Cash Flow News and Stock Price Dynamics

• Author: RICCARDO SABBATUCCI,DAVIDE PETTENUZZO,ALLAN TIMMERMANN
• DOI: http://doi.org/10.1111/jofi.12901
• Published date: 01 August 2020
• Date: 01 August 2020

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 4 •AUGUST 2020

Cash Flow News and Stock Price Dynamics

DAVIDE PETTENUZZO, RICCARDO SABBATUCCI, and ALLAN TIMMERMANN∗

ABSTRACT

We develop a new approach to modeling dynamics in cash flows extracted from daily

firm-level dividend announcements. We decompose daily cash flow news into a persis-

tent component, jumps, and temporary shocks. Empirically,we find that the persistent

cash flow component is a highly significant predictor of future growth in dividends

and consumption. Using a log-linearized present value model, we show that news

about the persistent dividend growth component predicts stock returns consistent

with asset pricing constraints implied by this model. News about the daily dividend

growth process also helps explain concurrent return volatility and the probability of

jumps in stock returns.

ONMOSTDAYS,A MULTITUDE of firms announces cash flow news, but the number

of firms, as well as the industries they belong to, can vary greatly over time.

Such variation gives rise to a highly irregular cash flow news process, which

complicates investors’ attempt to infer the underlying growth rate of cash flows

for individual firms, industries, and the economy as a whole. This is important

because cash flow news is a key determinant of investors’ forecasts of future

cash flows and of their risks, and hence ultimately of how stocks are priced.1

∗Davide Pettenuzzo is at International Business School, Brandeis University. Riccardo Sab-

batucci is at Department of Finance, Stockholm School of Economics. Allan Timmermann is at

Rady School of Management, UC San Diego. We are grateful to the editor, Stefan Nagel, an As-

sociate Editor, and two anonymous referees for many valuable suggestions on our paper. We also

received helpful comments from Jules van Binsbergen, Chris Polk, Federico Bandi, and partici-

pants at the Utah Winter Finance Conference 2019, CEPR Gerzensee 2018, BI-SHoF 2018, and

conferences in Toronto (WesternUniversity, April 2018), Toulouse (Toulouse School of Economics,

May 2018), and University of Chicago (May 2019). We are grateful to INQUIRE Europe for fi-

nancial support. We have read The Journal of Finance disclosure policy and have no conflicts of

interest to disclose.

Correspondence: Allan Timmermann, University of California, San Diego, Rady School of Man-

agement, 4S146 Otterson Hall, MC 0553, 9500 Gilman Drive, La Jolla, CA 92093-0553; e-mail:

atimmerm@ucsd.edu

1Patton and Verardo (2012) develop a rational learning model to explain the observed patterns

in firms’ betas around earnings announcements. Their model contains unobserved firm-specific and

common earnings innovation terms, and investors’ extraction of these components is modeled as

a Kalman filtering problem. Savor and Wilson (2016) develop a learning model in which investors

decompose cash flow news into firm-specific and market-wide components. Positive covariances

between the cash flow processes of individual firms and the broader market imply that bad (good)

news on individual firms’ cash flows result in decreased (increased) forecasts of aggregate cash

DOI: 10.1111/jofi.12901

C2020 the American Finance Association

2221

2222 The Journal of Finance R

While information extracted from firms’ cash flow announcements is critical

to understanding investors’ cash flow expectations and, in turn, movements

in stock prices, relatively few studies analyze the predictability of cash flows,

in most cases focusing instead on quarterly or annual changes in aggregate

dividends or earnings.2However, data aggregated in this manner may conceal

rich dynamic patterns in cash flows recorded at a higher frequency that reduces

our ability to study important questions such as how fast can cash flow growth

respond to changes in the underlying state of the economy.

Several challenges complicate attempts to measure daily cash flow dynamics.

First, the cash flows of most firms have a pronounced seasonal component

related to weather patterns and holiday sales. Second, the number of firms

announcing cash flow news on any given day can fluctuate between as little as

0 to more than 100 firms. Third, the particular date on which a firm announces

dividends can vary widely from year to year, requiring that care be taken in

constructing daily proxies that account for firm-specific effects. Fourth, there

is considerable heterogeneity across individual firms’ cash flow processes. The

combined effect of these various factors is that daily cash flow news tends to be

very lumpy.

To address these challenges, we develop a new approach that allows us to

measure and model dynamics in high-frequency (daily) cash flows. Specifically,

we account for firm-level heterogeneity and seasonality effects by taking a

bottom-up approach that starts from changes in individual firms’ dividends on a

given day relative to their value over the same quarter during the previous year.

In contrast to conventional smoothed estimates, only data on those firms that

announce dividend news on a given day are used to update the dividend growth

estimate, which ensure that our measure picks up changes in the cash flow

process in a timely fashion.3Moreover, by computing a dollar-weighted growth

estimate, we account for variation in the size of the firms that pay dividends on

any given day.4Ourdividend growth measure uses dividend announcements by

publicly traded firms as opposed to dividend payments, which form the basis for

flows. In turn, this cash flow learning channel implies that the stock returns of announcing firms

and of the aggregate market are positively correlated, justifying an announcement risk premium

for exposure to individual firms’ cash flows. These models do not allow for jumps in cash flows,

although, in practice, this is an important feature of earnings and dividend data.

2Cochrane (2008) finds little evidence of dividend growth predictability, while van Binsbergen

and Koijen (2010), Kelly and Pruitt (2013), and Jagannathan and Liu (2019) find that growth in

dividends is predictable.

3To illustrate the loss of information from the common practice of aggregating cash flow news

over the most recent 12-month period and updating this on, say, a monthly basis, suppose that

firms’ announcement dates are uniformly distributed across calendar dates. Each month when the

cash flow estimate is updated, the same weight is assigned to firms announcing cash flows close to

the cutoff date and firms whose announcement occurred almost one year previously.This weighting

automatically makes the resulting growth estimate stale and also introduces serial correlation in

the estimate.

4Cash flow news is often announced after the regular trading sessions in the stock market have

closed and thus aggregating across firms that announced cash flows within a 24-hour interval—as

opposed to modeling, say, hourly cash flow news—seems appropriate.

Cash Flow News and Stock Price Dynamics 2223

the Center for Research in Security Prices (CRSP) measure of dividend growth

conventionally used in the finance literature. This distinction is important

because dividend announcements precede dividend payments by several weeks,

which gives our dividend measure a significant timing advantage and more

closely aligns our measure with movements in market prices following dividend

news.5

To account for lumpiness in daily values of year-over-year changes in firm-

matched cash flows, we decompose cash flow news into a slowly evolving com-

ponent, which picks up time variation (predictability) in the mean of the cash

flow process, a transitory component, with volatility that changes over time,

and large jumps, the probability of which depends on the number of firms that

announce cash flow news on a given day. All three components turn out to

be important in capturing predictability in the dividend growth process and

evolution in the uncertainty surrounding growth in cash flows.

An important test of our approach is whether it can be used to generate more

accurate forecasts of cash flows than existing methods. Empirically,we find that

our estimate of the persistent dividend growth component is a strong predictor

of future dividend growth. Moreover, the predictive power of our approach

compares favorably to alternative predictors of dividend growth proposed by

van Binsbergen and Koijen (2010) and Kelly and Pruitt (2013). We also find

that our measure of the persistent dividend growth component is a positive and

significant predictor of future GDP growth and aggregate consumption growth.

In sharp contrast, “raw” dividend growth, as well as the individual jump or

transitory shock components, are very noisy and turn out not to have any

predictive power with respect to future dividend growth. Our results suggest

that firms closely monitor the state of the economy and adjust their dividend

policies in anticipation of changes in slow-moving economic indicators such as

GDP and consumption growth.

A key limitation of empirical tests of asset pricing models is that while

high-frequency data are available on movements in individual and aggre-

gate stock prices, individual firms’ cash flows are observed at much lower

frequencies—typically quarterly. The absence of high-frequency cash flow data

reduces researchers’ ability to estimate and test asset pricing models that rely

on the joint dynamics of stock prices, expected returns, and cash flow growth

expectations.

In the second part of our paper, we address these challenges by using our

new daily cash flow estimates to study predictability of stock returns in the

context of the log-linearized present value model developed by Campbell and

Shiller (1988a). The present value model offers several advantages. As shown

by Cochrane (2008) and van Binsbergen and Koijen (2010), the present value

model implies a set of cross-equation restrictions that tie the coefficients on

the variables in the prediction model for dividend growth to the coefficients on

the corresponding variables in the return prediction model. In particular, any

variable that helps predict future dividend growth should also have the ability

5Announced dividends precede actual dividend payments by approximately 42 days, on average.