Do Cash Flows of Growth Stocks Really Grow Faster?

• Author: HUAFENG (JASON) CHEN
• DOI: http://doi.org/10.1111/jofi.12518
• Published date: 01 October 2017
• Date: 01 October 2017

THE JOURNAL OF FINANCE •VOL. LXXII, NO. 5 •OCTOBER 2017

Do Cash Flows of Growth Stocks Really Grow

Faster?

HUAFENG (JASON) CHEN∗

ABSTRACT

Contrary to conventional wisdom, growth stocks (i.e., low book-to-market stocks) do

not have substantially higher future cash-flow growth rates than value stocks, in both

rebalanced and buy-and-hold portfolios. Efficiency growth, survivorship and look-back

biases, and the rebalancing effect help explain the results. These findings suggest that

duration alone is unlikely to explain the value premium.

GROWTH STOCKS,DEFINED AS STOCKS WITH LOW book-to-market ratios, have lower

future returns than value stocks with high book-to-market ratios. But do

growth stocks have substantially higher future cash-flow growth rates and

longer cash-flow durations? While this question is interesting in its own right,

it is important for the following reason. Several recent papers provide an

influential duration-based explanation for the value premium (Lettau and

Wachter (2007,2011), Croce, Lettau, and Ludvigson (2010)). This explana-

tion has two key ingredients: the term structure of equity is downward slop-

ing (long-duration assets earn lower expected returns), and growth and value

stocks differ substantially in the timing of cash flows, in that cash flows of

growth stocks grow faster than cash flows of value stocks. This duration-based

∗Huafeng (Jason) Chen is with PBC School of Finance, Tsinghua University, and Mays Busi-

ness School, Texas A&M University. Previous drafts circulated under the titles “What Does the

Value Premium Tell Us about the Term Structure of Equity Returns?” and “The Growth Pre-

mium.” I thank two anonymous referees; the Associate Editor; Kenneth Singleton (Editor); Ravi

Bansal; Jonathan Berk; Jules van Binsbergen; Oliver Boguth; John Campbell; Tarun Chordia;

John Cochrane; George Constantinides; Zhi Da; David De Angelis; Peter DeMarzo; Eugene Fama;

Adlai Fisher; George Gao; Anisha Ghosh; John Heaton; Ravi Jaganathan; Ralph Koijen; Martin

Lettau; Stefan Nagel; Stavros Panageas; ˇ

Luboˇ

sP

´

astor; Lawrence Schmidt; Jessica Wachter; Toni

Whited; Motohiro Yogo;Lu Zhang; Pietro Veronesi; and seminar participants at UBC, UBC summer

conference, Shanghai Advanced Institute of Finance, Cheung Kong GSB, City University of Hong

Kong, HKU, Chinese University of Hong Kong, HKUST, University of Iowa, Iowa State Univer-

sity, Stanford University, Duke/UNC asset pricing conference, Northwestern University, the first

ITAM conference, Western Finance Association Meeting, NBER Summer Institute Asset Pricing

workshop, Northern Finance Association Meeting, Pacific Northwestern Finance conference, UT

Dallas, University of Toronto,McGill University, TexasA&M, Penn State, University of Cambridge,

University of Oxford (economics), Peking University,Tsinghua University (PBCSF), Simon Fraser

University, University of Florida, and the Lone Star Finance Conference for comments. I thank

Haibo Jiang, Pablo Moran, Alberto Romero, and Kairong Xiao for excellent research assistance. I

have read the Journal of Finance’s disclosure policy and have no conflicts of interest to disclose.

DOI: 10.1111/jofi.12518

2279

2280 The Journal of FinanceR

explanation seems promising given that Binsbergen, Brandt, and Koijen (2012)

find a downward-sloping term structure of equity in the market portfolio and

among the leading asset pricing models that they review, only the model of Let-

tau and Wachter (2007) generates a downward-sloping term structure.1This

evidence raises the question of whether the difference between the timing of

growth and value stocks’ cash flows is sufficient to explain the value premium.

Existing empirical evidence on whether the cash flows of growth stocks grow

faster is puzzling. While several authors find that the dividends of value stocks

grow faster in rebalanced portfolios, conventional wisdom holds that in buy-

and-hold portfolios (or at the firm level), growth stocks have substantially

higher future cash-flow growth rates than value stocks. This view is suggested

by the name “growth stocks” and is apparently backed by empirical results.2

Yetit is puzzling because both buy-and-hold and rebalanced portfolios are valid

ways of looking at the data. The two kinds of portfolios give rise to two streams

of cash flows that have the same present values and the same first-year returns,

analogous to two dividend streams in a Miller and Modigliani (1961) setting.

Rebalanced portfolios tend to be used in empirical asset pricing (e.g., Fama

and French (1992)), and are likely to be more homogeneous over time, whereas

buy-and-hold portfolios correspond to firm-level behavior. Theoretical explana-

tions of the value premium typically start by modeling firm-level behavior and

therefore have direct implications for buy-and-hold portfolios. I explore both

approaches.

My results on cash-flow growth rates are as follows. Consistent with existing

studies, I find robust evidence that, in rebalanced portfolios, cash flows of

value stocks grow faster than growth stocks. Contrary to conventional wisdom,

however, I find that, in buy-and-hold portfolios, cash flows of growth stocks do

not grow substantially faster (and in fact often grow more slowly) than value

stocks. I provide four pieces of evidence on buy-and-hold portfolios. First, in

the modern sample period (after 1963), dividends in the growth quintile grow

only a little faster than those in the value quintile. The difference in long-run

growth rates is about 2% per year, which is substantially smaller than the 19%

assumed by duration-based explanations of the value premium. Second, in the

early sample period (before 1963), dividends of value stocks grow faster than

those of growth stocks, at least in the first 10 years after portfolio formation. The

difference is statistically significant at the 10% level. In the full sample period,

growth and value stocks have approximately the same dividend growth rates

1Giglio, Maggiori, and Stroebel (2015) and Lustig, Stathopoulos, and Verdelhan (2013) find

a downward term structure of discount rates in the housing and currency carry trade markets,

respectively. Boguth et al. (2012) and Schulz (2016) argue that the results in Binsbergen, Brandt,

and Koijen (2012) are driven at least in part by microstructure issues and taxes, respectively.

2A number of authors, including Chen (2004), have expressed views in line with the conventional

wisdom. Dechow, Sloan, and Soliman (2004)andDa(2009) find that growth stocks have a longer

cash-flow duration, a construct that is related to long-run cash-flow growth rates. For a classic

paper on the value premium, see Fama and French (1992). Extant literature shows that rebalanced

portfolios of value stocks have higher dividend growth rates (see Ang and Liu (2004), Bansal,

Dittmar, and Lundblad (2005), Hansen, Heaton, and Li (2008), Chen, Petkova, and Zhang (2008)).

Do Cash Flows of Growth Stocks Really Grow Faster? 2281

in buy-and-hold portfolios. Third, in the modern sample period, earnings of

value stocks grow faster than those of growth stocks, although the difference is

sometimes not statistically significant. Finally,in regressions of future dividend

growth rates on the book-to-market ratio, the coefficients are mostly positive

after I account for survivorship bias. When I reconcile the different results

between rebalanced and buy-and-hold portfolios, I find that rebalanced growth

rates should be higher than buy-and-hold growth rates for value stocks, while

the opposite is true for growth stocks, under mild conditions.

The conventional wisdom is widely held for at least four reasons. First, Gor-

don’s formula, P

D=1

r−g, suggests that, all else being equal, stocks with higher

prices should have higher cash-flow growth rates. Second, Fama and French

(1995) show that growth stocks have persistently higher returns on equity than

value stocks, even five years after they are sorted into portfolios. Third, in stan-

dard firm-level regressions of future dividend growth rates on book-to-market,

the coefficients are highly negative, even for dividend growth rates 10 years in

the future. Finally, Dechow, Sloan, and Soliman (2004)andDa(2009) find that

growth stocks have substantially longer cash-flow durations.

I address each of these four reasons in turn. First, when we compare value

stocks with growth stocks, all else is not equal. If we consider that value stocks

have higher expected returns than growth stocks, valuation models actually

imply that growth stocks have similar growth rates to value stocks in buy-

and-hold portfolios and lower growth rates than value stocks in rebalanced

portfolios.3Second, the results in Fama and French (1995) pertain to the be-

havior of the return on equity, which is relevant for studying the growth rate

of book equity, but do not imply that cash-flow growth rates for growth stocks

are higher. In fact, back-of-the-envelope calculations suggest that the results in

Fama and French (1995) imply that growth stocks have lower earnings growth

rates than value stocks initially.Changes in the return on equity (i.e., efficiency

growth) help explain this result. Third, the dividend growth rate regression is

subject to survivorship bias. After I account for survivorship bias, high book-to-

market equity no longer predicts a lower future dividend growth rate.4Finally,

Dechow, Sloan, and Soliman (2004)andDa(2009) are biased toward finding

longer cash-flow durations in growth stocks.

This paper builds on previous work that examines growth rates. Lakonishok,

Shleifer, and Vishny (1994) show (in their Table V) that equal-weighted port-

folios of extreme growth stocks have higher growth rates in some of the three

accounting variables they examine (earnings, accounting cash flow, and sales)

over the very short term, but often have lower growth rates from year 2 to

year 5 than extreme value stocks, an important result that has largely been

overlooked by the literature. Part of my contribution is to extend their work

3Interestingly, in studying the time series of the aggregate stock market, most authors (see

references in Cochrane (2011)) find that the dividend-price ratio does not predict the future dividend

growth rate. My finding provides cross-sectional evidence on this relation.

4Chen (2004) focuses on the forecasted future dividend growth rates from firm-level regressions

and thus his analysis is subject to survivorship bias.