THE IMPACT OF PRICE‐LEVEL ADJUSTMENT ON THE MEANING OF ACCOUNTING RATIOS

• DOI: http://doi.org/10.1111/j.1468-5957.1980.tb00207.x
• Author: Daniel G. Short
• Date: 01 September 1980
• Published date: 01 September 1980

THE

IMPACT

OF

PRICE-LEVEL ADJUSTMENT ON THE MEAN-

ING

OF

ACCOUNTING RATIOS

Daniel

G.

Short*

IlltroduetiOn

As

alternatives to historical cost accounting are being considered, it is import-

ant to recognize that inflation adjustments may alter the underlying meaning of

accounting ratios. Accounting ratios are often defined in terms of a particular

financial characteristic of a firm.

In

addition, ratios are often grouped on the

basis of measuring the same financial characteristics. The alternatives to histor-

ical cost accounting may alter the interrelationships among common accounting

ratios, resulting in a change in the meaning or grouping of the ratios.

The process by which inflation adjustments affect the relationship between

financial ratios may be seen

in

the case of the quick ratio and the current ratio.

These ratios are both considered measures of liquidity(a financial characteristic).

The quick ratio is relatively unaffected by price-level adjustment because both

the numerator and denominator of the ratio contain almost exclusively

monetary items, The current ratio includes inventory (which in nonmonetary)

and

will

show a greater change

as

the result

of

price-level adjustment. This

change may be sufficient to alter the relationship between the current and

quick ratio

so

that they will not behave as if they both measure the same financ-

ial characteristic

of

liquidity. The purpose of this study is to examine the impact

of price-level adjustment

on

the relationship between accounting ratios.

Testing the impact of price-level adjustment

on

the meaning of accounting

ratios may permit one to come to a better understanding of prior research

on

the relevancy of price-level data to decision making. Behavioral studies

(Dyckman,

1969;

Heintz,

1973)

indicate that price-level adjustments have

little influence

on

investment decisions. These studies are difficult to evaluate

because

no

evidence is provided

that

participants understood how to use price-

level data. If it can be shown that the meaning

of

ratios is altered by price-level

adjustment it is reasonable to question whether participants in prior behavioral

studies understood the basic meaning

of

price-level data.

A finding of a change in the measurement characteristics of accounting

ratios would have implications for future studies which attempt to evaluate the

relevancy of price-level data. For example, consider

a

replication of the Beaver,

Kettler, Scholes study

(1970)

of the relationship between market and account-

ing measures of risk, The original study examined several financial charactedst-

ics (such as liquidity and leverage). If this study were to be replicated in order

to compare the predictive ability

of

historical and price-level ratios, it would

'The author

is

Assistant

Rofessor

at the Graduate School

of

Business, Crniver-

sity

of

Texas

at Austin. (Paper received JuQ

I9

79)

Journal

of

Business Finance

&

Accounting 7,3(1980) 377

be important

to

demonstrate that each ratio which was compared did in fact

have a comparable meaning. The present study presents

a

methodology to

examine potential changes in the relationships among ratios produced by

different accounting procedures. The paper consists of three sections.

A

discussion of factor analysis and the procedures used to estimate price-level

data are presented in the' methodology section. The empirical findings are

presented

in

the second section and a summary of the study

is

contained in

the final section.

Methodology

The primary research question examined in this study is whether price-level

adjustment will affect the relationships between accounting ratios. The relation-

ships can be measured with correlation coefficients but it is obviously difficult

to discern a pattern of relationships

in

an extremely large correlation matrix,

Factor analysis is a useful technique to overcome this problem,

Factor analysis is used here to sort a set of data patterns (i.e. correlation

coefficients) into subsets, so that each subset contains data that are as similar

as

possible. These subsets are known as factors. Factors may be interpreted to

show the underlying relationship of the items

in

each subset and inferences

may

be

drawn about the conceptual dimensions that underlie the specific items.

Stated more concretely, factor analysis can be used in an attempt to uncover the

characteristics measured by accounting statements. The subsets produced by

factor analysis are expected to be similar to the groups formed when ratios are

classified according to the financial characteristics that they purportedly

measure.

AnaIytical Considerations

In general, it is possible to extract a number of factors equal to the number

of variables under study.

In

such a case, factor analysis would offer little insight

into the relationship between the variables. The analyst must decide at what

point additional factors account for a trivial amount of variance. For this study,

the Kaiser criterion was employed. This criterion includes

all

factors with an

eigenvalue greater than one. The practical meaning of this criterion is that a

factor must explain more variance than that produced by a single standardized

variable.

Factor analysis is sensitive

to

data which are distinctly not normal, but

Harman

(1967)

suggests that considerable latitude might be allowed in including

variables that deviate from a normal distribution. Histograms for the ratios

included in the study were examined and many right skews were found. Natural

log and square root transformations are appropriate normalizing transformations

when

right

skews are present.

In

order to test the robustness of the factor model,

sensitivity analysis was performed by making normalizing data transformations

and comparing solutions based on transformed and untransformed data. The

solutions were found to be essentially the same. The minor differences which

378

Daniel

G.

Short