Friday, August 05, 2005

Principal Economic Assumptions: What are they?

I constantly talk about "Productivity", but of course it is not the only number series involved. If we take Tables V.B1 Principal Economic Assumptions and V.B2 Additional Economic Factors we see a total of 11 columns:
Productivity
Earnings as a percent of Compensation
Average Hours Worked
GDP Price Index
Average annual wage in covered employment
Consumer price index
Real wage differential

Average annual unemployment rate
Average increase in Labor Force
Average increase in Total Employment
Average increase in Real GDP
Average annual interest rate

I focus on Productivity for a few reasons. One the Trustee's front it in their discussion, they talk about it in section B.1, they put it in the left hand column of the table. And they characterize it so "The rate of change in total productivity is a major determinant in the growth of average earnings." No they don't say "the major", but "a major" suggests that we should place close attention.

Two, the effects of Productivity on Trust Fund exhaustion are direct: a bigger future economy will have an easier time financing a fixed pool of boomers. This is true however we structure the financing and the payout.

Three, the effects of the other number series are harder to understand. The problem we have is that wage increases and inflation work on the Trust Fund in different ways. Wage increases boost contributions. Inflation increases boost cost. And the interaction is pretty complex. It works this way. Your ultimate initial retirement check depends on a combination of your income history and the overall rate of real wage increases. If workers as a whole do better you do better. But there the linkage stops, adjustments to your retirement check after that are determined by CPI: consumer price inflation. Okay lets unravel this a bit.

Higher inflation means higher payouts both short and long term. So all things being equal a jump in projected inflation is an immediate drag on the Trust Fund via increased payouts. Bad for Solvency (but oddly due to the disconnect from CPI and medical inflation not necessarily bad for current retirees. Topic for another time).

Increases in hours worked and average wage will increase contributions and so inject money into the Trust Fund. On the other hand they will increase your retirement check when the time comes and so increase costs. I have had posters elsewhere claim that this offsets itself, but I don't think so. The time shift between contribution and retirement check should make this a net gain for Solvency in a paygo system and the effect should roll forward. Like I said the effect some of these other number series are harder to understand. But people are free to bring numbers.

From the standpoint of solvency is seems clear that inflation is bad and real wage differential is good. But without a lot of study and more numeric skills than I have it is difficult to determine the significance of any particular number. But what I do know is that if productivity year in and year out comes in above both the Intermediate Cost and Low Cost projection Solvency is increasingly likely. And if we convince ourselves it will permanently come in above those points we are justified in asking skeptics to identify the particular number series that will offset that, and how they specifically work on Trust Fund balances.

Because I just don't think the long-range trends in this table are reversible EPI: Changes in Trustees Projections over time

The differences between projected 2004 and real 2004 (that is from the 2005 Report Table V.B1)
Productivity 2.7% 3.3% Earnings -.3 -.5 Hours worked (increase) .0 .0 GDP Price 1.6% 2.2%
Average wage 3.6% 3.8% CPI 1.2% 2.6% Real Wage Differential 2.4% 1.2%

Productivity is the Loneliest Number but still no. 1

I had an interesting e-mail from an economist who wanted to remain anonymous. He/she pointed out that Productivity is only one variable and that the Trustees had a stochastic analysis that validated the current exhaustion date. Let me make a couple of points here.

One, Productivity is shorthand for the whole range of economic numbers. It is true that the other numbers can vary, but many of them, like real-wage growth, have been tied to Productivity. True there are rumblings that this linkage is breaking down but it is incumbant on others to show how that offsets the huge gap between projected and actual Productivity. If Productivity was coming in at 2.2%, barely above the 2.1% used by Low Cost, I would still have a good case that we were on the road to solvency under the "if this goes on" theory, but I wouldn't be so cocky. My advice would still be to do nothing, if we are beating the model we are beating the model. But in the real world Productivity is coming in above 3.0% and now we are entering "what if" territory. "What if" 2005 ends up with a number over 3.0%? My answer is Solvency. Others can knock me off that mountain, but just pointing out that the other numbers of Low Cost are similarly optimistic compared to Intermediate Cost doesn't buy anything. Show me that they are being undershot by enough to offset the dominant variable. I am open to argument.

Two, I have never claimed to be an economist. But embedded in that stochastic analysis (which is studded by so many qualifiers and warnings about methods and reliability of projections to start with) is this sentence: "Each time-series equation is designed such that, in the absence of random variation, the value of the variable would equal the value assumed under the intermediate set of assumptions" (2005 Report p. 159). Near as I can tell all this analysis is doing is allowing variation around the assumed ultimate numbers under Intermediate Cost for the out years and seeing what happens. Given that in past years the Trustees have asserted that variations in the out years have little influence of ultimate results, it is not surprising that the results of this admittedly experimental model validate that. This is a relatively new part of the Report, introduced in 2003, and in this humble bloggers opinion is more intended to confuse and give some greater validity to Intermediate Cost than is warranted. You assume Intermediate Cost numbers as your point of departure and you would expect the outcome to vary around the projected result.

But it is near term numbers that are the big drivers, small changes at the beginning of a curve have outsize effects on its ultimate values. I have been begging professionals to weigh in from day one. Show me reasons why I should not just take Low Cost at face value. This was a nice, if private first step. Not entirely persuasive, but welcome.

Thursday, August 04, 2005

Scoop just ate my homework: MyDD diary

I have been taking up too much space at Economist's View (but Mark Thoma should be a daily stop for anyone discussing Social Security.) So I want to return to the topic of my previous diaries: Life After Solvency.

Social Security Solvency with no changes in benefits, retirement age or payroll tax is not an impossible dream. Each year the Trustees lay out a set of economic numbers that would would produce that result. This dataset, called Low Cost never gets a bit of attention. But that doesn't make it go away. You can get a little (okay a lot) of background at my website starting with What does Low Cost mean? More below the fold.
Welcome back, if you left at all. Over the last ten years Low Cost consistently returned the same result: flat trust fund ratio. What does that mean? For a fuller explanation you can check outThe Trust Fund Ratio explained. In brief the Trust Fund is a lot closer to a checkbook than a savings account. Contributions and interest earned go in, checks go out. The Trust Fund ratio is simply your balance expressed as a function of time. For example the Trust Fund ratio at the end of 2004 stood at 305 which means 3 years 18 days Table VI.C6.—Operations of the Combined OASI and DI Trust Funds
in Fiscal Years 2000-14
That is our current reserve and the direction of the curve is headed up under all three alternatives: Intermediate, High Cost and Low Cost Figure II.D7.—Long-Range OASDI Trust Fund Ratios Under Alternative Assumptions and if you look at our old friend Intermediate Cost (II) you see familiar dates like 2017 (when the curve peaks) and 2041 (Trust Fund Depletion). But what's up with curve (1): no drawdown until 2023? a slight dip then we sail through the 75 year window with a 450 Ratio? Are the economic numbers of Low Cost so optimistic that this is just pie in the sky? Judge for yourself. Personally I think 2.1% for 2005, 2.2% for 2006 and no number higher than that in the out years is more than doable.

Payroll vs Productivity: What would it take
2005 Report: Economic Assumptions

Low Cost is doable. In my view it is already done. You plug current growth numbers into this model and that ratio just keeps on rising.

If anyone responds maybe we can talk about what this would all mean. Meanwhile you might want to check out the terrific Rock the Vote flash Social Security: Don't get played and maybe follow that up with Lee Arnold's animation Social Security: The Real Connections. Lots to chew on. Mangia.