CBO: Social Security Policy Options (2015)
Social Security Policy Options, 2015
Update of 2010’s Social Security Policy Options
36 Options for Social Security. Scored by CBO. Numbers! Graphs! Open Thread! (Because I haven’t read it yet)
Social Security Policy Options, 2015
Update of 2010’s Social Security Policy Options
36 Options for Social Security. Scored by CBO. Numbers! Graphs! Open Thread! (Because I haven’t read it yet)
Krasting actually got to this first on the open thread.
http://angrybearblog.strategydemo.com/2015/12/open-thread-dec-15-2015.html#comment-2714975
Credit where credit is due.
Figure 1 which compares the options head to head is astonishingly hard to interpret without aid. Even if it didn’t require putting a ruler up against the screen to figure out the number represented by each option.
So a little key. CBO projects an actuarial deficit of 1.4% of GDP and each option is scored against an axis that runs from -0.4% (adds cost) to 1.2% (reduces costs/adds revenue)
So in order to get a total fix you would need to combine two options adding up to 1.4% while ignoring interaction effects. For example the straight ‘Lift the Cap, No New Benefits’ gives you the biggest bang. (As of course it would – all revenue no cost).
There is no option that corresponds closely to Coberly’s basic plan that underlies Northwest – steady phased in increases of around .2% of payroll over enougy years to meet the gap. Instead the Coberly/NW Plans would be an extension of Option 2 and an intensification of Option 3, with the effect of filling that much more of the gap.
Although we should note right up that Coberly and NW whether considered as one plan or two assume SSA numbers and not these much more pessimistic CBO ones.
So lots to chew on. Bon appetit!
Webb – Pessimistic CBO? How about Optimistic SSA?
If you start with a 4.4% I&P then the NW plan would have have an increase of 6%. This would mean a 0.2% increase every year for the next 30! Clearly not a viable plan.
As I’ve said to you for years, there is no single solution for SS. It has to be a combo approach. The combo solution will change the program forever.
Anyway, this report will not even be looked at. Nor will the one in 2016. Nothing will happen to SS until at least 2017. There will be only 12 years left to construct a fix by then, and the I&P will be north of 5%.
It is interesting that everything is calculated over the 75-year period, but the exhaustion of the Trust Fund is much sooner. The tax increases would have immediate,positive effect on the Trust Fund. The benefit reductions, however, take a long time to produce their effects (because current and near beneficiaries are excluded). So I think a combined solution would have to include significantly more tax increases than benefit cuts to avoid Trust Fund exhaustion. This should be more true the longer we wait.
Krasting by accepting CBO at face value you are totally rejecting early evidence as discussed in my piece from 2004, err I mean my piece from LAST WEEK.
http://angrybearblog.strategydemo.com/2015/12/us-life-expectancy-flat-for-third-year.html
Last year CBO drastically changed its projection for mortality trends. Okay fine. But these things are subject to tests of logic and data. And CBO didn’t give any more cogent argument for their change than “The future is unknowable so lets assume mortality improvements will continue in the future as they have in the past”. Okay that was literally debateable on issuance and duly debated. And because the future is unknowable in detail still so. On the other hand the future started the day after CBO changed its assumptions and so became the past and testable. Projection turned to data. And what did the (admittedly very early data) show?
“Life expectancy in the United States has stalled for three straight years, the government announced Wednesday.
A child born last year can expect to make it to 78 years and 9 1/2 months — the same prediction made for the previous two years.”
If this trend continues then we can stick CBO’s I&P of 4.4% on the dustbin of history.
You are free to disagree. But not free to simply dismiss my adherence to SSA numbers as being Pollyanna-ish foolishness. Exactly nothing validates your adherence to CBO assumptions. Except your proven tendency to grab onto every single date point that seemingly validates your position.
Hey how is that Atlanta Fed growth projection tool that you touted so highly working? Did they nail 3rd quarter growth in the way you asserted that they must? Or did you move right to “Nothing to see here folks, move right along” messaging after the car crash? Odd how things that were presented as dispositive instead become disposed of.
Mike B all of that is true. But a couple of caveats.
Under all projections the situation improves in the second half of the 75 year window, that is ‘crisis’ is front loaded and the numbers wouldn’t change much if we were using a 38 year window that took in the projected date of Trust Fund depletion in 2029 or 2034. It is worth keeping the 75 year window just to keep historical projections comparable but it doesn’t in itself really effect the policy options in the meantime. Because if they work at all on the 75 year number they will mostly work for the 38 year number.
The second objection is to the “sooner is better than latter” and “the longer we wait argument”. This has in fact been a constant theme of the Trustees Reports ever since I started reading them. In 1997. Every year the Trustees tell Congress that they have to move on this SOON. Like YESTERDAY. And the natural assumption is that years of inaction and foregone revenues and cost savings would have balooned the future cost in the way say BK projects.
But in point of fact they haven’t. In 1997 the Trustees projected Social Security Trust Funds to go dry by 2029. After 18 years of inaction they now project it good until 2034. And CBO had to radically change its demographic projections to just get their date back to that same 2029. What this means is that 18 years of inaction on Social Security not only cost us as a society nothing in future costs, it left money in our pocket from the ‘non-fix’.
Which is why years ago I published a series of posts with the titles “Nothing as a Plan for Social Security” and “The Cost of Inactivity”. In the later I presented a little gambling game where you could bet the cost of inactivity in the current year against the projected future cost in the next year. In Krasting’s terms should we have taken an I&P increase in payroll of 4.4% last year? (or accepted equivalent future benefit cuts). Or was it worthwhile to pocket that 4.4% and bet that the future value of Inactivity didn’t swamp our short term tax savings. And on balance the Cost of Inactivity since 1997 has been decidedly negative, the problem has not been getting bigger and more expensive, the worst you can say is that it is getting closer. But the fact remains that if we had taken the I&P approach back in 1997 we would have vastly overshot the revenue needed even as we had every worker take an immediate hit in the pay-stub.
That is my plan of ‘Nothing’ as promoted a decade ago has proven the be the numerically proven plan. Workers have pocketed a decade of foregone FICA increases (and in this age of stagnating wages that is a big deal) at a Future Cost of ‘Nothing’. Not a bad deal.
Yet all the way I have had lil’ Krastings nipping at my heels and telling me I was ignoring clear reality because “Even the Trustees say—” or now “Even CBO says—“. Well too bad, so sad. I ignored them for the last decade, or better discounted their projections in light of something I call “numbers” and “data” and am on the whole fairly satisfied.
One other thing I’d like to learn more about is the likeliest effect of an eventual increase in the minimum wage. This country is due a significant pay increase and I expect it to become a serious issue in the coming election.
I know Bruce Webb has commented in the past that these effects are difficult to model and likely to amount to a wash – in that the increased contributions helping with solvency also drive higher benefits at retirement. Okay.
But I wonder if the stall in life expectancy cited in another post might reasonably be modeled to have a more significant impact at the lower end of the income scale. So higher wages for minimum wage workers actually end up contributing to the overall solvency of the plan partially due to their unfortunate lower likelihood to draw benefits.
Net net I believe public opinion polls already support a significant and sustained increase in minimum wages. A projected improvement in SS solvency/sustainability could bolster those economic arguments.
“Live better now and have a more secure retirement to boot!” Senator Bernie Sanders white courtesy phone please….
The “Cost of Inactivity” Game was a little complex. Because the bet depending a lot on your demographic cohort.
Lets take an extreme example. The I&P cost of a Social Security fix is 4% in the current year. Gambler A is one year from retirement, Gambler B 45 years from retirement. We do nothing and the next year I&P cost is 4.06%. Which is in fact the baseline, you get .06 just from changing the baseline of the 75 year window. Gambler A pocketed his 4% of foregone tax. Clear winner. Gambler B pocketed his 4% at the cost of 0.06% for each of the next 45 years assuming a new policy of moving on Social Security. How many years does it take for Gambler B to have lost this bet once you figure in the PV of pocketing that 4%.? Well the specific answer to that is complicated and beyond my skills. But not beyond those of any financial analyst with a financial calculator and a solid economic projection model.
But what if the next years I&P stays at 4.0%. You don’t need any calculator at all. Everybody is a winner for that year. Moreover if you assume the P (‘permanent”) in I&P the closer you are to retirement the greater your savings/winnings. Because we know that once they ratchet up FICA there will be no clawbacks. Congress won’t give up that free borrowing from the Trust Fund.
It is a complicated game. But one that Boomers clearly won over the last 20 years. They didn’t bow down to I&P and so pocketed a lot of foregone FICA. But didnt’ really leave anyone in the bag. Gen-X is that much closer to retirement and have exposed themselves to not that much downside risk, while Millennials have at worst held their own. The game is risky, because just like Roulette you have the problem of the 0 and 00. The Trustees tell us that all things being equal the simple cost of doing Nothing increases by 0.06% a year. Which eats away at the present value of doing Nothing, in gamblers terms that is the ‘vig’. On the other hand every year that pocketing the I&P and its FV exceeds the next year’s I&P is a year that betting on ‘Nothing’ paid off. Even though betting on Nothing is going against the House and the apparent Iron Law of Probability. Yeah unless the dice are in fact loaded YOUR way. Something you learned by observation and analysis. Then you win.
“I know Bruce Webb has commented in the past that these effects are difficult to model and likely to amount to a wash – in that the increased contributions helping with solvency also drive higher benefits at retirement. Okay.”
Well you don’t know that. Because while I have conceded SOME offset I have never gone all the way to ‘wash’. On the other hand some of my friends and collaborators like Arne have placed more weight on that offset. And have better math skills than me. But haven’t convinced me.
Because on my analysis of the numbers (primitive as that might be) the key variables for Social Security solvency are Real Wage and Total Employment. Each drive cost via higher retirement benefits (but that is a big beneficial feature rather than a bug) but each drives up revenues even faster. Which leaves the question of whether Minimum Wage increases actually drive UP Real Wage and Total Employment. History and logic say yes, ideological theory says no: increased Minimum Wage causes enough offsetting losses in Total Employment to swamp out Real Wage. Hmm, well a little hard data would be nice here. Because that is a mighty convenient conclusion for the employers whose profits would be squeezed by increased labor share and whose only defense agains “Who gives a shit about your profits” is “Hey we are looking out for your best interests by restricting pay increases because somebody might lose a job and Rising Tides lift all Boats”.
Unless your Boat has in fact been stuck in the mud since 1980 and that tide is now lapping over the gunwhales.
So no, my proposed policy of MJ.ABW More Jobs. At Better Wages. Has plenty of room for minimum wage increases and direct federal investment in jobs. Each of which I am convinced will drive Social Security solvency quite directly.
Bruce – Actually, I’m fine with ‘do nothing’ for now, since I prefer tax increases to benefit cuts and I think that is more likely (assuming doing something is ultimately required) the longer we wait. However, I think the values of the various potential fixes (especially those with benefit cuts) using a 38-year horizon would be very different than the 75-year numbers. The benefit-cut options are almost always increasing over time in terms of their annual effects on the balance – often more that linearly. (For example, Table 2, option 18, the difference between outlays and revenues is 0.8% of GDP in 2050 and 2.4% in 2080. For the 75-year period, it is 1.1%. All values before 2050 are less than this, so the value for the 35-year period until 2050 should be much less than 1.1%, I think. By the way, I think Table 1 has an error for this option, only showing the disability bar and not the longer OASI line.) That is not true for most revenue increases, which tend to be flat (except for option 3 which has tax increases spread over 60 years). So for them the shorter horizon wouldn’t have much effect. Therefore, I think using the shorter horizon would make the tax increase options relatively more attractive.
Webb – are you fooling with me again? another of those ‘bloody chum’ comments you are fond of? You say:
Hey how is that Atlanta Fed growth
projection tool that you touted so
highly working?
Huh? Either you are being stupid, or baiting. The Atlanta Fed does not forecast GDP. The Atlanta Fed provides a running summary of the components of GDP as and when each data point is released.
It has been remarkably accurate in measuring GDP in real time. This info is used by the Press again and again as it has been so accurate. For you to think anything to the contrary shows your lack of understanding of what this is about.
Oh – and since you bring up the Fed, please note that the Fed increased rates today by RAISING IOER, exactly as I said they would months ago. I hope you recall disputing this with me back then.
With each future increase of % rates the IOER will rise. As it does, it will cost the Fed more to maintain its inventory of bonds, The result will be that the Fed’s income will decline.
Do us a favor, man up and admit that you had this wrong.
Screw you Krasting.
You touted a particular tool from the staff at the Atlanta Fed staff called GDP Now https://www.frbatlanta.org/cqer/research/gdpnow.aspx?panel=1 and tried to claim that its projection of 3rd quarter GDP was dispositive and that anyone relying on any other projection was a moron.
If you would like to bring back the particular forecast you were gleefully relying on a few months ago and compare it to actual 3rd quarter GDP and claim that it validated your position then fine. Show me the money.
You say “it has been remarkably accurate”. Yeah show us. Because you are the one who is wrong here. Or you can prove otherwise. Because your bare assertions are worth nothing.
Plus you were alone among about 10 gazillion ‘forecasters’ in projecting this rate increase. And you will not find a single word from me saying that there would be none such by the end of this year.
Don’t dislocate yourself patting yourself on the back because you predicted the sun would rise at least one day this week.
You say “The Atlanta Fed does not forecast GDP”. But since official GDP numbers are reported retrospectively and then revised a self-described tool called “GDP Now” is by definition a projection. You are typically doing a bunch of logic slicing and salad spinning. But just making a typical hash of things.
“18 years of inaction on Social Security not only cost us as a society nothing in future costs”
While it may have left us with more money in our pockets, the “nothing in future costs” is logically and analytically wrong. There would be more in the TF if taxes had been raised earlier. The exhaustion date would be farther out.
I suggest that if you put your nothing bet into a spread sheet and run it for all the years where the P&I number increased, you might find many more losses than you think. Although I understand that you did not intend to cherry pick, that fact that you started looking in 1997 means that you did so anyway.
“the key variables for Social Security solvency are Real Wage and Total Employment”. I disagree. It comes down to the ratio of workers to beneficiaries. Improvement in Real Wage improves the ratio that the program needs, but life expectancy changes W/B faster.
A decade ago my analysis (based on IC numbers) said we needed to have tax increases in the early 2020s. Learnings since then tell me we should have increased DI by itself some time ago, but otherwise IC has been a better projection than I first thought.
Given who was in Congress, nothing has been as good as we could get, but your numerical analysis is not convincing to me.
Arne:
And no productivity gains? This was always the argument to this: “the ratio of workers to beneficiaries.” I do not know it having changed much.
So the core assumption is that workers will not be able to produce enough goods and services for themselves and retirees at some point in the future? Furthermore, we won’t be able to import any shortfall in goods at some point in the future?
Folks, Social Security is solvent unless you can prove the above will happen. CBS estimates are baselines not macro projections.
“While it may have left us with more money in our pockets, the “nothing in future costs” is logically and analytically wrong. There would be more in the TF if taxes had been raised earlier. The exhaustion date would be farther out”
Arne you are the engineer, I am the medieval historian, so lets just say the computational chops are heavily weighted in your direction.
That said it seems to me you have built in some assumptions that it would have been a better idea to have more income in the Trust Fund and so more money lent to the Federal government by workers than we had. I don’t believe this to be correct because of the distortionary effects of interest on interest and believe that the 1983 deal would have been much more sound in the payroll tax increases had been slightly higher and more spread out so that the TF Ratio never got as high as it has. That is if Social Security had stayed on a Pay Go rather than prefunded (in part) basis.
But my point about no harm was a little different. From 1997 (and it is not entirely arbitrary) to 2005 the cost of an extended fix from a standing start actually declined. That is if we had put Northwest in place in 1997 we would have been adjusting the schedule of future FICA increases down each year. And might have made the choice of delaying the start of the series. Which in fact is the actual result. We didn’t start Northwest (or any other plan) and the cost per year going forward didn’t start going up until around 2010. That’s the productive money in our pockets that didn’t need to be committted to this purpose.
And I am late getting back to work.
“it seems to me you have built in some assumptions that it would have been a better idea to have more income in the Trust Fund”
Not at all. I do assume that is possible that Congress will negotiate a combination of tax increases and benefit reductions. In such a case, higher benefits could be paid longer if the TF had more money in it.
Run, by my analysis using Social Security report Intermediate Cost numbers, in the early part of the century it takes 2.9 workers to generate payroll taxes which will pay benefits for one worker. In the latter part of the century productivity improvements (and other factors?) will reduce that to 2.7 workers per beneficiary. MJ@BW will (as Bruce noted) increase benefits as well as wages, so it will have only a second order impact on the viable W/B ratio.
There are many factors impacting the adjustments that SS will need to stay in balance, but W/B is big. Longer lives is the biggest factor in the reports sensitivity analysis because living longer means more beneficiaries which is a first order impact on W/B.
I don’t disagree with Bruce about how to design adjustments to SS, or that MJ@BW is good policy. I just don’t think his analysis that the latter obviates the former is correct.
Arne:
This is where you and I are going to disagree. Productivity gains are more than just payroll taxes. You are bringing forth a Krasting and ABriggs argument on SS which is not true. We are not just talking payroll taxes, we are talking whole people being split between Labor and Capital which results in higher income.
The ratio of workers to retired is not a true measure and never has been. You are moving to the dark side. By that measurement, we would have been in trouble decades ago when the ratio was 10 or 20 to one and reduced to 5 to 1.
Arne has better number skills than me. In fact I do a lot of my arithi-math in my head limited to first and second order top down effects. Which means I get a lot of details wrong. But I got a fairly good feel for interactions.
On MJ.ABW. It may be true that it doesn’t do huge things for solvency when that is measured by the ratio of income to cost. But it does do some good things for the nominal and real result of Rosser’s Equation.
https://en.wikipedia.org/wiki/Rosser's_equation
This equation, named by umm me, measures the gap between scheduled benefits and payable benefits at Trust Fund depletion against real benefits earned by today’s retirees. And the result is surprising, so surprising that Prof. Rosser’s students failed to predict it in tests. At the point of Trust Fund Depletion current law requires a cut in benefits from Scheduled to Payable. And that cut has been fairly steady in the 25% range for years. And for both sides of the discussion, but for very different reasons, it is this cut that is the nut of “Social Security Crisis”
But what economists Dean Baker and Barkley Rosser Jr understood and to the best of their ability tried to MAKE understood is that the cut at Trust Fund Depletion would actually leave the bereft retirees with a basket of goods in real terms better than current retirees. By double digit percentages.
So in my mind the number that has real world importance is Payable as measured in Real Basket of Goods. And not so much Scheduled.
Now on a first order top of one’s head basis it is clear that a policy of MJ.ABW drives both Payable and Scheduled. What is not at ALL clear is whether it serves to close the gap. So a lot depends on whether you focus on the cut between Scheduled and Payable (the actuarial gap) or the actual level of Payable.
Me I go with Barkley Rosser and his eponymous equation (as named by yours truly). The real standard is Standard of Living compared to today. And the metric there is Payable. Which doesn’t mean shooting for Scheduled is a bad idea. Au contraire mon frere, I am all for it. Just that we should not allow everything to be measured by the actuarial gap between Scheduled and Payable while ignoring the actual level of the latter.
Does MJ.ABW drive Payable? I think that indisputable. Though God knows AB is full of people who have the chops (and imagined chops) to dispute it. Does it actually narrow the percentage of GDP gap between Scheduled and Payable? Maybe not. But we eat from the Basket of Payable. And the 12.4% slice taken from those MJ.AB Wages is what drives that.
Webb says above (re the cut to benefits at the year of TF exhaustion):
“that cut has been fairly steady in the 25% range for years.”
Not according to CBO……
December 16, 2015
Report
Under current law, CBO projects, Social Security’s trust funds, considered together, will be exhausted in 2029. In that case, benefits in 2030 would need to be reduced by 29 percent from the scheduled amounts.
In fact, CBO is saying that at exhaustion the cuts to DI will be 26% and the cuts to OASI will be 31%.
See exhibit 5 of:
https://www.cbo.gov/publication/51047
Krasting is 20-30% a “range” around 25%? Given the actual probability distribution around the IC projection from LC to HC?
You have gone from logic chopping to picking at nits here.
“By that measurement, we would have been in trouble decades ago when the ratio was 10 or 20 to one and reduced to 5 to 1.”
Take a look at the real numbers. https://www.ssa.gov/oact/TR/2015/lr4b3.html
W/B has not been above 5 since 1960, a mere 25 years into SS’s history. The ratio was high because we were in the start up period of a Pay As You Go program. W/B did not mean anything then, but it does now. It is unclear exactly how to compare W/B numbers over time since rules are still changing (retirement age not yet at 67), but by 40 years in it was down to 3.2.
When you aggregate the numbers you can determine how many workers payroll taxes it takes to pay each beneficiaries benefits. The number is not the same as the actual ratio of workers to beneficiaries, but a comparison of the two tells you (absent interest earnings) whether the TF is being drawn down or being increased. The TF balance does not care about fairness.
Arne:
You are misapplying the rule governing productivity gains as they still exist. Ratio of workers to retired has little to do with the issues we have today. You need to look elsewhere as to why it does not work as well today as it did then. The gains did not just disappear and they are still there. All your table does is show what I said about numbers of workers. Do you do throughput analysis? What took place?
https://www.ssa.gov/oact/TR/2015/V_A_demo.html#271410 Table V.A2.—Social Security Area Population on July 1 and Dependency Ratios,
Calendar Years 1945-2090 may be even more illuminating. Seniors are not the only people being supported by the working age population, there is also a metric called “Total Dependency” that includes children. You could call this the “mouths to feed per working age” metric. And it is never projected to be as high over the 75 year actuarial window than it was in 1960. (.889 in 2090 vs .902 in 1960).
Plus no one to my knowledge has ever folded in the number of beneficiaries in the 50s and 60s still receiving benefits under Social Security Title 1. All these numbers are based on beneficiaries under Title 2 what we know as Social Security today. But which covered fewer beneficiaries at a lower cost as late as 1950 and still had legacy effects a decade later.
The conservative solution always seems to be the reduction of wages and/or benefits to the workers (beneficiaries in the case of SS).
Run,
I am not sure what you are saying. I think we are looking at two different problem statements. When the basic question is “How many workers does it take to provide for each beneficiary” or perhaps whether that question is meaningful, what does a throughput analysis mean? The answer applies at a single point in time, there is a projection of a long term downward trend, so I don’t understand where the question “What took place?” comes from.
Arne:
When you improve throughput, productivity (gains) go up. My whole life has been cutting Labor to a refined level and now it is less than 10% of the total cost. Answer the question, where did the productivity gains go Arne. They just did not dissipate into thin air. Did they go to the left over Labor, did they go to Capital, were they left on the table, did the government take them. The resulting reduction in Labor resulted in value and the value for 3 today is the same as 30 – 30 years ago because of what?
Bruce,
In 2006, http://economistsview.typepad.com/economistsview/2006/09/the_myth_of_the.html, you said “By what I call Rosser’s Rule 75% of 160% means 120% better relative check than my Mom gets today.” The dark side of Rosser’s Equation is that that 160% gets lower every year until it gets to 100% and 75% of 100% is a cut.
I pulled up a 2012 spreadsheet that already has the inflation adjusted data,
https://www.ssa.gov/OACT/TR/2012/lr5c7.html ,
and added one more column:
1991 160%
1992 156%
1993 156%
1994 159%
1995 157%
1996 157%
1997 152%
1998 154%
1999 155%
2000 154%
2001 149%
2002 141%
2003 139%
2004 137%
2005 133%
2006 131%
2007 135%
2008 127%
2009 129%
2010 133%
2011 128%
2012 130%
2013 128%
2014 129%
2015 131%
2016 129%
2017 127%
2018 125%
2019 122%
2020 119%
2021 116%
2022 114%
2023 112%
2024 110%
2025 109%
2026 108%
2027 106%
2028 105%
2029 104%
2030 103%
2031 102%
2032 101%
2033 100%
Anything less than 133% means that right after depletion, those folks will not be doing better than their mothers. Sorry.
I took the basic concept and most of the numbers from Dean Baker and Barkley Rosser. And neither has backed off from this particular claim.
Perhaps Barkley will chime in hear as to whether you captured the right numerator and denominator. There is no question however that the result of Rosser’s equation had dropped significantly since 2004-5 when Trust Fund Depletion was projected to be in the 2040s. That is you are now talking a larger cut from a lower baseline. But even by your numbers a 25% cut equates to a zero cut in basket of goods terms compared to today. Hardly a case of my mother robbing me.
“Hardly a case of my mother robbing me.”
Agreed. But it is a case against doing nothing. This post,
http://adastra1960.blogspot.com/2013_03_01_archive.html ,
presents the implications of Rosser’s Equation in a graphical form. Doing nothing takes us along the red line, which would mean a drop in the basket of goods until it gets back to the same level again in 2055. Smoothing out the spend down of the TF requires doing something.
Of course, the NW Plan does do something.