INTRODUCTION as a doctoral student at Pitt was the

INTRODUCTION

I
would like to thank the MAS for selecting me as the recipient of the 2018
Lifetime Contribution to Management Accounting Award and the AICPA and CIMA for
sponsoring it. I am honored and happy to receive this Award.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

THANKS

I
want to thank several people who have helped me, because without their help I
would not have made the contributions to management accounting that I have made.
Let me begin by thanking my wife, Janet, and children, Jesse and Lilly. They
have provided me with lots of love and support that have kept me motivated.

I
also thank my 37 coauthors for helping me learn about management accounting and
research. I especially want to thank five who have been wonderful friends as we
learned about management accounting and how to improve our research.

My
time as a doctoral student at Pitt was the best of times! I chose Pitt because
Jake Birnberg was one of the few experts in what I wanted to study—the relation
between management accounting and psychology. We began and continued to work on
projects related to psychology theory and experiments. Jake was a fabulous
teacher, mentor, and friend.

My
research expanded by working with two other coauthors in the 1980s and 1990s. Mark
Young and I began with an experimental project based on judgment heuristics and
biases and then a survey study motivated by agency theory. We next moved to the
field to do interview and survey studies of emerging management accounting practices.
I worked with Chee Chow doing experimental studies based on agency theory, and
surveys using cross-cultural psychology theories.

In
the 1990s I began to work with Ranjani Krishnan and Joan Luft. We did
experimental research that was motivated by economics and psychology theories.
Joan and I also worked on studies that developed models of causes and effects
of management accounting and strategies for improving management accounting
research.

Finally,
I want to thank Chris Chapman and Anthony Hopwood for our editing of Accounting, Organizations and Society and
the three volumes of the Handbook of Management
Accounting Research.

CHANGES IN POSITIVIST MANAGEMENT
ACCOUNTING RESEARCH

DURING THE LAST FOUR DECADES

 

The
nature of this Award almost invites me to reflect on how positivist management
accounting research has changed during my lifetime. I received my PhD in 1978,
forty years ago. My reflection will focus on positivist research—research that
aims to provide general causal explanations of a management accounting practice
that are common to many instances of it and validated by quantitative
hypothesis testing—hereafter, management accounting
research (Luft and Shields 2014). I will describe this research as it
existed 40 years ago and how it has and has not changed and note implications
for research.

I will draw heavily on three articles as a primary basis
for my description. First, Luft and Shields (2003) develop maps of causal
models of published theory-consistent management accounting research from its
beginning in 1967 through 2001 in which management accounting is the cause
and/or effect. They also offer guidelines for constructing complete and valid
causal models. Second, Herschung, Mahlendorf, and Weber (forthcoming) update
those maps from 2002 through 2012 and make observations regarding how well research
has followed the guidelines in Luft and Shields (2003). Finally, Lachmann,
Trapp, and Trapp’s (2017) analysis of management
accounting research published from 1980 through 2012 provides evidence on the
diversity of topics researched, theories and research methods, and the validity
of that research.

Four
Decades Ago

When
I was a doctoral student in the 1970s, management accounting research was in
its infancy and the management accounting scholarly community was learning how
to do research. For example, before I started my doctoral program, there were
only nine management accounting research articles published in English-language
accounting journals (Luft and Shields 2003). Personal computers and Wi-Fi did
not exist, instead there were typewriters and paper revisions were
cut-and-paste with scissors and scotch tape. Data were analyzed with punch
cards processed at central computing facilities. Libraries were places where
people regularly went to search the stacks for books and journals and read at
tables. There were three English-language accounting journals—Abacus, Journal of Accounting Research, and The Accounting Review—with Accounting,
Organizations and Society and Journal
of Accounting & Economics starting up while I was a doctoral student.
There were only a few management accounting researchers: When I graduated in
1978 I recall there were only two other people in the management accounting job
market in the U.S.—Dave Larcker and Ken Merchant.

With
hindsight, in 1978 two indicators of drivers of future major changes in
management accounting research appeared. These changes stem from assumptions made
about human behavior (e.g., optimize vs. satisfice, cooperate vs. compete,
value monetary vs. monetary and nonmonetary payoffs) in the theories used in
management accounting research (Luft and Shields 2010; Luft 2016). One
indicator was the publication in the accounting literature of the first agency
theory article related to management accounting, in which Demski and Feltham (1978)
developed analytical models of the economic demand for budgetary control
systems. Agency theory assumes individuals are self-interested, optimizing, and
only value monetary payoffs. The other indicator in 1978 was Herb Simon winning
the Nobel prize in economics for his
pioneering research into decision-making processes within economic
organizations. He assumes that individuals are boundedly rational, satisfice,
and value monetary as well as nonmonetary payoffs: these assumptions are key foundations
of behavioral economics. Agency theory and behavioral economics,
separately and jointly, have had big effects on management accounting research
during the last four decades (Luft and Shields 2010).         

The
few management accounting articles published forty years ago tended to have
similar characteristics. Almost all articles addressed one management
accounting practice, often budgeting or organizational control (Herschung et
al. forthcoming; Luft and Shields 2003; Van der Stede 2015). They frequently
had short literature reviews because there was not much management accounting literature
to review: There were only 24 management accounting research articles published
in English-language accounting journals from 1967 through 1978, which is an
average of 2.0 per year (Luft and Shields 2003).

Causal
models usually were simple, often with few variables and linear additive forms
(Luft and Shields 2003). Interaction models were just beginning to appear: they
were used to try to account for conflicting results across studies and for conflicts
between predictions and results within studies, by investigating if or how these
conflicts were conditional on context or another management accounting practice.
Linearity was predicted in almost all studies. Almost all models were
unidirectional, focusing on causes or effects of management accounting at a
single level of analysis, most often at the individual or organization levels. Small
convenience cross-sectional samples were common, research designs and
operationalized constructs often created threats to validity, and statistical
tests of hypotheses frequently were bivariate.

Four
Decades Later

The
topical focus of published research has changed during the last forty years, from
planning (budgeting) to control (performance measurement, evaluation and incentives)
(Herschung et al. forthcoming; Ittner 2017; Luft and Shields 2003; Van der
Stede 2015). In contrast to this research trend toward control, practitioners
have different priorities: The Financial Executives Research
Foundation/Protiviti 2016 Financial Priorities Survey indicates that financial
executives’ top three priorities are forecasting, planning, and budgeting, and an
AICPA survey finds that Chartered
Global Management Accountants’ two top research needs are (1) business
planning, budgeting, and forecasting, and (2) strategic planning (Ittner 2017).

Many
articles continue to address one (or a very limited) number of management
accounting practices (Herschung et al. forthcoming), which assumes they are not
complementary with other practices (Grabner and Moers 2013). The number of
articles published per year has increased and so has the number of link-study
pairs (a link-study pair is the combination of a link in a causal model and the
associated study that provides evidence on the link) (Herschung et al. forthcoming).
The number of different variables used in management accounting research has increased
and so has the percentage of those variables used in only one study (Herschung
et al. forthcoming), making studies more unique and thus less directly
comparable.  

Causal-model
forms have changed in some ways (Herschung et al. forthcoming). They are less
likely to be additive only, more likely to include intervening variables, while
the inclusion of interactions has remained at a constant level. The increased
use of intervening-variable models is good because they can provide evidence on
the validity of assumed causal mechanisms. Causal-model forms still are almost
always linear even though social science theories often predict nonlinear
effects. They are also almost always unidirectional, which ignores causal
forces in the opposite direction. For example, in research on the management
accounting-performance link, almost all studies provide evidence on how management
accounting influences performance but almost no studies provide evidence on how
performance affects management accounting.

Almost
all studies continue to be at a single level of analysis, with the portion of
studies at the individual and sub-unit levels decreasing and increasing at the organization
and beyond-organization levels (Herschung et al. forthcoming). There continues
to be a lack of cross-level models, which limits explanations, for example, of
how organization-level management accounting influences individuals and how individuals influence that management
accounting. Of the relatively few cross-level models, most are invalid because
they do not have an interaction at the level of the dependent variable (Luft
and Shields 2003).

The portion of studies that are replicated
has always been low and, unfortunately, has decreased (Herschung et al.
forthcoming). The decrease could be due to researchers’ belief that
replications are not rewarded, are associated with low status, and/or the
journal review process rejects replications (Shields 2015). In addition, there
may be fewer incidental or partial replications (e.g., a link that was examined
as the primary hypothesis in one study becomes a control-variable relation in a
later study), because of the decreasing re-use of variables across studies and
the increase in variables that appear only in a single study (Herschung et al.,
forthcoming). Although the proliferation of variables can result from valuable
trends in research, such as an increase in the variety of research questions
asked and theories employed, it also has the negative side effect of reducing
comparability across studies. We should aim to have studies that are
reproducible and replicable to increase
confidence that the results of studies are not artifacts due to validity
threats. Good studies are likely to be reproducible and replicable because they
have effectively addressed threats to their validity (Shields 2015).

There
is evidence that some threats to validity have decreased, but also evidence
that other threats have not (Lachmann et al. 2017). Construct validity has
increased as indicated by a larger portion of studies having more measures of
constructs, more objective and less subjective measures, and multiple compared
to single data sources. In a related development, a higher portion of studies
report statistical tests of convergent and discriminant validities.
Statistical-conclusion validity is also higher as indicated by the increased
percentage of archival and survey studies that address multicollinearity, correlated
omitted variables bias, simultaneity bias, self-selection bias,
heteroscedasticity, or outliers. Most studies, however, do not address these
issues or control for endogeneity, for example, by using two-stage
least-squares estimations and propensity scoring analyses. Little evidence
exists that external and internal validities have increased (Lachmann et al.
2017).

Developing
complete and valid models of management accounting requires using diverse
theoretical perspectives and theories as well as diverse research methods. Every
research method (e.g., archival, experimental, survey) has strengths and
weaknesses in reducing threats to validity. Thus, it is good when diverse
research methods investigate the same causal relation because they have
complementary strengths and weaknesses (Birnberg, Shields, and Young 1990; Luft
2016). For management accounting research, research method diversity is at a
high and increasing level (Lachmann et al. 2017).

Diverse
theoretical perspectives are necessary to develop complete models of management
accounting. Different theoretical perspectives and theories make different
choices about which issues to study, and consequently they make different
choices about which issues to simplify away, for example by using maintained
assumptions that eliminate, hold constant, or substitute simpler versions of
non-core issues (Luft and Shields 2014). For management accounting research, diversity
in the use of theoretical perspectives has been at a high and constant level
since 1990 (Lachmann et al. 2017). The most commonly used theoretical
perspectives are economics, organizational behavior, psychology, and sociology,
and they are used with similar levels of diversity in research on planning and on
control.  

While
theoretical diversity is at a high level, it is almost always between studies. We would develop more
complete models of management accounting if more diverse theoretical
perspectives were used within studies.
Theory integration, however, must be done carefully to develop valid models.
Covaleski, Evans, Luft, and Shields (2003) identify four interrelated criteria
and illustrate how to apply them within
studies to selectively integrate theories from economics, psychology, and
sociology.

Besides
integrating diverse theories, research would have higher validity and develop
more complete causal models of management accounting if it followed the
guidelines in Luft and Shields (2003). Herschung et al.’s (forthcoming) analysis
of management accounting research identifies guidelines that could be adhered
to better. These include:

1.    
disentangling the multiple meanings of
practice-defined variables,

2.    
using theories that predict nonlinear
effects,

3.    
indicating why only additive effects and
not interaction or intervening-variable effects are predicted,

4.    
stating whether interactions involve only
independent or also moderator variables,

5.    
including bidirectional relations,

6.    
identifying the causal interval to
indicate the expected time required for a cause to have an observable effect,

7.    
aligning the level of theory (what is
being explained), level of variable measurement (source of evidence), and level
of data analysis (unit of data),

8.    
separating the effects of independent
variables at different levels of analysis, and

9.    
having cross-level models with an
interaction at the level of the dependent variable. 

FOUR DECADES INTO THE FUTURE

There
have been many improvements to management accounting research during the last
40 years, which has increased its validity, the completeness and validity of
causal models, and established management accounting knowledge. Yet there are many
opportunities to further improve research. These include:

1.    
increasing research on topics other than
control such as budgeting and planning,

2.    
including complementary management
accounting practices within studies, 

3.    
adhering to the guidelines (e.g.,
bidirectional, nonlinear, cross-level models),

4.    
reducing threats to construct, external, internal,
and statistical-conclusion validities,

5.    
increasing the production of comparable studies
to facilitate providing evidence on the reproducibility and replicability of
research, and

6.    
increasing theory diversity within studies.

I
look forward to attending this conference in 2058, 40 years into the future. At
that conference I will learn how the validity of management accounting research
will have improved. I will also learn how this research will have developed
more complete and valid causal models and established management accounting
knowledge.