Join Early Retirement Today
Reply
 
Thread Tools Search this Thread Display Modes
Deflation and SS: Possible Issue with FIREcalc?
Old 04-20-2015, 12:45 PM   #1
Dryer sheet wannabe
 
Join Date: Dec 2013
Posts: 21
Deflation and SS: Possible Issue with FIREcalc?

First, let me just say that I think FIREcalc is an excellent resource, and that Early Retirement.org is providing a tremendous service by making it available and maintaining it. The simulator's use of actual historical returns is far superior to Monte Carlo models that fail to account for mean reversion.

I'm writing because I think there may be an issue in the way the model handles Social Security payments during deflationary periods. I haven't been able to find any discussion of this in past threads. If it is an issue, (and I'm not sure it is) then FIREcalc may be significantly overestimating the size of withdrawals that should be taken from retiree portfolios during periods characterized by deflation. I've run a number of FIRECALC simulations and looked at the Excel results spreadsheet for a number of periods covering the 19th century, and the 1920s and 1930s, when we experienced deflation. It appears, based on the results I saw for Social Security payments in Column D, that these payments are being deflated in years characterized by deflation. My understanding is that, by law, the Social Security COLA adjustment cannot cause payments to decrease. Instead, payments are held constant in years when the CPI goes down instead of up. (In fact, this happened, I think in 2009 or 2010 (based on the deflation of 2008). It was in the news, because seniors complained that they weren't receiving a COLA adjustment, and I believe the Obama Administration actually floated the idea of a small upward adjustment despite the decline in the CPI.)

If this is an issue, it can make a significant difference in the numerator of the withdrawal rate. For example, from 1929 to 1933 deflation reduced the cost of living by about 25%. Say a person retired in 1929 with total expenses of $10,000, and Social Security of $5,000 (the example is hypothetical, since 1929 pre-dates Social Security). If Social Security is assumed to decline along with total expenses by 25%, then by 1933 we have the numerator in the withdrawal rate = 0.75 x $10,000 - 0.75 x $5,000 = $3,750. But if SS is instead held constant at $5,000, we have 0.75 x $10,000 - $5,000 = $2,500. Instead of a 25% reduction in the numerator, there is a 50% reduction.

I stumbled on the impact of the treatment of SS during deflation when playing around with my own (crude) historical simulator. I was running a test based on the above example, to see the impact of delaying SS from age 62 to age 70. I was stunned at what I saw for the 1929-59 time series. With the $5,000 SS payment delayed, it increased by 76% (plus the inflation that occurred after 1933) to $9,620, while total expenses declined to $8,248--so the numerator in the withdrawal rate actually went negative 8 years into retirement, and remained negative for the rest of the retirement! In other words, a retiree would no longer have to make withdrawals from his or her portfolio, but would instead be able to save money throughout retirement (or go to Vegas? )

Having looked at the FIRECALC results spreadsheets, and recalculating the SS payments during deflationary years, in addition to 1929-59 it appears that periods beginning in the early years of the 20th century (between roughly 1901 to 1920) could be significantly affected. This was a real rough patch for stocks (Panic of 1907, WWI and high inflation), but in my tests I saw that a number of retirements that began in these years and ran out of money in the 1920s and 1930s at a starting withdrawal rate of 5% were saved from bankruptcy when the SS payments were held constant during the deflationary years. The combination of the stock market boom plus the deflation during the 1920s helped keep the portfolios from failing, once SS payments were no longer deflated along with expenses.

I apologize if this isn't an issue or has been previously discussed.

Chris M
__________________

__________________
Chris M is offline   Reply With Quote
Join the #1 Early Retirement and Financial Independence Forum Today - It's Totally Free!

Are you planning to be financially independent as early as possible so you can live life on your own terms? Discuss successful investing strategies, asset allocation models, tax strategies and other related topics in our online forum community. Our members range from young folks just starting their journey to financial independence, military retirees and even multimillionaires. No matter where you fit in you'll find that Early-Retirement.org is a great community to join. Best of all it's totally FREE!

You are currently viewing our boards as a guest so you have limited access to our community. Please take the time to register and you will gain a lot of great new features including; the ability to participate in discussions, network with our members, see fewer ads, upload photographs, create a retirement blog, send private messages and so much, much more!

Old 05-02-2015, 01:41 PM   #2
Recycles dryer sheets
Theseus's Avatar
 
Join Date: Aug 2013
Posts: 484
Chris - I was myself actually searching on this topic and had not found much about it, but did find your recent post has been unanswered as of yet. My interest was due to researching how inflation/deflation might impact retirement finances. I view the volatile nature of inflation to be nearly as important, possibly even more important, of a concern when trying to do the educated guess calculations. Running a round number SS benefit scenario in FIRECalc with a ridiculously long retirement I pulled the SS and CPI numbers for 1935-2015 and found FIRECalc did show negative SS adjustments for 1938, 1939, 1949, & 1954, and 0% for 2008. From the SS site I did find reference to COLA adjustments having been introduced through legislation (which included the provision to not be adjusted downward) in 1975, and that prior adjustments were individually made through legislative action. I've not yet found a historical summary of what those pre-1975 adjustments were, I'm guessing they may have included downward adjustments for severe deflationary times?
Attached Images
File Type: png COLA.png (21.4 KB, 7 views)
__________________

__________________
Theseus is offline   Reply With Quote
Old 05-02-2015, 05:55 PM   #3
Give me a museum and I'll fill it. (Picasso)
Give me a forum ...
haha's Avatar
 
Join Date: Apr 2003
Location: Hooverville
Posts: 22,386
Quote:
Originally Posted by Chris M View Post
First, let me just say that I think FIREcalc is an excellent resource, and that Early Retirement.org is providing a tremendous service by making it available and maintaining it. The simulator's use of actual historical returns is far superior to Monte Carlo models that fail to account for mean reversion.

I'm writing because I think there may be an issue in the way the model handles Social Security payments during deflationary periods. I haven't been able to find any discussion of this in past threads. If it is an issue, (and I'm not sure it is) then FIREcalc may be significantly overestimating the size of withdrawals that should be taken from retiree portfolios during periods characterized by deflation. I've run a number of FIRECALC simulations and looked at the Excel results spreadsheet for a number of periods covering the 19th century, and the 1920s and 1930s, when we experienced deflation. It appears, based on the results I saw for Social Security payments in Column D, that these payments are being deflated in years characterized by deflation. My understanding is that, by law, the Social Security COLA adjustment cannot cause payments to decrease. Instead, payments are held constant in years when the CPI goes down instead of up. (In fact, this happened, I think in 2009 or 2010 (based on the deflation of 2008). It was in the news, because seniors complained that they weren't receiving a COLA adjustment, and I believe the Obama Administration actually floated the idea of a small upward adjustment despite the decline in the CPI.)

If this is an issue, it can make a significant difference in the numerator of the withdrawal rate. For example, from 1929 to 1933 deflation reduced the cost of living by about 25%. Say a person retired in 1929 with total expenses of $10,000, and Social Security of $5,000 (the example is hypothetical, since 1929 pre-dates Social Security). If Social Security is assumed to decline along with total expenses by 25%, then by 1933 we have the numerator in the withdrawal rate = 0.75 x $10,000 - 0.75 x $5,000 = $3,750. But if SS is instead held constant at $5,000, we have 0.75 x $10,000 - $5,000 = $2,500. Instead of a 25% reduction in the numerator, there is a 50% reduction.

I stumbled on the impact of the treatment of SS during deflation when playing around with my own (crude) historical simulator. I was running a test based on the above example, to see the impact of delaying SS from age 62 to age 70. I was stunned at what I saw for the 1929-59 time series. With the $5,000 SS payment delayed, it increased by 76% (plus the inflation that occurred after 1933) to $9,620, while total expenses declined to $8,248--so the numerator in the withdrawal rate actually went negative 8 years into retirement, and remained negative for the rest of the retirement! In other words, a retiree would no longer have to make withdrawals from his or her portfolio, but would instead be able to save money throughout retirement (or go to Vegas? )

Having looked at the FIRECALC results spreadsheets, and recalculating the SS payments during deflationary years, in addition to 1929-59 it appears that periods beginning in the early years of the 20th century (between roughly 1901 to 1920) could be significantly affected. This was a real rough patch for stocks (Panic of 1907, WWI and high inflation), but in my tests I saw that a number of retirements that began in these years and ran out of money in the 1920s and 1930s at a starting withdrawal rate of 5% were saved from bankruptcy when the SS payments were held constant during the deflationary years. The combination of the stock market boom plus the deflation during the 1920s helped keep the portfolios from failing, once SS payments were no longer deflated along with expenses.

I apologize if this isn't an issue or has been previously discussed.

Chris M
I was not able to closely follow your discussion. However, since as you pointed out SS did not exist in those years of heavy deflation, how does this matter? IMO, in the modern US of everyone wanting everything for free, and everyone being franchised to vote, how could there be true deflation of any meaningful amount, lasting for any meaningful period of time? I don't think it can happen, and I don't think it will happen- no matter all the Fed head fakes trying to keep us from going all in with trying to front run what they are clearly committed to doing.

Many people are expecting a conservative turn in the next election. Remains to be seen of course, but IMO a left turn is at least as likely. No matter how weak is the US economy, inflation continues to destroy the dollar. CPI Jan 2000: 168.8; CPI Jan 2015: 233.707.

However, I think it is a valuable service to do what you have done- try to find flaws or inconsistencies in the algorithms, and get to the bottom.

Software that people are well paid to troubleshoot is never flawless. So how could this work be?

Ha
__________________
"As a general rule, the more dangerous or inappropriate a conversation, the more interesting it is."-Scott Adams
haha is offline   Reply With Quote
Old 05-04-2015, 10:24 AM   #4
Dryer sheet wannabe
 
Join Date: Dec 2013
Posts: 21
Quote:
Originally Posted by Theseus View Post
Chris - I was myself actually searching on this topic and had not found much about it, but did find your recent post has been unanswered as of yet. My interest was due to researching how inflation/deflation might impact retirement finances. I view the volatile nature of inflation to be nearly as important, possibly even more important, of a concern when trying to do the educated guess calculations. Running a round number SS benefit scenario in FIRECalc with a ridiculously long retirement I pulled the SS and CPI numbers for 1935-2015 and found FIRECalc did show negative SS adjustments for 1938, 1939, 1949, & 1954, and 0% for 2008. From the SS site I did find reference to COLA adjustments having been introduced through legislation (which included the provision to not be adjusted downward) in 1975, and that prior adjustments were individually made through legislative action. I've not yet found a historical summary of what those pre-1975 adjustments were, I'm guessing they may have included downward adjustments for severe deflationary times?
Theseus, thank you for your reply. In answer to your question, the last long, severe bout of deflation we experienced ended in 1933. This pre-dated the introduction of Social Security, so we can’t know if or how Social Security payments might have been adjusted back then. However, even though FIRECalc is a historical simulator, keep in mind that if you input a value for Social Security payments then FIRECalc will apply those payments to all of the time periods it covers—even the early 20th Century and late 19th Century periods, when Social Security didn’t exist. In doing so, I think it should follow the COLA rules as they exist today—because the ultimate purpose of FIRECalc is to help us plan for the future, not tell us precisely what would have happened in the past (if the latter were the purpose, then Social Security payments should be reset to zero for retirements beginning in the early 1930s and earlier). If we were to experience an episode of severe deflation like that of the early 1930s at some point in the future, we know that Social Security payments would not decline. This is I think is really the key point. Deflation in the future would not affect retirees in the same way as deflation in the past (before Social Security) would have impacted them. We think of Social Security as protecting us against inflation, because of the COLA adjustment, but it will do an even better job of protecting us against severe deflation.

The legislative environment has changed since 1933. FIRECalc is applying that new legislation—the introduction of Social Security—back before 1933. But it isn’t handling the COLA adjustment correctly, according to the law as it exists now and will exist in the future. As a result it’s overestimating the probability that retirees will go broke.
__________________
Chris M is offline   Reply With Quote
Old 05-04-2015, 12:30 PM   #5
Dryer sheet wannabe
 
Join Date: Dec 2013
Posts: 21
Quote:
Originally Posted by haha View Post
I was not able to closely follow your discussion. However, since as you pointed out SS did not exist in those years of heavy deflation, how does this matter? IMO, in the modern US of everyone wanting everything for free, and everyone being franchised to vote, how could there be true deflation of any meaningful amount, lasting for any meaningful period of time? I don't think it can happen, and I don't think it will happen- no matter all the Fed head fakes trying to keep us from going all in with trying to front run what they are clearly committed to doing.

Many people are expecting a conservative turn in the next election. Remains to be seen of course, but IMO a left turn is at least as likely. No matter how weak is the US economy, inflation continues to destroy the dollar. CPI Jan 2000: 168.8; CPI Jan 2015: 233.707.

However, I think it is a valuable service to do what you have done- try to find flaws or inconsistencies in the algorithms, and get to the bottom.

Software that people are well paid to troubleshoot is never flawless. So how could this work be?

Ha
Ha, I agree with you that deflation is not likely (although I wouldn’t bet the security of my retirement that it’s impossible). But here is the issue: regardless of whether you or I think deflation can or can’t happen in the future, it has happened in the past. And because FIRECalc runs off of historical data including data for deflationary times, this means that FIRECalc is in effect assuming that it can happen again in the future. The probability estimates of portfolio survival calculated by FIRECalc take into account what would have happened to retirements that include the deflationary years of the 1920s and early 1930s. The problem (and I’m now convinced it is a problem) is that FIRECalc is not correctly handling Social Security payments during these deflationary years. As a result, FIRECalc is overestimating the probability that retirees will go broke.

To get a rough sense of how big the overestimations might be, I ran a test case. I ran FIRECalc using a 30-year retirement, total savings of $100,000, annual spending of $10,000, and Social Security payments of $5,000. So the initial withdrawal rate is 5%. FIRECalc indicated failure would occur 32 out of 115 cycles, for a success rate of 72.2%. I then pulled up the Excel result spreadsheets for all retirement start years that would have included all or some of the deflationary period from 1920 through 1932. (For example, I pulled up the spreadsheet for a retirement beginning in 1895, since that retirement would have extended up to 1925.) This way, I was able to identify those retirement start years that ended in portfolio failure, but that might not have ended that way had FIRECalc held Social Security payments constant during years with deflation. I identified the following 16 retirement start years (half of the 32 that failed) as having been possibly led to portfolio failure due to FIRECalc’s problematic treatment of the COLA adjustment:

1899
1901
1902
1903
1905
1906
1907
1909
1910
1911
1912
1913
1914
1916
1929
1930

Then, using the Excel results spreadsheet for each of the above retirement start years, I went in and fixed the Social Security COLA adjustment so that it would be applied only in years characterized by inflation. For years with deflation, I held the Social Security adjustment constant. I found that in 8 of the above 16 years (namely, in 1901, 1903, 1905, 1910, 1913, 1914, 1916 and 1930), the portfolio would not have failed had Social Security payments been held constant during deflationary periods. This means that, instead of the 32 portfolio failures shown by FIRECalc for my test case of a 5% withdrawal rate and 30-year retirement, the actual number of failures should have been only 24, and the probability of success should have been 79.1% instead of 72.2%. FIRECalc thus overestimated the number of failed cycles by 8/24 or 33%, and it underestimated the success rate by 6.9/79.1, or 9%. These error estimates apply only to the specific test case I ran--the errors will vary depending on the relative values of the Social Security payment, total spending, total savings, and retirement length.

Again, I think overall FIRECalc is a great tool, and the issue I'm pointing out in no way changes my view of that. I agree with you bugs are possible in all software, and my finding of a 9% underestimate of the success rate suggests this particular problem, though not insignificant, is also not huge.

Beyond this I’m not sure what more to do with this information. I’d be happy to share my detailed calculations and help whoever is in charge of maintaining FIRECalc to look into this issue more deeply, if desired. But failing that, I'm hoping to at least alert users of FIRECalc to this issue. It is probably best, in light of all this, to view FIRECalc's estimates of success rates as being conservative--and conservative is not a bad thing to be, when planning for retirement.
__________________
Chris M is offline   Reply With Quote
Reply


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Firecalc Second Spreadsheet Issue? wduff FIRECalc support 0 09-25-2014 04:17 PM
New Issue and Reopened Issue Bonds CaptainO FIRE and Money 15 08-22-2013 11:16 AM
TIPS funds and deflation wabmester FIRE and Money 26 12-18-2006 06:06 PM
Good deflation Arif Young Dreamers 5 02-10-2006 01:11 PM
Possible Deflation Scenario Cut-Throat FIRE and Money 13 08-30-2003 02:59 AM

 

 
All times are GMT -6. The time now is 01:29 AM.
 
Powered by vBulletin® Version 3.8.8 Beta 1
Copyright ©2000 - 2017, vBulletin Solutions, Inc.