Like many others, I believe that equities will out-perform bonds over the long term (NOT necessarily over short periods). For defined benefit pension schemes, the trustees should feel able to invest in equities (or property) if, and only if, they are confident that they have enough time before they have to secure a major tranche of the liabilities. This crucial aspect really should be discussed with the sponsor as part of agreeing the Statement of Investment Principles.
As I understood it, financial economists normally claim that there is no such thing as an equity risk premium over bonds. At its simplest, any additional return is fully counterbalanced by the risk. However, during an Institute of Actuaries seminar held on 18 May 2000, a “financial economist actuary” agreed that an allowance for an extra return can sometimes reasonably be made. When it comes to the long term, financial economics needs to be thought of as flawed economics.
So how much are we talking about? Does it vary over period measured? Finally, does DVR indicate anything radically different from MVR? Over all periods of 15 years falling within 1968 and 2018, measured by actual MVRs, the mean UK [US] risk premium was 2.6% pa [3.2% pa]. Assessed in relation to theoretical DVRs, the corresponding result for the UK [US] risk premium was 4.1% pa [3.6% pa].
Each chart (smaller or larger) has a slider for period length, a country radio selector, a portfolio radio selector and a statistic radio selector.
Depending upon the purpose of the assessment for which a risk premium is needed, one may want to build in a margin of prudence. However, one can only do that with some idea of the best estimate. A different approach for estimating long-term return is suggested here.