Idiot Switching

I read an article recently that showed that timing never works, regardless of the details. I marvel at how much cleverness it takes, possibly unknowingly, to force such a conclusion.

This got me to thinking about how our results could be distorted and whether our procedures would catch the errors.

I set about building a ridiculous algorithm. I call it Idiot Switching. I put it through its paces. This is what I found.

The Algorithm

I chose to be 100% in stocks whenever P/E10 fell below 14 and 100% in TIPS at a 2% (real) interest rate whenever P/E10 rose above 14. I named this Sw14T2.

Such an all or nothing approach is as ridiculous as I could come up with. Using P/E10 is consistent with our general procedures. Setting the threshold to 14 is a natural choice. It is close to the median P/E10 of the modern era.

Baselines

You can contrast these results with more sensible conditions throughout this site. I recommend Current Research A and Current Research B as a starting point. I also recommend the Accumulation Stage: Edited.

Current Research A
Current Research B
Accumulation Stage: Edited

In reality, we have moved far beyond these earlier studies. I would prefer our Latch and Hold approaches for varying stock allocations (i.e., switching). I prefer our Dividend-Based Strategies overall. And never forget that we have our standard TIPS-only baselines. You never have to accept the inferior results of the older, traditional, fixed allocation portfolios.

For Accumulators

I compared owning 100% stocks (HSWR100T2, including T2 nomenclature even without any TIPS) with the Sw14T2 algorithm. I set expenses to 0.0%. I set the initial portfolio balance to $1000. I made no deposits and no withdrawals.

I looked at portfolio returns at years 10, 20 and 30 starting from 1923-1980.

At year 10, switching did better from 1923-1946 and 1964-1979. [Dummy data may have created misleading results for 1976-1980.]

At year 20, switching did better from 1923-1941, 1943, 1964-1971 and 1973.

At year 30, switching did better from 1923-1934 and 1937.

Plots versus the percentage earnings yield 100E10/P (that is, 100%/ [P/E10]) showed very weak correlations. At year 10, the line was flat. At year 20, the slope was down slightly. At year 30, the slope was up slightly.

This is consistent with what I reported in Accumulation Stage: Edited. A careful analysis reveals important insights into the upside potential when starting from different valuations.

A superficial analysis simply points out that, as the time period is made longer and longer, eventually buy-and-hold does best. A time period of 30 years is not long enough to provide a guarantee. The long run really is the long run.

For Retirees

I set expenses equal to 0.2% and calculated 30-Year Historical Surviving Withdrawal Rates for 1923-1980. I added HSWR50T2, which has 50% stocks and 50% TIPS at a 2% interest rate and which is rebalanced annually.

[Note: HSWR50T2 is a better choice than HSWR80T2, which has 80% stocks, in times of high valuation.]

It turns out that Sw14T2 did best.

Those who like to nitpick might point out that the lowest 30-Year Historical Surviving Withdrawal Rate of HSWR50T2 was 4.2% in 1965 and 1966. The lowest 30-Year Historical Surviving Withdrawal Rates of Sw14T2 was 4.1% in 1959, 1960 and 1961. In a limited sense, with a handful of data points, they are right. In a more general sense, when we take advantage of all of the data, they are wrong.

[Note: The lowest 30-Year Historical Surviving Withdrawal Rate for HSWR100T2, which consists entirely of stocks, was 3.7% in 1966.]

Here are the regression equations:

Sw14T2:
y = 0.7597x+1.9823 plus 3% and minus 2%.
R-Squared = 0.4704.

HSWR50T2:
y = 0.4105x+2.9047 plus 1.5% and minus 0.5%.
R-Squared = 0.7218.

HSWR100T2:
y = 0.89x+0.6515 plus and minus 2%.
R-Squared = 0.6993.

Here are the comparisons at today’s valuations (P/E10 = 27):

Sw14T2:
Safe Withdrawal Rate = 2.8%.
Coin Toss Rate = 4.80%.
High Risk Rate = 7.8%.

HSWR50T2:
Safe Withdrawal Rate = 2.6%.
Coin Toss Rate = 3.05%.
High Risk Rate = 4.6%.

HSWR100T2:
Safe Withdrawal Rate = 2.0%.
Coin Toss Rate = 3.95%.
High Risk Rate = 6.0%.

Here are the comparisons at typical valuations (P/E10 = 14):

Sw14T2:
Safe Withdrawal Rate = 5.4%.
Coin Toss Rate = 7.41%.
High Risk Rate = 10.4%.

HSWR50T2:
Safe Withdrawal Rate = 5.3%.
Coin Toss Rate = 5.84%.
High Risk Rate = 7.3%.

HSWR100T2:
Safe Withdrawal Rate = 5.0%.
Coin Toss Rate = 7.01%.
High Risk Rate = 9.0%.

Here are the comparisons at low valuations (P/E10 = 8):

Sw14T2:
Safe Withdrawal Rate = 9.5%.
Coin Toss Rate = 11.48%.
High Risk Rate = 14.5%.

HSWR50T2:
Safe Withdrawal Rate = 7.5%.
Coin Toss Rate = 8.04%.
High Risk Rate = 9.5%.

HSWR100T2:
Safe Withdrawal Rate = 9.8%.
Coin Toss Rate = 11.78%.
High Risk Rate = 13.8%.

Conclusions

It turns out that even Idiot Switching has a lot to offer. All that it takes is a decent analysis procedure.

For retirees, the advantages of switching are readily apparent. Varying allocations with P/E10 (i.e., switching) helps, even when crudely done.

It takes a more detailed analysis to learn the lessons needed by accumulators. The general rule is that you will do better by dollar cost averaging into an all-stock portfolio when starting out, but that you need to preserve capital when you get within 15 years of retirement.

Even this rule can fail.

At times such as these, when valuations are especially high, you face a huge downside risk. Even if you are just starting out today, you may give up on stocks entirely. It takes two decades to be sure that stocks will do better than TIPS. A better choice is to invest in TIPS or another secure investment until stock valuations return to a more reasonable level (e.g., P/E10 below 20).

Have fun.

John Walter Russell
June 20, 2006