Conventional wisdom has it that young investors should allocate a large portion of their savings to stocks – a high return, high risk asset class – when they first start investing their savings, and then slowly reduce their stock allocation as they approach and then surpass retirement age.  By the time they reach their eighties, their allocation to stocks should be very low relative to bonds and other less volatile asset classes.  Indeed, almost every target date fund in 401(k) and other types of retirement plans is structured this way.  However, Wade Pfau, Professor of Retirement Income at The American College, and Michael Kitces, Director of Research at Pinnacle Advisory Group, have co-written a paper (not yet published) recommending exactly the opposite.  That is, allocation to stocks should increase during one’s retirement years in order to maximize the probability that the investment portfolio is not prematurely depleted.

Although counterintuitive, the study’s conclusion is logical when you consider what causes a retirement portfolio to “fail” (i.e. run out of money prematurely).  First, there’s the situation involving an extended period of poor returns during the early part of retirement.  This kind of performance threatens portfolio sustainability.  A strategy of reducing equity exposure under these circumstances will result in the portfolio containing the least amount of stocks just when the good return years finally show up (assuming, of course, that the entire retirement period does not experience poor returns). Conversely, a rising equity glide path – allocating more to stocks during the later retirement years – would provide higher equity exposure by the time returns improve.  In the other case where the early retirement returns are good but the later returns are bad, the portfolio grows so well during the early years that subsequent asset allocation choices have a much lower impact on achieving the original retirement goal.

Another way to think about it is that your nest egg is largest at the beginning of your retirement.  That is the point at which you stand to lose the most.  As you draw down your savings, even if you increase the percent of your portfolio in stocks, the dollar amount at risk is either the same or less.

Just what are the chances of a prolonged downturn in stocks during one’s retirement?  I looked at annual US large cap stock returns and inflation since 1926.  If you define a poor return as being below the inflation rate, I could find only seven multi-year periods having poor returns (1929-31, 1939-41, 1946-47, 1969-70, 1973-74, 1977-78, 2000-02), and none of them exceeded three years.  So the likelihood of experiencing prolonged losses in retirement is pretty low, at least based on history.  But it is important to recognize that the historical worst case scenario is to retire just before a market downturn.  “Market performance in the first 10 years of retirement predicts 80% of final outcomes,” Pfau says.

How much should you allocate to stocks throughout your retirement?  According to the study, the optimal allocation is 20%-30% in stocks at retirement, increasing to 50%-60% or even 70% in at the end, assuming a retirement period stretching from age 65 to age 95.  That would mean over one’s lifetime an equity allocation that starts high, declines until the point of retirement, and then increases again until death.  However, such an allocation need to be modified to take into account your personal risk tolerance.  If you’re not the type to be able to stomach watching your portfolio take a hit whenever the market takes a dip, an overall reduced allocation to stocks would certainly improve your enjoyment of your remaining retirement years.

Remember in the end that it can be just as risky to allocate too little to stocks as too much.  While a reduced stock allocation may feel more comfortable, if you’re not getting the returns you need, especially early in your retirement, your portfolio may become so stressed that it becomes crucial for you to allocate more to stocks to ensure the portfolio can support you through to the end.

Here’s a link to the paper: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2324930