How TSM simulated way back to 1929?

[lsbcal]

I posted the FIRECalc results below (see quote) in another forum. The purpose was to look at how a value tilted portfolio compared to a portfolio based on TSM (total stock market). Now I'm wondering how TSM was constructed since, of course, there was no TSM funds in 1929. Could someone enlighten me? I just would like a rough idea of where the data comes from for the different options in the "How to Invest" tab of the advanced FIRECalc tool. Thanks, it's a great resource.

Les
====================

I ran FIRECalc (with no withdrawals) to look at various allocations for the depression years 1929-1932 and also for the bear market of 1973-1974. The results are below. Portfolios 2,3,4 are just for reference. I realize this data just focuses on worst case declines.

Start value = 100,000
ER = 0.2
CPI adjusted, rebalanced yearly
Time period 1: 1929 to 1932 (3 years)
Time period 2: 1973 to 1975 (2 years)

Port1: 60% TSM, 40% 5yr treasury
Port2: 100% TSM
Port3: 100% 5yr treasury
Port4: 100% 30yr treasury
Port5: 15% SP500, 15% LV, 15% S, 15% SV, 35% long treasury, 5% 1mo treasury
Port6: 60% SP500, 35% long treasury, 5% 1mo treasury
Port7: 40% SP500, 20% SV, 35% long treasury, 5% 1mo treasury
Port8: 40% SP500, 20% LV, 35% long treasury, 5% 1mo treasury

Period..1....2

Port1 76768 68656
Port2 48880 53972
Port3 132440 93925
Port4 131239 92673
Port5 54122 62569
Port6 71841 63862
Port7 63430 60637
Port8 65502 66790

So for example, Port1 based on total stock market did dramatically better then Port5 with a big value tilt which lost almost half it's value (going from 100,000 to 54122).

[dory36]

FIRECalc looks at the total market performance, rather than any specific fund that is attempting to match the total market behavior.

FIRECalc uses the market behavior data compiled by Yale professor Robert J. Schiller. He reported the annual series of January values of the Standard and Poor Composite Stock Price Index starting in 1871, citing Standard and Poor's Statistical Service Security Price Index Record as the source.

[lsbcal]

Dory, thanks for your reply. Perhaps my premiss that I could use FIRECalc to compare TSM and a value tilted portfolio during severe market downturns is only roughly correct? What do you think?

BTW, when I look at the TSM from FIRECalc versus VTSMX (vanguard total stock market) I see the following:

 
      TSM  VTSMX
1993  8.8  10.6
1994 -1.7  -0.2
1995 31.3  35.8
1996 23.2  21.0
1997 25.7  31.0
1998 29.1  23.3
1999 12.4  23.8

Assuming I did this right the Schiller series is quite a bit different then what I get from Vanguard's TSM (ER=0.19 presently).

Les

[dory36]

I'd treat all the results as "roughly correct" -- every fund has a slightly different way of doing things, and also a different way of reporting things. FIRECalc/Schiller data separates out dividends and reports them separately. Does the fund?

Also, as far as results from FIRECalc are concerned, remember you are mostly looking at how one allocation versus another performed in the worst cases, not the typical cases.

[lsbcal]

Dory, from what you are saying it looks like it would be reasonable to use the FIRECalc data as I did above. Since some people focus quite heavily on average returns and standard deviations in arguing to tilt to value I was just trying to see what a major stress test does to the concept.

Seems that in the depression years drawdowns were more severe for a value tilted portfolio (46% decline) versus one that is roughly balanced as in TSM (23% decline). On the other hand, the downturn in 1973-74 was not as dramatically bad a comparision (37% vs 31%).

When I think of TSM I'm generally thinking of one like Vanguard's that has 70/20/10 allocation to large/mid/small and roughly balanced in value/blend/growth categories.

Les