Firecalc-test the mortgage-payoff decision?

[TimSF]

Like others here, I have questioned whether or not to pay off the mortgage in ER. I decided to run the payoff/not-payoff scenarios through Firecalc. The results were interesting. Here's what I did:

I took the length of time remaining on my mortgage (14 years) and ran two sets of Firecalc calculations out for 14 years. In the pay-off scenario, I reduced my portfolio value by the amount required to pay-off the mortgage (about $180k) and also reduced my annual expenses by the mortgage payments (about $17k). I then compared the result ranges of portfolio values after 14 years.

The not-payoff scenario produced a higher average portfolio value and higher top portfolio value, by about 2%. However, the not pay-off scenario also produced a significantly lower worst-case scenario, by about 10%. From this, I deduced that paying-off the mortgage was the safer call with only limited additional potential upside.

Has anyone else tried this?

 

[rodi]

I did the same calculation, sort of. We were closer to the payoff date, with lower payoff amount. I modeled it as my total spending (including mortgage payment) - and then had a pension come online equal to my mortgage payment, at the payoff time. I ran my normal 30-35-40 year runs. (I'm 52 - so I like to run all three just to make sure I'm including some bad years, even though I'm planning for a 40 year retirement.

I then re-ran it with the nest egg smaller, and the spending smaller... no fake mortgage payment reduction pension.

In my case, I had fewer failures with the payoff. (Portfolio has since grown to no failures in either scenario.)

Like you - the average portfolio value was higher without the payoff - but the lowest points came closer to zero.

I ended up splitting the difference in real life. Paid off my mortgage, but just opened a HELOC... which I will use if I need to.

 

[pb4uski]

Tim, I think you need to be careful because if you include mortgage payments in your spending Firecalc will inflate your spending (including the mortgage payments) each year which is not realistic. I would suggest putting in expenses ignoring your mortgage payments and include the mortgage payments as non-COLAed off-chart spending followed by a non-COLA pension of the same amount starting when your mortgage ends.

I like to model the mortgage problem as if one (for example) took out a $100,000, 4% 15 year mortgage and put the proceeds in a 60/40 portfolio and took the $8,876 annual payments out of the portfolio to service the mortgage. It says:

FIRECalc looked at the 129 possible 15 year periods in the available data, starting with a portfolio of $100,000 and spending your specified amounts each year thereafter.

Here is how your portfolio would have fared in each of the 129 cycles. The lowest and highest portfolio balance throughout your retirement was $-54,923 to $161,087, with an average of $36,905. (Note: values are in terms of the dollars as of the beginning of the retirement period for each cycle.)

For our purposes, failure means the portfolio was depleted before the end of the 15 years. FIRECalc found that 22 cycles failed, for a success rate of 82.9%.

For me, its a risk I'm willing to take.

 

[ChiliPepr]

The not-payoff scenario produced a higher average portfolio value and higher top portfolio value, by about 2%. However, the not pay-off scenario also produced a significantly lower worst-case scenario, by about 10%.

Just curious, since you compared the Avg to Avg (Payoff wins by 2%), and the Worst to Worst (Not-Payoff wins by 10%).... how did the Best to Best scenerio go?

 

[ERD50]

Two problems with this.

1) The mortgage expense should only be the interest portion (and as pointed out - this needs to be a separate non-inflation adjusted expense, not lumped in with 'spending'), the increasing principal payments are going back to you. A transfer of net worth, in effect, not an expense.

2) For some scenarios, your current mortgage rate would not have been available at the start of that scenario, so it's not really a valid back-test. A 4% mortgage would have been very valuable in 1982 (when I got a 17% !) mortgage (adjustable, which was a life-saver - payments went down steadily). But they were nowhere to be found.

-ERD50

 

[TimSF]

Tim, I think you need to be careful because if you include mortgage payments in your spending Firecalc will inflate your spending

pb4uski, I will try your method, but I do not understand your objection that Firecalc inflates my spending under my method. I am using actual numbers over the remaining life of the mortgage - additional expenses I will actually pay if I do not pay the mortgage off. Of course, the saved annual expenses are set off against the reduced portfolio starting value

It may be that I am asking a different question than you are: What are the range of portfolio values I will have in 14 years if I pay off the mortgage today vs. not paying it off? Obviously, I am keeping everything else, such as AA, consistent in the comparison.

 

[TimSF]

Chilippr, the best-to-best comparison came out with about a 3% advantage to no-payoff. I should emphasize that these numbers are particular to my circumstances. Different inputs will yield different comparisons.

 

[pb4uski]

Sorry Tim, I did not realize that you were using the manual entry option so my observation would not apply. I thought you were putting your spending on the Start Here page and it increases that amount each year as instructed by the Spending Models page.

 

[TimSF]

Two problems with this.

1) The mortgage expense should only be the interest portion (and as pointed out - this needs to be a separate non-inflation adjusted expense, not lumped in with 'spending'), the increasing principal payments are going back to you. A transfer of net worth, in effect, not an expense.

2) For some scenarios, your current mortgage rate would not have been available at the start of that scenario, so it's not really a valid back-test. A 4% mortgage would have been very valuable in 1982 (when I got a 17% !) mortgage (adjustable, which was a life-saver - payments went down steadily). But they were nowhere to be found.

-ERD50

I understand that as I pay off a mortgage, I am paying myself a larger and larger portion of the principal - but I do not think that is relevant to the question I am asking of Firecalc. In other words, Firecalc doesn't "care" what portion of your inputted annual expenses are principal or interest or some other category. In both scenarios my principal is paid in the calculation - either as a lump sum at the start, or as a portion of the mortgage payments over time.

Not suggesting this method would work as a back-test, but as of way posing a question to Firecalc that may be instructive for actual mortgage scenarios.
Am I missing something?

 

[ERD50]

I understand that as I pay off a mortgage, I am paying myself a larger and larger portion of the principal - but I do not think that is relevant to the question I am asking of Firecalc. In other words, Firecalc doesn't "care" what portion of your inputted annual expenses are principal or interest or some other category. In both scenarios my principal is paid in the calculation - either as a lump sum at the start, or as a portion of the mortgage payments over time.

Not suggesting this method would work as a back-test, but as of way posing a question to Firecalc that may be instructive for actual mortgage scenarios.
Am I missing something?

OK, on second thought I guess #1 is a wash. #2 is still valid.

And I'm still not 100% sure, from your replies, that you actually entered the mortgage payment separate from your spending. That's big. You said:

... and also reduced my annual expenses by the mortgage payments ...

In FIRECalc, your annual expenses ('spending' - on the first tab) are inflation adjusted, but mortgage payments are not. So did you enter the mortgage payments separately, on the "Other Income/Spending" tab, as 'Off Chart Spending' and deselect 'Inflation adj?' ?

To put it another way - keep 'spending' on the first page the same in both cases. For the mortgage case, you increase your portfolio, and add in non-inflation adjusted 'Off Chart Spending' to match the annual mortgage payment.

-ERD50

 

[ripper1]

I did the same calculation, sort of. We were closer to the payoff date, with lower payoff amount. I modeled it as my total spending (including mortgage payment) - and then had a pension come online equal to my mortgage payment, at the payoff time. I ran my normal 30-35-40 year runs. (I'm 52 - so I like to run all three just to make sure I'm including some bad years, even though I'm planning for a 40 year retirement.

I then re-ran it with the nest egg smaller, and the spending smaller... no fake mortgage payment reduction pension.

In my case, I had fewer failures with the payoff. (Portfolio has since grown to no failures in either scenario.)

Like you - the average portfolio value was higher without the payoff - but the lowest points came closer to zero.

I ended up splitting the difference in real life. Paid off my mortgage, but just opened a HELOC... which I will use if I need to.

Well, why would someone pay off a mortgage then open up a HELOC. It doesn't make sense to me to use cash reserves to pay off a home and then have to borrow it back and pay interest on it. I guess in your case it might just be piece of mind as is the case for a lot of mortgage burners here. I just like to have the banks money at these historic interest rates. I have better things to do with my cash.

 

[Cobra9777]

I ran the same analysis described by pb4uski last year before deciding to payoff my mortgage. The success rate was only 55%. I just re-ran the numbers to confirm. I think there are a couple reasons for the low percentage in my case: (1) I had made some sizable 1-shot principal payments over the prior 5 years, so the mortgage balance was very low compared to the payments, and (2) the interest rate was 5.25% (I never refinanced in anticipation of payoff at ER). I ran some what-if's assuming no 1-shot principal payments and current market rates, and the success rates were indeed much higher.

I suppose I could have refinanced at 4% for 15 years and left the money in my portfolio, with 80% chance of making a few bucks. But I already have enough money. I like the idea of reducing risk and uncertainty where I can, reducing expenses, and simplifying/grossing-down the balance sheet. I didn't like the idea of spending a few $K on closing costs just as I was entering ER.

Apparently I'm a little more risk averse than many on this forum. I also elected the annuity option over lump sum on my DB pension so that a larger percentage of spending is covered by guaranteed income vs portfolio withdrawals. A similar FIRECalc analysis at the time of that decision yielded only a 25% failure rate on the lump sum option. But that's 25% I no longer have to be concerned about.

In most cases, the raw math says keep the mortgage and don't worry about guaranteed income. But the actual decisions, at least in my case, were based more on risk tolerance and the bigger picture. If my overall financial preparedness had been more borderline, I might have made different decisions. When the war is won, it's time to stop fighting and enjoy peacetime.

 

[pb4uski]

Well, why would someone pay off a mortgage then open up a HELOC. It doesn't make sense to me to use cash reserves to pay off a home and then have to borrow it back and pay interest on it. I guess in your case it might just be piece of mind as is the case for a lot of mortgage burners here. I just like to have the banks money at these historic interest rates. I have better things to do with my cash.

I think that makes perfect sense - he said he opened a HELOC not that he drew against it. Since he didn't draw against it there is no interest. The HELOC is like an additional no-cost resource if you need cash quickly.

 

[rodi]

I think that makes perfect sense - he said he opened a HELOC not that he drew against it. Since he didn't draw against it there is no interest. The HELOC is like an additional no-cost resource if you need cash quickly.

This.

I've tried to account for every expense I can imagine in my spending, but I'm also hoping for ACA subsidies. If I were forced to deplete my taxable accounts sooner, and have to take money from an IRA - that counts against my MAGI for ACA. It's an insurance policy on cheap insurance.

I actually got the idea of a HELOC as an income smoother in high expense years from folks here.

 

[donheff]

Interesting thread. It is clearly important to make sure that the mortgage expense is flat and not calculated as an increasing expense by FC. As to the principal payback, while that is important for net worth, as pointed out by others it may or may not not be of use for specific payoff/don't payoff evaluations. Often the principal in the house is irrelevant to ER decisions. If you plan to age in place and don't want to consider a reverse mortgage, then the total expenses are what matters to issues of portfolio survival. If portfolio survival is not an issue (e.g, pensions cover all critical expenses, or a house sale/reverse mortgage are in the picture) it looks like the don't pay off option is better. From an investment perspective that is probably what I should have done. But, despite being well positioned via pensions, I was more focused on the possibility of increased portfolio spend down in the early years since I wanted to be able to travel and engage in other non-essential (spend thrift) activities even in the face of a significant downturn.

 

[Cobra9777]

1) The mortgage expense should only be the interest portion (and as pointed out - this needs to be a separate non-inflation adjusted expense, not lumped in with 'spending'), the increasing principal payments are going back to you. A transfer of net worth, in effect, not an expense.-ERD50

From a pure accounting perspective, this is correct. When you make a loan payment, only the interest portion is recorded as expense. The principal portion simply reduces the liability balance, which is neutral to net worth. However, the principal payment is certainly a cashflow, and that's what matters in this analysis. The essence of the OP's question is: WHEN to make the principal payment.

BTW, if you were to use only the interest portion in this analysis, it is not a flat amount. It decreases over time as the principal balance is reduced. Only the total payment is flat.

2) For some scenarios, your current mortgage rate would not have been available at the start of that scenario, so it's not really a valid back-test. A 4% mortgage would have been very valuable in 1982 (when I got a 17% !) mortgage (adjustable, which was a life-saver - payments went down steadily). But they were nowhere to be found.-ERD50

I think this is valid concern. The high success rates of these analyses seem to be driven by today's very low mortgage rates being compared against historical investment returns with significantly higher interest rates. Obviously we can't restate FIRECalc history to conform to a 4% environment. But a quick Google search landed me at the Freddie Mac website which says the average mortgage rate from 1971 to today is 8.4%. If you use that figure instead of 4% in the analysis described by pb4uski, the success rate drops from 82.9% to 27.1%. There might be a better long-term average for a more apples-to-apples comparison, but I just wanted to quickly quantify ERD50's point.

 

[pb4uski]

....I think this is valid concern. The high success rates of these analyses seem to be driven by today's very low mortgage rates being compared against historical investment returns with significantly higher interest rates. Obviously we can't restate FIRECalc history to conform to a 4% environment. But a quick Google search landed me at the Freddie Mac website which says the average mortgage rate from 1971 to today is 8.4%. If you use that figure instead of 4% in the analysis described by pb4uski, the success rate drops from 82.9% to 27.1%. There might be a better long-term average for a more apples-to-apples comparison, but I just wanted to quickly quantify ERD50's point.

While I see ERD50's point, on the other hand at any point in time mortgage rates, bond returns and equity returns all implicitly start with a risk free rate (which was high in the 80s and is now historically low) and then premiums for credit risk, equity risk, liquidity risk, etc.) and all those would be reflected in the firecalc data. I think it is fair to say that in low interest rate environments like we have now that taking out a mortgage and investing the proceeds is likely to be attractive and that in high interest rate environments (like in the early 80s when we took out our first mortgage at 13%) it would be downright crazy.

 

[Cobra9777]

While I see ERD50's point, on the other hand at any point in time mortgage rates, bond returns and equity returns all implicitly start with a risk free rate (which was high in the 80s and is now historically low) and then premiums for credit risk, equity risk, liquidity risk, etc.) and all those would be reflected in the firecalc data. I think it is fair to say that in low interest rate environments like we have now that taking out a mortgage and investing the proceeds is likely to be attractive and that in high interest rate environments (like in the early 80s when we took out our first mortgage at 13%) it would be downright crazy.

I just look at it very simply... historic low interest rates today, mean they will likely rise at some point during the remaining term of your mortgage. That's not good for bonds or stocks. Yet your (and the OP's) analyses measure against FIRECalc history which includes the 30-year rally in stocks and bonds since rates peaked in the mid 80s. Not to mention the post-war prosperity of the 50s and 60s. I won't get into baby boom demographics and such.

Obviously no one can predict the future. It just seems like a better measurement of the success rate would result from either: (a) matching the average mortgage rates in effect during the period from which investment returns were derived, or (b) picking 15-year periods in the past starting with mortgage rates similar to today.

From my quick-and-dirty testing, either of those changes result in significantly lower success rates for retired mortgage holders.

 

[ERD50]

...
I think this is valid concern. The high success rates of these analyses seem to be driven by today's very low mortgage rates being compared against historical investment returns with significantly higher interest rates. Obviously we can't restate FIRECalc history to conform to a 4% environment. But a quick Google search landed me at the Freddie Mac website which says the average mortgage rate from 1971 to today is 8.4%. If you use that figure instead of 4% in the analysis described by pb4uski, the success rate drops from 82.9% to 27.1%. There might be a better long-term average for a more apples-to-apples comparison, but I just wanted to quickly quantify ERD50's point.

It depends what question you are trying to answer. Personally, I've never said that carrying a mortgage is a reasonable idea in average market conditions. But when we are at near historic lows in mortgage rates, I think it very well could be a reasonable thing.

I think the only way to test this with FIRECalc is to find time periods with low mortgage rates, and then do the single year analysis with the spreadsheet for those years. I'm not sure about the accuracy of that ss output though, so carry some salt with you.

-ERD50

 

[pb4uski]

I just look at it very simply... historic low interest rates today, mean they will likely rise at some point during the remaining term of your mortgage. That's not good for bonds or stocks. ....

I'm not sure your hypothesis is correct. If/when interest rates rise then it will mean that the Fed believes that the economy has strengthened sufficiently to raise rates and while increase rates are a negative factor for bonds and stocks, the stronger economy is good for stocks.

Besides, mortgage holders who also own stocks and bonds are so far ahead of the game after the last few years that there is probably no way short of some sort of economic Armageddon that we could end up behind. As an example, I refied in early 2012 and since the refi my portfolio has earned 14.25% annualized which far exceeds my 3.375% mortgage rate so it is very unlikely that I will lose out on that deal.

 

[Cobra9777]

If/when interest rates rise then it will mean that the Fed believes that the economy has strengthened sufficiently to raise rates and while increase rates are a negative factor for bonds and stocks, the stronger economy is good for stocks.

Your portfolio is 40% bonds and we both agree that will be negative. Stocks... yes, the first few years should be positive due to a strong economy. After that, it's negative as high rates take a toll on corporate earnings and growth. You can see this pattern following the low post-war rates. Going forward, how that plays out in any given situation depends on the timing of the stock turn vs the remaining mortgage period and how bad bonds are hit.

Besides, mortgage holders who also own stocks and bonds are so far ahead of the game after the last few years that there is probably no way short of some sort of economic Armageddon that we could end up behind. As an example, I refied in early 2012 and since the refi my portfolio has earned 14.25% annualized which far exceeds my 3.375% mortgage rate so it is very unlikely that I will lose out on that deal.

Those numbers are from the tail-end of the QE cycle and resulting asset inflation. I'm talking about the upcoming period when all that goodness unwinds. The past is in the books. Everyday, you make a decision to continue holding debt based on expected future market returns.

Bottom line: I think ERD50's point is valid. One should be a bit cautious relying too heavily on raw FIRECalc history to justify holding a mortgage. There are some simple ways to adjust the analysis to get more reasonable success rates. Then, it's a matter of risk tolerance, timing, and how one feels about bond and stock performance in a rising rate environment.

I'm not against the concept of holding a mortgage in retirement and leaving the would-be payoff funds in a 60/40 portfolio. Especially anything sub-4% for 20-30 years. It's hard to imagine how that could go wrong. But if your timeframe is 10-15 years or less, and your rate is north of 4.0-4.5%, I think the decision analysis should include more than a quick FIRECalc run.

 

[jkern]

I paid off my mortgage in 2004 with 9 years remaining. Balance $180K at 4.75%. For the heck of it, I did some calculations to see how my decision turned out. For my case using my assumptions, they came out very close to equal with a slight edge to paying off the mortgage.

The one variable that I could not predict is would I have stayed the course during the stock market crash? Since I have never had a large sum in an after tax account loose 50%, I cannot guarantee that I would not panic and sell at the wrong time. It was easy not to panic in a before tax retirement account that I don't plan to access for a long time.