How do you interpret variances?

With extreme caution and after a thorough investigation. Variances are not evidences, but ambiguous cues (even third-level variances) and you cannot make right away a judgement about their causes, who might be responsible, and even whether they are a good or a bad signal for the organization.

To understand this, imagine the following scenario: to make smoothies, purchasers bought fruits which were very cheap compared to the budget (favorable material price variance) but of varying quality. As a consequence, the production department had to discard and waste a lot of them (unfavorable usage variance) and the sales people had to reduce prices because of the unreliable quality (unfavorable price variance).

This example illustrates some very important ideas. First, a favorable variance (here the material price) is not necessarily a good signal; it can be evidence of some other performance dimension being sacrificed, or even of slack introduced in the budget (budgeted material price was too high). Second, the cause of a variance and the person responsible are not necessarily where the variance is observed: even if each variance is more strongly associated with one particular manager, it is usually a shared responsibility. In our example, the cause and responsibility for the material usage variance and selling price variance are at least in part in the purchasing department. Third, it is misleading to interpret a variance in isolation, as it can result from, or have the same underlying cause as, another variance.

This leads to a few recommendations about how to interpret variances. First, detail them as much as possible to avoid mutually cancelling variances to hide one another. Second, look at the overall pattern of variances and look for plausible explanations, i.e. narratives which are consistent with what you observe. This will usually generate several hypotheses about what might cause the variances:

  • Were budgeted standards and targets set too low (or too high)?
  • Were the wrong (right) persons selected for a task?
  • Were employees more (or less) motivated?
  • Were employees well (or not well) trained;
  • Had employees better (or worse) working conditions, tools, or materials to work with?
  • Was the environment more (or less) favorable than expected?
  • Did employees sacrifice (worked on) performance dimensions not captured by the budget?
  • Were the decisions made conform to the strategy?
  • etc.

Once you have a set of plausible hypotheses, investigate, collect information which might support or falsify your hypotheses until only the best possible explanations remain.

Variance analysis is subject to the same cost benefit trade-off as all other phases of management accounting. Investigations are costly and therefore only large variances justify them (management by exception). Some random fluctuations are to be expected, if only because of the reliance on imperfect estimates. Large variances however suggest that something significantly different from what was planned happened or that some systematic process is at work.

Variances signal that expectations were not met. More detailed and fine-grained decompositions of variances also suggest where to look for reasons explaining why expectations were not met. But variances are estimations, not explanations. Once you have identified the major causes of variations in profit, you still need to investigate. You just know where and have a better idea of what to look for. Variances orient questions, they do not provide answers. So once you have identified the major sources of variances between actual and budgeted profit, it is important to ask managers what happened (the variance analysis suggests the questions to ask), and only suggest corrective actions based on their explanations. It is also important to check in the next cycle whether the corrective actions were effective. If they were not it might suggest that the assumed sources of variance were not the actual causes. In that sense, variance analysis fosters learning.

Please indicate how clear and understandable this page was for you: