This paper examines causal reasoning, applying the theories to financial reporting. Causal reasoning involves diagnosis (determining the cause of an effect) and prediction (vice versa). These are important and commonplace amongst analysts, investors and management regarding company earnings and share prices. However, there is very little recent research employing causal reasoning theories to this field. Attribution theory describes how causes are attributed to past events.
This may be dispositional (attributed to a person) or situational. In a single instance of an outcome, correspondent inference theory suggests diagnosis is based on choice, expectations and intent. Expected, freely chosen behaviour is considered to be dispositional, whereas unexpected, forced behaviour is attributed to situation. Attributing intent is simplest when considering behaviour with only one positive effect.
When an outcome occurs numerous times, covariation theory describes possible cause attribution. This involves looking at cases where the suspected cause is present/absent and matching them to instances where the outcome is present/absent, utilising information on consensus (how others behave), distinctiveness (how the individual behaves in different circumstances) and consistency (how the individual has behaved in similar circumstances).
Consistency should be high to make a good attribution judgement. When consensus and distinctiveness are also high, the attribution is likely to be situational whereas if consensus and distinctiveness are low, a dispositional attribution is likely. Counterfactual reasoning is identifying a cause by considering whether alternate events would have led to the same outcome. It is commonly used by a person who has been directly involved when undesired, abnormal but avoidable events have happened.
Studies show that using weighted statistical averages of causal factors produces a more accurate prediction than human judgement; however it is not widely used in financial reporting. Covariation, ordering, similarity and contiguity are all logical cues-to-causality, allowing the strength of a potential connection between cause and effect to be judged: •Covariation – the presence of the cause with the effect and absence otherwise •Ordering – the cause must precede the effect •Similarity – the cause and effect should be similar in size •Contiguity – the effect should occur soon after the cause
Heuristics speed up causal reasoning at the expense of accuracy: •Representativeness – judging a situation’s outcome based on its similarity to others •Availability – making a judgement based on the information which is most accessible •Confirmation/desirability bias – focusing on information which confirms a previously held belief or desired outcome •Ignoring regression to the mean – ignoring that abnormally high or low performance tends to be followed by more average performance
The construal level theory suggests that more distant predictions will be less accurate, yet made with greater confidence, due to the absence of context such as feasibility of outcomes. Prior diagnosis is often used in a prediction of future results whilst previous predictions will be used to diagnose observed effects. Future research on causal reasoning in financial reporting is necessary to study the resulting positive and negative carryover effects. Such research would provide significant insight to individuals making judgements on financial reports.