How much corpus is enough? The 4% rule

Everyone approaching retirement has the same question. How much is enough? To answer that question with any degree of confidence, it'd help if we could look back to years gone by and learn from how things turned out in the past. Although past performance is not a prediction of the future, just to humour Mr. Twain, "history doesn't repeat itself, but it often rhymes".

If one tries to conduct a historical longitudinal study of retirement outcomes, she'd quickly realise that India has limited historical data. The first Index in India, the BSE Sensex debuted in 1986(the base year was 1979). The NIFTY 500 was launched in 1995, the Nifty 50, the Index we all track today was instituted in 1996. US on the other hand, has a much longer and richer recorded history of capital markets. Started in 1896, Dow Jones index is today more than 125 years old. S&P, constituted in 1923 is more than 100 years old.

Given US has much richer data, we lean on the US studies for insights. That's where the 4% rule originated. Back in 1990s an MIT Aeronautical Engineer turned wealth advisor, Bill Bengen came up with a study on Retirement Corpus and postulated the 4% rule*.

Safe Withdrawal Rate (SWR). The origins of the 4% rule

For the study, Bill assumed a 30 year retirement period and traced all 30 year periods starting 1926 onwards e.g. 1926 to 1956, 1927 to 1957 and so on. He assumed one was invested in Equities and Bonds (60%-40% respectively). He used the index as a benchmark and Consumer Price Inflation CPI data for his calculations.

The calculations begin with a corpus and a withdrawal rate. The withdrawal rate is the first year withdrawal as a percentage of the corpus, e.g. a corpus of Rs. 1Cr(100 lacs) and a withdrawal of Rs. 4lacs would give 4/100 = 4%. The Corpus itself was divided into 2 parts. The first part was invested in equities index, like S&P and the other was invested in intermediate US treasury bonds. In order to keep things simple, there was no bucketing of short term vs long term goals, no separate emergency corpus and no tax complications. Everything was assumed to be paid for from the corpus itself. The withdrawals and re-balancing were done annually such that the proportion of stocks and bonds remains the same over the years. Every year the previous year's withdrawal was adjusted to inflation and then withdrawn from the corpus. The above steps (Inflation adjusted withdrawal and rebalancing) were repeated for 30 consecutive years. If there was any corpus left at the end of the period, the same exercise is repeated with a higher withdrawal rate. The aim was to find the highest withdrawal rate for which the balance remaining at the end of 30 years was zero.The same exercise was done for all 30 year periods to identify max withdrawal rate for each 30 year period.

Let’s take an example. Mr. and Mrs. Smith retired in 1951 with a corpus of $500K and invested that corpus in S&P for equities and Medium term govt securities and bonds with a 60%-40% split. The withdrawal rate was 5% of the corpus in the first year. At the end of the first year they withdrew $25K. The next year they adjusted this $25K for inflation (lets say inflation was 5%) and withdrew $26,250 and so on, up until 1981 where after the withdrawal their corpus was fully depleted.

As one would imagine, the withdrawal rates were higher for those 30 year periods where the markets performed well and lower when there were major downturns along the way. From all these withdrawal rates, he picked the lowest rate and called it the “Safe Withdrawal Rate” (SWR). That rate came to 4%. This rate applied to period starting 1968 and end in 1998. And thus, 4% came to be known as the Safe Withdrawal Rate. In 1998, a Trinity study reaffirmed it and since then 4% withdrawal rate became the standard “Safe Withdrawal Rate”(SWR)

Given the availability of historical data for US, Bill could draw up 500, 30 year periods to analyse the data covering periods like the 1929 depression, WWII, Oil shock of 1973 etc.

The India Context

If someone was to do the same study in India, they would be hamstrung by lack of historical data. But where there isn’t enough historical data, simulations can help bridge the gap. I was fortunate to find a couple of well researched articles on the subject by Ravi Saraogi.

The first study scrutinises the 4% rule on the lines of what Bill Bengen did. It uses 40% equity(BSE) and 60% Bonds (Combination of FD rates and a debt mutual fund) starting in 1980. The first study reaffirms the 4% rule for India. The SWR comes to ~4.7%. He however observes that the trend line for returns(Both Equities and Bonds) is on a decline(1980-90 had much higher returns than 2010-2020) and therefore in conclusion the paper suggests that the 4% rule may be challenged sometime in the future. In the later part of the paper author turns to simulation as there were only 12 historical retirement paths available starting 1980-2010 to 1991-2021, too small a dataset. Also, accounting for falling yields over decades, he uses data only post year 2000. He concludes by revising the SWR to 3%.The simulation studies demonstrate a 92% confidence on a 3% withdrawal.

The second paper is almost entirely based on simulations. Based on these simulations he brought down the Safe Withdrawal Rates for India to 3% - 3.5%. He did his analysis using MonteCarlo and Bootstrap simulation (resampling). I'd urge anyone who wants to get a closer look to read up the papers here. They are a treasure trove for historical insights.

Impact of negative shocks on Retirement Corpus

For most retirees, the initial years of retirement are the go-go years. They are excited stepping into this new phase where they own their time and freedom. They want to travel the world, live experiences they always wanted to but could not due to work commitments. All in all, the early years are the ones where a freshly retired couple is likely to spend the most. Over time things quieten a bit and in the later years, medical expenses exceed travel, dining out and leisure.

On the other hand, all the studies conducted on SWR highlight that the retirement corpus is most vulnerable to negative shocks in the early years. If in the early years of retirement, the investments decline due to market volatility, or a bear market, it greatly increases the overall chances of “Portfolio Failure”. If the initial years are positive, the corpus creates its own margin of safety that helps weather any market downturn in later years. In our discussion on Volatility, the corpus of Rakesh and Sunita could not last the planned time even when their “Average Annual Returns” were the same as that of Gurmeet and Harleen, as they faced a market downturn in the first year of their retirement. The risk of negative returns early on is called “Sequence of Returns” and I discuss the implications in greater detail here.

Longevity is one of the biggest factors impacting the success of Retirement Planning. I’ve discussed it here. If a couple is healthy, it’s possible that one of them may live to be a centurion. Based on life expectancy at the time of retirement, one may need to plan for more than 30 years. As the longevity increases, the SWR declines. Similarly, Volatility brings in “Sequence of Events” Risks. All of these need to be factored when preparing a Retirement Plan.

The 4% rule should just be a guiding principle. Yet it's a great starting point. It translates to the corpus being 25 times the annual expenses. Retirement plans are highly personalised, reflecting a retired couple’s personal circumstances. For example, someone expecting a pension post retirement or an inheritance may have an SWR higher than 4%. If one is flexible, she may go for dynamic withdrawal, cutting down during downturns and living it up during a bull run.

But remember, initial years of retirement is like walking a tight rope, watching out for any ROI slippages. If things go well in the initial years, the anxiety ebbs away as time passes.

* 4% Rule Bill Bengen https://web.archive.org/web/20120417135441/http://www.retailinvestor.org/pdf/Bengen1.pdf