Todays theme: Accounting experiments

1
Todays theme Accounting experiments

• Key characteristic Manipulates some independent
  variable(s) (or factor(s)) under ceteris paribus
  conditions to compare the resulting dependent
  variable(s) (usually judgments, decisions,
  actions, or negotiations).
• Query Is a survey a type of experiment?

2
Todays theme Accounting experiments

• Key strength If experimental participants are
  randomly assigned to treatment conditions, any
  observed difference in judgments, decisions, or
  actions is likely caused by the manipulated
  factor(s).
• Key limitation Experimental settings usually
  abstract away from other real-world factors that
  can also influence judgments and decisions.
• Analogy Chemistry experiments

3
Overview Maines, Salamon, and Sprinkle (BRIA,
2006)

• The goal of accounting research is to create
  legitimate, consequential belief revision about
  accounting-related decision settings (p. 85).

4
Overview Maines, Salamon, and Sprinkle (BRIA,
2006)

• The goal of accounting research is to create
  legitimate, consequential belief revision about
  accounting-related decision settings (p. 85).

5
Overview Maines, Salamon, and Sprinkle (BRIA,
2006)

• Legitimate intellectually honest
• By intellectual honesty, we mean the need to
  reveal findings even if they appear inconsistent
  with either the authors primary articulated
  views on the research issue or the current body
  of research results (p. 89).
• Authors example Ball and Brown (1968) and
  post-earnings announcement drift.

6
Overview Maines, Salamon, and Sprinkle (BRIA,
2006)

• Consequential The research question must
  matter to a decision maker (p. 89).
• Bad motivation No one has investigated the
  effect of X on Y.
• Good motivation A better understanding of the
  effect of X on Y is important to ltfirms,
  investors, regulators, auditors, etc.gt because of
  ltreasongt.

7
Overview Maines, Salamon, and Sprinkle (BRIA,
2006)

• Advantages of experiments for consequential
  research
• Timing (ex ante policy implications).
• Inclusiveness (can examine alternatives even if
  they do not occur in natural settings with
  sufficient frequency)
• Comparing levels of behavior, such as
• Judgments vs. decisions
• Individual judgment vs. interactive negotiations
  / laboratory markets
• Experiments can examine decision processes and
  the intervening links between independent and
  dependent variables.

8
Overview Maines, Salamon, and Sprinkle (BRIA,
2006)

• Belief revision To achieve significant belief
  revision, an experiment needs to create ex
  ante tension in the research hypothesis (pp.
  92-93).
• Useful questions to ask of any study
• What might one do differently based on this
  research?
• Why might the predicted results not occur?

9
Overview Maines, Salamon, and Sprinkle (BRIA,
2006)

• Advantages of experiments for revising beliefs
• Contrasting theoretical perspectives and
  predictions (e.g., economics vs. psychology).
• Query Is this potential unique to experiments?
• Internal validity and causal inferences
• With an experiment, one obtains a greater level
  of assurance that if X varies, all else equal, Y
  will result.
• Construct validity, theoretical precision, and
  intentional experimental artificiality
• Example Separating the influences of
  disclosure format vs. strategic inferences in
  Hobson and Kachelmeier (TAR, July 2005).

10
External validity limitations

• Key question Does ltinsert your realism
  objection heregt interact with the observed
  influence of X on Y?
• Important note A systematic difference does not
  necessarily imply an interaction.
• Example Using students instead of experienced
  professionals as experimental participants.

11
Demand effects

• Key question Are participants responding to the
  treatments or to the researcher?
• Example Heavy-handed instructions in an
  experimental study on the willingness to share
  ABC information.

12
Sprinkle (AOS, 2002) Managerial accounting
experiments

• It is useful to distinguish two roles for
  managerial accounting (from Demski and Feltham)
• Decision-influencing role (controlling employees)
• Decision-facilitating role (improving decisions)

13
Where do surveys fit in?

• In defending controlled laboratory experiments,
  Sprinkle observes that archival-empirical and
  field-based approaches may be contaminated
  because their effects cannot be disentangled from
  other effects, including self-selection biases
  and sample-selection biases (p. 288).
• Are surveys also susceptible to these limitations?

14
Complementarities between experiments and
analytic models (p. 289)

• Economic models (such as agency theory) provide a
  useful, simplified structure for understanding
  the uses and consequences of management
  accounting information.
• Experiments can test a models predictions and
  pit those predictions against competing
  psychological perspectives.
• Example Kachelmeier, Reichert and Williamsons
  (2007) study of the effects of performance-based
  compensation for both quantity and creativity.

15
Experimental tests of the decision-influencing
(i.e., control) effects of managerial accounting

• Mitigating hidden information (adverse
  selection)
• Budgetary participation vs. slack (pp. 291-292)
• Using compensation contracts to screen potential
  employees and infer private information (pp.
  292-293)

16
Experimental tests of the decision-influencing
(i.e., control) effects of managerial accounting

• Mitigating hidden action (moral hazard)
• Using compensation contracts to motivate effort
  and productivity (p. 293).
• Perhaps somewhat surprisingly, only about half of
  the experiments in this area find a significantly
  positive effect of financial incentives on
  performance (see Bonner et al., JMAR 2000).
• Tournament schemes in particular often produce
  undesirable results.

17
Experimental tests of the decision-influencing
(i.e., control) effects of managerial accounting

• Mitigating hidden action (moral hazard),
  continued
• Framing effects (e.g., Lufts JAE 1994 study of
  the different motivational effects of contingent
  bonuses vs. contingent penalties).
• Relative performance schemes (which are not
  necessarily tournaments) to filter out
  uncontrollable state of nature effects.

18
Experimental tests of the decision-influencing
(i.e., control) effects of managerial accounting

• Current and future directions
• Social motives, values, peer pressure, gift
  exchange, trust, and perceptions of fairness
  (pp. 294-298).
• Examples Kachelmeier and Towry (TAR July 2002)
  Hannan (TAR January 2005) Coletti, Sedatole, and
  Towry (TAR April 2005) among others.
• Performance schemes for multi-dimensional tasks
  (pp. 298-299)
• Example Quantity vs. creativity?

19
Experimental tests of the decision-influencing
(i.e., control) effects of managerial accounting

• Current and future directions, continued
• Motivating the willingness to take risks (p.
  299).
• Using group and team incentive schemes without
  inducing free-riding behavior (p. 300).
• Example Towry (TAR, October 2003)
• Objective measures vs. subjective performance
  evaluation (p. 301)
• Example Fisher, Maines, Peffer and Sprinkle
  (TAR, April 2005)

20
Experimental tests of the decision-facilitating
(i.e., improving decisions) effects of managerial
accounting

• Judgment biases relative to normative models
  (e.g., Bayes Rule) (pp. 302-303).
• Usefulness of management accounting information
  and information overload (p. 303).
• Presentation format effects (e.g., making a
  balanced scorecard system easier to understand).
• Exploring the effects of ABC and variable vs.
  fixed product costing systems on pricing
  decisions (e.g., Cardinaels, Roodhooft, and
  Warlop, JMAR 2004).

21
Experimental tests of the decision-facilitating
(i.e., improving decisions) effects of managerial
accounting

• Current and future research directions
• Ex post use of financial and non-financial
  performance measures to improve evaluators
  decisions (pp. 304-306).
• Example Factors that affect the usefulness of a
  balanced scorecard system (e.g., Libby, Salterio,
  and Webb, TAR, October 2004).
• Optimal distribution and use of information in
  multiperson and multiperiod settings (pp.
  306-307)
• Is it useful or harmful to share information
  openly within the organization?
• Expertise issues, building off auditing research
  progress in this area (pp. 307-308).

22
Integrating the decision-influencing and
decision-facilitating roles of managerial
accounting (pp. 308-310)

• Example (p. 309)
• Participative budgets can serve both to control
  production (decision influencing) and to plan the
  most effective allocation of resources (decision
  influencing).
• Thus, divisions that build slack into their
  production budgets run the risk of fewer
  allocated resources in future periods (Fisher,
  Maines, Peffer, and Sprinkle, TAR, October 2002).

23
Next week

• Finish accounting experimentation (Kachelmeier
  and King article)
• Opportunity to discuss your research
• A broader discussion of academic research in
  accounting (with thanks to Colleen Hayes)
• What exactly is the process by which editors
  evaluate submissions to top-tier academic
  accounting journals?
• What standards are in place to ensure
  objectivity?
• Has the accounting research agenda been captured
  by those interested primarily in the economic
  effects of financial accounting in capital
  markets?
• Should intellectual authority in our discipline
  be shared equally among all qualified academics?
• What exactly is our role and responsibility as
  accounting academics?
• How are things changing over time, and where is
  the academic accounting community headed?