This article originally appeared in PROJECT M, Allianz’s award-winning magazine for the investment, retirement and insurance markets.

70–80% is a rule of thumb used by financial planners to set an income replacement rate for individuals to maintain the same standard of living in retirement as they had while working – but has it met its use-by date?

Michela Coppola: Welcome to you both. Marike, ladies first, the 70% to 80% rule – is it still valid?

Marike Knoef: On average it is a good approximation. However, there is a lot of heterogeneity. Some people have paid off their mortgage, others have not. Some people have children and spend relatively more money during their working life, others don’t. Some people have a low income and will need a 100% replacement rate to be able to fulfill their basic needs, while others have a high income and can do with less than 70%.

James Moore: It’s a heuristic and if you are a retirement planner dealing with laypeople without knowledge of the life-cycle model, you want something they can anchor on. The 80% rule allows you to make certain behavioral assumptions. First, when you go from pre- to post-retirement, you no longer save for retirement, so the replacement ratio can be reduced by that amount. So, if people were saving 10% pre-retirement, that brings you down to 90%. Second, if you pay down a home mortgage during your working years, you prefund a portion of your shelter consumption. That is 20%–25% of expenditure you can reduce. Third, differential tax treatment of retirement earnings as opposed to pre-retirement earnings plays a particular role in the U.S.. So it is pretty straightforward to understand the rationale why 70% to 80% is a good indicator.

Coppola: Recently some economists have concluded the income replacement rate is of little use in the modern world. While it may have been suitable when retirement income came largely from social security and DB schemes, it is ill suited for a time of DC and where people use other assets, such as savings or housing, to finance retirement. Is this too harsh?

Moore: Its weakness concerns how consumption bundles change during the life cycle. Consumption needs change towards and into retirement. Early in retirement, retirees no longer have work-related expenses; time is substituted for expenditure, so needs are fewer. But the further and further into retirement you go, increasing demand for medical care can raise that fraction of the mix substantially. So, the interesting question is whether 80% makes sense. There needs to be more research done on what the consumption bundle looks like 15 to 25 years into retirement.

Knoef: Well, the strength of the rule is its simplicity, but it is also a weakness. The world is full of very different people and households, so no one rule covers them all. In this sense, income replacement rate is of little use. In the Netherlands we find that the self-employed have very low income replacement rates, but relatively high savings and housing wealth. It is important to take this into account.

A second point is that while replacement rates show the situation at retirement, the period after is also important, as Jim mentioned. Pensions may decline after retirement because of cuts or because people opt for a high pension at the beginning and a lower pension later in retirement. Expenditures can also change during retirement, for example after one spouse is widowed. Expenditures also increase when health problems emerge or when the house needs maintenance. There is a lot of variation that the rule does not capture.

Coppola: Jim, you mentioned the life cycle. Some economists, such as Scholz and Seshadri, or more recently Hurd and Rohwedder, suggest using a closer approximation of the life-cycle consumption model to judge retirement preparedness of households, which might imply lower optimal replacement rates for many groups. What do you think of the merits of this?

Moore: It depends on the audience. Eighteen years ago, when I was doing research I found that roughly a third of the population was on track for retirement, a third was drastically undersaving and a third oversaving. As Marike noted, there is heterogeneity and any heuristic is not going to be appropriate for everybody, but it should be a first-order approximation. The work of a good financial planner is to think about the individual household and why you want to deviate from the rule.

Furthermore, there is heterogeneity in the workforce. At the lower income end, social security and other programs may replace more of my income. If I’m wealthier, I may have to save more to replace more of my income. The question is, at what point do you make decisions based on an absolute level of consumption needs versus a relative level of consumption needs. The problem is that no model provides the golden answer for every household. For those who can deal with sophisticated life-cycle models then the approach is good. I am thinking here of Larry Kotlikoff’s planning software as an example: it’s incredibly sophisticated, but unfortunately conceptually beyond the grasp of many who could benefit from it, without some guidance.

Knoef: I like that the work of Scholz and the others takes heterogeneity into account because, as Jim rightly says, there is no one measure for everyone. But assumptions still have to be made with regard to the parameters of the model and with regard to whether spending patterns of the future elderly will follow a similar path to that of current retirees. Since we haven’t developed a crystal ball, we can’t do without assumptions.

Coppola: Marike, I know in your work in the Netherlands you’ve examined subjective “minimum” and “preferred” target replacement rates. How do your results compare with the “rule of thumb” and the “life-cycle” approaches?

Knoef: On average, subjective “minimum” and “preferred” target replacement rates are comparable with the 80% rule, but there is a lot of variability. Subjective target replacement rates range from about 100% in the lowest income quintile to about 60% in the highest income quintile. However, in our research, we were not so much focused on minimal and preferred target replacement rates as on minimal and preferred expenditures during retirement, and then we compared them to current income. We found a lot of heterogeneity even within income quintiles, with part of this relating to income. People with a lower income prefer a higher replacement rate than people with a high income.

Coppola: So when people reported their preferred level, how do you interpret an average over the whole retirement period? At the beginning, you may need a higher replacement rate, maybe 100% even for the wealthy; then a 70% replacement rate later may be fine because your consumption is going to decline.

Knoef: It is a good point and we asked respondents about the situation during retirement. You really can’t say “at retirement” because it is an average throughout retirement. This is a weakness of the 80% rule, which only really looks at the moment at retirement, but there is also a whole period after, which is also important.

Michela: I agree … and the interesting thing is that the replacement rate – this rule or threshold – is more or less constant across different countries. However, I think health expenditures vary greatly in different countries, particularly in the U.S. compared to Europe, and that might have an effect on the optimal replacement rate. Jim, your view?

Moore: I think there needs to be more viable markets for long-term care, but a lot of insurances offering long-term care 10 to 20 years ago have pulled back dramatically from doing this because of the actuarial uncertainty associated with forecasting demographics, costs and technological advances on the medical side. To me, this is the biggest wild card in discussing retirement over the next 10 to 20 years. In most phases of the economy, technology is a deflationary force, but in medicine, technology tends to be an inflationary force: the bundle provided is always changing, in that if you have new technological advances they get added to the standard course of treatment. This could create uncertainty in two ways. It could reduce the costs of care and volatility, but if it increases the standard of care procedures it could work the other way.

Coppola: Marike, you mentioned that the strength of the 80% rule is the simplicity. In comparison, a life-cycle consumption model would seem complex to implement from an industry perspective. Or is it more practical than imagined?

Knoef: Well, Jim would be better placed to answer this, but from an industry perspective I imagine this could be difficult since you need information from different sources such as savings, housing, mortgages, household composition and pension wealth in other funds. From an individual perspective, this may be doable. I was involved in updating a pension tool for Nationale Nederlanden, an insurance company. People can fill in all their financial resources, choose a retirement age and a retirement expenditure goal. It appears people find it difficult to define a retirement expenditure goal. If these tools were extended with a life-cycle consumption model, it may help, but it is still important for people to make their own decision because life-cycle models may not always accurately reflect the decision process and heterogeneous preferences of real households.

Coppola: Jim, a quick last question before our time is up: What should a good retirement product offer retirees?

Moore: The essential thing, which is particularly difficult in a low-return world, is to provide some measure of capital protection and a competitive return for a reasonable amount of risk. If you start trying to bundle healthcare products in there, then there is need for better healthcare products owing to their complexity.

Coppola: Thanks to you both.

More on Retirement Investing:
View the results of our recent DC Consultant Survey.