Variance investigation
Variance investigation is the decision process for determining whether a reported variance is significant enough to justify the cost of identifying its cause; frameworks include a simple materiality threshold, statistical control charts.
FrameworkManagement by exception
See it move
Variance investigation weighs two dimensions: size and pattern. A small but systematic variance is monitored; a large, systematic one is investigated without much debate. A small random variance is ordinary noise and is ignored, while a large random variance calls for a cost-benefit comparison between the expected saving from correction and the cost of investigating it.
The formula
Variables
- estimated probability that the process is genuinely out of control and the variance reflects a correctable problem (decimal)
- present value of the total loss that will accumulate if the out-of-control process is left uncorrected (€)
- cost of corrective action once the root cause is identified (€)
- cost of the investigation itself (€)
The left side is the expected net saving from correction; the right side is the certain cost of investigating. When the expected saving exceeds the investigation cost, investigation is worthwhile. In practice P is difficult to estimate, so organisations use materiality thresholds or statistical control charts as proxies.
Check yourself
A management accountant receives a variance report showing that the direct materials price variance for March is €2,100 adverse — equal to 2.3% of standard materials cost for the period. The company's policy states: 'Investigate any variance that exceeds 5% of standard cost or falls outside the statistical control limits calculated from twelve months of historical data.' No investigation is initiated. Which principle best explains the decision not to investigate?