Veritas propter investigationem [Truth through research]
TitleRetirement sorted? The adequacy and optimality of wealth among the near-retired (2014)
AuthorsRowena Crawford
 Cormac O’Dea
InstitutionInstitute for Fiscal Studies
TopicsRetirement income adequacy
 Saving for retirement
 Life cycle model
 Pension replacement rates
CountryUnited Kingdom
Date Published2014
Date posted on PR28 Apr 2015
Crawford, R O’Dea, C, (2014). Retirement sorted? The adequacy and optimality of wealth among the near-retired (2014) Institute for Fiscal Studies,

PensionReforms’ summary and comments

Commentators worry about whether their fellow citizens are saving enough for retirement.  Fewer people belong to formal retirement savings schemes; Defined Benefit schemes are disappearing; state pensions are scaling back despite longer life-expectancies and so we need to do something to counteract citizens’ seeming myopia.


Often the test of adequacy centres on so-called ‘retirement income replacement rates’ that usually compare compulsory pension benefits with a notional full-career earner.  For an example of these from the OECD see here.


“In this paper, we present the first assessment of the optimality of households’ saving, focusing on couple households born in the 1940s [and ‘observed’ in 2002-3].  We use a life-cycle model of consumption and saving to determine what households ought to have saved, given their life circumstances, if their objective is to maintain their living standards over time, and compare that optimal wealth with their observed wealth holdings.”


And by maintaining “living standards over time” the authors include after retirement and until death.


The report looks at wealth holding data from the English Longitudinal Study of Ageing (as did this 2006 report here) and the Wealth and Assets Survey.  ELSA data was linked to administrative data from National Insurance records.


The results were surprisingly comforting:

“We find that the vast majority of households (92%) hold more wealth than our model suggests is optimal, and that that would still be true for 75% of households were we to exclude housing wealth from observed wealth holdings.”


The report wondered why these households, overall, had more than seemed needed.  There may be a reason:  “…bequest intentions, precautionary saving against long-term care costs, households expecting their well-being from consumption to change in retirement” were some of the possibilities but PensionReforms suggests there are other possibilities, including that households may simply not know what might be ‘enough’ or ‘too much’ or ‘when to stop’.


The report compared its numbers with the more ‘traditional’ measures of replacement rates.


“We find that, once non-pension wealth is included, only 2% of households would be unable to replace 67% of their average lifetime earnings at age 65, only 5% would be unable to replace 80% and only 3% would be unable to achieve the replacement rate assumed to be adequate by the Pensions Commission benchmark.”


But those measures are, in the report’s view “a more demanding test of the adequacy of [individuals’] resources than a comparison with the optimal wealth suggested by our model.”


The report noted a significant range of results around what it regards as ‘optimal’:

“One-quarter of households are found to have an optimal replacement rate of less than 40%, while one-quarter have an optimal replacement rate of over 70%.”


The report therefore cautions against “simple replacement rate benchmarks…[as]

Such analysis could give a misleading picture of the preparedness of households for retirement, since it cannot capture the vast heterogeneity in households’ circumstances.”


PensionReforms thinks that the report offers a more ‘realistic’ view of individuals’ needs and objectives but suffers the problems encountered by all models to a greater or lesser extent.  There is so much to assume; so many decisions that have to be made on behalf of the survey subjects and behaviours to guess at between age 20 and death.  All pension assets are annuitised at inflation-linked ‘actuarially fair’ rates.  Even the savings options are simplified to allow for manageable calculations.  The model’s ‘optimal’ calculations are then compared with the data from ELSA (the longitudinal survey).


Where this report differs from most is that it takes a group of ‘real’ people who have already been working and saving for about 35 years at the date of observation.  That makes the calculations less subject to the impact of assumptions on the outcomes over very long periods.  (File size 933 KB; 52 pp) 721