Using the historical return data provided in earlier post it is instructive to see effects of the variability of returns. Lets start with a simple case of having investable funds of $1,000, a planning horizon of 20 years and all the funds are invested in the SP500. Does it matter to the ending balance as to when the investment was started and to make it comparable inflation factors that would have been applicable to that period will also be considered.
YRWINDOW=20 STARTAMT=1000 STARTYEAR=1928 rollSP<-rollapply(as.vector(histBack$SP500), width=YRWINDOW, function(dummy) { return(dummy)}) rollSP<-rollSP + 1 rollCPI<-rollapply(as.vector(histBack$CPI), width=YRWINDOW, function(dummy) { return(dummy)}) rollCPI<-rollCPI + 1 ### to plot ending values over the rolling window year by year nyr = nrow(rollSP) grw_sp5<-apply(rollSP, 1, prod) * STARTAMT grw_cpi<-apply(rollCPI, 1, prod) xlim=seq(STARTYEAR, by=1,length.out=nyr) plot(grw_sp5, main="Invetsment Growth\nSequence Risk Plot", xlab="Start Year", xaxt="n", ylab="Ending Balance" ) ### sets up plot pl=seq(1, nrow(rollSP), by=10) axis(side=1, at = pl, labels=xlim[pl], las=2) lines(grw_sp5, col="blue", lwd=2) ## shows SP lines(grw_cpi*STARTAMT, col="red", lwd=4) ## shows CPI in relation lines(grw_sp5/grw_cpi, col="green", lwd=2) ## shows infl adj cur value legend(1, max(grw_sp5), legend=c("Nominal", "Adjusted", "CPI"), col=c("blue", "green", "red"), lty=1: 3, cex=0.7)
One should get this plot from which we note that if one had begun investment 1928-29, or around 1858 or early 1990s, investments would not really have grown as much as we’d have liked. In fact, the years where the green lines fall below the red line, our investment would have actually lost value factoring inflation into account. On the other hand, early years of the WWII, and late 70s / early 1980s seemed to have been very good years if one were starting out to invest.
In the simple case above, all eggs were in the equity basket. Immediately alternatives to this strategy suggest themselves. Many possibilities exist and it would be helpful if spreading out assets in some manner or following some other scheme makes much of a difference. Some of the other scenarios can be listed as follows:
- Allocate some percentage of assets to fixed bond instruments at start of investment period. Compute ending balances and examine how varying allocation ratios affect outcome.
- As above, but at start of every year during the planning horizon, that is to say, if horizon was 20 years, for each simulation at end of year 1, rebalance allocation to original starting ratio, repeat for year2, year3 and so till year 20. Examine results.
- Set aside an amount in cash at start, just invest remaining and do each of the two tests above. Cash of course does not grow (even on trees) but acts as cushion.
Thanx for reading
One other point. I checked source of some of returns showed, and it would appear that return data include dividends reinvested.
If that is the case, then calculations are in fact quite accurate and reflective of actual performance – all subject to accuracy of data.
To answer your question ‘ whether it matters when the investment was initiated ‘, it does matter but only initially since market returns historically negotiate through peaks and valleys. However, since a large chunk of returns are attributable to dividend reinvestments, market volatility is blunted by the dollar cost averaging if one stays invested over a longer time horizon.
Yes market returns at start time become very important which was the point of the plots. Dividend reinvestments definitely make a difference and could be big over the long term. But sequence risk is important to analyze for especially when you start withdrawal scenarios. A big downtown initially would be fatal.
I will have some code later specifically looking at withdrawal scenarios. For now I was looking at returns and their variability.
Thanx for input.