Page 1 of 1

TIPS in Portfoliovisualizer's Monte Carlo simulation

Posted: Sun Sep 23, 2018 7:48 pm
by NoHeat
I'm experimenting with the Monte Carlo simulation in portfoliovisualizer.com, and I came across something I don't understand: the performance of a simple portfolio consisting 100% of TIPS, and nothing else.

I would expect that every simulation run would end up with the same "portfolio end balance (inflation adjusted)" if I withdraw a fixed amount that's inflation adjusted. But that doesn't happen. The outcomes for 90th percentile, 75th percentile, etc. all end up very different.

For example, $1 million initial amount, $40k amount withdrawn (inflation adjusted) annually. I've tried all kinds of inputs: 5 years vs 10 years vs 30 years for the simulation period. Historical returns vs statistical returns for the simulation model. Historical inflation vs what I think is a constant inflation rate ("parameterized inflation" with zero percent volatility). No matter what which combination of these inputs I choose, the inflation-adjusted end balance always has a very wide dispersion among the percentiles shown, and I don't see why it should vary at all, given the nature of TIPS.

I'm not sure what's going on. User error is always a likely problem. Maybe I just don't understand the simulation and its user interface. Or maybe the simulation is just doing something wrong, since it relies on historical data from Vanguard Inflation-Protected Security Fund (VIPSX), which it could be using in a way that's not consistent with inflation data for the same period.

Any ideas?

Re: TIPS in Portfoliovisualizer's Monte Carlo simulation

Posted: Sun Sep 23, 2018 8:34 pm
by JBTX
I've never used portfolio visualizer.

Two guesses:

Taxes - interest including inflation adjustment is taxable if in taxable account.

Real interest rate. Real interest rates can and do go up and down with tips.

Re: TIPS in Portfoliovisualizer's Monte Carlo simulation

Posted: Sun Sep 23, 2018 8:47 pm
by vineviz
JBTX wrote:
Sun Sep 23, 2018 8:34 pm
Real interest rate. Real interest rates can and do go up and down with tips.
This.

TIPS prices have had serious volatility, with annual returns swinging between +16% and -9% depending on maturity. In fact, in real (inflation adjusted terms) most major TIPS funds have not regained their losses from 2013.

Re: TIPS in Portfoliovisualizer's Monte Carlo simulation

Posted: Sun Sep 23, 2018 10:28 pm
by willthrill81
vineviz wrote:
Sun Sep 23, 2018 8:47 pm
JBTX wrote:
Sun Sep 23, 2018 8:34 pm
Real interest rate. Real interest rates can and do go up and down with tips.
This.

TIPS prices have had serious volatility, with annual returns swinging between +16% and -9% depending on maturity. In fact, in real (inflation adjusted terms) most major TIPS funds have not regained their losses from 2013.
+1

Many mistakenly believe that they can buy TIPS whenever they want and be guaranteed a positive real return. That is completely false. Take a look at the chart from the Fed here. From April, 2011, until September, 2013, five year TIPS' real yield was negative. Real yields then see-sawed back and forth above and below a 0% real yield until April of 2017. Since then, they have been positive and climbing.

While you are guaranteed to earn the real yield stated at the time of purchase, TIPS do not guarantee that the real yield will be positive.

By bond standards, TIPS are very volatile. But that and the currently small price you pay for the insurance they provide against both unexpected (not expected) inflation and deflation is worth it to a lot of people.

Re: TIPS in Portfoliovisualizer's Monte Carlo simulation

Posted: Mon Sep 24, 2018 12:36 am
by JoMoney
It may be unrelated, but one thing to keep in mind is that if using historical returns in PV's Monte Carlo simulation the data sources you use limit the available information for it to randomize. Since the TIPS data source is Vanguard's mutual fund, and only has data since 2001, it limits the history being randomized to that period.
Performance for stocks was historically pretty bad since 2001, but if you include even 1% of TIPS, then that's the time period being used and randomized, change the TIPS to a bond asset with a longer history and you get much different results.