The Effects of Financial Statement and Informational Complexity on

Cash Flow Forecasts

Leslie Hodder

Patrick E. Hopkins

David A. Wood

Kelley School of Business

Indiana University
Bloomington, IN 47405-1701

December 23, 2006

Email: lhodder@indiana.edu, peh@indiana.edu, & woodda@indiana.edu. We thank the

International Association for Accounting Education and Research (IAAER) and the KPMG and
University of Illinois Business Measurement Research Program for their generous research
support. This project also benefited from funding provided by the Kelley School of Business and
BKD LLP and the research assistance of Yvonne Lee. We thank Mary Barth and Katherine
Schipper for their suggestions during the development of this project and Kris Allee, Vicki
Dickinson, Susan Keenan, Laureen Maines, Mark Nelson, Derek Oler, Kenny Reynolds, Mike
Staub, Mike Tiller, participants at the IAAER Reporting Financial Performance workshops in
Bordeaux and New York City, and participants in the workshops at the University of Florida,
Indiana University, New York University, and the 10th World Congress of Accounting Educators
in Istanbul, Turkey for their comments.

Electronic
Electronic
Electronic
copy ofcopy
copy
this paper
available
available
is available
at:
at:https://ssrn.com/abstract=953454
http://ssrn.com/abstract=953454
at: http://ssrn.com/abstract=953454
 Abstract

We characterize the operating-activities section of the indirect-approach statement of cash flows

as backwards because it presents reconciling adjustments in a way that is opposite from the

intuitively appealing, future-oriented, Conceptual Framework definitions of assets, liabilities and

the accruals process. We propose that the reversed-accruals orientation required in the currently

mandated indirect-approach statement of cash flows is unnecessarily complex, causing increased

cash-flow forecast error and dispersion. We also predict that the mixed pattern (i.e., +/–, –/+) of

operating cash flows and operating accruals reported by most companies also impedes investors’

ability to learn the time-series properties of cash flows and accruals. We conduct a carefully

controlled experiment and find that (1) cash-flow forecasts have lower forecast error and

dispersion when the indirect-approach statement of cash flows starts with operating cash flows

and adds changes in accruals to arrive at net income and (2) cash-flow forecasts have lower

forecast error and dispersion when the cash flows and accruals are of the same sign (i.e., +/+,–/–)

with the sign-based difference attenuated in the forward-oriented statement of cash flows. We

also conduct a quasi-experiment to test our mixed-sign versus same-sign hypotheses using an

archival sample of publicly available Value Line cash-flow forecasts. We find that Value Line

analysts’ cash-flow forecasts exhibit the same pattern of forecast error as documented in our

experiment.

Electronic
Electronic
Electronic
copy ofcopy
copy
this paper
available
available
is available
at:
at:https://ssrn.com/abstract=953454
http://ssrn.com/abstract=953454
at: http://ssrn.com/abstract=953454
 The Effect of Financial Statement and Informational Complexity on Cash Flow Forecasts

I. Introduction

In this study, we investigate whether the currently required structure of the operating-

activities section of the indirect-approach statement of cash flows impedes investors’ ability to

learn the time series properties of operating cash flows and operating accruals. We propose that

the current structure of the indirect-approach statement of cash flows, which starts with reported

earnings and reverses changes in non-cash and non-operating items to arrive at operating cash

flows, presents information in a way that is opposite of the intuitive, future-oriented perspective

taken in the Conceptual Framework definitions of assets (Financial Accounting Standards Board

(FASB) 1985, para. 25), liabilities (para. 35), and the accruals process (para. 141). Prior research

suggests that information presentation that does not conform to the semantic, sequential or causal

structure of a data-generation process can negatively influence humans’ ability to learn the

parameters of forecast-relevant data-generating processes (Luft and Shields 2001). Thus, we

predict that the “reverse-process” format, currently required in the indirect-approach statement of

cash flows, interferes with financial-statement users’ learning, resulting in systematically higher

cash-flow forecast errors and greater forecast dispersion.

We also investigate whether the offsetting (i.e., “mixed sign”) pattern of operating cash-

flows and operating accruals information reported by most companies also impedes investors’

ability to learn the time series properties of operating cash flows and operating accruals.1 Prior

research suggests that this offsetting pattern of positive and negative forecast-model inputs will

likely lead to increased incidence of forecast errors over the errors observed for same-sign

1
Our analysis shows that in the years 2003-2004, 68 percent of firms in the Compustat Annual database have
positive operating cash flows and negative changes in current operating net assets.
1

Electronic copy available at: https://ssrn.com/abstract=953454

inputs. Overall, we predict that same-sign operating cash flows and operating accruals

information will result in lower forecast errors and dispersion than mixed-sign operating cash

flow and operating accruals information. However, related to our format-related predictions, we

expect that presenting cash flow and operating-accruals-change information in a more accruals-

process-congruent (i.e., “forward”) format will result in improved learning of the model

parameters, causing lower forecast errors and dispersion.

We test these predictions in a computerized experiment in which 50 second-year business

graduate students, with prior training in financial forecasting and residual-income valuation

models, sequentially forecast year-two cash flows from operations (CFO) for 16 different

companies after receiving year-one CFO and year-one changes in non-cash, current net operating

assets (to help exposition, we refer to this change as CNOA). After each of the 16 judgments, we

provide participants with cumulative feedback including the inputs to their forecast tasks (i.e.,

year-one CFO and CNOA for each preceding forecast) and accuracy-related information for each

preceding forecast (i.e., the matched set of “actual” year-two CFO, their forecasts of year-two

CFO, and the percentage forecast error). For each of the companies, we provide half of the

participants with highly summarized indirect-approach statement of cash flows information in

the currently mandated reverse order (i.e., net income – CNOA = CFO), and we provide the

other half with the same information in forward order (i.e., CFO + CNOA = NI). We also

counterbalance the signs of the year-one CFO and CNOA information, so that each subject

receives each of the following year-one CFO/CNOA sign combinations four times: +/+, +/–, –/+,

–/–.2

2
We also counterbalance the order of the companies by rotating the companies and by reversing each of the rotated
orders. This counterbalancing scheme controls for order effects and also results in all of the companies appearing
with equal frequency across the 16-company forecast sequence.
2

Electronic copy available at: https://ssrn.com/abstract=953454

 Consistent with our predictions, we find that the underlying structure of the operating

section of the indirect-approach statement of cash flows significantly affects the extent to which

participants learn the time-series model that generates year-two CFO. Compared to the forward-

order format for the indirect approach statement of cash flows, we find that forecast errors and

forecast dispersion are significantly higher under the currently mandated reverse-order format. In

addition, we find that mixed-sign forecast inputs lead to significantly higher cash-flow forecast

errors and forecast dispersion across both forecast conditions, but that the forecast errors from

the mixed-sign inputs are worst in the currently mandated reverse-order indirect approach

statement of cash flows.

In recent years, financial analysts have published forecasts of CFO with increasing

frequency (DeFond and Hung (2003). Motivated by our theoretical development and

experimental findings, we also conduct a quasi-experimental analysis designed to determine

whether analysts’ published CFO forecasts exhibit a pattern of forecast errors that is consistent

with the errors observed in our experiment. Because all publicly traded non-real-estate and non-

financial US companies must prepare a reverse-order indirect approach statements of cash flows,

we cannot test with archival data whether the forward-order indirect-approach statement of cash

flows yields lower forecast errors. However, publicly available CFO and CNOA data exhibit

sufficient variability to allow a quasi-experimental test of our mixed-sign hypotheses. As we

predict, after controlling for economic factors previously associated with analysts’ forecast

accuracy, we find that our sample of Value Line CFO forecasts exhibit significantly higher year-

two forecast errors when year-one CFO and year-one CNOA have mixed signs (i.e., +/–, –/+

versus +/+, –/–).

Electronic copy available at: https://ssrn.com/abstract=953454

 Although we focus on cash-flow forecasts, our study is related to prior research

suggesting that analysts and investors systematically ignore forecast-relevant information

contained in publicly released performance statements (i.e., statements of income, statements of

cash flows). For example, Elgers, Lo, and Richardson (2003) and Bradshaw, Richardson, and

Sloan (2001) provide evidence that sell side analysts’ earnings forecasts do not sufficiently

reflect the transitory nature of current-period accruals, resulting in predictable patterns of

earnings-forecast errors. These studies demonstrate that current-period analysts’ earnings-

forecast errors are positively related to prior-period earnings accruals (i.e., difference between

operating cash flows and earnings. Overall, these results are consistent with financial-statement

users’ systematic information-processing inefficiency when combining two types of performance

information (i.e., cash flow and accruals) contained in the statements of cash flows for publicly

listed US companies.3

Our experimental results suggest that two features of the indirect-approach statement of

cash flows reporting regime—financial-statement structure and information configuration—can

significantly affect financial-statement users’ ability to detect time-series patterns in reported

accounting information. The conceptual frameworks of the United States’ Financial Accounting

Standards Board (FASB) and the International Accounting Standards Board (IASB) identify

understandability as a necessary qualitative characteristic of financial information. Financial-

statement structures that are more understandable should increase comprehension of the time-

3
Statement of Financial Accounting Standards (SFAS) No. 95 (Financial Accounting Standards Board [FASB]
1987) allows companies to prepare the statement of cash flows using the (1) indirect approach or (2) the direct
approach with indirect approach information also reported. Thus, some form of the accruals-based-income-to-cash-
flows reconciliation in the indirect approach is provided by all companies. According to the American Institute of
Certified Public Accountants (2005), 592 out of 600 companies (98.7%) only report cash flows from operations
using the indirect approach.
4

Electronic copy available at: https://ssrn.com/abstract=953454

series properties (e.g., persistence) of reported performance information, leading to higher-

quality financial forecasts.

Whether understandability is improved through mandated, economy-wide reformatting of

the primary financial-statements (e.g., SFAS 130) or via user-level decision aids (c.f. Bonner

1999) depends on the perceived costs and benefits of each. Our analysis provides evidence on a

potential benefit of statement reformatting (i.e., improved investor understanding of time-series

properties of performance measures). Asset pricing theory indicates that market efficiency

increases as the level of available information increases and as investor knowledge becomes

more homogenous (Easley and Ohara, 2004). Thus, financial disclosure that is equally

accessible to more- and less-sophisticated investors can increase market efficiency, even if the

number of value-relevant facts reported in the financial statements is unchanged.

We organize the remaining paper in the following way. In the next section, we discuss

current cash-flow reporting and, after considering relevant judgment and decision making

research, make our research predictions. We describe our experiment and results in Sections III

and IV. We describe our quasi-experimental analysis of Value Line cash-flow forecasts in

Section V and provide a summary and conclusion in Section VI.

II. Theory and Predictions

Financial-statement analysis and forecasting is an iterative, evaluative process through

which financial-statement users engage in information discovery, expectation formation,

hypothesis testing, and conditional prediction. During this process, financial-statement users

attempt to understand patterns and relations in reported company-specific information, often with

the goal of identifying the portion of performance that is expected to recur in future periods.

Prior research suggests that the format of information presented in financial statements can

Electronic copy available at: https://ssrn.com/abstract=953454

influence the way company-specific information is processed and used by novices (e.g.., Maines

and McDaniel 2000) and experts (e.g., Hirst, Hopkins and Wahlen 2004). Thus, financial-

statement format also likely affects financial-statement users’ time-series-pattern learning and

resulting financial-forecasting judgments.

An important structural feature of the currently prescribed format for the indirect-

approach statement of cash flows is its “reverse” orientation; that is, increases in CNOA are

reversed to reconcile current-period earnings to current-period CFO. For example, an increase in

an operating asset account, like accounts receivable, is subtracted from earnings to arrive at

CFO. Likewise, an increase in an operating liability account, like accounts payable, is added to

earnings to arrive at CFO. However, the valence of each of these adjustments is opposite (i.e.,

reverse) from the conditional expected future cash-flow effects of each item (i.e., compared to

current period CFO, we expect additional future cash flows for larger accounts receivable

balances and additional future cash outflows for larger accounts payable balances). Indeed, the

logic underlying the operating-section adjustments in the indirect-approach statement of cash

flows occasionally leads our most gifted teaching colleagues to implore confused students and

executives to simply memorize the reverse relation because the reverse-accruals logic is

difficult.4

As modeled in Dechow, Kothari, and Watts (1998), Barth, Cram, and Nelson (2001) and

Francis and Smith (2005), current-period CNOA (i.e., or “current-period accruals”) are

positively associated with future cash flows. The positive, inter-temporally dependent relation

between current-period accruals and future cash flows conforms to the accrual-accounting

4
The books we reviewed often provide little discussion of accruals-process congruent future-period cash flow
effects, and instead offer highly detailed comprehensive lists of the operating net asset accounts along with
individual instructions for whether an increase or decrease in each account is added to, or subtracted from, earnings
to arrive at CFO.
6

Electronic copy available at: https://ssrn.com/abstract=953454

perspective offered in most introductory and intermediate accounting courses (e.g., Phillips,

Libby, and Libby 2005; Revsine, Collins, and Johnson 2004) and provides an intuitive and

parsimonious explanation for the empirical observation that current period earnings (which

includes the current-period CNOA) explain more variation in next period’s operating cash flows

than current-period operating cash flows (Dechow, et al. 1998; Barth, et al. 2001). Further, the

definitions of assets and liabilities in Statement of Financial Accounting Concepts No. 6 (FASB

1985) suggest that the relation between current-period change in operating accruals and future

cash flows is positive and causal.

Psychology research offers some insights into the likely effects of reverse-presentation of

information in financial-information comprehension and performance forecasting tasks. At its

most basic level, the process of using financial statements to forecast future performance is one

in which financial statement users learn probabilistic relations between dependent and

independent variables and make out-of-sample predictions of a future-value-correlated measure,

like future earnings or operating cash flows. A robust finding in prior psychology research is that

learning and subsequent judgment accuracy are greater for prediction scenarios with a positive

relation between independent and dependent variables than for prediction scenarios with a

statistically equivalent negative relation (Naylor and Clark 1968; Deane, Hammond, and

Summers 1972; Brehmer 1973, 1974, 1979; Brehmer, Kuylenstierna, and Liljergren 1974).

Interestingly, positive relations (e.g., y = a + bx) in multiple-cue probabilistic learning

tasks result in superior learning and judgment performance for tasks in which individuals have

no prior knowledge of the signs of the model parameters (i.e., they must learn the sign and

weight of parameters based solely on the data) and for tasks in which individuals are provided

with pre-experiment information about the signs of the parameters. Although, Muchinski and

Electronic copy available at: https://ssrn.com/abstract=953454

Dudycha (1975) and Sniezek (1986) provide some evidence that learning and judgment accuracy

improves in context-specific tasks with labeled variables, the settings in which prior research

demonstrates improved negative-relation learning are fairly straight-forward (e.g., predicting

college grades based on high-school grades in related and unrelated subjects).

While these context-related findings provide some support for financial-statement users

having the ability to adjust for the negative-accruals presentation format in the indirect-approach

statement of cash flows, we believe the use of cash flow and accruals information in forecasting

is sufficiently complex to mitigate the potential benefits of contextual familiarity. Thus, we

propose that the currently implemented indirect-approach structure for the statement of cash

flows will impede learning of persistence of the CFO and CNOA components of year-one

earnings, leading to higher levels of year-two forecast errors and higher levels of year-two

forecast dispersion. More formally, we propose the following hypotheses (in alternative form):

H1A: Investors’ year-two operating-cash-flow forecast errors will be higher when increases
(decreases) in year-one operating net assets are presented as negative (positive)
reconciling items in an earnings-to-cash-flows reconciliation, as compared to the opposite
presentation in a cash-flows-to-earnings presentation.

H1B: Investors’ year-two operating-cash-flow forecast dispersion will be higher when

increases (decreases) in year-one operating net assets are presented as negative (positive)
reconciling items in an earnings-to-cash-flows reconciliation, as compared to the opposite
presentation in a cash-flows-to-earnings presentation.

Prior research also suggests that CFO and CNOA typically have a +/– “mixed sign”

configuration (e.g., Dechow, et al. (1998) report average values of approximately 1.63 per share

for CFO and -0.50 per share for CNOA while Sloan (1996) reports this configuration for 7 out of

10 deciles). Given the cognitive complexity of learning a historical trend of year-one CFO, year-

one CNOA, and year-two CFO, and of forecasting out-of-sample CFO, mixed-sign variation in

model inputs to correctly generate one-period-out forecast judgments will further overload the

Electronic copy available at: https://ssrn.com/abstract=953454

processing capacity of individuals. In particular, humans quickly reach computational capacity in

complex tasks (like trend learning), leaving little additional mental capacity for manipulation of

data or complex calculations (Miller 1956). Further, even in high-stakes real-world settings (e.g.,

investing), individuals will attempt to use schematic problem representations and take

computational shortcuts (Hong, Stein and Yu 2005; Simon 1982). This leads us to the following

(alternative form) predictions about the effect of mixed-sign versus same-sign inputs into cash

flow forecasts.

H2A: Compared to forecasts made when year-one changes in operating net assets and year-one
operating cash flows have the same sign, investors’ year-two operating-cash-flow
forecast errors will be higher when year-one changes in operating net assets and year-one
operating cash flows have different signs.

H2B: Compared to forecasts made when year-one changes in operating net assets and year-one
operating cash flows have the same sign, investors’ year-two operating-cash-flow
forecast dispersion will be higher when year-one changes in operating net assets and
year-one operating cash flows have different signs.

III. Experimental Design

We investigate the effects of cash-flow statement format and forecast-relevant

information configuration (i.e., sign of model inputs) in a 2 x 2 mixed-design experiment (format

is the between-subjects factor varied at two levels: forward versus reverse; sign is the within-

subjects factor varied at two levels: mixed versus same). Fifty second-year masters-level

business students participated in our experiment.

Masters-level business students are appropriate participants for two reasons. First, our

theory and hypotheses are related to a general psychology of inference and pattern learning in a

financial-forecasting setting. Because we have no ex ante reason to believe that general

forecasting experience will eliminate the effects of format or input-sign configuration on forecast

accuracy or dispersion, we decided to avoid unnecessarily consuming limited available analyst

Electronic copy available at: https://ssrn.com/abstract=953454

resources. Second, Elliott et al. (2007) find that second-year masters students (i.e., similar to our

participants) are suitable proxies for nonprofessional investors, an important constituent group

for accounting standard setters and regulators.

As indicated in Table 1, participants previously completed 10 (standard deviation = 4)

accounting and finance courses, with one of those courses a three-credit hour MBA-level

financial-statement-analysis course that included an emphasis on residual-income valuation and

financial forecasting. Participants had, on average, 14.5 (standard deviation = 30) months of

prior work experience. As indicated in Table 1, we found no significant differences (all p-values

> 0.10) between treatment groups for demographic information, including current degree

program, prior work experience, accounting/finance classes taken, the number of participants

who plan to buy or sell equity securities, or the number of times participants previously

performed fundamental analysis on the financial information of publicly traded companies.5

Overall, our participants appear to be knowledgeable about the relation between cash flows and

accrual accounting and have experience in financial forecasting.

Procedure for Experiment

Participants completed the experiment during a single monitored session in a computer

lab. Computer administration allows us to control the flow of information to participants,

monitor the amount of time participants spent viewing individual screens, and to collect process-

related data (e.g., the number of times participants viewed previous screens or the feedback

pages). In addition, computer administration allows us to prevent participants from changing

their forecasts after learning the actual outcomes. At the beginning of the experiment session, we

provided participants with written and oral instructions describing how to access the

5
We present a subset of demographic information in Table 1. None of the demographic question (tabulated and
untabulated) yielded significant differences across treatment conditions (all p-values > 0.10).
10

Electronic copy available at: https://ssrn.com/abstract=953454

computerized experiment, how to enter their responses during the session, how to receive

compensation upon completion of the experiment, and prohibited forms of assistance during the

experiment (e.g., computational aids, like Microsoft Excel).

Participants completed a series of pre-experiment questions designed to gauge their

accounting knowledge, their perceived difficulty of the accounting knowledge questions, and

their perceptions of the time series relation between Year1 CFO, Year1 CNOA, and Year2 CFO.

After finishing the pre-experimental-treatment questions, participants began the experimental

task. The experimental task consisted of viewing Year1 CFO and Year1 CNOA information for

16 different companies and predicting Year2 CFO for those companies. Figure 1 illustrates the

primary computer screens for operationalizing the forward-order and reverse-order formats.

We intentionally incorporate a highly simplified statement of cash flows into our

experiment because we wish to conduct the cleanest possible test of cash-flow-statement format

effects. Including additional real-world features (e.g., multiple line items of reconciling

information) will uniformly increase the complexity of the time-series learning task. However,

we could identify no theoretical reason to expect increased complexity to attenuate the

hypothesized differential effect of format on time-series learning. However, an increase in

complexity would certainly increase the noise in our data, thereby also increasing the likelihood

of making a Type II error in our inferences.

In the Reverse Order (RO) condition, participants view three lines of information on the

Statement of Cash Flows screen, in the following order: “YEAR1 Operating Earnings” (Year1

OE) “YEAR1 (Increase) Decrease in Non-Cash Operating Net Assets” (Year1 CNOA) and

“YEAR1 Cash Flows from Operations” (Year1 CFO). The actual amount of Year1 OE is

11

Electronic copy available at: https://ssrn.com/abstract=953454

masked by a black box.6 When ready, participants proceed to the Judgment screen on which

they provide their estimate of Year2 CFO. After participants enter a forecast amount, they are

taken to a screen that includes only their Year2 CFO forecast and the actual Year2 CFO

realization. When ready, the subject can proceed to the Feedback screen that presents

cumulative, comprehensive summary information that includes Year1 CFO, Year1 CNOA,

Year2 actual CFO, their Year2 Forecast of CFO, and their percentage forecast error for each of

their previous forecasts.

Our second between-subjects condition, Forward Order (FO), differed from RO in two

ways. First, the Statement of Cash Flows screen includes company data in the following order

(i.e., opposite of the reverse order): Year1 CFO, Year1 Increase (Decrease) in NOA, and Year1

OE (again blackened out). Second, the Feedback screen includes cumulative, comprehensive

summary information that includes the same information as the RO condition, except it is

presented in the same order as Statement of Cash Flows screen for the FO condition.

After completing their forecasts for each of the 16 companies, participants completed

several post-experiment questions designed to elicit perceived task difficulty and judgment

confidence. Participants were asked to assign weights to the values of CFO and NOA based on

how these characteristics related to future CFO in the last few predictions they made. Finally,

participants were asked to supply answers to several demographic questions.

Upon completing the experiment, participants left the computer lab and were

compensated based on the absolute value of their forecast errors. Compensation rates were

determined based on average forecast errors of students who pre-tested the experiment. All

6
Providing an explicit amount for OE introduces a potential confounding factor into the experiment. OE is a
summation of Year1 CFO and CNOA and, therefore, represents redundant information. If subjects believe OE was
more relevant to forecasting Year2 CFO, they can compute it by summing the Year1 CFO and NCOA.
12

Electronic copy available at: https://ssrn.com/abstract=953454

participants were eligible to earn between $6 and $20, with actual payoffs spanning that range,

and a mean payoff of $13.

Company Data Included in the Experiment Materials

Information for 500 companies was generated through a simulation based on Barth et al.

(2001). Specifically, Year2 CFO was modeled as a linear function of Year1 CFO and Year1

CNOA. Panel A of Table 1 provides the parameters of the generating function. We seeded a

zero-mean, normally distributed error term into the simulation to add a small amount of noise to

the subsequent realization of CFO.

We chose 16 firms from the model-generated observations. We intentionally selected

companies to obtain a balanced design across all possible combinations of positive and negative

operating cash flow and operating asset changes (+/+, +/–, –/+, –/–), crossed with randomly

assigned positive and negative model errors. In Panel B of Table 1, we present the task

information for the sixteen firms we include in our experiment. Panel C of Table 1 presents

descriptive statistics for the 16 selected firms. Although selection of sample firms was not

random, parameters estimated from regressing sample values of Year2 CFO on Year1 CFO and

Year1 CNOA do not differ from theoretical model parameters.7

7
Across the two conditions, we presented the 16 firms in eight unique sequences. Four of the sequences were based
on a four-firm rotation. Specifically, the firm randomly assigned as Firm 1 in Sequence 1, is Firm 5 in Sequence 2,
Firm 9 in Sequence 3 and Firm 13 in Sequence 4. The remaining four sequences are an exact reversal of Sequences
1 through 4. Therefore, the firm randomly assigned as Firm 1 in Sequence 1 is Firm 16 in Sequence 5, Firm 12 in
Sequence 6, Firm 8 in Sequence 7 and Firm 4 in Sequence 8. These eight unique sequences assures that every firm
appears with equal frequency in Trials 1-4, 5-8, 9-12, and 13-16 This also results in equal representation of the
positive and negative operating cash flow and operating asset changes (+/+, +/–, –/+, –/–) across the trials in each
condition.
13

Electronic copy available at: https://ssrn.com/abstract=953454

 IV. Results of the Experiment

Pre-Experiment Knowledge Questions

Prior to beginning the forecasting task, participants completed an accruals knowledge

test, and assessed the difficulty of the test. Participants were required to successfully complete

the knowledge test before they could proceed to the 16-company forecasting task. Table 2,

Panel B indicates that the almost all participants in both treatment groups correctly answered the

required test questions on the first attempt. Participants rated these questions as relatively easy

with a combined mean difficulty rating of 4.61 on a 15-point scale, increasing in perceived

difficulty. Perceived difficulty of the tutorial task did not differ across groups (Z = -0.73 and Z =

-0.95, both p-values >0.10). Overall, the results reported in Panels A and B suggest that random

assignment of participants across conditions successfully distributed knowledge and abilities

across the two treatment groups.

Panel C of Table 2 presents participants’ post-experiment self-assessments of forecast-

task difficulty and forecasting confidence. Neither the participants’ perceived difficulty of the

forecast task (Z = 0.55, p-value >0.10) nor their forecast-related confidence levels differed across

conditions (Z = 0.55, p-value >0.10) despite the differential treatments. These results suggest

that participants’ perceptions of task difficulty and participants’ confidence in their forecasting

skills are independent of their assigned treatments.

Hypothesis Tests – Late-Round Forecasts

Table 3, Panel A reports participants’ median and mean forecast accuracy statistics, along

with standard deviations of participants’ forecast judgments (i.e., our proxy for forecast

dispersion), separately presented for the crossed combinations of format (two levels: reverse

14

Electronic copy available at: https://ssrn.com/abstract=953454

order, forward order), input sign (two levels: mixed, same) and forecast judgment timing (two

levels: first half, last half).8

Before discussing the fully interacted analysis of the complete data set, we first examine

whether our predicted forecast-error and dispersion effects occur in the later rounds (i.e., last

half) of the experimental trials. We first test for differences in the late-round subset of data

because we did not provide estimation-relevant information prior to the experiment (i.e.,

participants will learn the model parameters after receiving trial-by-trial feedback) and we made

no explicit ex-ante predictions of differences in participants’ pre-experiment forecasts. While not

definitive, significant differences in later rounds suggest that the predicted effects are fairly

robust and are more likely to occur in natural settings.

Consistent with hypothesis H1A, we find that in the last half of the experimental trials the

median forecast errors of participants in the reverse-order condition are significantly higher than

those of participants in the forward-order condition (see shaded boxes in Table 3, Panel A; 0.29

compared to 0.17; F = 11.86; p-value = 0.001; one-tailed; comparison statistic untabulated).9

Consistent with hypothesis H1B, we find that in the last half of the experimental trials, the

dispersion in forecast errors was significantly greater for participants in the reverse-order

condition than for participants in the forward-order condition (see shaded boxes in Table 3, Panel

8
We focus our discussion on median forecast accuracy statistics to mitigate the influence of extreme observations
on readers’ inferences. In addition, given that our dispersion hypotheses predict differences in variance across
treatment conditions (i.e., violation of a distributional assumption in parametric tests), we report all of our primary
inferences on the basis nonparametric (i.e., rank-transformed) tests. As usual, parametric tests yielded similar
inferences, with more extreme test statistics and related p-values.
9
Each individual gave multiple responses (i.e., the within-subject responses are not independent from each other),
so we conduct a repeated-measure ANOVA on the rank-transformed dependent variable to compare medians.
Because conventional parametric tests of differences in sample variance assume normal distributions, we also test
differences in dispersion by conducting a repeated-measure ANOVA on a rank-transformed variability measure.
Our rank transformed variability measure is generalized from the nonparametric test for differences in dispersion
suggested by Fligner and Killeen (1976) (as modified by Conover et al. 1981). For this test, the dependent measure
is computed by transforming the absolute value of the difference between each observation and its corresponding
cell median. The transformation used is Φ-1(0.5+i/2(N+1)) where Φ-1 represents the inverse standard normal
distribution function, i is the ranked absolute value, and N is the combined sample size of all experimental cells.
15

Electronic copy available at: https://ssrn.com/abstract=953454

A; 1.70 compared to 1.07; F = 28.44; p-value <0.001; one-tailed; comparison statistic

untabulated). Interestingly, we observe larger errors and higher dispersion in the reverse-order

condition despite the investment of equivalent amounts of time by participants in both groups.10

This result suggests that improvements in accuracy associated with the forward-order treatment

are due to lower task complexity rather than greater effort by forward-order participants.

We also find, in the last half of the experimental trials, that the median forecast errors for

mixed-sign trials are significantly higher than those same sign trials (see unshaded boxes in

Table 3, Panel A; 0.25 compared to 0.19; F = 3.72, p-value <0.027; one-tailed; comparison

statistic untabulated), thereby supporting our prediction in H2A. With respect to hypothesis H2B,

we find that in the last half of the experimental trials, the dispersion in forecast errors was

significantly greater for mixed sign trials than for same sign trials (see unshaded boxes in Table

3, Panel A; 1.95 compared to 0.46; F = 24.68, p-value <0.001; one-tailed; comparison statistic

untabulated). Taken together, these results suggest that the all of our predicted effects survive in

the later rounds of cash flow forecasts, after all participants had equal opportunity to learn the

parameters of the data-generating model.

Hypothesis Tests – All Forecast Judgments

We extend these analyses to examine differences over the entire range of participants’

responses. To further test our forecast-error-related research hypotheses, we conduct a repeated-

measures analysis of variance (ANOVA) on the rank-transformed forecast errors and we report

that analysis in Table 3, panel B. To further test our dispersion-related research hypotheses, we

conduct a repeated-measures analysis of variance (ANOVA) on a rank-transformed variability

10
In prior psychology studies, a common proxy for effort is the amount of time spent on a given judgment task. In
the present study, the average amount of time spent per forecast did not differ across any of the sample partitions (all
p-values > 0.10).
16

Electronic copy available at: https://ssrn.com/abstract=953454

measure and we report that analysis in Table 3, panel C. Panels A and B of Figure 2 provide

graphic representation of the interactions from these two respective analyses.

Consistent with hypothesis H1A, the significant main effect for “Method” indicates that,

across all forecasts provided by participants, median forecast errors are significantly greater in

the reverse-order condition than in the forward-order condition (0.30 compared to 0.22; F = 8.66,

p-value < 0.01). Consistent with hypothesis H1B, the significant main effect for “Method”

indicates that dispersion is higher in the reverse-order condition (standard deviation of 2.26

compared to 1.43; F = 29.97, p-value < 0.01).

To explore the learning process in each treatment, Table 3, Panel A presents forecast

errors and dispersion statistics for the first and last eight forecast trials. The significant main

effect for “Half” in Table 3, Panel B indicates that median forecast errors are significantly

greater in the first eight trials than in the last eight trials (0.36 compared to 0.22; F = 12.18, p-

value < 0.01). Consistent with differential learning evidenced by participants across the two

methods, forward order showed a significant decline in median forecast errors over the last eight

trials compared to the first eight (0.37 versus 0.17), while the forecast error remaining

statistically flat in the reverse order condition ( 0.35 versus 0.29) (see middle figure of Panel A,

Figure 2 for graphical presentation). This differential rate of learning across the two methods is

confirmed by the significant Method x Half interaction term reported in Panel B of Table 3 (F =

5.36, p-value = 0.02).

Table 3, Panels B and C also presents the results of tests that compare participants’

accuracy and dispersion across same-versus-mixed signs of CNOA and CFO. Hypotheses H2A

and H2B predict that forecast error and dispersion increase with task complexity. In our cash-

flow forecasting context, we propose that task complexity increases when data inputs have

17

Electronic copy available at: https://ssrn.com/abstract=953454

conflicting signs. Consistent with hypothesis H2A, the significant main effect for “Mixed” in

Table 3, Panel B (untabulated medians are 0.36 compared to 0.22; F = 21.41, p-value < 0.01)

indicates that forecast errors are higher for company-observations of mixed signs compared to

same-sign company-observations. Consistent with hypothesis H2B, the significant main effect

for “Method” indicates that dispersion is higher for company-observations with mixed signs

(standard deviation of 2.59 compared to 0.47; F = 32.67, p-value < 0.01).

An alternative way to think about our research hypotheses is to compare the model

weights implicitly evidenced by the 16 trials of forecasts provided by participants. As reported

in Table 4, we extracted implicit subject model weights for CFO and NCOA by regressing

subject forecasts of future cash flows on current levels of CFO and NCOA. The mean (median)

weight placed on CFO by the forward order participants is 0.90 (0.91) compared to 0.96 (0.91)

for the reverse order condition. Although the means differ, the identical medians confirm that

the difference is not statistically significant. Therefore, it appears that participants in both

conditions underweight CFO by the same amount relative to the theoretical model weight. The

insignificant difference in subject model weights on CFO is not surprising given that CFO was

displayed with the same sign in both conditions. Interesting differences across conditions are

revealed by the subject weights on CNOA. Specifically, our evidence shows that participants

robustly underweight CNOA in the reverse order condition (mean weight of 0.41 compared to

the theoretical value of 0.55) while participants in the forward-order condition assign an average

weight of 0.50 to CNOA, which does not statistically differ from the theoretical value. This

pattern of results is even more pronounced for median implied CNOA weights.

18

Electronic copy available at: https://ssrn.com/abstract=953454

 This pattern of results is consistent with reverse order participants failing to fully adjust

for the sign reversal inherent in the reverse order presentation.11 Cross-subject dispersion is also

much higher in the reverse order condition. Implied model weights for CFO in the reverse order

condition range from 0.13 to 2.88 (standard deviation of 0.45) compared to the implied CFO

model weights in the forward order condition, which range from 0.26 to 1.29 (standard deviation

of 0.25). The relatively high dispersion across participants in the reverse order condition is also

apparent in participants’ weights on CNOA which range from -0.37 to 2.06 (standard deviation

of 0.44). In contrast, subject weights on CNOA in the forward order condition range from 0.05

to 1.24 (standard deviation of 0.23). These results provide additional support for hypothesis H1B

that dispersion is higher among participants in the reverse-order condition.

V. Quasi-Experiment with Value Line Cash Flow Forecasts

Quasi-Experimental Design

In this section, we provide evidence on the effects of informational complexity in the

context of passively observed cash flow forecast data collected from the Value Line Investment

Survey. The Value Line Investment Survey is a comprehensive source of information and

ratings on approximately 1,700 publicly-traded companies compiled by over 90 independent

analysts. Each company report contains, among other things, Value Line's proprietary

performance ranks and financial forecasts for the coming 1 to 5 years. We hand-collect cash

flow forecast and actual data from company reports, published monthly over the period 2003-

2005. We exclude from our sample firms in the banking and insurance industries because

measures of operating accruals for these firms are not comparable to those in other industries.

11
Recall that increases (decreases) in operating assets are displayed as negative (positive) numbers in the
reconciliation of income to CFO. Subjects evaluating the reverse-order format must translate a number subtracted in
arriving at CFO in the current period into an increase in future CFO.
19

Electronic copy available at: https://ssrn.com/abstract=953454

 We further limit our analysis to calendar-year firms for which cash flow forecasts are

available as of April 30 of each year. We choose April 30 as a cutoff date for forecasts to allow

sufficient time for analysts to incorporate into their current year forecasts previous cash flow

information contained in annual reports published during the first quarter. We collect realized

(actual) cash flows from subsequently published Value Line reports to ensure that predicted and

realized cash flows are reported on consistent measurement bases. We also collect Value Line’s

proprietary PREDICTABILITY and STABILITY indexes for each firm. All other financial

statement data is collected from Compustat by individually cross-referencing company names

and stock tickers in Compustat to those provided by Value Line. Our data collection procedure

results in 893 firm-year observations with sufficient data to complete our analyses.12

In contrast to our experimental research design, we are unable to control (by assignment)

factors likely to influence forecast accuracy that may also be correlated with our proxy for

informational complexity. Specifically, in our quasi-experiment we cannot impose a consistent

cash-flow-generating process across conditions and across companies. Therefore, we must

augment our archival research design with proxies that control for differences in firms’

underlying data-generating processes that may affect the predictability of future cash flows.13

We use the life-cycle construct developed by Dickinson (2006) to control for firm

attributes that affect the generation of future cash flows. Dickinson (2006) shows that firms’

life-cycle stages are systematically associated with future levels of profitability and cash flows.

Extending prior research, she develops a proxy based on patterns of firms’ cash flows to identify

12
Forecasts made in 2003 and 2004 are paired with actual realizations reported in 2004 and 2005, respectively.
13
We are also unable to control through randomization or assignment, factors associated with analysts that may
influence forecast accuracy. Therefore, we must rely on an assumption either that a) Value Line analysts are equally
competent at their forecasting tasks, or b) that if competency differs across Value Line analysts, less competent
analysts are not systematically assigned to firms with greater informational complexity. We have insufficient
analyst data to test either of these assumptions.
20

Electronic copy available at: https://ssrn.com/abstract=953454

five life-cycle stages (Startup, Growth, Mature, Shake-out and Decline). The jointly determined

pattern of operating, financing, and investing cash flows represents firms’ strategies and

capacities for obtaining and using resources.

Dickinson (2006) defines Startup firms as those with negative cash from operations

(CFO), negative cash from investing (CFI), and positive cash from financing (CFF). This pattern

reflects the initial investment necessary to bring profitable ideas to market. Startup firms have

higher information risk and experience greater operating variance across industries. Growth

firms also have negative CFI and positive CFF, reflecting continuing investment. However,

growth firms have strongly positive CFO as they realize monopoly rents that accrue to

innovators. Growth firms experience increasing profit margins, although variance is high.

Mature firms have positive CFO and negative CFI, but CFF is negative, reflecting return

of capital to investors. Mature firms have the most predictable profitability and cash flows. In

contrast, as their businesses contract, Decline firms have negative CFO, positive CFI and

negative CFF. Dickinson (2006) classifies all other patterns as Shake-out firms. Shake-out firms

have variable patterns of cash flows that reflect either rejuvenation through structural change, or

progression to Decline.

Because life-cycle stage influences firms’ cash and profit-generating processes, we

estimate model (1), controlling for life-cycle STAGE and other variables thought to influence

forecast accuracy.
4
ERRORt = β 0 + β1MIXEDt −1 + ∑ β 2,i STAGEi ,t −1 + β 3 ASSET _ GROWTH t −1 + β 4 RANK _ | INCOME t −1 |
i =1
+ β 5 RANK _ | CNOAt −1 | + β 6CFOt −1 + β 7 PREDICTABI LITYt −1 + β 8 STABILITYt −1 + β 9σCFOt −5,t
(1)
12
+ β10σCNOAt −5,t + β11SIZEt −1 + β12 PROFITABIL ITYt −5,t + ∑ β13, j INDUSTRY j ,t −1 + ε t
j =1

21

Electronic copy available at: https://ssrn.com/abstract=953454

 ERRORt is the absolute value of the difference between the actual cash flow per share

reported by Value Line for a given year and the forecasted cash flow per share predicted by

Value Line analysts as of April 30 of that year. We divide the difference by the absolute value of

the actual operating cash flow per share for the year (CFO). We use the absolute value of the

ERROR because our primary interest is in relative magnitude of forecast errors, rather than the

sign. We include MIXEDt-1 as an explanatory variable to proxy for informational complexity.

MIXEDt-1 is set to 1 when CFO and changes in operating assets (CNOA) are of different signs

(positive CFO and negative CNOA, or negative CFO and positive CNOA). We predict a

positive coefficient on MIXEDt-1. We include dummy variables to represent four of the five life-

cycle stages (Start-up, Growth, Shake-out, and Decline). Because the coefficients capture the

effect on ERROR relative to mature firms, we predict positive signs for each of the four STAGE

variables consistent with these firms having more variable operations.

We augment the regression with ASSET_GROWTH because high growth firms may

experience structural shifts in their operating and information environments that render forecasts

more difficult and less accurate. We expect the coefficient on GROWTH to be positive. We

compute GROWTH as the change in assets (ASSETSt – ASSETSt-1)/ASSETSt-1 ). Prior research

finds that relatively extreme realizations of income or accruals are less likely to recur (e.g.

Freeman and Tse 1992; Sloan 1996). If analysts fail to consider mean-reverting tendencies of

extreme performance realizations, then forecast errors will be higher following such realizations.

We include RANK_|INCOMEt-1| and RANK_|CNOAt-1| to control for these hypothesized

effects. RANK_|INCOMEt-1| is the decile rank of the absolute value of income before

extraordinary items divided by assets. RANK_|CNOAt-1| is the decile rank of the absolute value

of the change in net operating assets (CNOA). We compute CNOA as the difference between

22

Electronic copy available at: https://ssrn.com/abstract=953454

INCOME and CFO, divided by ASSETS. We rank the absolute values of INCOME and CNOA

by year, within the population of Compustat firms. This identifies with higher ranks those

observations falling within the extreme positive or negative portions of the cross-sectional

distribution. We expect positive coefficients on each of these rank variables. We also include

the prior period reported realization of CFO, deflated by assets, to control for potential

differences in firm life cycles. Because firms with relatively high levels of operating cash flows

are likely to have more stable operations, we predict a positive coefficient on CFOt-1.

Value Line reports two proprietary measures related to forecasting. The first is a

predictability index that measures the reliability of forecasts based on the stability of year-to-year

comparisons. The index ranges from 5 to 100, where 5 represents the lowest level of

predictability and 100 the highest. We expect a negative coefficient on PREDICTABILITYt-1.

The second proprietary measure is the price stability index—a measure of the stability of the

stock’s price. This index also ranges from 5 to 100, with 5 representing the lowest level of

stability. According to Value Line, STABILITY includes sensitivity to the market (beta) as well

as the firm’s inherent volatility. We expect a negative coefficient on STABILITYt-1 to the extent

that systematic and unsystematic risk factors are negatively related to cash flow predictability.

In addition to Value Line’s proprietary predictability and stability indexes, we include

historical measures of CFO and CNOA variability. We measure the standard deviation of each

over the five year period ending with the forecast date. We expect positive coefficients on

σCFO and σCNOA because greater variability is associated with lower predictability.

We include the decile rank of assets, RANK_SIZEt-1, to control for potential structural

differences associated with firm size. Larger firms may have more diverse operating segments,

making prediction of future cash flows difficult. Alternatively, larger firms may have richer

23

Electronic copy available at: https://ssrn.com/abstract=953454

information environments that assist analysts in accurately forecasting cash flows. Because

RANK_SIZE can operate to either increase or decrease forecast accuracy, we do not predict a

sign for this variable. Similarly, we include PROFITABILITY, the average return on assets over

the five years ending with the forecast period, to control for aspects of the information

environment not captured by the other variables. Firms that consistently report high levels of

profitability may obscure the underlying data generating process for cash flows through earnings

management, leading to greater forecast errors. Alternatively, highly profitable firms may have

richer information environments and be followed by more skilled analysts. Because

PROFITABILITY may operate to increase or decrease forecast accuracy, we do not predict a

sign for this variable. Finally, because Barth et al. (2001) suggests that the time series properties

of cash flows are influenced by industry, we include twelve industry control variables consistent

with their classification.

Results of the Quasi-Experiment

Panel A of Table 5 presents descriptive statistics for the sample of Value Line firms used

in the archival analysis. Means and quartiles are presented in the first three columns and means

by life-cycle stage are presented in the last four columns. The average cash flow forecast error

for the sample of Value Line firms is 28.9% and varies by life-cycle stage consistent with

Dickinson (2006). Specifically, cash flow forecast errors are lowest for firms in the Mature stage

(25.1%) and highest for firms in the Startup stage (50.6%). Firms in the Shake-out and Decline

stages experience forecast errors of 40.2% and 41.9%, respectively.14 The proportion of firms

with MIXED cash flows and accruals is highest in the Mature life-cycle stage (94%), followed

14
The equally-weighted forecast error across the life-cycle stages of 37.12% is lower than the average experimental
forecast error in the last eight trials of the experiment (68.0%). We observe equally-weighted average forecast
errors for mixed sign (same sign) observations of 47.3% (29.0%) in the archival data and 79.0% (35.0%) in the last
eight trials of the experimental forecasting task.
24

Electronic copy available at: https://ssrn.com/abstract=953454

by the Growth stage (87.1%). Overall, the values of variables across the life-cycle state

partitions suggest that MIXED is positively correlated with firm attributes that result in more

predictable cash flows, and that failure to control for these attributes will bias the coefficient on

MIXED toward zero.

Across all Value Line observations, the average rank of the absolute value of income and

the average rank of the absolute value of accruals is 3.8 and 4.0, respectively, indicating that the

magnitudes of income and accruals as a fraction of total assets are less extreme than the

population of Compustat firms.15 The mean sample PREDICTABILITY and STABILITY index

values of 48.7 and 50.5, respectively, are lower than the midpoints of the scale, which ranges

from 5 to 100 for each measure. Compared to firms in other life-cycle stages, firms in the

Mature stage have higher PREDICTABILITY and STABILITY and lower variance in cash

flows, accruals and net income across the preceding five years. These findings are consistent

with the life-cycle theory of the firm discussed in Dickinson (2006).

For the overall sample, the rank of SIZE is 6.68, which is higher than the midpoint of the

Compustat rank, and consistent with Value Line’s propensity to cover larger, better-known

firms. The average standard deviation of CFO (CNOA) over the five year period ending with the

forecast date is 4.7% (6.0%), and the average standard deviation of income is 5.4%, consistent

with the negative correlation between CFO and CNOA observed in other studies (e.g. Dechow,

et al 1998). This smoothing pattern for accruals holds only for the Mature firm group.

Panel B of Table 5 presents descriptive statistics for the Compustat population over the

same period. The majority of observations in the Compustat population comprise Mature firms

and Growth firms (42.4% and 27.2%, respectively). These groups are also the most highly

15
Values are ranked from 0-9; therefore, the median rank is between 4 and 5.
25

Electronic copy available at: https://ssrn.com/abstract=953454

represented in the Value Line sample (52.1% and 31.1%, respectively). Firms in the Value Line

sample are more profitable than those in the Compustat population and the mean variability of

income, accruals and cash flows is lower. However, the pattern of variability across life-cycle

groups is consistent between the Compustat population and the Value Line sample. For

example, INCOME, CFO, and CNOA are each least variable in the Mature firm group, and most

variable in the Startup and Decline groups.

Barth et al. (2001) suggests that the time series properties of cash flows are influenced by

industry. Panel C of Table 5, shows the distribution of sample firms across each of 13 industries

defined by Barth et al. (2001) and across life-cycle stages defined by Dickinson (2006). A

significantly higher percentage of Startup and Decline firms are in the Computer and

Pharmaceutical industries. Because the Startup and Decline groups contain lower proportions of

firms with MIXED accruals and cash flows, we include INDUSTRY dummies in our regression

to control for potential effects of INDUSTRY on forecast errors.

The correlation matrix of regression variables presented in Table 6 shows that forecast

error is positively correlated with GROWTH (Pearson Correlation = 0.650), σCFO (Pearson

Correlation = 0.180), σCNOA (Pearson Correlation = 0.104). These relations are as expected—

variability in historical performance measures appears to make forecasting cash flows more

difficult, as does GROWTH in assets. Extreme changes in net operating assets (CNOA) in the

period of the forecast are positively associated with cash flow forecast errors (Pearson

Correlation = 0.107). This is consistent with Bradshaw et al’s 2001 conjecture that analysts fail

to anticipate lower levels of persistence associated with extreme values of accruals. In contrast,

forecast error is negatively associated with Value Line’s PREDICTABILITY index (Pearson

Correlation = -0. 300) as well as the STABILITY index (Pearson Correlation = -0.233). Most

26

Electronic copy available at: https://ssrn.com/abstract=953454

important to our research question, we find that MIXED is positively correlated with forecast

errors (Pearson Correlation = 0.190).

Coefficients from the estimation of equation (1) are presented in Table 7. The third and

fourth columns show the estimates obtained from regressing |ERROR| on MIXEDt-1 and life-

cycle STAGE without any of the control variables in the model. The intercept is positive and

significant (coefficient of 0.149; t-value of 2.74). The coefficient on MIXEDt-1 is also significant

and positive (coefficient of 0.108 ; t-value of 2.06). The coefficients on the life-cycle indicator

variables are positive and mostly significant, consistent with theory that Start-up firms, Growth

firms, Shake-out firms and Decline firms experience higher forecast errors than Mature firms.

Holding the effects of MIXED constant, fitted values can be obtained by adding the

intercept to the coefficient for each STAGE. Results suggest that the average error is highest in

the Startup group (45.8%), followed by the Decline group (37.3%) and the Shake-out group

(31.2%). The average error in the Growth group (18.3%) is not statistically different from the

average error for Mature firms (14.9%). These findings provide support for the hypothesis that

informational complexity is associated with higher forecast errors across firms at different life-

cycle stages. In addition, our results suggest that the effect of MIXED is economically

significant. Specifically, MIXED increases forecast error between 23.5% and 72%, depending on

life-cycle group.

The center two columns present coefficients from estimation of the model with industry

controls. In the presence of industry controls, the intercept becomes insignificant. Four

industries are significantly associated with higher forecast errors: (1) Textiles, printing and

publishing, (2) Extractive industries, (3) Computers, and (4) Transportation. The effects of

INDUSTRY are incremental to life-cycle STAGE, suggesting that INDUSTRY and STAGE

27

Electronic copy available at: https://ssrn.com/abstract=953454

capture different factors contributing to forecast errors. Inferences about MIXED and life-cycle

STAGE are unchanged when INDUSTRY is included in the regression.

The fully augmented regression coefficients are presented in the two right-most columns

of Table 7. In the presence of other control variables, both the intercept and the coefficients on

STAGE and INDUSTRY variables become insignificant. However, in the fully augmented

regression, the coefficient on MIXEDt-1 retains its sign and becomes more significant. Moreover,

the explanatory power of the model increases significantly (the R2 increases from 0.037 for the

INDUSTRY-augmented regression to 0.122 in the fully augmented model). These results

suggest that the control variables subsume the explanatory power of STAGEt-1 and INDUSTRY,

but provide stronger support for the hypothesis that informational complexity is associated with

higher forecast errors. In particular, the deleterious effect of informational complexity on

forecast errors is incrementally and economically significant in the presence of many other

forecast-error explanatory variables.

The coefficients estimated for the control variables have the predicted signs, and are

generally significant. The coefficient on ASSET_GROWTH is positive and marginally

significant, consistent with analysts having greater difficulty forecasting future cash flows when

firms are undergoing expansion, either through internal growth or acquisition. Consistent with

the significant positive correlation between forecast error and RANK_|CNOAt-1|, the coefficient

on RANK_|CNOAt-1| is marginally significant (coefficient 0.013; t-value 1.72). The coefficient

on RANK_|INCOMEt-1| is not significant. Therefore, there is no evidence that extreme values of

income mean-revert in an unanticipated and systematic way that incrementally contributes to

forecast errors.

28

Electronic copy available at: https://ssrn.com/abstract=953454

 In contrast, the coefficient on CFOt-1 is negative and significant, consistent with the

notion that low cash flow levels are more difficult for analysts to project into the future

(coefficient -1.052; t-value -3.22). The coefficient on Value Line’s proprietary

PREDICTABILITY index is negative and very significant (coefficient = -0.004; t-value -5.26),

suggesting that realized forecast errors are higher for firms that, ex-ante, analysts perceive more

difficult to forecast. The coefficient on Value Line’s STABILITY index is also negative and

significant (coefficient = -0.002; t-value = -2.17). Factors adversely affecting price stability also

seem to adversely affect analysts’ ability to forecast future cash flows. Incremental to

PREDICTABILITY and STABILITY, the five-year standard deviation of cash flows (σCFO) is

significant and positively associated with forecast errors (coefficient = 1.147; t-value 2.45). The

five-year standard deviation of accruals (σCNOA) is positively associated with forecast errors,

but only marginally so (coefficient = 0.405; t-value = 1.70). This suggests that variation in

operating cash flows and accruals is associated with future forecast errors, and is incompletely

incorporated into Value Line’s PREDICTABILITY index. We did not predict signs for the

coefficients on SIZE and PROFITABILITY; however, we find each is incrementally significant

and positively associated with forecast errors (coefficients = 0.031 and 0.647; t-values = 2.48 and

2.03).

VI. Summary and Conclusion

Plumlee (2003) reports that analysts’ forecast errors are positively related to the

complexity of forecast-relevant information. We extend her work by proposing that two features

of the information contained in the operating-activities section of the indirect-approach statement

of cash flows can contribute to forecasting complexity. First, we show that the currently

required structure of the operating-activities section of the indirect approach statement of cash

29

Electronic copy available at: https://ssrn.com/abstract=953454

flows⎯when compared to an alternative, forward-order format that conforms more closely to the

positively related, inter-temporal behavior of accruals and cash flows⎯causes lower levels of

learning and higher levels of forecast error and dispersion.

Second, we show that the offsetting (i.e., “mixed sign”) pattern of operating cash-flows

and operating accruals information reported by most companies also impedes investors’ ability to

learn the time-series properties of operating cash flows and operating accruals. While we find

that mixed-sign forecast inputs lead to significantly higher cash-flow forecast errors and forecast

dispersion across both forecast conditions, the forecast errors from the mixed-sign inputs are

highest in the currently mandated reverse-order indirect-approach statement of cash flows.

We also report the results of a quasi-experimental analysis of publicly available cash-

flow forecasts provided by Value Line analysts. Inferences from this analysis are consistent with

inferences drawn from the experiment and support our hypotheses about the complexity of

mixed-sign forecast inputs. Specifically, after controlling for firms life-cycle stage and other

economic factors commonly associated with the accuracy of analysts’ forecasts, we find that

forecast errors are higher for companies that report mixed-sign operating cash flows and

operating accruals information. These results suggest that cognitive limitations that impede

information acquisition manifest as systematic errors in a market setting.

This study extends prior research on the effects of alternative financial-reporting

classification, measurement, and presentation schemes on financial-statement users’ learning and

judgment processes (e.g., Dietrich, Kachelmeier, Kleinmuntz, and Linsmeier 2001). In this

study, we provide evidence that is directly relevant to questions regarding the form and content

of an important performance statement. Obtaining a better understanding of financial statement

users’ information acquisition processes will help to inform the deliberations of standard setters

30

Electronic copy available at: https://ssrn.com/abstract=953454

seeking to weigh the costs and benefits of proposed financial-statement disclosure requirements.

Interestingly, the “reverse” presentation format for the operating section of the indirect approach

statement of cash flows is often derided as confusing by students, accounting educators, and

users of financial statements (see, e.g., Stice, Stice, and Skousen 2004, 245; Stickney and Weil

2003, 184)16 and is opposite of the predictive, future-oriented perspective taken in most

conceptual descriptions of the balance sheet elements and the accruals process (e.g., FASB

1985).

Our study also illustrates an unintended consequence of giving primacy to reported net

income in the reconciliation between cash flows and accruals. The results of this study should be

useful to standard setters as they reconsider the structure of performance reporting and the

prominence currently afforded accrual-basis net income across currently required performance

reports. In particular, accrual-basis net income is (1) the focus of the statement of earnings, (2)

the numerator in the only ratio required by generally accepted accounting principles (i.e.,

earnings per share), (3) the first item included in the determination of comprehensive income,

and (4) the first item presented in the indirect approach statement of cash flows. However,

because net income is not defined in the conceptual framework, it is one of an infinite number of

potentially reportable, arbitrary partitions of the change in net assets during a given period.

Although we believe the comparative advantages of experiments are particularly well-

suited to the research questions in this study (Libby, Bloomfield and Nelson 2002), critics of

experiments-based research occasionally suggest that documented empirical findings in

experiments are caused by under-motivated participants in idiosyncratic, vacuum-like laboratory

16
In some cases, textbooks will supplement the detailed lists with a discussion of the rationale for indirect-approach
statement of cash flows operating-section adjustments that are unrelated to the expected timing of cash receipts and
disbursements. For example, Harrison and Horngren (2004) provide this explanation for the “reverse” adjustment
for increases in operating assets: “An increase in a current asset other than cash indicates a decrease in cash. That’s
because it takes cash to acquire assets.”
31

Electronic copy available at: https://ssrn.com/abstract=953454

settings producing results that cannot sustain in market settings. Our analysis attempts to

partially address these concerns by (1) providing our participants with accuracy-based

compensation and cumulative forecast-by-forecast feedback and (2) supplementing our

experimental results with a quasi-experimental archival analysis of publicly released cash-flow

forecasts (Libby, Bloomfield and Nelson 2002). The outcome-feedback mechanism

operationalized in our experiment is more transparent than those appearing in real-world

financial-forecasting settings and is more comprehensive and informationally complete than the

outcome feedback appearing in similar accounting-related experiments (e.g., Luft and Shields

2001; Bloomfield, Libby and Nelson 2003). Indeed, the form of outcome feedback included in

the present study has informational properties (i.e., single source, no delay, and linear) that

typically yield the highest levels of learning in prior research (Diehl and Sterman 1995).

32

Electronic copy available at: https://ssrn.com/abstract=953454

 References

American Institute of Certified Public Accountants. 2005. Accounting Trends and Techniques,
59th Edition (Y. Iofe, Editor), New York: AICPA.

Barth, M. E., D. P. Cram and K. K. Nelson. 2001. Accruals and the prediction of future cash
flows. The Accounting Review 76 (1): 27-58.

Bloomfield, R. J., R. Libby and M. W. Nelson. 2003. Do investors overrely on old elements of
the earnings time series? Contemporary Accounting Research 20 (Spring): 1-31.

Bonner, S. E. 1999. Judgment and decision-making research in accounting. Accounting

Horizons 13 (December): 385-399.

Bradshaw, M. T., S. A. Richardson, and R. G. Sloan. 2001. Do analysts and auditors use
information in accruals? Journal of Accounting Research 39 (1): 45-74.

Brehmer, B. 1973. Single-cue probability learning as a function of the sign and magnitude of the
correlation between cue and criterion. Organizational Behavior and Human Performance
9 (3): 377-395.

Brehmer, B. 1974. Hypotheses about relations between scaled variables in the learning of
probabilistic inference tasks. Organizational Behavior and Human Performance 11 (1):
1-27.

Brehmer, B. 1979. Note on hypothesis testing in probabilistic inference tasks. Scandinavian

Journal of Psychology 20 (3): 155-158.

Brehmer, B., J. Kuylenstierna, and J. Liljergren. 1974. Effects of function form and cue validity
on the subjects' hypotheses in probabilistic inference tasks. Organizational Behavior and
Human Performance 11 (3): 338-354.

Conover, W. J, M. E. Johnson, and J. K. Loebbecker. 1981. A comparative study of tests for

homogeneity of variances, with applications to the outer continental shelf bidding data.
Technometrics (November): 351-361.

Conover, W. J., 1999. Practical Nonparametric Statistics, 3rd Ed., New York: John Wiley and
Sons, Inc.

Deane, D., K. Hammond, and D. Summers. 1972. Acquisition and application of knowledge in
complex inference tasks. Journal of Experimental Psychology 92 (1): 20-26.

Dechow, P. M., S. P. Kothari, and R. L. Watts. 1998. The relation between earnings and cash
flows. Journal of Accounting and Economics 25 (1): 133-168.

33

Electronic copy available at: https://ssrn.com/abstract=953454

DeFond, M. L. and M. Hung. 2003. An empirical analysis of analysts’ cash flow forecasts.
Journal of Accounting and Economics 35 (April): 73-100.

Dickinson, V. 2006. Future profitability and the role of firm life cycle. Working paper, Fisher
School of Accounting, University of Florida.

Diehl, E. and J. D. Sterman. 1995. Effects of feedback complexity on dynamic decision making.
Organizational Behavior and Human Decision Processes 62 (May): 198-215.

Dietrich, J. R., S. J. Kachelmeier, D. N. Kleinmuntz, and T. J. Linsmeier. 2001. Market

efficiency, bounded rationality, and supplemental business reporting disclosures. Journal of
Accounting Research 39 (September): 243-269.

Easley, D. and M. O'Hara. 2004. Information and the cost of capital. The Journal of Finance
59(4): 1553-1584.

Elgers, P. T., M. H. Lo, and R. J. Pfeiffer Jr. 2003. Analysts’ vs. investors’ weightings of
accruals in forecasting annual earnings. Journal of Accounting and Public Policy 22: 255–
280.

Elliott, W. B., F. Hodge, J. J. Kennedy, and M. Pronk. 2007. When are graduate students a
reasonable proxy for nonprofessional investors? The Accounting Review 82 forthcoming.

Financial Accounting Standards Board. 1985. Elements of financial statements. Statement of

Financial Accounting Concepts No. 6. FASB: Norwalk CT.

Fligner, M. A. and T. J. Killeen. 1976. Distribution-free two-sample tests for scale. Journal of
the American Statistical Association (March): 210-213.

Francis, J. and M. Smith. 2005. A reexamination of the persistence of accruals and cash flows.
Journal of Accounting Research 43 (3): 413-451.

Freeman, R. N. and S. Y. Tse. 1992. A nonlinear model of security price responses to

unexpected earnings. Journal of Accounting Research 30 (Autumn): 185-209.

Harrison Jr., W. T. and C. T. Horngren. 2004. Financial Accounting, 5th Edition. Upper Saddle
River, NJ: Pearson Education, Inc.

Hirst, D. E., P. E. Hopkins & J. M. Wahlen. 2004. Fair values, comprehensive income reporting,
and bank analysts’ risk and valuation judgments. The Accounting Review (April): 455-
474.

Hong, H., J. C. Stein, and J. Yu. 2005. Simple forecasts and paradigm shifts. Unpublished
Working Paper: Princeton University.

34

Electronic copy available at: https://ssrn.com/abstract=953454

International Accounting Standards Board. 2004. Project Update: Performance Reporting
(Reporting Comprehensive Income), revised October 24, 2004, www.iasb.org/.

Libby, R., R. J. Bloomfield, and M. W. Nelson. 2002. Experimental research in financial

accounting. Accounting Organizations and Society 27 (October): 775-810.

Luft, J. and M. Shields 2001. Why does fixation persist? Experimental evidence on the judgment
performance effects of expensing intangibles. The Accounting Review 76 (October): 561–
587.

Maines, L. A. and L. S. McDaniel. 2000. Effects of comprehensive-income characteristics on

nonprofessional investors' judgments: The role of financial-statement presentation format.
The Accounting Review (April): 179-208.

Miller, G. A. 1956. The magical number seven, plus or minus two: Some limits on our capacity
for processing information. Psychological Review 63: 81-97.

Muchinsky, P. M. and A. L. Dudycha. 1975. Human inference behavior in abstract and

meaningful environments. Organizational Behavior and Human Performance 13: 377-391.

Naylor, J. and R. Clark. 1968. Intuitive inference strategies in interval learning tasks as a
function of validity magnitude and sign. Organizational Behavior and Human Performance
3 (4): 378-399.

Phillips, F., R. Libby, P. A. Libby. 2006. Fundamentals of Financial Accounting. New York:
McGraw Hill/Irwin.

Revsine, L., D. W. Collins, and W. B. Johnson. 2005. Financial Reporting & Analysis. Upper
Saddle River, NJ: Pearson Prentice Hall

Simon, Herbert A. 1982. Models of Bounded Rationality: Behavioral Economics and Business
Organizations, Vol. 2, (Cambridge, MA: MIT Press).

Sloan, R. G. 1996. Do stock prices fully reflect information in accruals and cash flows about
future earnings? The Accounting Review 71 (July): 289-316.

Sniezek, J. 1986. The role of variable labels in cue probability learning tasks. Organizational
Behavior & Human Decision Processes 38(2): 141-161.

Stice, E. K., J. D. Stice, and K. F. Skousen. 2004. Intermediate Accounting, 15th Edition. South-
Western: Mason, OH.

Stickney, C. P. and R. L. Weil. 2003. Financial Accounting: An Introduction to Concepts,

Methods and Uses, 10th Edition. South-Western: Mason, OH.

35

Electronic copy available at: https://ssrn.com/abstract=953454

 FIGURE 1
Format of Financial Information, Judgment Elicitation and Feedback Screens across
Conditions

FORWARD ORDER REVERSE ORDER

Statement of Cash Flows Screens

YEAR 1 Cash Flows from Operations XXX YEAR 1 Operating Earnings

YEAR 1 Increase (Decrease) in YEAR 1 (Increase) Decrease in

Non-Cash Operating Net Assets YYY Non-Cash Operating Net Assets (YYY)

YEAR 1 Operating Earnings YEAR 1 Cash Flows from Operations XXX

Judgment Screens

Forecast of YEAR 2 Cash Flows Forecast of YEAR 2 Cash Flows

from Operations ZZZ from Operations ZZZ

Feedback Screens

Cash Flows from Operations XXX YEAR 1 (Increase) Decrease in

Non-Cash Operating Net Assets (YYY)
YEAR 1 Increase (Decrease) in
Non-Cash Operating Net Assets YYY Cash Flows from Operations XXX

Your forecast of YEAR2 Cash Your forecast of YEAR2 Cash

Flows from Operations ZZZ Flows from Operations ZZZ
ACTUAL value of YEAR2 Cash ACTUAL value of YEAR2 Cash
Flows from Operations ACT Flows from Operations ACT
Percentage Error %ERR Percentage Error %ERR

(continued on next page)

36

Electronic copy available at: https://ssrn.com/abstract=953454

 FIGURE 1 (continued)

___________________
Participants were randomly assigned to one of two format conditions. After completing the pre-
experiment questions, participants provided forecasts of year-two cash flows from operations for
16 different companies. We counterbalanced the order of the companies (eight unique orders:
four different rotated orders with reversals of those orders) to ensure that all 16 companies were
equally represented across all 16 trials. For any given company, the information provided across
the conditions was identical. However, the presentation of the information differed in a manner
consistent with the “Statement of Cash Flows Screens” and “Feedback Screens” illustrated in the
Figure. After each judgment, participants were provided with comprehensive and cumulative
feedback information for all of their judgments (e.g., immediately prior to their 16th forecast,
participants could view the information on the “Feedback Screens” for all 15 of their preceding
forecasts. Upon completing forecasts for the 16 different companies, the participants provided
responses to an identical set of post-experiment questions (e.g., demographic questions).

%ERR = (ACT – ZZZ)/ACT

37

Electronic copy available at: https://ssrn.com/abstract=953454

 FIGURE 2
Experiment Results – Graphs of Interaction Terms for Forecast Errors and Dispersion

Panel A: Median Forecast Errors

Method x Mixed Interaction Method x Half Interaction Mixed x Half Interaction

0.45 0.45 0.45
0.40 0.40 0.40
Electronic copy available at: https://ssrn.com/abstract=953454

0.35 0.35 0.35

0.30 0.30 0.30
0.25 0.25 0.25
0.20 Reverse Reverse 0.20 Mixed
0.20
0.15 Forw ard 0.15 Forward 0.15 Same
0.10 0.10 0.10
0.05 0.05 0.05
0.00 0.00 0.00
Same Mixed First Half Last Half First Half Last Half

Panel B: Forecast Dispersion

Method x Mixed Interaction Method x Half Interaction Mixed x Half Interaction

3.50 3.50 3.50

3.00 3.00 3.00

2.50 2.50 2.50

2.00 2.00 2.00

Reverse Reverse Mixed
1.50 1.50 1.50
Forw ard Forw ard Same
1.00 1.00 1.00

0.50 0.50 0.50

0.00 0.00 0.00

Same Mixed First Half Last Half First Half Last Half

__________________________
The interactions graphed correspond to the interaction effects tested in Table 3, Panel B and Panel C.

38
 TABLE 1
Summary of Demographic Information and Pre- and Post-Experiment
Participant Assessments

Formatc
RO FO
(n = 25) (n=25) Comparisona
b
Panel A: Demographic Information
Number pursuing MBA degree (versus MPA) 19 19 χ2 = 0.00
Mean months work experience 10 19 Z= 0.95
Mean courses in finance and accounting 10 10 Z= 0.04
Number who plan to buy or sell stock in future 22 23 χ2 = 0.22
Times performed fundamental analysis on public data χ2 = 4.34
Category detail: - Once 2 2
- Once to five times 10 14
- Six to ten times 5 6
- More than ten times 8 2

Panel B: Pre-experiment itemsd

Accounting question 1 correct in first attempt 25 23 χ2 = 2.08
Perceived difficulty of accounting question 1 4.00 3.84 Z = -0.73
Accounting question 2 correct in first attempt 20 21 χ2 = 0.14
Perceived difficulty of accounting question 2 5.64 4.96 Z = -0.95

Panel C: Post-experiment itemse

Difficulty of forecasting task in experiment 8.16 8.68 Z = 0.55
Confidence in forecasts made in experiment 6.96 7.40 Z = 0.55

_____________________________
a = None of the comparisons are statistically significant. Comparisons of categorical (i.e., frequency)
data incorporate χ2 tests. Comparisons of interval data incorporate Wilcoxon/Mann-Whitney tests
(Conover 1999, 271-286). Parametric comparisons yield a similar pattern of insignificant
differences between groups.
b = After all forecasting-related questions (including DIF3 and CONF), we presented participants
with a series of demographic questions.
c = RO = Reverse order condition, FO = Forward order condition
d = Results of two accounting questions posed to all participants. We provide data for the number of
participants who correctly answered each question in their first attempt, along with the perceived
difficulty of each question. The difficulty scale ranged from 1 (Not at all Difficult) to 15
(Extremely Difficult). The first question required participants to determine cash flows from
customer sales given a set of facts related to sales and accounts receivable. The second question
required participants to determine cash flows from operations given a hypothetical company’s
operating earnings and beginning and ending balances for receivables, inventories, and payables
e = Participants difficulty assessment for the cash-flow forecasting task presented to them in the
experiment. Post-experiment Confidence Score = Participants confidence assessment for the
cash-flow forecasting task presented to them in the experiment. The scale range from 1 (Not at
all Confident) to 15 (Extremely Confident)

39

Electronic copy available at: https://ssrn.com/abstract=953454

 TABLE 2
Experimental Materials: Company Information

Panel A: Model Used to Generate Firm Observations in Experimental Materials

Change in
Cue-Generating Model Parameters: CFO t=1 NOA t=1
CFOt=2 = β1CFOt=1 + β2CNOAt=1 + εi β1=1.00 Β2=0.55

Panel B: Values for Firm Observations in Experimental Materials

Company CFO t=2 CFO t=1 CNOA t=1

A (254) (198) (142)
B (245) (64) (385)
C (727) (629) (75)
D (176) (131) (72)
E 108 (406) 899
F 370 (60) 787
G (171) (320) 351
H (633) (932) 653
I 3,178 3,795 (1,286)
J 186 278 (216)
K 1,756 1,962 (396)
L 348 504 (293)
M 646 566 152
N 422 76 664
O 282 8 581
P 324 125 437

Panel C: Descriptive Statistics for Firm Observations in Experimental Materials

CFO t=2 CFO t=1 CNOA t=1

Descriptive Statistics:
Mean 338.38 285.88 103.68
Standard Deviation 920.93 1,096.64 554.95
Minimum -727.00 -932.00 -1,286.00
Maximum 3,178.00 3,795.00 899.00
(continued on next page)

40

Electronic copy available at: https://ssrn.com/abstract=953454

 TABLE 2 (continued)

_______________________________
CFO t=2 = cash flows from operations during period 2
CFO t=1 = cash flows from operations during period 1
CNOA t=1 = change in net operating assets during period 1

We used a model to generate the company information included in the experimental materials.
The model parameters are based on the general relation between the time-series properties of
operating cash flows and changes in operating assets in Barth et al. (2001). The error component
is normally distributed with a mean of zero. Contemporaneous changes in operating assets are
negatively correlated with operating cash flows (Dechow et al., 1998).

*, **, *** – (two-tailed), p-value ≤ 0.10, 0.05, and 0.01 level of significance respectively

41

Electronic copy available at: https://ssrn.com/abstract=953454

 TABLE 3
Experiment Results: Forecast Errors and Dispersion of Forecast Judgments

Panel A: Median, (Mean), and [Standard Deviation of Forecast Judgments] of Forecast Errors

Reverse Order Forward Order Across Orders

Electronic copy available at: https://ssrn.com/abstract=953454

First Half Last Half Overall First Half Last Half Overall First Half Last Half Overall
Mixed 0.41 0.29 0.39 0.49 0.20 0.35 0.43 0.25 0.36
(1.48) (1.16) (1.32) (1.17) (0.70) (0.94) (1.32) (0.93) (1.13)
[3.72] [2.30] [3.09] [2.28] [1.49] [1.95] [3.07] [1.95] [2.59]

Same 0.26 0.28 0.27 0.21 0.17 0.18 0.24 0.19 0.22
(0.44) (0.49) (0.47) (0.37) (0.27) (0.32) (0.41) (0.38) (0.39)
[0.54] [0.57] [0.55] [0.40] [0.30] [0.35] [0.47] [0.46] [0.47]

Overall 0.35 0.29 0.30 0.37 0.17 0.22 0.36 0.22 0.27
(0.96) (0.82) (0.89) (0.79) (0.47) (0.63) (0.87) (0.65) (0.76)
[2.70] [1.70] [2.26] [1.71] [1.07] [1.43] [2.26] [1.43] [1.89]

Panel B: Repeated Measures Analysis of Variance on Rank Transformed Forecast Errors

Factors df F Value Pr > Fa

Method 1 8.66 <0.01
Mixed 1 21.41 <0.01
Half 1 12.18 <0.01
Method x Mixed 1 0.28 0.60
Method x Half 1 5.36 0.02
Mixed x Half 1 2.52 0.11
Method x Mixed x Half 1 0.13 0.72
Error 792
(Continued on next page)

42
 TABLE 3 (continued)

Panel C: Repeated Measures Analysis of Variance on Rank Transformed Variability Measureb

Factors df F Value Pr > Fa

Method 1 29.97 <0.01
Mixed 1 32.67 <0.01
Electronic copy available at: https://ssrn.com/abstract=953454

Half 1 19.02 <0.01

Method x Mixed 1 4.57 0.03
Method x Half 1 15.82 <0.01
Mixed x Half 1 0.92 0.34
Method x Mixed x Half 1 0.09 0.77
Error 792

___________________________
Forecast Error = The absolute percentage error subjects response deviated from CFOt=2 calculated by taking the absolute value of the
difference between actual CFOt=2 and subject’e estimate of CFOt=2 divided by the model estimate CFOt=2.
Mixed/Same = Dichotomous variable equal to 1 (0) if the sign of CFO t=1 and the sign of CNOA t=1 were (were not) the same.
First/Last Half = Dichotomous variable equal to 1 (0) if the subject estimated CFO t=2 in the first (last) 8 trials of the experiment.
Method = Dichotomous variable equal to 1 (0) if the presentation format was RO (FO).
a
Reported p-values represent two-tailed tests of significance.
b
Rank transformed variability measure is generalized from the nonparametric test for differences in dispersion suggested by Fligner
and Killeen (1976) as modified by Conover et al. (1981). The dependent measure is computed by transforming the absolute value of
the difference between each observation and its corresponding cell median. The transformation used is
Φ-1(0.5+i/2(N+1)) where Φ-1 represents the inverse standard normal distribution function, i is the ranked absolute value, and N is the
combined sample size of all experimental cells.

43
 TABLE 4
Extracted Model Weights by Condition

Forward Reverse Comparison of Forward

Order Order Order to Reverse Order
Subject Weights on CFOt-1:
Mean 0.90# 0.96 (Wilcoxon Z-stat = 0.12)
Median 0.91## 0.91## ( Median Z-stat = -0.28)
Standard Deviation 0.25 0.45 *** (Folded F-value = 3.38)
Minimum 0.26 0.13
Maximum 1.29 2.88
Mean deviation from
theoretical value -0.10 -0.04 (Wilcoxon Z-stat = 0.12)

Subject Weights on CNOAt-1:

Mean 0.50 0.41# ** (Wilcoxon Z-stat = 2.17)
Median 0.50# 0.34### ** (Median Z-stat= 1.96)
Standard Deviation 0.23 0.44 *** (Folded F-value = 3.70)
Minimum 0.05 -0.37
Maximum 1.24 2.06
Mean deviation from
theoretical value -0.05 -0.14 ** (Wilcoxon Z-stat = 2.17)

___________________________
Subject weight parameters are obtained by estimating the following model for each subject over
16 forecast trials:

CFOt,s = βsCFOt-1 + βsCNOAt-1.+ εs (A)

Mean deviations from the theoretical parameter values are calculated using the same model
weights implicit in the generated data (1.0 for CFOt-1 and .55 for CNOAt-1).

Nonparametric comparisons were made using the Wilcoxon-Mann-Whitney test and Median
tests where appropriate.

#, ##, ### – amount varies from its theoretical value at (two-tailed) p-value ≤ 0.10, 0.05, and
0.01 level of significance respectively

*, **, *** – differences across conditions are significant (two-tailed), p-value ≤ 0.10, 0.05, and
0.01 level of significance respectively

44

Electronic copy available at: https://ssrn.com/abstract=953454

 TABLE 5
Archival Data Descriptive Statistics

Panel A: Descriptive Statistics for Regression Variables

All Observations Means by Sign Life Cycle Stagea

Shake-
Variable Q1 Mean Q3 Startup Growth Mature out Decline
ERRORt 0.052 0.289 0.310 0.506 0.278 0.251 0.402 0.419
MIXEDt-1 (%) 87.12% 43.75% 87.10% 94.00% 82.78% 41.93%
GROWTHt -0.002 0.122 0.179 0.040 0.158 0.116 0.104 0.004
RANK_|INCOMEt-1| 2.000 3.814 6.000 4.813 3.540 3.839 3.541 5.645
RANK_|CNOAt-1| 2.000 4.011 6.000 4.875 3.601 4.301 3.287 4.483
CFOt-1 0.050 0.089 0.133 -0.077 0.091 0.118 0.071 -0.124
PREDICTABILITYt-1 25.000 48.701 70.000 32.033 48.273 53.012 39.425 31.129
STABILITYt-1 25.000 50.515 75.000 23.910 47.823 57.075 44.713 20.000
σCFOt-5 to t 0.020 0.047 0.060 0.090 0.044 0.042 0.048 0.089
σCNOAt-5 to t 0.021 0.060 0.061 0.126 0.066 0.048 0.056 0.134
σINCOMEt-5 to t 0.013 0.054 0.048 0.138 0.063 0.037 0.049 0.151
RANK_SIZEt-1 6.000 6.628 8.000 5.501 6.608 6.839 6.448 5.323
PROFITABILITYt-5 to t 0.010 0.026 0.069 -0.110 0.023 0.047 0.035 -0.143
N 893 32 278 465 87 31
(100.00%) (3.58%) (31.13%) (52.07%) (9.74%) (3.47%)

Panel B: Descriptive statistics for Compustat population for the same periods:

All Observations Means by Sign Life Cycle Stagea

Shake-
Variable Q1 Mean Q3 Startup Growth Mature out Decline
σCFOt-5 to t 0.026 0.051 0.094 0.177 0.062 0.056 0.085 0.167
σCNOAt-5 to t 0.026 0.050 0.108 0.221 0.076 0.064 0.115 0.255
σINCOMEt-5 to t 0.017 0.041 0.111 0.290 0.076 0.054 0.116 0.313
PROFITABILITYt-5 to t -0.044 0.019 0.052 -0.326 0.011 0.035 -0.040 -0.320
9,759 1,036 2,602 4,057 1,050 815
N (100.00%) (10.83%) (27.22%) (42.44%) (10.98%) (8.53%)
(continued on next page)

45

Electronic copy available at: https://ssrn.com/abstract=953454

Panel C: Industry Distribution by Life Cycle Stage

Shake-
Industry SIC Codes Startup Growth Mature out Decline
Agriculture, mining, 0001-1999, except 1300-1399 3.13 2.88 0.86 4.60 0.00
construction
Food 2000-2111 0.00 1.44 2.15 1.15 0.00
Textiles, printing and 2200-2799 3.13 2.52 7.53 4.60 3.23
publishing
Chemicals 2800-2824, and 2840-2899 12.50 5.04 5.81 4.60 9.68
Pharmaceuticals 2830-2836 25.00 5.40 1.29 2.30 32.26
Extractive industries 2900-2999, and 1300-1399 0.00 2.16 3.87 1.15 0.00
Durable manufacturers 3000-3999, except 3530- 21.88 19.06 18.92 18.39 3.23
3579, and 3670-3679
Computers 7370-7379, 3570-3579, and 15.63 10.79 10.97 6.90 22.58
3670-3679
Transportation 4000-4899 3.13 5.40 6.88 10.34 0.00
Utilities 4900-4999 6.25 16.55 12.47 16.09 0.00
Retail 5000-5999 3.13 5.76 6.67 5.75 9.68
Services 7000-8999, except 7370-7379 6.25 19.78 19.35 21.84 12.90
Other >9000 0.00 3.24 3.23 2.30 6.45

__________________________
a
GROUP IDENTIFIERS:
CFO is Cash from Operations, CFI is Cash from Investing, and CFF is Cash from Financing
Startup = negative CFO, negative CFI, positive CFF
Growth = positive CFO, negative CFI, positive CFF
Mature = positive CFO, negative CFI, negative CFF
Decline = negative CFO, positive CFI, negative CFF
Shakeout = all other combinations, not defined above

ERRORt = absolute value of the Value Line cash flow forecast error divided by the actual cash flow per
share reported by Value Line for period t. Cash flow forecast error is the actual cash flow per
share reported by Value Line for period t minus the Value Line forecasted cash flow per share as
of April 30 in period t.
NEG_CFOt-1 = Dummy variable equal to 1 if operating cash flow is negative and 0 otherwise.
ASSET_GROWTHt-1 = (ASSETSt – ASSETSt-1)/ASSETSt-1 (Compustat data6).
RANK_|INCOMEt-1| = Decile rank of the absolute value of income before extraordinary items
(Compustat data123 divided by Compustat data6). |INCOME| is ranked within the population of
Compustat firms.
RANK_|CNOAt-1| = Decile rank of the absolute value of the difference between INCOME and CFO
(Compustat data123 - Compustat data308) divided by ASSETS (Compustat data6). |CNOA| is
ranked within the population of Compustat firms.
CFOt-1 = CFO (Compustat Data308) divided by ASSETS (Compustat data6).
PREDICTABILITYt-1 = Value Line’s earnings predictability index. A measure of the reliability of
forecasts based on the stability of year-to-year comparisons. Ranging from lowest predictability
of 5 to highest predictability of 100.
(continued on next page)

46

Electronic copy available at: https://ssrn.com/abstract=953454

STABILITYt-1 = Value Line’s price stability index. A measure of the stability of a stock’s price. It
includes sensitivity to the market (beta) as well as the firm’s inherent volatility. Ranging from
lowest stability of 5 to highest stability of 100.
σCFOt-5 to t = Standard deviation of CFO over the five years ending with the forecast year.
σCNOAt-5 to t = Standard deviation of CNOA over the five years ending with the forecast year.
RANK_SIZEt-1 = Decile rank of ASSETS in the period of the forecast. ASSETS are ranked within the
population of Compustat firms.
PROFITABILITYt-5 to t = Average profitability over the five years ending with the forecast year of
INCOME divided by ASSETS.

47

Electronic copy available at: https://ssrn.com/abstract=953454

 TABLE 6

Pearson (above the diagonal) and Spearman (below the diagonal) Correlations (N=893)

NEG ASSET_ RANK_ RANK_ PREDICT- STABILITY σ RANK_ PROFIT-

ERRORt _CFOt-1 MIXEDt-1 GROWTH |INCOME|t-1 |CNOA|t-1 CFOt-1 ABILITYt-1 t-1 σCFO CNOA SIZEt-1 ABILITY

ERRORt 0.127 0.190 0.650 -0.010 0.107 -0.096 -0.300 -0.233 0.180 0.104 -0.013 -0.120
Electronic copy available at: https://ssrn.com/abstract=953454

NEG_CFOt-1 0.138 -0.365 -0.081 0.147 0.063 -0.597 -0.191 -0.280 0.307 0.153 -0.217 -0.365

MIXEDt-1 -0.047 -0.365 0.038 -0.227 0.217 0.374 0.035 0.179 -0.165 -0.093 0.132 0.166
ASSET_
0.018 -0.122 0.024 0.126 0.020 0.190 -0.002 0.005 0.015 -0.027 0.133 0.127
GROWTH
RANK_
0.034 0.125 -0.213 0.222 0.168 0.209 0.214 -0.161 0.204 0.026 -0.271 0.098
|INCOME|t-1
RANK_
0.093 0.054 0.219 0.004 0.149 0.274 -0.203 -0.220 0.164 0.056 -0.082 -0.094
|CNOA|t-1

CFOt-1 -0.129 -0.472 0.390 0.300 0.412 0.377 0.194 0.161 -0.129 -0.089 0.105 0.493

PREDICT-
ABILITYt-1
-0.301 -0.192 0.032 0.087 0.217 -0.203 0.236 0.469 -0.251 -0.205 0.063 0.345

STABILITYt-1 -0.309 -0.285 0.182 0.029 -0.156 -0.215 0.125 0.464 -0.360 -0.272 0.391 0.310

σCFO 0.169 0.278 -0.190 0.035 0.215 0.162 -0.011 -0.295 -0.415 0.339 -0.357 -0.293

σCNOA 0.219 0.258 -0.195 -0.047 0.131 0.155 -0.129 -0.335 -0.502 0.735 -0.140 -0.750

RANK_SIZEt-1 -0.149 -0.197 0.119 0.090 -0.259 -0.070 -0.123 0.052 0.392 -0.379 -0.345 0.102

PROFIT-
ABILITYt-1
-0.200 -0.337 0.088 0.336 0.438 -0.092 0.601 0.545 0.301 -0.082 -0.2345 -0.040

(continued on next page)

48
Table 6 Notes:

Correlations shown in bold are significant at the .05 level or better.

ERRORt = absolute value of the Value Line cash flow forecast error divided by the actual cash flow per
share reported by Value Line for period t. Cash flow forecast error is the actual cash flow per
share reported by Value Line for period t minus the Value Line forecasted cash flow per share as
of April 30 in period t.
NEG_CFOt-1 = Dummy variable equal to 1 if operating cash flow is negative and 0 otherwise.
ASSET_GROWTHt-1 = (ASSETSt – ASSETSt-1)/ASSETSt-1 (Compustat data6).
RANK_|INCOMEt-1| = Decile rank of the absolute value of income before extraordinary items
(Compustat data123 divided by Compustat data6). |INCOME| is ranked within the population of
Compustat firms.
RANK_|CNOAt-1| = Decile rank of the absolute value of the difference between INCOME and CFO
(Compustat data123 - Compustat data308) divided by ASSETS (Compustat data6). |CNOA| is
ranked within the population of Compustat firms.
CFOt-1 = CFO (Compustat Data308) divided by ASSETS (Compustat data6).
PREDICTABILITYt-1 = Value Line’s earnings predictability index. A measure of the reliability of
forecasts based on the stability of year-to-year comparisons. Ranging from lowest predictability
of 5 to highest predictability of 100.
STABILITYt-1 = Value Line’s price stability index. A measure of the stability of a stock’s price. It
includes sensitivity to the market (beta) as well as the firm’s inherent volatility. Ranging from
lowest stability of 5 to highest stability of 100.
σCFOt-5 to t = Standard deviation of CFO over the five years ending with the forecast year.
σCNOAt-5 to t = Standard deviation of CNOA over the five years ending with the forecast year.
RANK_SIZEt-1 = Decile rank of ASSETS in the period of the forecast. ASSETS are ranked within the
population of Compustat firms.
PROFITABILITYt-5 to t = Average profitability over the five years ending with the forecast year of
INCOME divided by ASSETS.

49

Electronic copy available at: https://ssrn.com/abstract=953454

 TABLE 7
Summary Statistics for Forecast Error Regression
4
ERROR t = β 0 + β 1 MIXED t −1 + ∑ β 2 ,i STAGE i ,t −1 + β 3 ASSET _ GROWTH t −1 + β 4 RANK _ | INCOME t −1 |
i =1
+ β 5 RANK _ | CNOAt −1 | + β 6 CFO t −1 + β 7 PREDICTABI LITY t −1 + β 8 STABILITY t −1 + β 9 σCFO t − 5,t
12
+ β 10 σCNOAt − 5,t + β 11 SIZE t −1 + β 12 PROFITABIL ITYt − 5,t + ∑ β 13, j INDUSTRY j ,t −1 + ε t
j =1

Variable Pred. Coeff. t-statistic Coeff. t-statistic Coeff. t-statistic

INTERCEPT ? 0.149*** 2.74 0.016 0.15 0.057 0.39
MIXEDt-1 + 0.108** 2.06 0.107** 2.03 0.135*** 2.45
STARTUPt-1 + 0.309*** 3.33 0.320*** 3.40 0.060 0.58
GROWTHt-1 + 0.034*** 0.92 0.051 1.36 0.010 0.26
SHAKEOUTt-1 + 0.163*** 2.85 0.172*** 3.02 0.074 1.30
DECLINEt-1 + 0.224*** 2.38 0.237*** 2.44 -0.048 -0.43
ASSET_GROWTHt-1 + 0.119* 1.80
RANK_|INCOMEt-1| + 0.012 1.39
RANK_|CNOAt-1| + 0.013 1.72
CFOt-1 - -1.052*** -3.22
PREDICTABILITYt-1 - -0.004*** -5.26
STABILITYt-1 - -0.002** -2.17
σCFOt-5 to t + 1.147** 2.45
σCNOAt-5 to t + 0.405* 1.70
RANK_SIZEt-1 ? 0.031** 2.48
PROFITABILITYt-5 to t ? 0.647** 2.03
Ag, Mining, Construction ? 0.143 0.96 0.006 0.04
Food ? 0.063 0.41 0.152 1.02
Textiles, printing and pub ? 0.248** 2.16 0.173 1.56
Chemicals ? 0.005 0.04 0.004 0.04
Pharmaceuticals ? 0.088 0.73 -0.062 -0.52
Extractive industries ? 0.271** 2.04 0.044 0.32
Durable manufacturing ? 0.157 1.59 0.103 1.08
Computers ? 0.224** 2.17 0.072 0.69
Transportation ? 0.268** 2.40 0.179 1.64
Utilities ? -0.004 -0.04 0.014 0.14
Retail ? 0.174 1.56 0.169 1.56
Services ? 0.091 0.93 0.042 0.44
N 893 893 893
2
Adj. R 0.017 0.037 0.122
(continued on next page)

50

Electronic copy available at: https://ssrn.com/abstract=953454

 TABLE 7 (continued)

__________________
Industry and year effects are not tabulated. Industries are defined based on four digit SIC codes used by
Fama and French (1997).

*, **, *** – (one-tailed), p-value ≤ 0.10, 0.05, and 0.01 level of significance respectively

See variable definitions in Table 6 Notes

51

Electronic copy available at: https://ssrn.com/abstract=953454