1

On Accounting Flows and Systematic Risk

Neil Garrod
University of Glasgow

Dusan Mramor
University of Ljubljana














Address for correspondence: Neil Garrod,
Department of Accounting and Finance,
University of Glasgow,
65-71, Southpark Avenue,
Glasgow G12 8LE,
Scotland,
U.K.

Tel: 00-44-141-330-5426
e-mail: [email protected]

2
On Accounting Flows and Systematic Risk


Abstract
The body of work that relates accounting numbers to market measures of systematic equity risk was largely
undertaken in the 1970s and early 1980s. More recent proposals on changes in accounting disclosure of risk
mean that a rigorous theoretical model of the relationship between accounting measures and market measures of
risk is timely. In this paper such a model is developed. In addition, the assumptions required to develop the
model are explicitly identified. By so doing it becomes possible to identify the potential cross-sectional
differences which drive the empirical relationship between accounting and market based measures of risk. The
model developed highlights a clear relationship between accounting and market measures of risk which can be
exploited in situations where accounting data alone is available. It also provides a framework within which the
environmental factors leading to cross-sectional differences between companies can be further explored.


3
On Accounting Flows and Systematic Risk

I. Introduction
Work that relates accounting numbers to market measures of systematic equity risk was largely undertaken in the
1970s and early 1980s (Ryan, 1997). More recent proposals on changes in accounting disclosure of risk
(Scholes, 1996) mean that a theoretically sound model of the relationship between accounting measures and
market measures of risk is timely. In addition, the finding that earnings variability is the accounting variable
related most strongly to systematic equity risk (Beaver et. al., 1970; Rosenberg and McKibben, 1973; Myers,
1977) suggests that a disaggregation of this number into the operational aspects of a firm which drive the
earnings number might improve the empirical relationship between accounting estimates of beta and its market
realisation. In this paper a rigorous theoretical model of the relationship between accounting flow variables and
systematic market risk of equity is developed.

Identification of this relationship is helpful on a number of fronts. Firstly, the instability of market betas over
time means that ex post measures of market risk are not good predictors of future risk. Identification of an
appropriate relationship between accounting variables and market risk could lead to improved predictive models
of future market risk. Secondly, financial models of risk (e.g. CAPM) do not identify the operational factors and
environmental contingencies which influence risk. An accounting model gets closer to the identification of
economic fundamentals which drive such relationships. Finally, interest in this relationship is further fuelled by
being of practical use in situations where market estimates of risk are unavailable. Conventionally these have
been considered as situations such as the estimation of risk for private companies, initial public offerings or
divisional capital budgeting. However, the transformation of former command economies to market economies
has now created a situation where theoretical models of risk assessment which can be used in the pricing of
companies for privatisation purposes are at a premium.

The remainder of the paper is set out as follows: in section II the previous literature in this field is briefly
reviewed; in section III the model is developed along with preliminary consideration of the relationship between
the theoretical model developed and its empirically testable equivalent; in section IV concluding remarks are
made.

II. Previous Research
Work in this field can be usefully divided between theoretical and empirical studies. The empirical work has,
largely, been unguided by a theoretical model (Foster, 1986). This has resulted in regressions of market
measures of market beta on various accounting measures of risk (Beaver, Kettler and Scholes, 1970; Pettit and
Westerfield, 1972; Breen and Lerner, 1973; Rosenberg and McKibben, 1973; Thompson, 1974; Lev, 1974; Lev
and Kunitzky, 1974; Bildersee, 1975; Beaver and Manegold, 1975) or the use of accounting number analogues
to market derived measures of risk ( Hill and Stone, 1980). Given the lack of rigorous theory underlying these
various models and the, often high, correlation between the accounting variables, these studies identify often
quite different significant explanatory variables. What does appear common across the studies, however, is that
earnings variability is the most significant accounting variable in explaining risk; that both accounting variables

4
and other information are useful in the assessment of risk and that substantial room remains for additional
research (Ryan, 1997).

The theoretical work began with Hamada (1972) and Rubinstein (1973) who identified the multiplicative impact
of financial leverage on the beta of the levered firm. Their, by now, well known results is that:

( ) = +

1
D
E


where = the levered firms common stock beta,

*
= the unlevered firms common stock beta,
= the corporate income tax rate,
D = the market value of debt, and
E = the market value of common equity.

Whilst
*
was called operating risk, Rubinstein recognised that it reflected the combined effects of operating
leverage, the pure systematic influence of economy wide events and uncertainty surrounding the firms operating
efficiency. Lev (1974) separated operating leverage from the other two variables and found it to be individually
significant.

Mandelker and Rhee (1984) explicitly incorporate measures of the degree of operating and financial risk into
their theoretical model and arrive at the following relationship:


j j
O
DOL)(DFL) = (

where:
j
= the levered firms common stock beta,
DOL = the degree of operating leverage
=
~
~
X
X
Sjt
Sjt
jt
jt
|
\

|
.
|
|

(
(

|
\

|
.
|

(
1
1
1
1

where: X
jt
= earnings before interest and taxes for company j in period t, and
S
jt
= sales for company j in period t,
and represents expectations
DFL = the degree of financial leverage

5
=
~
~

jt
jt
jt
jt
X
X

|
\

|
.
|
|

(
(
|
\

|
.
|
|

(
(
1
1
1
1

where:
jt
= earnings after interest and taxes for company j in period t, and

j
O
= the intrinsic business risk of common equity of company j
=
( )
Cov
S
S
E
R
Var R
jt
jt
jt
jt
mt
mt

|
\

|
.
|
|
|
.
|
|
|
\

(
(
(
(
(
(
1
1
1
~
,
~
~
.
.

where: E
jt
= the market value of common equity of company j in period t
R
mt
= the rate of return on the market portfolio for period t-1 to t
A major contribution of the Mandelker and Rhee (1984) model over Hamada and Rubinstein type models is that
it utilises leverage values based on accounting flow numbers (degree of operating and financial leverage) rather
than market stock numbers (level of operating and financial leverage). In the Hamada model, for example, both
the value of debt and equity are stock measures and, theoretically, should be market values. However, Bowman
(1980) found that the market value of debt was not significant in assessing the effect of financial leverage on
risk, but this may be attributable to the noise in his estimates of the market value of private debt (Ryan, 1997).
The difficulty in finding a market value of debt in many cases has led researchers (e.g. Chance, 1982) to use
accounting book (stock) values of debt in leverage estimates. The use of book values is a major limitation on the
subsequent leverage measures as it effectively constrains the leverage measure to be a static one which is unable
to respond and reflect the changing relative costs of equity and debt. The use of flow equivalents avoids this
problem even when using accounting data and ensures that the resultant leverage measures are dynamic and
responsive to changes in the economic environment. Defining the degree of total leverage (DTL) as the
percentage change in net income which results from a 1% change in sales, the degree of financial leverage (DFL)
as the percentage change in net income which results from a 1% change in earnings before interest and taxes, and
the degree of operating leverage (DOL) as the percentage change in earnings before interest and taxes (operating
income) which results from a 1% change in sales, we have, by definition, that
DTL = DFL*DOL
Unfortunately, the Mandelker and Rhee model suffers from two problems as a rigorous, accounting based
theoretical model of levered . Firstly, the impact of utilising accounting proxies for market measures of return
are not explicitly recognised within the model and, secondly, their measure of the intrinsic business risk of the
company incorporates both an accounting measure of profit and a market measure of value. The former is
subject to accounting manipulation under different codes of generally accepted accounting principles (GAAP),
whilst the latter is a non accounting measure of value. Our intention in the next section is to develop a rigorous

6
model which defines basic business risk utilising only published accounting data as free from accrual
manipulations as possible.


III. Model Development
We start our model development with the basic accounting equality:
NI = (S - VC - FC - I)(1-) ...........(1)
where: NI = net income,
S = sales,
VC = variable costs,
FC = fixed costs,
I = interest payments, and
= the company average and marginal tax rate.
and thus, taking present values, leads us to:
PV(NI)= [PV(S) - PV(VC) - PV(FC) - PV(I)](1-) ........(2)
Applying the linear additivity of systematic risk ( Brealey and Myers, 1993) and replacing NI/(1-) by earnings
before interest and tax (EBIT) minus interest, equation (2) can be expressed as:
[ ]

E S VC FC
PV S PV VC PV FC PV I
PV EBIT PV I
=

{ ( ) ( )} { ( ) ( )}
[ ( ) ( )]
+ ......(3)
D
1

Where:

E
= Cov(change in dividend adjusted value of equity, change in dividend adjusted value of total market equity)
Variance( change in dividend adjusted value of total market equity)

S
= Covariance( change in sales, change in dividend adjusted value of total market equity)
Variance( change in dividend adjusted value of total market equity)

VC
= Covariance( change in variable costs, change in dividend adjusted value of total market equity)
Variance( change in dividend adjusted value of total market equity)

FC
= Covariance( change in fixed costs, change in dividend adjusted value of total market equity)
Variance( change in dividend adjusted value of total market equity)

D
= Covariance( change in debt value, change in dividend adjusted value of total market equity)
Variance( change in dividend adjusted value of total market equity)

Under the normal (see e.g. Brealey and Myers, 1993) simplifying assumptions that:
1.
FC
=
I
= 0, and
2.
VC
=
S
,
equation (3) simplifies to

Ef S S
PV S PV VC
PV EBIT PV I
PV EBIT PV FC
PV EBIT PV I
=

=
+

( ) ( )
( ) ( )
( ) ( )
( ) ( )
......(4)

7
The coefficient of
S
represents 1 plus a stock measure of total leverage. In order to convert this model into one
which uses the flow measure of degree of total leverage, two further assumptions need to be made:
3. the discount rate on all variables is equal, and
4. the growth rate on all variables is equal.
Given the presence of interest and fixed costs, the appropriate discount rate would appear to be the risk free rate
and the presence of interest would indicate a zero level of growth. However, it is only the equality of these rates
and not their values which is important, and under these conditions equation (4) simplifies to:

EfDTL S f S
EBIT FC
EBIT I
DTL =
+

= ......(5)
where:
DTL
f
= the degree of total leverage based on actual accounting data and assuming riskless debt and fixed costs
Thus we now have a model for levered based upon disclosed accounting variables. The model is, of course,
very similar to the Mandelker and Rhee model except that their measure of intrinsic business risk has been
replaced in our model by a measure of sales risk. These can readily be shown to be equivalent but our
formulation has the advantage of relying only on sales and not profit and does not include any market based
variables. In addition, we have explicitly identified the assumptions which are necessary in order to arrive at this
accounting based estimate of risk. By so doing we are in a position to investigate the likely impact of each of the
assumptions on our estimate of systematic equity risk.
Assumption 1:
FC
=
D
= 0
The possibilities of bankruptcy mean that debt is not totally riskless and thus DTL
f
overestimates the true impact
of total leverage on common equity. The extent of the overestimation will be directly, and linearly, related to the
bankruptcy risk of debt. With regard to fixed costs their composition is likely to be dominated by asset charges
(depreciation) and (un)employment costs. Whilst the former are unlikely to vary with general market
movements, the latter risks will be absorbed by the workforce to a greater or lesser extent dependent upon the
extent of employment protection legislation. Additionally, in capital intensive industries any employment
influences are likely to be dominated by depreciation charges so that any impact on the estimate of equity risk is
likely to be small. The net effect of risky fixed costs would again be to overestimate
E
. The required correction
will be an additive adjustment to DTL which is proportional to measures of solvency and liquidity and inversely
proportional to corporate employment legislation protecting employees against unemployment.
Thus:
E
= (DTL - k
1
)
S

Assumption 2:
VC
=
S
On the assumption that prices include an element for variable costs, fixed costs and profit the risk of sales
revenue is likely to be greater than that of variable costs. The extent to which it exceeds
VC
will depend upon
the competitive nature of the industry in which a firm is operating. In highly competitive industries the
importance of fixed costs and profit in the pricing equation will be smaller than in less competitive industries and
thus any difference between
VC
and
S
would not be large. In any event,
S
exceeding
VC
will lead to an
underestimate of the true risk of common equity by using equation (5). Again the adjustment will be inversely
proportional to industry competition and an additive adjustment to
S
.
Thus:
E
= (DTL - k
1
+ k
2
)
S


8
Assumption 3 : the discount rate on all variables is equal
Whatever the risk of debt and fixed costs, by definition, the risk and, therefore, the discount rate on EBIT will be
larger. Thus both the numerator and denominator of the coefficient of
S
in equation 5 will decrease by the same
amount. As the coefficient in equation (5) is greater than 1 this change will result in a decrease in the coefficient.
In this case the impact on our estimate of
E
is multiplicative on
Ef
and inversely proportional to (r
EBIT
- r
FC
).
The most convenient indicator of this difference is likely to be the degree of operating leverage itself.
Thus:
E
= {(1 + k
3
)DTL - k
1
+ k
2
)
S
Assumption 4: the growth rate on all variables is equal
Improving efficiency and economies of scale should ensure that the growth of earnings before interest and taxes
should exceed the growth in fixed costs and debt. This will result in a decrease in the numerator and denominator
of the coefficient of
S
in equation 5 and, thus, a decrease in the coefficient itself. The impact on the estimate of

E
is again multiplicative on
Ef
and inversely proportional to (g
EBIT
- g
FC
). Appropriate predictors of this value
could be GNP, industry growth levels or historical company growth levels in earnings before interest and tax.
Thus:
E
= {(1 + k
3
- k
4
)DTL - k
1
+ k
2
)
S

IV. Concluding Remarks
In this paper we have developed a theoretically valid model to estimate the systematic risk of a companys
equity. By commencing the analysis from the fundamental accounting equality we are able to generate a
forecasting model which utilises accounting measures to the fullest extent. This is considered important because
of the number of significant situations where market measures are unavailable. As it turns out the model still
contains a measure of systematic sales risk which depends upon a market, but not a company specific, measure
of return. It is clearly an empirical, rather than a theoretical, issue for further research as to whether accounting
proxies of such a measure provide suitable and accurate measures of systematic sales risk such that a pure
accounting model can be developed.

The model generated turns out to be similar to that of Mandelker and Rhee (1985). However, their model
weights the degree of total leverage by a measure of intrinsic business risk rather than sales risk. Whilst the
former turns out to be a simple linear function of the latter, the inclusion of this additional factor necessarily
leads to increased measurement error on their measure of intrinsic business risk over our simple sales risk. In
addition Mandelker and Rhee do not identify the specific assumptions made in order to convert market measures
included in the fundamental equity beta definition into accounting proxies. By developing our model from the
fundamental accounting equality we are able to identify at each stage the assumption which needs to be made to
arrive at our equivalent of their estimation model. This allows further theoretical refinement to identify the
specific adjustments required. The identification of suitable proxy measures for these adjustments will be the
subject of future empirical work.



9
REFERENCES

Beaver, W., P. Kettler and M. Scholes; 1970; The Association Between Market-Determined and Accounting-
Determined Risk Measures; The Accounting Review, October, pp. 654-682.

Beaver, W. and J. Manegold; 1975; The Association Between Market-Determined and Accounting-Determined
Measures of Systematic Risk: Some Further Evidence; Journal of Financial and Quantitative Analysis; June,
pp. 231-284.

Bildersee, J.; 1975; The Association Between a Market-Determined Measure of Risk and Alternative measures
of Risk; The Accounting Review; January, pp. 81-98.

Bowman, R.; 1980; The Importance of a Market-Value Measurement of Debt in Assessing Leverage; Journal
of Accounting Research; Spring, pp. 242-254.

Brealey, R. A. and S. C. Myers; 1996; Principles of Corporate Finance; McGraw Hill.

Breen, W.J. and E.M. Lerner; 1973; Corporate Financial Strategies and Market Measures of Risk and Return;
Journal of Finance; May, pp. 339-352.

Chance, D.M.; 1982; Evidence on a Simplified Model of Systematic Risk; Financial Management; Autumn,
pp. 53 - 63.

Hamada, R.; 1972; The Effects of the Firms Capital Structure on the Systematic Risk of Common Stocks;
Journal of Finance; Vol. XXVII: pp. 435-452.

Hill, N.C. and B.K. Stone; 1980; Accounting Betas, Systematic Operating Risk and Financial Leverage: A Risk
Composition Approach to the Determinants of Systematic Risk; Journal of Financial and Quantitative
Analysis; September, pp. 595-633.

Lev, B.; 1974; On the Association between Operating Leverage and Risk; Journal of Financial and
Quantitative Analysis; September: pp. 627 - 641

Lev, B. and S. Kunitzky; 1974; On the Association Between Smoothing Measures and the Risk of Common
Stock; Accounting Review; April, pp. 259-270.

Mandelker, G. and S. Rhee; 1984; The Impact of Degrees of Operating and Financial Leverage on Systematic
Risk of Common Stock; Journal of Financial and Quantitative Analysis; March, pp. 45-57.

Myers, S,.; 1977; Determinants of Corporate Borrowing; Journal of Financial Economics; November, pp.
147-175.

Pettit, P.R. and R. Westerfield; 1972; A Model of Capital Asset Risk; Journal of Financial and Quantitative
Analysis; March, pp. 1649-1668.

Rosenberg, B., and WW. McKibben; 1973; The Prediction of Systematic and Specific Risk in Common
Stocks; Journal of Financial and Quantitative Analysis; March, pp. 317-333.

Rubinstein, M.; 1973; A Mean-Variance Synthesis of Corporate Financial Theory; Journal of Finance; Vol.
XXVIII: pp. 167-181

Ryan, S.; 1997; A Survey of Research Relating Accounting Numbers to Systematic Equity Risk, with
Implications for Risk Disclosure Policy and Future Research; Accounting Horizons; Vol. 11, No. 2: pp. 82-95.

Scholes, M.; 1996; Global Financial Markets, Derivative Securities, and Systematic Risks; Journal of Risk and
Uncertainty; May, pp. 271-286.

Thompson, D.J.; 1974; Sources of Systematic Risk in Common Stocks; Journal of Business; April, pp. 173-
188.

10