676 10.234 11.409 10.951 11.632 13.897
(3.786) (4.264) (4.600) (4.748) (4.756) (4.908) (4.505) (4.544) (4.864) (0.176)
ˆ t
epb 0.096 0.132 0.159 0.178 0.235 0.313 0.431 0.567 0.669 0.913
(2.519) (2.516) (2.444) (2.358) (2.474) (3.138) (4.244) (4.907) (5.336) (3.781)
dt − et -0.019 -0.023 -0.005 0.013 0.087 0.115 0.301 0.598 0.712 -0.889
(-0.312) (-0.284) (-0.050) (0.114) (0.553) (0.674) (1.545) (2.813) (3.220) (-1.906)
RRELt -1.298 -1.987 -1.952 -1.770 2.293 5.483 8.660 11.667 12.413 16.610
(-1.362) (-1.443) (-1.068) (-0.867) (1.108) (2.119) (3.079) (4.784) (4.685) (3.603)
DEFt 1.147 1.317 2.252 3.534 7.612 15.044 25.944 38.181 46.303 63.741
(0.476) (0.406) (0.540) (0.712) (1.081) (1.927) (2.672) (3.910) (4.725) (6.417)
TRM t -0.156 -0536 -1.009 -1.251 -1.042 1.377 4.450 4.407 2.032 13.119
(-0.168) (-0.426) (-0.645) (-0.632) (-0.364) (0.513) (1.391) (1.195) (0.570) (3.139)
[0.211] [0.283] [0.311] [0.346] [0.437] [0.501] [0.527] [0.532] [0.503] [0.699]
♦ -4.491
9 ˆ t
cay 2.640 3.874 4.880 6.010 8.830 9.768 9.050 9.513 11.701
(3.266) (3.619) (3.817) (3.893) (3.909) (3.886) (3.311) (3.524) (4.556) (-1.366)
dt − pt 0.089 0.132 0.170 0.196 0.315 0.403 0.545 0.753 0.977 1.413
(2.262) (2.479) (2.567) (2.554) (3.289) (3.695) (5.240) (6.434) (9.426) (5.350)
dt − et -0.079 -0.113 -0.118 -0.117 -0.132 -0.162 -0.046 0.165 0.226 -1.179
(-1.249) (-1.318) (-1.152) (-1.005) (-0.836) (-0.887) (-0.225) (0.791) (1.122) (-3.861)
RRELt -1.970 -2.969 -3.204 -3.219 -0.151 2.595 4.889 6.685 6.996 10.732
(-2.023) (-2.154) (-1.791) (-1.656) (-0.073) (1.069) (1.788) (2.557) (2.802) (2.580)
DEFt -2.447 -4.068 -4.722 -4.533 -5.946 -2.086 3.188 7.547 8.018 18.543
(-0.863) (-1.128) (-1.061) (-0.852) (-0.756) (-0.257) (0.345) (0.794) (0.866) (1.812)
TRM t 0.236 0.099 -0.183 -0.305 0.351 3.117 6.779 7.345 6.132 17.988
(0.248) (0.077) (-0.116) (-0.154) (0.127) (1.170) (2.076) (2.106) (1.844) (4.205)
[0.199] [0.281] [0.319] [0.358] [0.480] [0.538] [0.569] [0.594] [0.607] [0.746]
Note: The table reports estimates from OLS long-horizon regressions of real stock returns on lagged variables.
For each regression, the t-statistics, listed in parentheses, rely on a Newey-West correction. Adjusted R² statistics
appear in square brackets. Significant coefficients at the five percent level are highlighted in bold. The sample
period is fourth quarter of 1952 to third quarter 1998. Significant coefficients at the 5% level are highlighted in
bold face. The sample period spans from fourth quarter of 1951 to the second quarter of 2003, except for
regression 7, 8 and 9 (indicated by ♦), which begins in the second quarter of 1953, the largest common sample
for which all the data are available.

41
 Table 13. Long Horizon Forecasts of Excess Returns:
Nested Models

k 1 2 3 4 8 12 16 20 24 48
Panel A : Excess Returns
ˆ t vs. C
Panel A: Reestimated epi
MSEu/MSEr 0.941 0.898 0.859 0.842 0.850 0.783 0.764 0.818 0.846 0.653
ENC-NEW 8.992 14.497 17.997 20.160 16.080 24.555 31.709 27.279 25.408 30.469
(p-value) (0.000) (0.004) (0.004) (0.008) (0.048) (0.040) (0.036) (0.053) (0.068) (0.045)

MSE-F 9.032 15.886 22.742 25.976 23.692 36.120 38.977 27.231 21.432 50.043
(p-value) (0.000) (0.001) (0.000) (0.002) (0.014) (0.012) (0.018) (0.035) (0.058) (0.028)

ˆ t vs. C
Panel A2: Fixed epi
MSEu/MSEr 0.926 0.879 0.824 0.812 0.800 0.754 0.803 0.921 1.030 1.193

ENC-NEW 16.208 25.715 32.222 36.028 27.683 33.378 32.769 21.657 11.852 2.871
(p-value) (0.000) (0.000) (0.000) (0.001) (0.021) (0.027) (0.048) (0.088) (0.170) (0.373)

MSE-F 11.135 19.294 27.620 31.929 33.578 42.406 30.924 10.417 -3.480 -15.195
(p-value) (0.000) (0.000) (0.000) (0.000) (0.007) (0.011) (0.032) (0.105) (0.307) (0.664)

Panel B : Real Returns

ˆ t vs. C
Panel B1: Reestimated epi
MSEu/MSEr 0.942 0.904 0.871 0.859 0.886 0.839 0.831 0.872 0.882 0.670

ENC-NEW 8.498 13.388 16.054 17.298 11.786 17.056 21.294 19.146 19.086 27.578
(p-value) (0.001) (0.004) (0.007) (0.012) (0.078) (0.069) (0.066) (0.091) (0.099) (0.054)

MSE-F 8.704 14.960 20.625 22.686 17.206 24.855 25.555 17.899 15.787 46.209
(p-value) (0.000) (0.001) (0.002) (0.004) (0.027) (0.024) (0.034) (0.062) (0.081) (0.033)

ˆ t vs. C
Panel B2: Fixed epi
MSEu/MSEr 0.920 0.870 0.827 0.808 0.817 0.789 0.824 0.908 0.993 1.099

ENC-NEW 17.221 26.920 32.798 35.589 24.175 26.965 26.811 19.425 11.454 3.302
(p-value) (0.000) (0.000) (0.000) (0.001) (0.027) (0.042) (0.061) (0.101) (0.184) (0.359)

MSE-F 12.288 20.979 29.146 32.820 29.992 34.731 26.985 12.361 0.828 -8.436
(p-value) (0.000) (0.000) (0.000) (0.000) (0.009) (0.015) (0.037) (0.100) (0.221) (0.533)
Note: The MSE-F statistic is used to test the null hypothesis that the MSE for the unrestricted model
forecasts is less than or equal to the MSE for the restricted model forecasts. The ENC-NEW statistic is
used to test the null hypothesis that restricted model forecasts encompass the unrestricted model forecasts.
The dependent variable in Panel A is the k-period log excess returns. In Panel B, the dependent variable is
the k-period log real stock returns. We estimate the cointegration parameters recursively in panel A1 and
B1 and using the full sample in panel A2 and B2. We consider a restricted (benchmark) model of constant
returns. The rows labeled “MSEu/MSEr” report the ratio of the root-mean-squared forecasting error of the
unrestricted model 1 to the restricted model. A number less than one indicates that the unrestricted model
has lower forecasting error than the restricted model. The initial estimation period begins with the fourth
quarter of 1953 and ends with the first quarter of 1968. The model is recursively reestimated until the
second quarter of 2003. The p-values are calculated using a bootstrap based on Kilian (1999). The p-value
provides a measure of the rate at which null hypotheses are rejected. Significant coefficients at the 5%
level are highlighted in bold face.

42
Figure 1. Excess Returns and transitory deviations from the common trend in the
earning-price ratio and inflation
40%
1,7

30%
1,2
Trend Deviation

Excess Return 20%

0,7

0,2 10%

-0,3 0%

-0,8 -10%

-1,3 -20%

-1,8 -30%
1951 1957 1963 1969 1975 1981 1987 1993 1999

Figure 2. recursively estimated coefficient on inflation

25

20

15

10

0
1967 1972 1976 1981 1985 1990 1994 1999 2003

43
 Figure 3. Predicted and realized excess returns at long-horizons

0,5 0,5
0,4 k =1 0,4 k =2
0,3 0,3
0,2 0,2
0,1 0,1
0 0
-0,1 -0,1
-0,2 -0,2
-0,3 -0,3
-0,4 -0,4
1952 1958 1964 1970 1976 1982 1988 1994 2000 1952 1958 1964 1970 1976 1982 1988 1994 2000

0,7 0,8
0,6 k =3 k =4
0,5 0,6
0,4
0,4
0,3
0,2
0,2
0,1
0
0
-0,1
-0,2 -0,2
-0,3
-0,4 -0,4
1952 1958 1964 1970 1976 1982 1988 1994 2000 1952 1958 1964 1970 1976 1982 1988 1994 2000

1,2 1,4
1 k =8 1,2 k = 12
0,8 1

0,6 0,8
0,6
0,4
0,4
0,2
0,2
0
0
-0,2 -0,2
-0,4 -0,4
-0,6 -0,6
1952 1958 1964 1970 1976 1982 1988 1994 2000 1952 1958 1964 1970 1976 1982 1988 1994 2000

1,6 1,8
1,4 k = 16 k = 20
1,6
1,2
1,4
1
1,2
0,8
1
0,6
0,8
0,4
0,2 0,6

0 0,4

-0,2 0,2
-0,4 0
1952 1958 1964 1970 1976 1982 1988 1994 2000 1952 1958 1964 1970 1976 1982 1988 1994 2000

Note: The figure shows realized and predicted excess returns at different horizons by an
ˆ t as the sole predictor. The predictions are based upon the log earning-
univariate model with epi
price ratio. The gray line represents the realized excess returns and the black line represents the
predicted excess returns.

44
Appendix 1 Additional Tables

Table A1. Unit root and stationary tests

variables in level variables in differences

ADF lags KPSS ADF lags KPSS
et − pt -2.03 1 3.37 -8.10 3 0.05
dt − pt -1.09 0 9.97 -13.38 0 0.09
pt -1.56 1 6.33 -13.03 0 0.16
e t
-2.16 7 1.95 -6.48 6 0.04
d t
-2.73 8 1.74 -5.47 7 0.30
πt -1.86 8 0.53 -8.48 7 0.05
tbt -2.07 7 1.10 -6.22 6 0.15
dt − et -3.49 5 0.46
rft -4.11 7 0.34
Note: The ADF test statistics highlighted in bold face type are above the
critical value at the five percent significance level. The KPSS statistics
highlighted in bold face type are below the critical value at the five percent
significance level. ADF and KPSS critical values are from Hamilton (1994).
The maximum lags considered is 8. AIC is used to selct the lag length. The
sample period is firth quarter of 1948 to firth quarter of 2004.

45
 Table A2. Forecasting Quarterly Excess Returns with different specifications for the
normalized stock prices

Constant e1t − pt e5t − pt e10t − pt d1t − pt d 5t − pt d10t − pt ˆt

epi ˆt
dpi
(t-stat)
R²
# (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat

Panel A: Excess Returns

1 0.211 0.049
0.061
(4.742) (3.800)
2 0.205 0.047
0.056
(4.560) (3.627)
3 0.192 0.043
0.047
(4.291) (3.351)
4 0.043 0.072
0.094
(8.669) (4.418)
5 0.043 0.061
0.085
(8.389) (3.922)
6 0.162 0.043
0.047
(4.499) (3.336)
7 0.156 0.041
0.042
(4.327) (3.6169
8 0.152 0.039
0.041
(4.321) (3.132)

Panel B: Real Returns

9 0.213 0.049
0.059
(3.878) (3.011)
10 0.206 0.046
0.053
(3.623) (2.801)
11 0.192 0.042
0.044
(3.450) (2.631)
12 0.046 0.074
0.096
(9.022) (4.381)
13 0.046 0.062
0.085
(8.703) (3.830)
14 0.165 0.043
0.046
(3.825) (2.802)
15 0.159 0.041
0.041
(3.596) (2.566)
16 0.156 0.039
0.041
(3.659) (2.594)
Note: The table reports estimates from OLS regressions of stock returns on lagged variables named at the head
of a column. The variables labeled “ ent − pt ” are the log ratio of n-year average earnings to stock prices. The
variables labeled “ dnt − pt ” are the log ratio of n-year average dividends to stock prices. dpi
ˆ t is the log dividend-
price inflation ratio. Regressions use data from the fourth quarter of 1951 to the second quarter of 2003. Newey–
West corrected t-statistics appear in parentheses below the coefficient estimate. Significant coefficients at the 5%
level are highlighted in bold face. The sample period is fourth quarter of 1951 to second quarter of 2003.

46
 Table A3. Forecasting Quarterly Excess Returns
with different specifications of the log earning-price ratio

Constant ˆt
epi ˆ t(12l )
epi ˆ t(4l )
epi ˆ t( Joh )
epi ˆ t( −1)
epi
# R²
(t-stat) (t-stat) (t-stat) (t-stat) (t-stat) (t-stat)

Panel A: Excess Returns

0.044 0.072
1 (8.807) (4.423)
0.094
0.017 0.070
2 (1.990) (4.366)
0.092
0.043 0.071
3 (8.638) (4.395)
0.093
0.043 0.073
4 (8.632) (4.408)
0.094
0.043 0.072
5 (8.669) (4.418)
0.094

Panel B: Real Returns

0.047 0.074
6 0.096
(9.146) (4.380)
0.019 0.072
7 0.095
(2.269) (4.352)
0.046 0.073
8 0.096
(9.000) (4.371)
0.046 0.074
9 0.096
(8.971) (4.353)
0.046 0.074
10 0.096
(9.022) (4.382)
Note: The table reports estimates from OLS regressions of stock returns on lagged variables
named at the head of a column. Regressions use data from the fourth quarter of 1951 to the
second quarter of 2003. The regressors are as follows: epi
ˆ t(12l ) and epi
ˆ t(4l ) are the log earning-
price ratio estimated respectively with 12 and 4 lead/lag lengths in estimating the DOLS
specification; epi
ˆ t( Joh ) is the log earning-price ratio with the cointegrating parameters obtained
in section 3 based on Johansen’s (1988) full information maximum likelihood approach;
ˆ t( −1) is the log earning-price lag inflation ratio (from the cointegration relationship among
epi
the earning-price ratio and one lag of inflation). Newey–West corrected t-statistics appear in
parentheses below the coefficient estimate. Significant coefficients at the 5% level are
highlighted in bold face.

47
Appendix 2 Out of sample tests statistics for nested models

The sample is divided into in-sample and out-of-sample portions. The in-sample
portion spans observations 1 to R. Letting P − k + 1 denote the number of k -step (1 ≤ k) ahead
forecasts, the out-of-sample observations span R + k through R + P. The total number of
observations in the sample is R + P = T.

Calculation/definition of test statistics for equal MSE

The McCracken (2004) MSE-F statistic is a variant of the Diebold and Mariano (1995)
and West (1996) statistic designed to test for equal predictive ability. The MSE-F statistic is
used to test the null hypothesis that the unrestricted model forecast MSE is equal to the
restricted model forecast MSE against the one-sided (upper-tail) alternative hypothesis that
the unrestricted model forecast MSE is less than the restricted model forecast MSE.
T −k
Let dˆt + k = (uˆ1,t + k ) 2 − (uˆ2,t + k ) 2 and d = ( P − k + 1) −1 ∑ dˆt + k = MSE1 − MSE2 , where
t=R

MSEi = ∑ t = R (uˆi ,t + k ) 2 , i = 1, 2 , the McCracken (2004) MSE-F statistic is given by:

T −k

MSE-F = ( P − k + 1) ⋅ d / MSE2 (1)

Under the null that the mean square error associated with model 1 is the same as that
for model 2, the expected difference between u1,t2 + k and u2,t
2
+ k is zero. Under the alternative

the mean square error associated with model 2 (unrestricted model) will be smaller than that
for model 1 (restricted model).

Calculation/definition of test statistics for forecast encompassing

The Clark and McCracken (2001) ENC-NEW statistic is a variant of the Harvey,
Leybourne, and Newbold (1998) statistic to test for forecast encompassing between two non-
nested models.

48
 T −k
Let c = ( P − k + 1) −1 ∑ cˆt + k and cˆt + k = uˆ1,t + k (uˆ1,t + k − uˆ2,t + k ) . The Clark and McCracken (2001)
t=R

ENC-NEW statistic is given by:

ENC-NEW = ( P − k + 1) ⋅ c / MSE2 (2)

Under the null that the forecast from model 1 (restricted) encompasses that of model 2
(unrestricted), the covariance between u1,t + k and u1,t + k − u2,t + k will be less than or equal to

zero. Under the alternative that model 2 contains added information, the covariance should be
positive.

The MSE-F and ENC-NEW statistics have key power advantages over the original
Diebold and Mariano (1995), West (1996) and Harvey, Leybourne, and Newbold (1998)
statistics according to extensive Monte Carlo simulations in Clark and McCracken (2001,
2004).
The limiting distributions of the MSE-F and ENC-NEW statistics are non-standard and
pivotal for k = 1 (Clark and McCracken, 2001) when comparing forecasts from nested
models. Since the remaining tests have non-standard and non-pivotal limiting distributions for
k > 1 that are usually dependent upon unknown nuisance parameters, we follow Clark and
McCracken (2004) in using a bootstrap similar to that in Kilian (1999) to estimate
asymptotically valid critical values and construct asymptotically valid p-values.

49
Appendix 3 Bootstrap Procedure

The p-values associated with the MSE-F and ENC-NEW statistics are estimated using
a bootstrap similar to that discussed in Kilian (1999) and used by Clark and McCracken
(2004). Following Rapach and Wohar (2004b), we postulate that the data are generated by the
following system under the null hypothesis of no predictability:
yt = α 0 + ε1,t (1)
p
xt = β 0 + ∑ βi ⋅ xt −i + ε 2,t (2)
i =1

where the number of lags, p, is determined using AIC.

We compute the OLS residuals from estimated equations (1) and (2) and sample them
with replacement to obtain a set of bootstrap residuals, ε1,t* and ε 2,t
*
. We create the bootstrap

series yt* and xt* recursively using these bootstrap residuals and the estimated coefficients.
For the bootstrap series, we calculate the two tests statistics outlined in appendix 2.
For each test statistic, critical values are simply computed as percentiles of the bootstrapped
test statistics. Following Kilian (1999), the number of bootstrap draws is 2000.

50