Adjusting for optimism bias

A5.6 The aim of adjusting for optimism bias is to provide a more realistic assessment of the initial estimates of costs, benefits and time taken to implement a project. As the appraisal develops, more accurate costing of project or programme specific risk management should be undertaken.

Accordingly, adjustments for optimism bias may be reduced as more reliable estimates of specific risks are made. Any reductions should be presented transparently and tested with sensitivity analysis where appropriate.

A5.7 Optimism bias adjustments are applied to reflect the level of risk identification and management that has taken place. In options where a higher level of risk has been factored into the calculation of costs at the outset, a lower optimism bias adjustment is appropriate. Steps 1 to 4 below explain how optimism bias adjustments should be applied:

  Step 1 - decide which category of adjustment it is most appropriate to use. Categories of generic spending are as follows:

O  standard building projects involve construction of buildings that do not require special design, e.g. most accommodation projects including offices, living accommodation, general hospitals, prisons and airport terminal buildings.

O  non-standard building projects involve construction of buildings that do require special design e.g. specialist innovative buildings, unusual output specifications, building with space constraints or complicated site characteristics. Specialist/ innovative buildings could include specialist hospitals, innovative prisons, high technology facilities and other unique buildings or refurbishment projects.

O  standard civil engineering projects involve construction of standard facilities or infrastructure e.g. most new roads and some utility projects.

O  non-standard civil engineering projects involve construction of specialist facilities or infrastructure e.g. innovative rail, road, utility projects, or upgrade and extension projects.

O  equipment and development projects involve provision of equipment and/or development of software and systems e.g. manufactured equipment, Information and Communication Technology (ICT) development projects or cutting-edge technology projects.

O  outsourcing projects involve provision of hard and soft facilities or management services e.g. ICT services, facilities management or maintenance projects.

Table 7 below provides adjustment percentages for these generic categories that should be used in the absence of more robust, organisation and project specific evidence. It is based on a study by Mott MacDonald into the size and causes of cost and time over-runs in past projects.

A project which includes several category types should be considered as a 'programme' with a number of 'projects' for assessment purposes. Each 'project' should be separately rated for optimism bias and risk. Where alternative options being considered involve different spending types this could result in different assumptions regarding optimism bias. If the effect is significant it should be clearly set out and justified within the presentation of results.

Table 7. Generic Optimism Bias Adjustment Percentages

  Step 2 - consider if the optimism bias adjustment can be reduced. Reduce the upper bound adjustment to the extent risk has been identified and included in cost estimates. If appropriate consider the extent to which the remaining contributory factors are mitigated and apply a mitigating factor. The mitigation factor has a value between 0, which means that contributory factors are not mitigated at all, and 1, which means all contributory factors are fully mitigated. The value selected between 0 and 1 will be an evidence-based judgement of the extent to which risk has been mitigated at the outset, and needs to be justified. In practice this will mean reducing the optimism bias adjustment from the upper bound to the extent that risk has been costed.

  Step 3 - apply the optimism bias adjustment determined in steps 1 and 2. The present value of costs should be increased by the appropriate optimism bias factor. For example, if costs are -£10 million and the optimism bias factor is 40%, this would lead to optimism bias adjusted costs of -£14 million (i.e. -£10 million*(1+0.4)). If an appropriate optimism bias factor is available the present value of the benefits should also be adjusted (which would lead to a lower level of estimated benefits). The results should be added to estimate the total Net Present Social Value (NPSV) adjusted for optimism bias. This is the best estimate of the social value of an option, allowing for risk and optimism bias.

  Step 4 - review the optimism bias adjustment at different stages of appraisal. Optimism bias adjustment should be reduced in proportion to risk avoidance or risk mitigation measures taken. Objective and transparent evidence of the mitigation of contributory factors should be observed and verified independently before reductions are made. Procedures for this include the Gateway Review process. Further information can be found on the Infrastructure and Projects Authority's assurance review toolkit webpages.

A5.8 Closer to implementation the optimism bias adjustment for a project can be reduced to its lower bound provided mitigating evidence is robust. This assumes that the cost of mitigation is less than the cost of managing any residual risks. The costs of risk avoidance should be built into the proposal in their entirety since they will be incurred irrespective of whether the risks materialise. The costs of mitigation are included as expected costs, which is cost of mitigation multiplied by likelihood of the risk occurring.

A5.9 Optimism bias should be applied to operating costs and benefits, as well as capital costs. Where there is no appropriate measurement of typical bias, the confidence intervals of key input variables can be used.