Monday, October 30, 2017

New article from the CBO: "Measuring Retirement Income Adequacy"

The Congressional Budget Office has a very nice new study titled "Measuring Retirement Income Adequacy: A Primer," which outlines the economic theory behind retirement income adequacy and the choices of calculation you need to make when applying that theory to data.

The study hits on a number of issues I've discussed in how to measure replacement rates, which are a key shorthand for measuring retirement income adequacy. I appreciate that the CBO cites my work in a couple of places.

The basic theory of retirement saving is the so-called "life cycle model," which -- in simplified terms -- predicts that people will tend to spend the same amount from year to year.

Two key points I'd make regarding how to measure replacement rates, which represent Social Security benefits or total retirement income as a percent of pre-retirement earnings.

First, pre-retirement earnings should be calculated in real, inflation-adjusted terms. These allow you to compare the buying power of retirement income to the purchasing power that the retiree had when he was working. That's how the life cycle model would tend to see things. Social Security's actuaries, by contrast, compare retirement benefits to the "wage-indexed" average of pre-retirement earnings. This overstates the real purchasing power of the retiree's pre-retirement earnings and inappropriately raises the bar on what counts as an adequate retirement income.

Second, if you're calculating replacement rates using administrative data -- meaning, real earning records rather than stylized earners -- you're faced with the issue of whether to include years of zero earnings in the measure of average pre-retirement earnings. The life cycle model says that you should: if people smooth their consumption across years, that means that their average spending will be a function of all their years of earnings, including years of zero earnings. The SSA actuaries include 'zero years' when they calculate replacement rate relative to career-average earnings. But when they calculate replacement rates relative to 'final earnings' -- meaning, earnings in the years approaching retirement -- they exclude zero years. Doing so raises the measure of pre-retirement earnings, and so makes Social Security replacement rates look lower. The actuaries' argument is that there are too many 'zero years' in the years approaching retirement. But as I showed using the actuaries' own data, zero years aren't that much more common in the years immediately preceding retirement than they are earlier in life, when people may leave the workforce due to education, unemployment or child raising.

Where does the rubber meet the road? Well, if you were to ask SSA, they'd tell you that the average person receives a Social Security replacement rate of about 40% and that they need a replacement rate of about 70% in order to maintain their standard of living in retirement. Properly measured, I believe the average Social Security replacement rate isn't 40% but something in the 50-55% range. That helps explain why most retirees say they're doing well, even if they don't seem to have much savings on top of their Social Security.

In any case, the new CBO primer is highly recommended. Many commentators and journalists write about how much is "enough" retirement income, but the reality is that you can't really know what your opinion is until you wrestle with the sorts of choices that the CBO lays out.

5 comments:

WilliamLarsen said...

"First, pre-retirement earnings should be calculated in real, inflation-adjusted terms. These allow you to compare the buying power of retirement income to the purchasing power that the retiree had when he was working. "

I would disagree with this. Using inflation would not provide you with the same level of spending in retirement as it did while working. "Replacement" rate is the rate at which you wish to escalate some present value by. SS-OASI benefits are escalated by the CPI. SS-OASI initial benefits are calculated using "Wage Growth' to determine the equivalent value at age 60.

Using inflation as your "replacement" rate allows the person to replace that basket of goods available in the year from which you begin using inflation to escalate your "theoretical" future needs. So any improvement in medicine, electronics or other goods and services which did not exist would not be available in terms of income to purchase. You would then need to make a trade off in where you would spend your money. A prime example is the mobile phone. I still use a land line (copper) and it costs me $29 a month with unlimited calling in the US. I am still using the same $19 phone I bought (GE) in the early 80's. The new I-phone I heard costs $1,000 and then you need a carrier. Clearly this new product is outside the CPI.

After keeping track of my spending since 1968 with all of it in digital archives since 1987 my experience is that you can use past spending to accurately predict next years spending. I never used inflation. I used the average of my wage maximum[average(growth, bank rate and cpi), wage growth] and I hit my budget within +/-0.2% a year.

WilliamLarsen said...

"if you were to ask SSA, they'd tell you that the average person receives a Social Security replacement rate of about 40% and that they need a replacement rate of about 70% in order to maintain their standard of living in retirement. "

This is not convincing to me. First of all, what is the base cost of everything you buy up until you retire? Housing is a large cost if you buy (purchase price, financing, insurance, maintenance, home size-family size-overall cost, vehicles/transportation, groceries, taxes, FICA taxes and more)

I have five kids with one still home. Cost of groceries are down 70%. Cos of utilities are down 40%, Retired - vehicle costs are down 60%, Not working eliminates 7.65% FICA tax. Not working eliminates 1/2 my clothing costs.

The single larges reduction in costs is housing. Do I look at my house as an investment or an expense? I do not pay rent, but I pay for maintenance at 2% a year and if a home is 3 times your income, that would be about 6% plus property tax at 1% of the homes value (=>3%). I view a paid off home in terms of an equivalent worth analysis. What would it cost me to rent v own? Clearly it cost much more to own over the years than to rent which increased my expenses during those years, reducing my savings. Now the home is paying maybe a dividend by being cheaper than an apartment? Then of course I am no longer saving for retirement in terms of taking money out of my pretax wage at the rate of 17% since 1984.

So adding up the expenses I no longer have that were paid for out of my income during my working lives I get a value much lower than 70%.

No FICA tax ... 7.65%
No retirement savings ...17%
Housing normally is 20% of income retired mine is <11% while retired, but was 21% while working. so a net reduction of 10%.

Eliminate the kids expenses ..... 15% minimum

I think a more realistic value than 70% is closer to 50%. Now I do travel more than I did while working just because I have the time to do so. But is traveling now based off past years spending experiences or is this a new category that was not available to me while working?

In the end everyone is different. I think government trying to determine what the right amount of SS at this point in time is, is just too late. It should have been done in 1935 if this is the point of SS. But the point of bringing up "Replacement" rates now with ten - twelve years of funds left in the SS-OASI trust fund is to rationalize a new goal for Social Security because the current goal of Social Security is unrealistic.

SS's value that 70% is what a retiree needs in terms of retirement income replacement and a new value being put out that is may be closer to 50% means that there is room to reduce Social Security benefits: change the rules, change expectations, and more.

Growing up no one liked the ruled changing once the game started. Every kid I knew that it was unfair.

Arne said...

"average of my wage maximum"

I don't understand what you mean.

Andrew G. Biggs said...

Arne, To be fair, 70% is a common figure among financial planners, which SSA cites. Some analysts argue for something lower than that, while others go higher.

I'm not sure where I wrote "average of my wage maximum," but it sounds like typo.

Arne said...

Sorry, Andrew. That was part of William's comment.

My own salary history rose rapidly for 10 to 15 years and dropped to less than CPI after that, so I understand that wage indexed how wage indexing produces a PIA higher than final wages. I don't have any zero years, but I did have zero wages for periods between jobs, so my history does not match a stylized worker.