Thursday, July 23, 2015

Summary of 2015 Trustees Report from the Committee for a Responsible Federal Budget

The CRFB has produced a detailed but still readable summary of the 2015 Social Security Trustees Report, which was released yesterday. The report projected a small improvement in the program's finances versus the 2014 Report, with the combined trust funds projected to last an additional year and the long-term actuarial deficit slightly reduced.

The the Trustees still project that Social Security isn't close to financially sustainable without reforms. In particular, the Disability Insurance trust fund is projected to run out next year.

Click here to read their whole report.

Fig. 1: Social Security Revenue and Benefits (Percent of Payroll)


WilliamLarsen said...

“Advocates for seniors say that gives policymakers plenty of time to address both programs without cutting benefits. But some in Congress note that the longer lawmakers wait, the harder it gets to address the shortfall without making significant changes.”

Plenty of time, what does this mean? How long does it take to save for retirement? The longer you wait, the more you have to save each year until you retire. In other words it gets harder the less time you have. So to put “plenty of time” into perspective here is how long it takes an individual to save.

The assumptions: Retire at age 67, life span 22 years, US Treasury Rate 5%, CPI 2.5%, Wage growth 2.0%. The percent to save of your income every year until age 67 to replace 100% of your life time indexed (wage growth) wages (the same as SSA-OASI benefit formula) is the following starting at the specified age.

Age 21 save 19.4%, 25 – 22.4%, 30 – 27.3%, 35 – 33.8%, 40 – 42.9%, 45 – 56.2%, 50 – 77.6%, 52 – 90.2%, 53 – 97.9, 54 – 106.8%, 55 – 117% and at age 61 -252.6% of wages.

Clearly the longer you wait to start saving for retirement, the more you have to save. However, this is on an individual level and some will live longer than 22 years and some will live far less.

How long does the United States have to save as a country in terms of Social Security? The analysis is only slightly different than that of individual. 22.0 years past age 67 is the projected average life expectancy of those born in 2000. With an individual there is a great range, but as you add individuals and it becomes a group, you now have shared risk. Social Security is a shared risk. This means the cohort of 2000 as a shared risk pool need to save 19.4% of their collective wages to fund their Social Security benefits in the future to replace 100% of their wages. However, SSA targets to replace 41% of wages which would require a payroll tax of 8.0%. The OASI tax is 10.6% which is 33% greater than is required.

Social Security already has 43 million retired beneficiaries and about 160 million workers. There ages span from age 16 to over 100. Generally those over age 65 are retired while those under are working. Taking the number who are retired multiplied by their collective years left to draw and the number of workers multiplied by their number of years left to work provides a weighted average as to how many years the United States has to save for retirement. Social Security is viewed as a retirement fund. The Trust fund is the amount that has been collectively saved by workers to fund their retirement and covers less than 5% of liabilities. Collectively there is less than six years to save. In simple terms, if Social Security were an individual who wanted to retire at age 67, it would be like the individual who is age 61 and is just now starting to save. It would require a payroll tax of 103.6% over the next six years.

Some will say we can reduce the tax and spread it over more years. The problem is that spreading the cost over more years requires borrowing and borrowing requires interest payments which adds to costs. There is no free lunch.

The charts show SS-DI, but SS-OASI has the same issue on a larger scale. Though 2034 is 19 years away, collectively we only have six years to fund SS-OASI and one year to fund SS-DI. If not why try, it only will get worse. This is exactly what happened to Greece.

“There is an easy fix available for the disability program: Congress could shift tax revenue from Social Security's much larger retirement fund, as it has done in the past.”

Do you think taking from a fund that is already insufficient is going to help?

Arne said...


I created a spreadsheet to replicate your results. I got the same numbers and found that they are insensitive to CPI and wage growth. To add one more number, at the 10.8 percent set-aside rate (with 2.5 percent real ROI) the expectation is for a 56 percent replacement rate.

You will recall that Andrew was concerned that the replacement rate provided by the SSA is not used the way most people would (or the way our spreadsheets do). Andrew reports it is around 54 percent for a "stylized medium earner".

The spreadsheet tells us 61 percent for a lifetime minimum wage earner ($10/hour), so SS is a good deal for those who need it most.

Arne said...

"spreadsheet tells us 61 percent"

Sorry, I added the bend points to my spreadsheet, but what I meant was that it you assume the Primary Insurance Amount is the same as the final salary and use the benefit formula, you get 61 percent.

Arne said...

Such an interesting spreadsheet you have caused me to create. If I assume an $60K salary for a 42 percent replacement rate along with working from age 26 to 62 with one year off (for job changes) and claiming at 67, I get that I need to set aside 11.6 percent.

For a median earner who went to grad school and had a realistic wage history, SS is still a really good deal.

Andrew G. Biggs said...

This paper from a few years back doesn't focus on lifetime returns, but we did find that for new retirees in 2005 the median IRR was about 3.5 percent. So a decent deal at the time, given that it's a safe benefit. Going forward you've got the rising retirement age and obviously the solvency issue which, however you solve it, will end up lowering returns.

For people entering the workforce today, the payable benefits level produces an IRR (for a typical stylized couple consisting of a medium and a low earner) of 2.43 percent, according to the SSA actuaries. But their stylized workers have long working careers and aren't exactly average earning, so the payable benefit level for an actual median household is probably a bit above that. Not terrible for a low-risk benefit, though still under the projected government bond rate (to the limited degree such comparisons are useful).

WilliamLarsen said...


A 2.43% return when you have a mortgage is a terrible way to create wealth. In essence you loan SS-OASI money at 2.43% while paying 4%. It can get worse, those who have student loans or credit card debt could easily see an effective rate of return significantly higher than 5%.

WilliamLarsen said...


I actually us the SSA formula when I determine the initial OASI benefit. If using a spreadsheet, it is easy enough using vertical lookup tables to pull any given cohorts bend points as well as the SSA wage index. It is also easy enough to use a random generated number to provide a variable wage though this may not mimic life, I have found that the law of averages after several hundred thousand calculations do not show an appreciable difference in the overall result.

When dealing with cpi, treasury rates and wage growth, there is an equivalent rate of return for working years and for benefit years.

Dealing with compound gradient formulas (two exponential growth rates cannot be mathematically subtracted linearly as many would).

The effective rate of return is (1 + return)/(1+replacement rate)-1

The replacement rate during working years is wage growth. This is the rate at which wages are growing year after year. The higher the rate of wage growth, the higher the payroll tax must be to pay higher benefits in the future. The replacement rate as a beneficiary has nothing to do with wage growth, but CPI which is COLA. The key is not the actual cpi or US treasury rate but the effective rate of return. As long as the effective rate of return remains fairly constant, it will and does cover a wide variation of cpi and corresponding treasury rates. However, the most critical value is wage growth.

The replacement rate brackets are 90%, 32% and 15%. Once you get past the first bend point, the sensitivity of earning more in wages only adds 32 cents on the dollar. Yes you can have a wide range of overall benefit rates, but looking at the overall benefits some have high replacement rates approaching 60% while others with high wages have low replacement rates. The rest of those fall within 3 sigma.

I have the same computer model I had in 1984. I have added new SSA population distribution data to my data base i use to determine potential workers and potential beneficiaries in any given year. The surprising thing is that my calculated date that SSA would be unable to pay scheduled benefits was 2036 - 2037. It remained this way for over 20 years. I do not tweak my assumptions. I add the new trust fund balance and that is it. Today is shows 2026-2027 because I reduced the labor participation rate. Because most of this reduction is due to people in their 50's losing their jobs, their overall initial SS-OASI benefit will not decrease that much, it is basically fixed due to the number of years they worked and the fact that when you reach age 55 your wages normally grow slower than the Average SSA wage base. This means that previous wages when indexed are worth more than current wages in many cased.

I will revisit my math and look for errors. Your comments are appreciated.

Arne said...


I am quite sure from your previous comments that you understand real rate of return and that you believe that the real rate of return for SS is negative, so I understand that you did not take Andrew's 2.43 percent as real return. But it is. And it matches my estimate.

Arne said...

"The effective rate of return is (1 + return)/(1+replacement rate)-1

I do not understand this. From the following paragraph "the replacement rate during working years is wage growth" and the one after "the replacement rate brackets are 90%, 32% and 15%" it seems you have more than one meaning.

Andrew has suggested that there are two definitions of replacement rate.
1- Final working salary divided by benefits.
2- Primary Insurance Amount divided by benefits.
From this comment "when you reach age 55 your wages normally grow slower than the Average SSA wage base", I can see you also know PIA is often higher than final salary, but I cannot understand what you are doing with the above formula.

WilliamLarsen said...

years ago when I was trying to figure out how much it would take to save and buy a boat when inflation as double digits identified that simply subtracting inflation from interest rate each year did not work. It was obvious why; two different compound growths. Sure for a back of an envelope calculation less than 5 years you will get a good number. However, when you you look past 5 years the error is compounded. This is compound gradient formula. It is mathematically precise/exact.

As wage growth (replacement rate while working) increases, the payroll tax must also increase exponentially unless both the income rate of return and the cola (replacement rate drawing benefits) which determines the effective rate of return remains the same.

If you go to you can find my derived formula as well as two java script retirement calculators; one how much you need to save and the other how much can I take out.

When I look at the IRR for SS-OASI using the 10.6% combined payroll tax (employers pay half thus increasing cost passing it on to workers who buy their products and services, or profits go down which reduce dividends and share price to owners - 401K, IRA, Mutual funds, etc) and the US Treasury RATE long term paid to OASI, I find that the payroll tax to duplicate the OASI benefit can be as low as 4.5% and as high as 9% depending on max and low wage earners. I use the life span at full retirement (some live longer and some live shorter while others do not draw anything. I compared a actuarial calculation to using the life span at Full retirement age and found that the difference was less than 0.5% which I believe is well within the realm of any errors in assumptions.

When I say that those born after 1985 can expect 29 cents on the dollar from OASI for each $1 combined payroll tax and income at the US Treasury Rate) I am using payable scheduled benefits under current law.

Current law reduced COLA to zero when the trust fund reaches less than 20% of any given years expenses. Since the trust fund is reported to be exhausted in 2033-2034, COLA will be zero. In addition using the SSA population file I was sent identifying the number of wormen and men from age 0 to 110 in groups of married, never married, widowed and divorced, I have created a worker to beneficiary ratio or more accurately the potential number of workers and beneficiaries by age. Based on this I calculate each cohorts initial OASI benefit which produces the each cohorts cost for every year. From this I hope to determine life time wages thus life time payroll taxes and determine the theoretical benefit and compare that to to OASI's payable scheduled benefit.

As for the OASI benefit it can be found at

The formula I derived using replacement rates because that is what I defined them in 1975, because the 1977 OASI benefit formula had not yet been created. you are correct that my replacement rates used in my formula are totally different from those in the OASI PIA formula.

My formula is used solely to gage the value of the SS-OASI benefit. Is SS-OASI a good deal? It is my opinion that OASI was a great deal to my parents. My dad and I disagreed over SS. To me and my siblings it is not a good deal and it is even worse for my children. I am a boomer.