Pinsent Masons have garnered a bit of press recently for their survey re the intentions of currently open final salary schemes to close or not. The accompanying headline was “Survey shows “death” of final salary schemes premature” and mentioned that out of 200 schemes surveyed none were planning to close. This is very much at odds with my experience of schemes, admittedly in the SME Sector. It did find that a large percentage of schemes were tinkering with accrual or contribution rates or flirting with CARE schemes, or rearranging the deckchairs on the Titanic as I think it should be known. None of those putative solutions really addresses the fundamental risks associated with Defined Benefit schemes. I have seen a number of companies who had tried such solutions eventually come to us to help them close their scheme as the tinkering hadn’t helped. Such solutions are clearly in the interests of consultancies with large numbers of actuaries and other professional staff to keep gainfully employed but are they in their clients interests? I find it really hard to believe that out of 200 (I presume randomly selected) businesses, none were considering closing their scheme. I re-read the press release very carefully. It said “A survey of 200 businesses found that no final salary schemes currently open to new members is planning to close” What’s not clear from the press release is how many of the 200 businesses actually had final salary schemes open to new members. If the answer is “none” then the finding starts to make sense. This highlights a number of flaws with surveys in general – but I can’t comment on the specific methodology of the Pinsent’s survey and wouldn’t presume to suggest that it suffers from any of the following common flaws. And if it did I’m sure they will have adjusted the results statistically (I always thought this sounded a bit like fudging them). I won’t dwell on the potential additional flaws this can introduce. Firstly surveys are rarely genuinely random and are often self selecting - perhaps someone who is struggling to fund their pension scheme as a consequence of running a struggling business doesn’t have time to fill in a survey. And who were on Pinsents mailing list? And who knew the name Pinsents and therefore didn’t immediately bin the survey? Was an incentive offered to respondents (by which I mean free access to the results, perhaps, rather than a Westminster style expense account or any other such questionable transaction.)? And would the incentive appeal to a particular group? Secondly, people lie in surveys all the time. In 1992, exit polls in Great Britain showed the Conservative party running roughly a point behind the Labour party, when Conservatives actually led by 7.6%. A report by the British Market Research Society attributed part of the error to a lower response rate by Conservative voters. British pollsters have named the phenomenon the "Shy Tory Factor," and some now routinely adjust their results to account for the skew. This basically states that the Tory party, rightly or wrongly, and certainly in 1992 was not a party that people wanted to be associated with, even in an anonymous poll, so they either lied about who they had voted for or declined to say. A similar economy with the truth is exposed in just about every survey ever published which asks men about their number of sexual partners. Thirdly, it often depends how you ask the questions. Contrast:- 1) Would you like an inflation proofed pension? 2) Would you like an inflation proofed pension if it means the pension you receive will initially be 40%-50% less than if you opted for no increases? Approximately. None of the above means that the Pinsents survey suffered from any of the above flaws and is rubbish. I’m sure it accurately reflects the answers the respondents (however they were selected, or self-selected) gave to the question they were posed in whatever particular way it was posed. And I am sure it provides great comfort to the likes of Mercer and Watsons, but if I was them, I wouldn’t go building my business model around it.