A Social Security Reform Debate
Earlier this week, I mocked a rather sensational editorial on Social Security reform that appeared in Glamour magazine. In particular, I was unimpressed with an Accurate Benefits Calculator that the story cited.
Although they didn't comment on the site, someone from the Center for Economic & Policy Research, which runs the calculator, took offense to my post. Well, actually, took offense is the wrong phrase. They questioned some of my criticisms. I welcome that kind of debate, but I prefer that it takes place on the site. One of the great things about a blog is that you can create a little community that discusses the issues, not just a single person on a digital pulpit. So, I thought I'd publish our email exchanges in the extended entry below and get everyone more involved in the debate.
"John Doe" from the CEPR:
Thank you so much for the plug. We appreciate that here.
It might have been nice to take a little more time noticing things with our calculator. Specifically,
1. The 4.35% return is explicitly documented as to how and why we use such a low rate of return. The growth in stocks is tied directly to growth in GDP. The fact is CBO projects a slowdown in GDP growth, and as a result there will either be a slowdown in the return to stocks or an ever-increasing price-to-earnings ratio. Neither CBO nor the Social Security Administration has been able to justify their 6.5-6.8% return to stocks.
2. Your benefit with the account may in fact be lower than without the account. You are expected to pay back the government every dollar diverted from Social Security, with interest. Plus, the account generates fees. Thus, your account must generate interest well in excess of the Treasury rate in order to produce a net benefit. This is particularly important for "older" workers, meaning those already in the labor force, as they will benefit less from the compounding of interest.
3. The calculator does take into account increases in pay over a working life. Rather than asking every user to input wages for every year, we use a stylized growth in wages. At the minimum, a worker's wages grow in step with the average wage index, and faster than that earlier in working life. The calculator also adjusts for the age of the worker, so a worker entering "36,470" (the estimated average for 2005) but not born until 2035 will still make about an average wage in 2070. This too, is spelled out in the technical notes.
Finally, a more political note. Suppose we all decide we want to retire with income some fraction of the income we receive during our working lives. (Yes, the fractions may vary from person to person, but let us assume that fraction should have no reason to fall over time. Expect people to retire in a style somehow connected to their working life.) Suppose also we live longer and longer. Then we have no choice but to save more and more of our wages in order to fund our retirements. There is nothing special about Social Security in this regard. What you describe is not a problem with the Social Security system than it is with your 401(k) or IRA or your home or any other savings mechanism. If you want more for your retirement, you need to consume less when you work.
You may agree or disagree on the last point, but it would be helpful to become remotely familiar with the calculator before comparing it to animal waste.
I didn't notice any technical notes that explained how wages over time were calculated. My apologies if I missed it.
I have two major problems with your calculator. First, you use the 4.35% rate of return as a default when the PLAN itself uses something different. Don't you think that's a bit misleading? I do. I read several articles about rates of return over time and while there is debate about what rate to estimate, many economists felt that the SSA's rate was certainly reasonable, perhaps not in perpetuity, but that it was a sustainable rate for many, many years. If you want people to evaluate the Bush plan, I think it's only fair to use the Bush estimates.
But finally, I object to the whole idea of comparing what you would get under the current plan to what you'd get under proposed reforms. I think that implies that we can merrily go along under current conditions indefinitely, which isn't true.
I think the calculator and the whole Glamour article are just a tad overly sensational and are built to lead people to a certain point of view, NOT to simply provide information. I have no problem with that (and frankly, I'm not sure what we should do about social security), but I think it's dishonest to pretend you're just disseminating information rather than a supporting a specific view.
No problem. I just thought you should know that there are detailed notes on the calculations.
The PLAN doesn't contain a specific rate of return on stocks. That's not within the purview of the Social Security Administration. Our point is the predicted return is mathematically inconsistent with the other economic assumptions they make, barring some new and wholly implausible assumptions about either dividends or stock prices. It is not a partisan issue-- it's simply unfair to use those estimates because they appear to be nonsense. If anyone at all can explain how to get to 6.5-6.8% a year through the planning period by projecting capital gains and dividends, we would most certainly love to hear it. We have an outstanding challenge to exactly that (The No Economist/Policy Analyst Left Behind Test)
As to comparison to current benefits, that is the only comparison that may be made. To use payable benefits (without additional taxes or borrowing) as the baseline means that Social Security faces no shortfall whatsoever at any time. It is in balance by definition. Furthermore, the President's plan mandates trillions of dollars in extra spending and trillions more in borrowing not reflected in the calculator. If payable benefits is the baseline, we are being hugely generous the other way, for additional cuts beyond the Pozen indexing would be necessary to bring the system into balance.
As to Glamour, I'm afraid I haven't seen the article, so I cannot comment specifically. I can only say that we have made a concerted effort to make the calculator as accurate as possible given the information brought to the table.
[An additional note... the S&P is up 6% in the last year, true. That's 6% before inflation. CBO projects 6.8% after inflation. If we throw in dividends, the yield on stocks could be as high as 5.8% after inflation. It's also true that 5.8% is almost 1.5% over our long-term projection of 4.35%. But CBO projects GDP growth to be some 2.5-3 times as fast this year relative to the long-term average. It's silly to compare our long-term projection (when growth is slow) to today's relatively fast growth. We don't dispute that stocks could grow faster in the short run, and we certainly take that into account in the calculator.]
At this point, I have to admit that I'm absolutely no expert in stock returns. Although, I did do some research on expected rates when I made my original post. Luckily, we do have someone here who is. And really, that's why I wish things like this took place entirely in the comments. I pulled in our own John Tant and got his take on the debate:
My first blush beef is over his statement "The growth in stocks is tied directly to growth in GDP. The fact is CBO projects a slowdown in GDP growth, and as a result there will either be a slowdown in the return to stocks or an ever-increasing price-to-earnings ratio."
Well, on the first point I disagree, mainly because I don't think there is a solid cause/effect relationship between the two. Correlation...maybe. But that's different than asserting flat out that one is tied to the other. In fact, savvy investors will tell you that when stock prices drop is when you need to start buying them up. Buy low, sell high...one of the most immutable laws of the universe, and one most often ignored by so-called professionals. So people
start buying up these stocks (supply, demand, price goes down demand goes up, etc)...and suddenly stocks are growing. Yes, even though GDP growth is still theoretically slowing down.
Second, why should we care WHAT the CBO is projecting? What's important is what the group is projecting. If they're using the CBO projection then they need to justify why they think that projection is more valid than others. Besides, after 24 straight months of positive job growth, the forecast of a GDP growth slowdown doesn't seem too realistic. I bet if we look at the CBO's forecast history, we'll see it more wrong than right over the years.
I think he also contradicts himself in the follow-up email, by the way. First he says CBO is projecting a GDP growth slowdown. He then turns around and says "But CBO projects GDP growth to be some 2.5-3 times as fast this year relative to the long-term average." Well, is it a slowdown or isn't it? Is the support for a slowdown based on fudging the base year of that "2.5-3 times" rate growth and then comparing it to the long term average? If so, I think that's a little
And his cite of a PE ratio is intellectual filler. He says a slowdown in GDP growth would possibly result in an increasing PE ratio. But in the 90s, GDP growth was increasing and PE ratios also increased. So either the 90s were an outlier or he is misunderstanding the relationship if any. I don't think it's an outlier, because PE ratios on a whole have been increasing for quite some time. It's one argument that is used to support the idea that the stock market is overvalued. I do have some sympathy for that argument, but on the other hand I have even MORE sympathy for the argument that "value" is only found in "what someone else will pay for something."
Next, let him complain about the S&P 6% growth being before inflation. If he's really thinking inflation is a big worry at this point in time, then he's damaged his own credibility.
For the wage growths (item 3), bottom line he seems to be assuming that if I make an "average" wage in 2005, I'll be making an "average" wage in 2035. I understand the limits of some models, but that single assumption is wrong enough to taint any results that come out of the other end, in my opinion. How does he know what an "average" wage will be in 2070? How does he know what the average wage growth will be between 2035 and 2070? On what basis does he assume that a younger worker will outstrip the average wage rate increase, and how much is he assuming that worker will outstrip it by? All things that he'll
say are explained in the technical notes (which I read, all six pages, and darned if I could see it in the detail he claims it's there...), but at the end of the day we're talking about assumptions plugged in by some research associate at a left-wing think tank.
I could go on and on, so I will.
I plugged in my income in the calculator and chose married. It came back with "This calculator assumes that the spouse makes 1/3 the wages of the primary earner." Huh?!? "If you expect to have more equal incomes, calculate separately as unmarried individuals." But if I do that, doesn't that mess with the other assumptions he cites?
I like the material misdirection at the bottom of the calculator too: "If you received and spent the benefits under the Bush Plan listed above, you could bequest nothing to heirs." Um, as opposed to the current system? Puh-leeze. And what's the point of saying that anyway? It's like saying "If you spend all the money in your bank account, you won't have anything to leave to your heirs. No shit, Sherlock. And if I'm married, projected expenses are going to be lower (don't have to pay two heating bills, you know). But oh no...to get a "realistic" money number I have to calculate myself as single and add in my wife...also calculated in as single.
Sorry this ran longer than the brief snark I was planning on. :)
And if that wasn't enough, John followed up with more info on the relationship between GDP growth & stock returns:
Food for thought on one of the points I brought up (relationship between GDP growth and stock returns):
The "Baker" in the quote is Dean Baker of the Center for Economic and Policy Research...the same think tank who came up with that calculator.
As baby boomer retirements and low birth rates reduce labor-force growth to just 0.2 percent annually, total economic growth will decline as well. The critics' argument is simple: Slower economic growth means lower corporate profits, and profits drive stock prices; hence, stocks can't possibly return more than 3.5 percent, placing their long-term returns below today's government bond rate. While even this performance would substantially exceed Social Security's paltry 2 percent return, higher stock returns would speed the transition to personal retirement accounts.
But it's easier to say "slower economic growth equals lower stock market returns" than it is to prove it. Research by Philippe Jorion, professor of finance at the University of California-Irvine, reveals both theoretical and empirical flaws in Baker's argument. Jorion acknowledges the bookkeeping idea that "asset prices should grow at the same rate as cash flows," but in the real world "this relationship. . . may be blurred by a number of factors." A more sophisticated theoretical model shows that returns on capital investments "should be related to real GDP growth per capita, instead of total GDP growth."
Jorion's empirical analysis confirms the theory. Drawing on research on global equity markets he conducted with Professor Will Goetzmann of Yale, Jorion examined the relationship between economic growth and stock returns for 31 countries, ranging from established markets to new economic powers to developing countries. The results directly contradict what Baker's theory would predict. While Jorion found "no observable relationship between stock market returns and GDP growth," statistical analysis revealed that "stock market returns are positively correlated to GDP per capita growth."
For instance, developing economies grew 1.4 percentage points faster than economies of developed countries, but their stock returns averaged 2.6 percentage points below those in the developed world. How could this be? Developing economies expanded through rapid labor-force growth, not productivity improvements. As a result, their GDP growth per capita—and their stock returns—lagged behind those of developed countries. Hence, Jorion concluded, "Lower capital gains are really associated with lower per capita economic growth," not lower total economic growth.
This link between per capita GDP growth and stock returns affects the debate over personal accounts, since the economic slowdown projected by Social Security's trustees stems almost entirely from reduced labor-force growth. Productivity increases – the other main component of economic growth – will remain at the 1969-98 average of 1.5 percent annually and GDP per capita growth will be respectable. Jorion found that a 1 percent change in per capita GDP growth correlates with a 0.7 percent change in equity returns. If true, we can expect future stock returns to be less than one percentage point below their 1926-97 average of 7.2 percent.
People mock blogs, but where else do you get this kind of information? John and the gentlemen from the CEPR both explain their position and back it up with other sources. You'll never see this depth of information and opinion in a newspaper. And maybe that's why their circulation continues to plummet.
Posted by at May 14, 2005 01:37 PM
The trackback entry for this page is : http://www.inthehat.com/mt/mt-tb.cgi/903
|# May 14th, 2005 9:04 PM secondclasscitizen|
|While I can be a policy wonk, and a debate is good to have, I'm sold on privatization and don't need the whole ball of wax. The current system is bad for a lot of people, especially unmarried black men working for low wages. (I'm two out of three.) Who wants to pay in for decades and die before you get your money back?
|# May 15th, 2005 12:00 AM BrianH|
|The people who are against personal accounts seem to be trying to convince us that the current PAYG system can "grow it's way out" of the problem. I think they're trying to change the subject (or maybe they really don't understand how the system works).
I've been arguing with the libs over at Matt Yglesias's site. They don't seem to understand that economic growth isn't the problem or the answer with Social Security since it doesn't really apply (except with regard to the employment rate).
The real problem is the ratio of workers to retired people. The rate of growth of the workforce (the birth rate + WORKING immigrants +/- changes in the employment rate) is slower than the rate of growth of beneficiaries. There is no way that a PAYG system can keep up with the current trends. The current worker to beneficiary ratio is about 5:1 (it was 7.2:1 in 1950). It's projected to be about 2.6:1 in 2050. In other words about half the number of workers will be supporting 1 retiree.
In a PAYG system, current receipts are paid out to current beneficiaries. In a perfect system the income and outgo would be equal. In our current system, about 1/3 of the payroll tax collected is excess above the current benefit requirements. This excess money is spent for other purposes. (This is a classic Ponzi structure.) There is no investment in the economy so economic growth has very little effect (except on the unemployment rate).
If the excess money were instead invested in the economy, then it would be possible to grow out of it with economic growth. Without this investment, any discussion of economic growth is a red herring.
|# May 15th, 2005 12:24 AM BrianH|
|Oh, and I didn't check it, but I suspect the calculator they're using is similar to the one Sen. Reid (D-asshat) used. FactCheck.org had a pretty good analysis of that calculator here:
|# May 15th, 2005 8:29 AM kris|
|Thanks for the factcheck.org link. This may not be the exact calculator that Reid uses, but it's close enough that the criticism is virtually the same. |
|# May 16th, 2005 10:37 AM JohnTant|
|First, I'm sorry for my prolonged absence. I've been involved in a somewhat lengthy job interview process for a company that is courting me.
Anyway, I first want to say I don't like any of the online calculators, and that even includes the one from Cato. The thing is, they all have their own assumptions built into the model, assumptions which can tailor the result to pretty much any policy stance desired. So bottom line, I'd take anything that comes out of an online calculator with a grain of salt. I certainly wouldn't use it to bolster a policy argument.
That said, I think the biggest material weakness in the CEPR calculator is the assumption of GDP growth having a direct influence on stock market returns. The Biggs column (written back in July 2000, showing the CEPR line has been around even before Bush's reform initiative) does a good job in pointing out the weaknesses. It even cites real-world data of the type that economists are generally criticized for not having.
I think the best stance from which to look at this should rest less on what potential spoils one will get from the system and instead focus on the morality (for lack of a better term) of the system and the alternatives.
|# May 16th, 2005 11:09 AM d_rosnick|
|I have been asked to continue this debate in comments, and I'll be happy to do so. I would like to kick things off by saying that I have been given an awful lot to which to respond. I have responsed to them all and have no intention of running away from any possible criticism. My main concern is with posts that will become so excessively long that the debate will be impossible to follow. So bear with me as I take things one bit at a time, and I will be happy to answer questions.
Mostly, the debate is over our projections for stock returns, so let me begin there. The cross-country relationship between stocks and per-capita GDP is shaky. That has absolutely nothing to do with our projections. We make our projections based on GDP (not per-capita.) Earnings in the United States (and therefore dividends) have grown in a tight relationship with GDP for decades. The ratio of after-tax profits to GDP has averaged roughly 6% since 1947. Thus, we project that earnings will grow in step with GDP in the future. We also project that dividends will continue to be a roughly constant share of earnings and that prices be a constant multiple of earnings. The last item is really the shakiest assumption, to which the stock bubble of the late 1990's will attest. However, the extremely high price to earnings ratio of the bubble years did predictably correct to more rational levels. We are not going to be in the habit of predicting manias and try to make predictions based on long-term conditions.
At any rate, those are the assumptions by which we make our projections. Based on dividends and stock prices, we compute the stock return. Going from dividends and prices to the return is not a matter of dispute. If the assumptions are reasonable, then so are the returns.
Our concern is that virtually no alternative projections for dividends and capital gains will yield the 6.8% return assumed by CBO (6.5% by SSA) without resorting to new and amazing assumptions. For example, a 6.8% return is possible in the long run if one assumes the PE ratio grows to levels far past those of the stock bubble. A 6.8% return may be available in the long run if the market today crashes by about 2/3. There are alternative assumptions one may make regarding dividends and prices. Write them down. If you can put down dividend and price paths that add up to a 6.8% return on stocks, let us know. Take the No Economist Left Behind Challenge. (http://www.cepr.net/publications/ss_economist_test.htm)
Center for Economic and Policy Research
(aka "John Doe" above, due to a miscommunication about anonymity...)
|# May 16th, 2005 11:39 AM d_rosnick|
|I am happy to see John agrees that the stock returns are the primary issue. However, he fails to recognize that there are three assumptions that we make in our projection, all of which apply in any reasonable sense to the United States.
It in no way surprises me that a cross country analysis fails to show a positive relation between per-capita GDP and stock returns. Here is one possible reason why: The United States is considered the safest place for putting money. That means the stock price one would pay for a dollar in dividends should be higher here than elsewhere in the world. (One should expect to pay a premium for the relatively low risk.) The dividend to price ratio is currently around 2.9% ($34.50 for $1 in dividends.) Suppose that the ratio in another country is 5% (only $20 for $1 in dividends.) Then 4.0% growth in the U.S. means (by our formula) a 7.0% stock return, while 4.0% growth in the other country means an 9.2% stock return. That's a big difference: a factor of 1.75 compared to 2.3.
There are many reasons why our assumptions would not as apply to other countries. We can hardly deny that under different assumptions one would get different results. This debate should be over what is going to happen here, based on CBO projections about the future.
|# May 16th, 2005 12:55 PM JohnTant|
|Well, as far as I'm concerned the main assumption in the model that growth in stocks is "tied directly" to growth in GDP is the main weakness and I've said as much (I "fail to recognize" other assumptions only in the sense that I haven't spent much if any time on them...). David, in your first post you blithely dismiss arguments to the contrary and in the second you basically say that's not the way things work in the US...again, rather blithely.
The person doing the study is one Philippe Jorion. He drew on research he did for this paper (http://www.gsm.uci.edu/~jorion/papers/century.pdf), and that research included data from the US. In fact, the column cited says:
Drawing on research on global equity markets he conducted with Professor Will Goetzmann of Yale, Jorion examined the relationship between economic growth and stock returns for 31 countries, ranging from established markets to new economic powers to developing countries.
Now are you telling me and everyone else that his research doesn't include the US? That the conclusions from Jorion's study do not apply to the US?
I think they do.
You then say:
It in no way surprises me that a cross country analysis fails to show a positive relation between per-capita GDP and stock returns.
Well, what's being said is that there is, actually, a relationship. What's also being said is that a more sophisticated model would use the per-capita GDP and not the overall GDP figure like CEPR director Dean Baker continually cites. So what we have here is empirical data showing per capita GDP to be the thing, not overall GDP. How does CEPR address the contridiction shown in Jorion's research aside from merely asserting that it isn't true for the US?
|# May 16th, 2005 1:33 PM d_rosnick|
|This is very simple. There are three assumptions we make.
1. Earnings grow with GDP.
2. Dividends grow with Earnings. (constant dividend payout)
3. Prices grow with Earnings. (constant price-to-earnings ratio)
If you accept these three things, and use them to compute the return to stocks, you get the formula we use in the calculator. The algebra is laid out in the technical notes. If the conclusion is invalid, it can only be a result of an error in assumptions or an error in algebra.
I haven't gone through the paper you cite with a fine tooth comb, but I haven't seen anything that disputes any of those assumptions.
In fact, the paper states quite explicitly that "the high U.S. equity premium seems to be the exception rather than the rule."
Where is the contradiction? Please cite a page number or something to point me to it, and I will be happy to respond.
|# May 16th, 2005 2:31 PM BrianH|
|I think the problem with your assumptions is assumption #1. Earnings can grow due to other factors such as reduced cost structures, not just production growth.
|# May 16th, 2005 2:42 PM JohnTant|
|There's a tendency when a policy wonk says "this is very simple" it actually turns out to be anything but.
You originally said "The growth in stocks is tied directly to growth in GDP." I then pointed out that way back in 2000 that theory was tested by Philippe Jorion and he found that it wasn't GDP but rather per capita GDP, and he pulled out a bunch of empirical data to bolster his case.
You turned around and said first that the research was "shaky" and then admitted per-capita GDP had nothing to do with your model. But then 30 minutes later you said your assumptions were only valid for the US and not for other countries. In other words, the empirical data cited by Jorion doesn't apply because we're talking about the US. But in the study, Jorion actually includes US data. So the argument that his conclusion for the Biggs article doesn't apply to the US market seems unsupported because Jorion did actually use US data in his research.
So when you ask where the contradiction is, I'll quote it again:
For instance, developing economies grew 1.4 percentage points faster than economies of developed countries, but their stock returns averaged 2.6 percentage points below those in the developed world. How could this be? Developing economies expanded through rapid labor-force growth, not productivity improvements. As a result, their GDP growth per capita—and their stock returns—lagged behind those of developed countries. Hence, Jorion concluded, "Lower capital gains are really associated with lower per capita economic growth," not lower total economic growth.
At the end of the day, we're left with CEPR's assertion vs. real world empirical data.
And I didn't cite the Jorion/Goetzmann paper to disprove your assumptions (nor did I ever say it did)...I cited it to show the scope of Jorion's research and to show it included the US. I cited the Biggs article to criticize Assumption #1, that stock returns are "directly tied" to GDP.
|# May 16th, 2005 3:18 PM BrianH|
|To further illustrate the problem with your assumption #1.
The definition of GDP is
"The monetary value of all the goods and services produced by an economy over a specified period. It includes consumption, government purchases, investments, and exports minus imports."
The earnings or profits of a company can be stated as:
Gross profit = value of good/services sold - cost of goods sold
Net profit = gross profit - overhead (SG&A) and taxes.
Now GDP includes the value of good sold. And it also includes that value of investment defined as the purchase of durable goods. That's not used in calculating profits. GDP does not take into account the cost of goods sold.
Cost of goods sold include a number of "fixed costs" and "variable costs". Fixed costs include the costs of the buildings, machinery, etc used in production that do not change no matter how much material is produced. Variable costs include materials, labor, power, etc that change depending on the amount of goods produced.
Because of this I can produce more goods (thus increasing GDP) without having any earnings. If I produce and sell enough goods to exceed my fixed costs, my earnings start to go from negative to positive and grow at a rate faster than my increase in production.
At the same time there are overhead expenses and taxes that can vary. Changing these variables will change earnings with no change in production and therefor no change in GDP.
What all this amounts to is GDP and earnings can (and DO) change at different rates.
|# May 16th, 2005 6:00 PM BrianH|
|I think our economist must have decided to go take a basic accounting class before answering.......
|# May 16th, 2005 9:25 PM d_rosnick|
|Ah. Assumption #1. Thank you. You are absolutely correct. If profits were to grow faster than GDP, then yes, the stock return would be higher. Now suppose that assumption #1 is incorrect. Let's work backwards in the algebra from the target stock return. What rate of growth in earnings would be necessary in order to sustain 6.8% growth?
A 6.8% real return inflates to 9.1% nominal (CBO assumes 2.2% growth in the CPI-W, the basis for determining COLAs) The stock return arithmetic results in a 6.1% nominal growth in profits, long-term. Nominal growth in GDP is projected to be 3.7% per year, long term. After-tax profits are currently at a record high of 8% of GDP, so after 25 years of such a differential, they will be 14% of GDP. After 50 years, 26%. After 100 years, after-tax profits will be 81% of GDP.
Note that historically, this has not been the case. After-tax profits were as high as 7% in 1949, but fell towards 4% in 1951. They reached nearly 8% in 1965, but tumbled below 4% in 1974. They again pushed up towards 8% in 1997, but by 2001 were barely 5% of GDP.
From where will such huge profits suddenly come?
Brian seems to suggest it is possible to obtain higher earnings without increasing production and therefore without increasing GDP. The problem with this argument becomes readily apparent when one moves from the Gross Value Added definition to the Total Income definition. If there are higher earnings without a corresponding increase in total income, then there must be a drop in some other type of income. That other type of income is dominated by compensation. This is certainly possible, but is it realistic? And if it is the most plausible scenario (as opposed to the trend indicated by historical data) then someone had better convince SSA and CBO because they do not project a fall in compensation large enough to make the arithmetic work. In fact, SSA expressly states in the 2005 Trustees' Report (section V.B.3.) that the "ratio of compensation to GDP is assumed to be stable." If the earnings, say, come out of compensation (currently 56% of GDP) then in 50 years, rather than 56% of GDP, compensation will be only 38%, or a fall of roughly 1/3 from projections. Over 50 years, this means lopping off more than 0.75% off of annual compensation growth. You are welcome to make projections different from SSA and CBO, but the calculator is designed based on CBO numbers. The 6.8% number we must ignore on the grounds that its inclusion appears to contradict CBO's previous assumptions.
As to the "contradiction" you cite, it contradicts nothing I have assumed or stated. The Biggs article just cites results: no source data, no assumptions. Without knowing the methodology behind the work, I can hardly take that as evidence against our work. I will be happy to look at his work of you provide some sort of citation. (I have written to Jorion asking for assistance tracking it down. Hopefully, someone will come through.) In any case, the bits that Biggs provide point to a cross-country analysis. I will be surprised if the analysis actually controls for the D/P ratio which one could easily expect to vary across countries. (We make no such claims, mind you, so a failure to control for all the variables in the formula would be a real problem, akin to arguing "2+3 can't be 5, because 1+3 is only 4, and 5+3 is 8")
|# May 17th, 2005 12:26 AM BrianH|
|Yes it is possible to increase earnings without an increase in production. It's also possible to increase production without an increase in earnings.
If for example a company replaces fully depreciated equipment with new equipment that has a lower fixed cost, that company can increase profits without increasing production. It has contributed to an increase in GDP due to the capital investment in the new production equipment, however that may not be proportional to the savings generated. Take computer equipment. An old computer would require not only the purchase cost of the equipment, but also massive outlays for cooling, electricity, and manpower to keep it running. The replacement equipment is cheaper (lowering the depreciation costs), smaller (requiring less floor space), uses less power (and therefor less cooling), requires less manual intervention, has less expensive supplies and as an added benefit has a higher production capacity (not necessarily utilized). The cost savings is much higher than the capital outlay. The result is a significant increase profits and a nominal increase in GDP.
Again, not a 1 to 1 corespondence.
It's also possible to increase production and increase profits at a faster rate than GDP. The same is true on the other side of the equation though. I can increase GDP and have no profits at all.
An overly simplified example:
Lets say I am starting a company that is producing widgets that I can sell for $20 per. I lease a building for $10,000/year and lease equipment for $5,000/year (to keep the fixed costs simple) which gives me a capacity to produce 10,000 widgets. The cost of the raw materials and electricity and labor to produce each widget is $10 per unit. If I make and sell 1000 of the widgets, I'll have gross sales of $20,000. My variable costs are $10,000 and my fixed costs are $15,000. I'll have increased GDP by $20,000 but have lost $5000 no profit at all for that GDP I produced.
Lets say in the second year I double production and sales. I sell 2000 widgets at $40,000. My fixed costs are still $15,000 but my variable costs are now $20,000. I've made a profit of $5,000. I've doubled my contribution to GDP, and went from a loss to a profit.
The third year I produce and sell at full capacity (it seems everyone wants to have a widget). I sell 10,000 for $200,000, my fixed costs are still $15,000. Variable costs have dropped to $9 per widget because I get a volume discount on the materials for a total of $90,000. My profit is now $85,000. I've had a 500% increase in my contribution to GDP and a 1,700% increase in profits. (Wall street gets interested at this point and I do an IPO that raises an unreasonable amount of capital that I can use to add to production capacity thus changing the production costs).
Not a 1 to 1 correspondence.
GDP and profits obviously do NOT have the same rate of growth (or decline). Since they only share some of the same variables, you can not make a hard assumption that one drives the other.
It's also unforunate that the valuation of a company often has less to do with its actual earnings than it does with some future guess at potential earnings. In the 90's stock valuations for internet companies had little or no basis at all in reality.
By the way, have you applied your GDP to earnings and stock valuation model to any historical market data? How close does it match?
|# May 17th, 2005 2:12 AM d_rosnick|
|This is ridiculous.
I don't have a stock valuation model. I want to know what stock returns will be in the future. Stock returns come from dividends and earnings, so I make projections for dividends and earnings consistent with CBO economic assumptions. Based on those projections I compute the stock return they imply.
Anyone who doesn't like my projections for dividends and prices is welcome to try to make his or her own. How fast do you think dividends and prices will grow over the next 75-100 years? Seriously.
Of course corporate earnings are only one component of income. Thus, it is possible for earnings to grow faster or slower than GDP. I don't dispute that.
So, if in making your dividend projections, you wish to assume earnings grow faster than GDP, fine. Put that down. The important thing is that you write down an alternative set of dividend and price projections.
Can anyone else manage it? Is it really that difficult?
|# May 17th, 2005 7:35 AM JohnTant|
As to the "contradiction" you cite, it contradicts nothing I have assumed or stated.
You say GDP is "tied to" stock returns. Jorion says no, it's per capita GDP.
The Biggs article just cites results: no source data, no assumptions. Without knowing the methodology behind the work, I can hardly take that as evidence against our work.
Well, Biggs cites the paper to which I provided a link as the source. Biggs did, after all, write rather plainly "Drawing on research on global equity markets he conducted with Professor Will Goetzmann of Yale..." A simple Internet search turned up the paper. Reading the paper showed US data included in the source.
And there's an interesting standard here. Thus far you've continued to assert that total GDP growth is "tied to" stock returns, but I haven't seen you support that assertion, certainly not on a par of what you demand of Biggs and Jorion.
I will be happy to look at his work of you provide some sort of citation. (I have written to Jorion asking for assistance tracking it down. Hopefully, someone will come through.)
Well, if we have an idea of the source data from his original paper, seems to me we could get in the ballpark of his finding that "developing economies grew 1.4 percentage points faster than economies of developed countries, but their stock returns averaged 2.6 percentage points below those in the developed world." And that is a contradiction from what CEPR theorizes about the GDP/stock returns relationship.
In any case, the bits that Biggs provide point to a cross-country analysis. I will be surprised if the analysis actually controls for the D/P ratio which one could easily expect to vary across countries. (We make no such claims, mind you, so a failure to control for all the variables in the formula would be a real problem, akin to arguing "2+3 can't be 5, because 1+3 is only 4, and 5+3 is 8")
How could you be "surprised" if a couple of minutes earlier you complained that you don't know the methodology behind the work?
On the D/P ratio (that would be dividend yield for those readers who don't want to wade through high level, unclear language), Jorion is talking about total return, not just dividend yield. And as for it being "controlled," isn't the point of the exercise to determine the relationship of a country's GDP growth to that country's stock return growth (or as Jorion found, lack thereof)? I think Jorion found it wasn't the GDP growth itself that accounted for increases in stock returns, but rather the nature of that growth...which contradicts CEPR's line.
|# May 17th, 2005 7:44 AM BrianH|
|"I don't have a stock valuation model. I want to know what stock returns will be in the future."
Uh... trying to calculate stock returns is a stock valuation model.
"This is ridiculous. "
It's not ridiculous. You're attempting to find the value and returns from corporations. You are basing your calculations on faulty assumptions.
1. Earnings grow with GDP.
2. Dividends grow with Earnings. (constant dividend payout)
3. Prices grow with Earnings. (constant price-to-earnings ratio)
Claim #1 is Earnings grow with GDP. That is demonstrably false. Earnings and GDP have different variables that drive them.
With regards to corporations, GDP is driven by the top line of a company's annual report + capital investments made by the company (which are only applied to earnings as depreciation expense). Earnings take this top line number and subtract various types of expenses, add non-operating gains and losses and subtract taxes (I left out SG&A and taxes in my contrived example to make it clear).
It is VERY easy to increase production (the major component of GDP) and decrease earnings by allowing expenses to get out of hand or by setting prices too low or due to unforseen competition. With good management, it's also possible to maintain a current level of production (or even reduce it) while at the same time increasing profits. The 2 items do not have a 1:1 relationship. Your basic assumption is wrong. Therefor your model is wrong.
If you can take 5 random starting points say 1940, 1945, 1950, 1955, 1960, and 1965 (the GDP, earnings, and stock return data is available), run your calculations and match the real return from the market (within a reasonable MOE) in every case, I'll bow to your greatness and start buying puts on the S&P 500.
I suspect your model will not match real data. If it does, you may be looking at a nobel prize.
As for a return rate to use for estimating stock investments:
A data point I've read from a number of sources is that the average return for the S&P 500 over any 20 year period since it's inception (pick any starting year) is 11%. I haven't checked the number against returns myself as I've seen similar claims from multiple places that I trust. I suspect the CBO and SSA are using that as the basis for their projections. I usually use 7% as an average ROI when I'm trying to calculate what will be needed but I also calculate at lower and higher rates to see what the changes would be. I also use retirement planning tools that run Monte Carlo simulations to find out what their high and low and most likely projections will be.
In other words, I don't believe that you can predict ANY exact returns on future investments. You can produce a range (based on historical performance) or you can pick a number in the middle of your expected return rates and use that as a target or you can pick a low end number, plan for that and consider any excess to be gravey.
"Is it really that difficult?"
Yes. People have been trying to build accurate market models to tell them when to and how invest for MANY years. Sometimes they get it right for a while, sometimes they don't. The most reliable model has been to ignore prices and dollar cost averaging (inherent in any PSA plan) into a diversified fund of good companies and holding on for a long term (20 years +).
|# May 17th, 2005 8:08 AM JohnTant|
|Brian, very good post.
I'm going to be out of pocket for the rest of the day, but I wanted to toss in a point.
I think the original point that Kris made was that these models are so reliant on assumptions that they are basically useless when it comes to policy prescriptions. This is because those assumptions can be easily gamed to produce a desired result. I tend to agree.
|# May 17th, 2005 8:28 AM BrianH|
I do have one question that I do think you'll know the answer to. Since the increased investment from PSAs will be stimulative to the economy, is this increased GDP growth due to the investments factored in to the CBO GDP estimates?
|# May 17th, 2005 8:55 AM BrianH|
|I found another place that the CBO may have obtained its 6.8% number.
The article below (I found the link from factcheck.org)
"The Short- and Long-Term Outlook for Stocks"
States that the average annual inflation adjusted return for stocks since 1802 has been 6.8%. This suggests that the 11% non-inflation adjusted number is probably accurate. Again I haven't checked the numbers myself, but I trust the source (UPENN).
|# May 17th, 2005 9:13 AM d_rosnick|
Look, I am not asking for any returns. I'm only looking for the most likely return consistent with other CBO (or, if you prefer, SSA) assumptions.
SSA says 6.5%, but they don't make an attempt to reconcile that with their assumptions.
CBO says 6.8%, but they don't make an attempt to reconcile that with their assumptions.
By "reconcile" I mean that the return is not something unto itself. Stock returns are computed based on dividends and prices. Whatever CBO says dividends and prices will be, they better be consistent with their other economic assumptions. Unfortunately, CBO refuses to do so. If I said our combined ages (the two of us) totaled 500 years, everyone would laugh. Why? Because no guess for our individual ages could both add to 500. It's the same way with a 6.8% stock return in the future. Not because 6.8% wasn't possible in the past, but because the numbers one must add to get there make no sense in terms of CBO's economic assumptions.
Try it. You will see.
Maybe you don't like CBO's economic assumptions. Well, I'm sorry, but that's what I'm basing my calculations on. You can't really criticize me for that.
Try it. You will see.
You don't like my assumptions. I keep saying that's fine. Make your own. If you can't make any predictions, then mine are just as good as anyone else's.
You still said you use 7%. That's fine. So how much is in dividend growth and how much is in capital gains?
You suggest using historical data for projecting the future. Fine. Let's do that. Rather than simply tell me the stock return, give me the dividends and capital gains. Yes, I have this data myself, but I'm not going to put numbers in your mouth. You pick the years and whatever method for averaging. I don't care. Just get back to me with the results.
Here is an example:
Every dollar in stocks produces 4 cents in dividends and 3 cents in capital gains. That's a 7% real return.
Here is another:
Every dollar in stocks produces 3 cents in dividends and 4 cents in capital gains. That's also a 7% real return.
Here is another:
Every dollar in stocks produces 12 cents in dividends and loses 5 cents in capital gains. That's also a 7% return.
Every dollar in stocks produces no dividend, but produces 7 cents in capital gains. That's also a 7% return.
Every dollar in stocks for the next five years produces 10 cents in dividends and loses 3 cents in capital gains. Every dollar thereafter produces 3 cents in dividends and 4 cents in capital gains.
(Just so you see you may make these as detailed as you wish. Break it down by year if you wish.)
Your turn. Just do it.
|# May 17th, 2005 9:36 AM d_rosnick|
I know CBO says it makes sense. But stock returns come from capital gains and dividends. Unless they come up with capital gains and dividends that add up to 6.8%, they're pulling it out of thin air. An appeal to history doesn't work because the conditions that supported the past growth in dividends and prices do not exist in CBO's assumptions about the future.
This has nothing to do with my assumptions. CBO's projections are like my saying I'll be dead in 50 years, but I've awakened every day since my birth, so I project I will wake up every day for the next 100 years. The assumptions are contradictory.
In CBO's case, any explanation of the stock return, based on the factors that mathematically make up the return are bound to look very strange in context.
That is, write down capital gains and dividends over then next 100 years. Any. Do dividends grow with GDP? I don't care! Just wrote it down. Does it add up to a 6.8% annual return? Good. Does it make sense in terms of CBO assumptions? We shall see.
At least my capital gains and dividends are consistent with CBO assumptions. You may not like them, but at least there is that. Give me an alternative set. That's all I ask.
|# May 17th, 2005 9:47 AM d_rosnick|
You like the S&P.
Fine. Here you go. From 1929 to 2004 GDP growth has averaged 6.5%. The mean dividend/price ratio over that time was 3.9-4.2% (geometric and arithmetic, respectively). If we plug 6.5% GDP growth and 4% D/P into my formula, we get 11%.
So what is your point?
|# May 17th, 2005 9:49 AM BrianH|
Frankly, it doesn't much matter to me if the 6.8% return over inflation is in dividends or capital gains. As you say, it's the combined returns that matter. By the way, the paper I referenced above states the historical average GDP growth is 2.2% over this same period where the real growth in stock over inflation is 6.8%. Again I have no reason to doubt their number.
You're calculating future returns based on a model you developed. I think the model is flawed due to faulty assumptions. If I provide data to fit into a flawed model it is meaningless. You say your assumptions are correct and your model is not flawed.
There is historical data that can be used to check your model. Why is it unreasonable for me to ask you to use the historical data to see if the model's predictions match real data?
If you can do so, it would prove your model is correct and we should all just shut up and go away.
I'm not asking for an exact match in every year. I'm not even asking you to produce exactly 6.8%. I'm asking you to check your model against the historical record over 40 year periods starting at a few randomly chosen years and get a number that's CLOSE to the real return from stocks (say within 1% which is a pretty big MOE when we expect 6.8%).
I'm also not trying to build a model (as I've stated, I don't think anyone has EVER produced an accurate model of the stock market). I'm basing my assumptions of future returns on historical returns. I expect variations over short term but expect very long term gains to match historical averages.
|# May 17th, 2005 11:44 AM d_rosnick|
|You think my assumptions are flawed. The problem is, you can't come up with a more reasonable alternative. If I am wrong about that, then correct them.
My point about it not being a model is only that once one makes assumptions about dividends and prices, there is nothing left. I agree. The only question is what we should assume.
Look above. I showed you that the model worked for the S&P over the period 1929-2004. 11%, just like you said.
You can assume future returns to be the same as historical returns if you wish, but that doesn't mean your assumptions are consistent with CBO's other economic assumptions. All I ask for is consistency. If you don't like CBO's other economic assumptions, you can't blame me.
If you think I've made a faulty assumption, then tell me what to assume in its place. If earnings are going to grow faster than GDP, how fast will they grow? If the dividend payout will be higher, what will it be? If the PE ratio will be lower, then what will it be? You can't keep saying the assumptions are flawed simply because they result in a conclusion you don't like.
Don't you understand?
I say earnings will grow in step with GDP, because that is consistent with CBO economic assumptions. I say the dividend payout will be 60% of earnings. I say the price to earnings ratio will remain stable at 21 (adjusting for the current business cycle)
If you agree with the second and third points, then the only way to get a 6.8% future return is to have different earnings. There is no other mechanism.
If you agree with the first and the second, then the only way to get a 6.8% return is to have a different PE ratio. There is no other mechanism.
If you agree with the first and the third, then the only way to get a 6.8% return is to have a different dividend payout. There is no other mechanism.
So what assumption is wrong, and why?
|# May 17th, 2005 12:28 PM BVBigBro|
|The original criticism remains. The calculator originally described and criticised ignores the need for tax increases or benefit reductions to pay the amount shown by the calculator. Using the calculator leads one to the conclusion that the current system, untouched, is the way to go, ignoring the repercussions of not acting.
I submit that Kris's original conclusion, that the calculator was intended to mislead the user is entirely correct.
|# May 17th, 2005 12:31 PM BrianH|
|"You think my assumptions are flawed."
I KNOW your assumptions are flawed. I have demonstrated that earnings are not directly tied to GDP.
They use one common component (gross sales) as one of the data points in their equations. The rest of the formulas have no common data points.
I could also argue that your assumptions 2 and 3 are too rigid and you should use a range of values for each, but if assumption 1 is wrong, the other 2 don't matter.
How do you reconcile a 6.8% after inflation return on equities when GDP averaged 2.2% in the same time frame?
"Don't you understand?"
I understand your formulas. I understand that if you assume earnings growth is directly linked to GDP that your predictions don't look good for the market. I just don't agree with your basic assumption for the reasons stated. It's good work, but good work based on bad assumptions will still give wrong answers.
I have come up with a reasonable alternative, historical averages. I won't try to make a more accurate prediction because I don't believe it's possible. If I had a better way of predicting future stock activity I'd be comfortably retired instead of worrying about Social Security getting cut by 27% about the time my son needs it (unless they raise the SS tax rates to absurd levels).
"There is no other mechanism."
There obviously MUST be another mechanism. I haven't really thought through all the variables, but a couple I can think of offhand are:
1. Changes in tax rates (which will effect GDP)
2. Changes in money supply and interest expenses (again it'll be partly reflected in GDP).
3. Changes in production efficiencies (which should be partly accounted for in GDP with increased production, but lowered costs are not).
4. Changes in SG&A methods that lower costs. For example changing from a mailed catalog business to an internet catalog (printing and postage costs go to $0, the internet system costs that replaced them are much lower).
5. Maybe changes in debt to capital ratios (I'll need to think about this one, it may already factor into GDP in the capital investments area).
And I don't think I'll be able to come up with a workable model any time soon.
|# May 17th, 2005 12:56 PM d_rosnick|
Let me spell it out for you again. I'll use SSA because they are more explicit about it.
1. SSA makes projections for GDP.
2. One component of GDP is worker compensation.
3. SSA makes projections for compensation.
4. Specifically, SSA states that in the long run, compensation remains constant as a share of GDP.
5. Another component of GDP is corporate earnings.
6. If earnings grow faster than GDP in the long run, compensation will have to grow slower than GDP in the long run.
7. Thus, we cannot reconcile faster earnings growth with SSA's projections for GDP and compensation.
8. Thus, IF we are going to project earnings based on SSA assumptions, THEN the most optimistic growth we may project in the long run is growth in line with GDP.
THAT is why I say earnings grow with GDP.
So if you think earnings will grow faster than GDP, you'll have to take it up with SSA. I'm only using what they've given.
|# May 17th, 2005 2:15 PM d_rosnick|
I found the Jorion/Goetzmann paper "Global Stock Markets in the Twentieth Centry" After all, you did provide a URL. I'm stubborn, but I'm not thick. What I cannot seem to find is the work based on that paper. Just because the paper includes US data doesn't prove I'm wrong.
What you say about a cross-country analysis doesn't contradict what I lay out, either.
SSA assumes that compensation grows in step with GDP. Is that true of the countries in the study? I don't know, but if it is not, then there is no contradiction.
I assume a constant dividend payout. Does each country in Jorion's analysis show a constant dividend to price ratio? If not, there is no contradiction. If so, does Jorion control for differences in the dividend to price ratio across countries? If not, there is no contradiction.
I assume a constant price to earnings ratio. Does each country in Jorion's analysis show a constant price to earnings ratio? If not, there is no contradiction. If so, does Jorion control for differences in the price to earnings ratio across countries? If not, there is no contradiction.
So you see, there are many questions that must be answered before anyone can say that Jorion's work contradicts mine. And yes, I will be greatly surprised if Jorion does make those controls.
The question is not how could I be surprised if I don't know Jorion's methodology. Obviously, if I knew the methodology, then I would not be surprised. I hope that much is clear.
|# May 17th, 2005 4:27 PM BrianH|
|OK, you're working the income side of the equations. GDP = GDI (+/- statistical errors)
You're claiming that if the change in corporate profits doesn't match the change in GDP then the change in personal income must vary inversely with corporate profits.
You are leaving out the other components of the GDI side of the equation. These include taxes, depreciation, interest expenses, etc. More or less the same items I mentioned above.
|# May 17th, 2005 5:13 PM d_rosnick|
|Oh good. I think we are making progress.
No. I'm not leaving them out. I'm saying that if compensation is to hold at 56% of GDP then corporate earnings plus everything else can't grow past 44% of GDP.
So you can get earnings all the way to 44% of GDP, but then you have zero rental income, zero interest income...
Just how much of a squeeze can you put on all other types of income?
I hope it is now clear what is at issue.
|# May 17th, 2005 6:26 PM BVBigBro|
|The single biggest beef I have with the model is that it assumes 20% of the private account would be in treasury bonds. Why would someone divert funds from treasury bonds and then then reinvest them in treasury bonds? I think a realistic number for this percentage would be zero. |
|# May 17th, 2005 6:41 PM BVBigBro|
|For that matter, the model assumes 30% private bonds for the account. Given the conservative nature of the investment as a whole, I should think anyone getting advice on the subject would get a recommendation of 100% equities for the private portion of the account. This, I suspect, will radically alter the results. |
|# May 17th, 2005 7:20 PM d_rosnick|
Again, CBO assumptions. The reason it might have been a decent idea at the time was Model 2 called for a 1% annual subsidy. So if Treasuries were 3%, the clawback would only be 2%.
|# May 17th, 2005 9:22 PM BVBigBro|
|Nevetheless, the 50% of investment in bonds is absurd. The 20% in treasury bonds has a 100% chance of losing more than 3%. The 30% in private bonds is also certain to lose money at an undetermined rate.
In short, your calculator assumes that investors will willingly invest 50% of their money in an investment that will lose money, and that they have total knowledge will lose money.
The only reasonable way to analyze the bush plan is with 100% in equities for the privitized portion. I submit that going back to original calculator and inserting a 100% equities investment would give the answer everyone expects.
|# May 17th, 2005 9:57 PM Laura|
|I stumbled across this ad via Ace, who complains that there is a media blackout on this story: 450 economists, including 5 Nobel Laureates have endorsed personal accounts for Social Security. |
|# May 18th, 2005 7:31 AM JohnTant|
|David, I think you're trying to change the subject.
You said overall GDP growth was "tied to" stock returns growth.
Jorion says it isn't GDP growth, but per capita GDP which should be used.
After pointing that out, you started to float all sorts of stuff, from saying that it didn't apply to the US (despite Jorion having used US data in his study) to then bringing up dividend yields (even though Jorion was talking total returns). You then go for the intellectually honest tack and say you can't comment on Jorion's conclusions until you see the specific work he gave to Biggs. Fair enough. But then that makes any "surprise" you have illegitimate. After all, at the end of the day that's floating a comment based on knowledge you say you do not have. And again, you can certainly be "surprised," but in context I don't think that surprise means what you seem to think it means.
With that, I think Jorion's conclusions pass the sniff test. It isn't GDP in and of itself that's the factor, but the nature of the growth. GDP growth based on increased labor force involvement doesn't increase returns as much as increased productivity of the existing labor force. Brian has been making an argument along these lines for awhile now. And Jorion has empirical data to back up his conclusions. All we have from CEPR is an assertion.
With that, I think we're starting to argue in circles. You're saying "uh huh" and I'm saying "nuh uh" at this point. You're welcome to the last word on this, as I think I've made my point.
|# May 18th, 2005 7:49 AM BrianH|
|I would not personally have any money in bond funds right now. We're at below normal rates and the trend is rising rates. That causes the market value of the bonds in the fund to drop and the NAV of the fund to drop. The opposite condition occurred when interest rates were falling and most of the money made in bonds was due to market value increasing...
But that's really beside the point. 50% in bonds is rediculous for a young person saving for retirement. Not so rediculous for someone near retirement. But that's also beside the point since as David said, he was using a sample portfolio used in a CBO example.
I do have many other problems with the calculator. The most obvious is the assumption that company earnings CAN'T grow faster than GNP (even though it often happens). We've beaten that to death and neither of us seem willing to back down.
I have problems with the expected benefits calculations under "current law" since current law requires a benefit cut when the so called trust fund runs out (estimated to be about 27% in the first year). I haven't checked the actual calculation with the SSA's calculators yet, but I'll assume David can do math when given a formula and that it's accurate assuming inflation (and CPI adjustments) and earnings growth match his projections. (I know David, you're going to say they're the CBO projections). But leaving out the cuts is missleading at best.
I have a problem with the limitations on contributions imposed (Yes I've read plan 2) which I don't believe will be enacted in any final version.
I have some problems with the default estimates of fees. Though to his credit, this is adjustable in the calculator.
I suspect the "clawback" provisions will be dropped. Without this provision there would be no contest in the accounts. If you don't know what the clawback is, one of the plans (plan 2) that Democrats usually point to as the "Bush Plan" includes a provision that when you retire, ALL of your contributions + compounded interest at Tbond rates would be deducted from your account balance before you are forced to buy an annuity (at David's 5% fee) with whatever remains.
I have a problem with assuming that a workers first 3 years of work are ages 22-25. This is a bit late in life compared to where most people start work. Many (most?) people have a part time entry level job in high school (I certainly did). Those who don't go to college will usually start work after high school. Say age 18 or 19. Many of those who DO go to college work part time while attending school. 22 assumes that someone went to college and didn't work until after graduation. This may sound a bit nit picky, but even a small amount compounded at David's nominal rate of 5.96% over 47 (from age 20) or 49 (from age 18) years grows to a large amount. ($100 grows to $1841.51 when compounded monthly).
That's it for the moment.
|# May 18th, 2005 8:09 AM pinguino|
|i have a problem with the fact that the calculator on the SSA.GOV site comes up with a number that is 60% of the number estimated by this site for the "current" system. this webpage is grossly overestimating the return of the "current" system.
|# May 18th, 2005 8:30 AM BrianH|
|Oh and I was also overlooking another large driver of corporate profits, International business. Most major corporations also have large overseas component these days. Any profits not produced in the US are not included in GDP. From the latest GDP report:
"GNP includes, and GDP excludes, net receipts of income from the rest of the
That's also going to drive corporate earnings faster than GDP. I had completely forgotten about this until I was reading my company's earnings announcement today.
|# May 18th, 2005 8:45 AM d_rosnick|
Nice try there, but you are totally misrepresenting the state of things.
If you'll excuse me for minding details, I never said "tied to" in this discussion or in the technical notes. I do say in the notes that we "see the nominal return on stocks must be tied directly to nominal growth in the economy. This is clearly a conclusion, based on the derivation and the assumptions leading to it. There are clearly other variables in the formula and that makes a difference. The point is that if one assumes earnings grow with GDP, and if one assumes a particular constant PE ratio and DE ratio (and therefore D/P ratio), then the return is tied to GDP. There is no good reason to assume that the D/P ratio is the same across countries, and I don't. I only assume that it will hold at approximately 2.9% in the U.S. That is why I talk about dividend yields. Dividend yields are part of total returns, and are an important part of any return formula. I fail to see what is inappropriate about discussing them in the context of stock returns. Failure to discuss dividend yields would leave out the part that ties to the real economy. Doing so is not a diversion-- it is going to the heart of the matter. Just because I discuss them does not mean I don't figure that the quotes of Jorion's work referred to total returns.
Now that we have gone over the difference between assumptions and conclusions, let me say that tying earnings to growth is not really my assumption. It is implicit in the exonomic projections for the reasons I explained above. I didn't want to get into an extended explanation in the technical appendix. The point there was to lay out the math in the calculator as best as possible.
Look, if you don't know what it means to be surprised, I don't know why I should bother. If I expect not to see something in his work, but then I see it, I will be surprised. I can't believe we are even discussing this. You say Jorion has empirical data to back up his conclusions, but I have no idea what his assumptions are. Do you understand how this works?
If I assume that GDP is constant, then GDP per-capita in the United States is shrinking over time. I can show this with actual data. That doesn't mean anything because the assumption is poor. I'm not saying Jorion made an assumption that bad, but the point is assumptions do in fact matter.
I have presented much much more than an assertion. I cited the Trustees' assumption and explained how it plays in to the analysis. (As far as SSA assumptions are concerned, I could use far more pessimistic assumptions about earnings without contradiction, but do not.) If you check out the BEA NIPA data, you'll see that profits in the US have actually grown with GDP in the long run. I've shown all the math that follows.
On the other hand, there's a study that nobody seems to have available, let alone an actual citation. Thus providing no opportunity for evaluation.
You may be convinced by Biggs' quotes, but that does not constitute a reasoned argument against anything I say.
|# May 18th, 2005 9:36 AM d_rosnick|
You are correct. Profits may in fact come from other countries, and so the connection to GDP is not quite as strict as I implied. However, I again request that you take a look at the size of these profits, and the size of the foreign component that would be required in order to sustain 6.5-6.8% growth. I think we will find this quickly leads to absurdities.
I very much appreciate your thinking on this matter, though.
|# May 18th, 2005 10:40 AM JohnTant|
|Well, David, I just have to say you were doing OK until this:
I never said "tied to" in this discussion or in the technical notes.
What was one of the very first things you said? In scrolling up, I see you say right off the bat in your note to Kris:
The growth in stocks is tied directly to growth in GDP.
So thanks for your congratulations of "nice try." To that I guess I just have to say "right back atcha!"
|# May 18th, 2005 11:39 AM d_rosnick|
|I asked for forgiveness for attention to detail, but "tied to" is not the same as "tied directly to." The meaning may be the same, but it is a misquote. I thought when I said "If you'll excuse me for minding details," that folks would understand I was not seriously upset about that. I guess not.
I'm not a fan of slipper slope arguments, but if I quoted John as saying "Jorion says GDP growth should be used" that's in a technical sense as much a direct quote as "tied to." Obviously, the meaning is changed in one and not the other, but again I was trying and failing to keep John honest by being comically nitpicky.
The fact John makes an admittedly minor misquote pales in relevance to the fact that he misrepresented the context of the words. My serious point, that it refers is a conclusion, not an assumption, remains.
|# May 18th, 2005 3:52 PM d_rosnick|
I bet you selected the constant-dollar calculator off SSA's site. Funny thing is, if you select current-dollar and deflate to current dollars by hand, you'll probably get a much, much closer number. This is because SSA makes a totally different projection for future wages for the constant-dollar calculator.
Just so you know.
|# May 18th, 2005 6:56 PM BVBigBro|
|The technical details aside, the calculator assumes you will pay back SS at 3% above the rate earned by treasury bonds. Given this assumption, you have to assume a portfolio of 100% equities. Anything else is indefensible. With this change alone, the calculator will produce a very different result.
To answer your original post, Kris, the calculator produces the result it does precisely because it is not investing all of your privitized portion in stocks, and is instead investing 50% of the money in bonds gauranteed to produce a negative return given the calculator's assumptions.
|# May 18th, 2005 9:37 PM d_rosnick|
You have it wrong. The calculator does not assume 3% over Treasuries. The White House backgrounder called for a clawback at Treasury rates (to maintain a sort of(*) revenue neutrality) and that is what I use in the calculator.
(*) I say "sort of" because nothing addresses what happens when the clawback is larger than the worker's defined benefits
|# May 19th, 2005 8:04 AM BrianH|
|I was trying to refind my sources so I could find the "clawback" references and discovered that what I had previously read was a CBO summary of the plan 2 documents produced for the Senate. I don't know if the summary is biased or not, but the summary is at:
As I said, this is the plan that is CLAIMED by Democrats to be the "Bush Plan". I don't think Bush has proposed this as the plan to use. As a matter of fact, I think the only things Bush originally had said is that he wants Congress to do Social Security reform, that it should contain a private accounts option, and that the tax RATE should not be increased. All other options are on the table. In the recent speech he also proposed progressive indexing. That is something the libs have always wanted, "Give the poor more than the rich", but they are having fits over this one.
Anyway, I don't believe the clawback will be part of any final legislation.
Frankly, what David's calculator is showing is the income derived from private investment GAINS which are above the treasury rate. Even using an assummed low rate of return and taking back ALL of the money you put in + 3% interest (more or less crippling the account) the private account comes close to matching the income from SS (and will beat it after you take the 27% benefit cuts that are scheduled) . Frankly, this clawback is the ONLY reasonable way that a PSA WON'T beat Social Security.
David, just for fun, can you remove the "clawback" from you calculator and rerun the numbers?
|# May 19th, 2005 9:29 AM d_rosnick|
|SS with accounts still don't actually beat the payable benefit because that comparison is biased toward the accounts. Why? Because the plan is NOT revenue neutral. The benefit cuts are insufficient to provide make Social Security solvent even if they apply to the disability side as well. At best, the private accounts do nothing toward solvency, and by all indications will
make hurt the solvency of the program.
Thus, to pay the full benefit promised under the Bush plan, Treasury will have to dedicate monies in addition to those needed to pay back the bonds in the trust fund from Treasury to Social Security. Not just a few bucks, either. SSA estimated $1.9 trillion based on the Pozen Plan (that is, the same price indexing, but a different account contribution limit (2% up to $3,000 price indexed, as opposed to 4% up to $1,000 wage indexed.)
That is, the relatively small cuts in the Pozen Plan mask an additional $2 trillion in outside funds needed to keep paying those benefits. The Pozen Plan falls short of it's own scheduled benefits starting in 2030.
Essentially, current law says "3.4 trillion in debt." The current law payable benefit says "3.4 trillion in benefit cuts." The Pozen Plan says "1.9 trillion in debt and 1.5 trillion in cuts."
If it is unfair to the Bush Plan to compare with current-law, it is unfair the other way to compare to payable benefits.
Brian, I'm not sure what it is you're asking me to do by removing the clawback. You want me to show what a great deal private accounts would be if the government simply gave everyone free money for their accounts?
Maybe I should assume that fairies will come with gold and pay all our debts as well.
Yes, it is what the calculator shows. I think it's a horrible idea. But it's not my idea. I didn't make it up-- it's what the White House put out.
In any case, the clawback is the difference between the benefit with no account and the guaranteed benefit with the account.
|# May 19th, 2005 11:01 AM d_rosnick|
|Just to clarify that the clawback is part of the Bush Plan, we have a transcript of the White House backgrounder.
"Now, the way that election is structured, the person comes out ahead if their personal account exceeds a 3 percent real rate of return, which is the rate of return that the trust fund bonds receive." [my emphasis]
The briefer is of course, wrong about who comes out ahead. The statement is only true if there are no fees on the account-- no annual fees, and no price to annuitize.
Now, one might ask, what if the real rate of return on Treasuries is not 3%? Well, the point is actually to claw back at Treasury rates (3% is SSA's projection for the long run) Why do I say that?
"There is a time lag in the cash flow, but in the net indebtedness of the federal government. If the question is, what is the overall impact on the government's debt situation, that would be neutral from the point of enactment."
This pretty clearly means that workers will be borrowing from the government at Treasury rates, or it simply is not true.
It happens to be deceptive in the sense that the White House is asking to have things both ways. They want to say it is revenue neutral (no net cost present value) but it adds quite a bit to the government's cash flow problem. The debt burden at any time is much larger than it would otherwise be.
It also happens to be deceptive because it is not clear what happens when the total to be clawed back exceeds the defined benefit. Presumably, if a worker is getting $1000/month and owes $1200/month, the worker will only pay back $1000/month. Social Security would subsidize the difference.
For the record, the ABC does assume this subsidy.
|# May 19th, 2005 12:30 PM BrianH|
|"You want me to show what a great deal private accounts would be if the government simply gave everyone free money for their accounts? "
Well right there is the crux of the problem. It's not the government giving free money. It's the government taking 4% LESS of the PEOPLE'S MONEY.
What I'm asking is for you to quit distorting the picture.
"Just to clarify that the clawback is part of the Bush Plan, we have a transcript of the White House backgrounder.
Did you actually read the whole quote? Or did you just take that out of context to distort the view?
If you read the 3 questions and answers that include your quote, it's pretty clear that the "SENIOR ADMINISTRATION OFFICIAL" is stating that you would have to get a 3% or better gain on your private account in order to make up for the cut in the guaranteed portion of the benefits. He's NOT saying that they're going to claw back all the money you put in PLUS 3% interest.
I think even you'll admit that we can expect better than 3% return.
""There is a time lag in the cash flow, but in the net indebtedness of the federal government. If the question is, what is the overall impact on the government's debt situation, that would be neutral from the point of enactment."
This pretty clearly means that workers will be borrowing from the government at Treasury rates, or it simply is not true. "
No, what you are missing is that as workers are paying into the system, the system is incuring a debt to pay them back. Moving to private accounts will affect short term cash flows, but will also reduce the Government's future obligations (debt).
If you insist that there is a clawback provision. At least break it out when you present the numbers or you are distorting what is occurring.
|# May 19th, 2005 1:30 PM d_rosnick|
Hahaha. No. See, today most OASDI payroll taxes go to pay current beneficiaries. If we instead divert 4 of 12.4 percentage points of those taxes into private accounts, that doesn't leave enough for current beneficiaries.
You can't pay current retirees and yourself with the same dollar. So the plan is for you to make up for the money the government will have to borrow to make up for it.
I read the whole transcript, and no, I am not at all taking it out of context. there are two cuts. There are the cuts associated with the Pozen indexing. Then there are additional cuts for folks taking the account. That's the clawback. I'm sorry to be the one to break it to you, but that's the way it works. The reason that there is only one cut described in the transcript is because it only outlined the account proposal, and steered away from solvency. Remember how proponents kept saying folks on the other side were being dishonest by assuming price indexing? That's why. The White House refused to talk about that part of things. The cut being discussed is the clawback. That is how revenue neutrality is (sort of) maintained.
The clawback is broken out. It is clearly visible in the graph (as the difference in guaranteed benefits under the Bush Plan with and without private accounts.) It is also listed under "Loss due to benefit offset" in the table at the bottom.
In any case, there is no distortion. Everyone is subject to the Pozen indexing. If you take the account, then your guaranteed benefits are reduced by an additional amount.
I know it looks bad, but that's the plan so far.
As far as the government debt goes, private accounts (if you include the clawback provision) do worsen short term cash flows, but (almost) reduce future obligations dollar-for-dollar (plus interest) for the debt incurred as a result of your putting $1000 into your account rather than paying current beneficiaries.
The problem is that there will always exist a real (not implicit) unpaid debt on the books. That is, there will always be people with future obligations to pay back the diverted payroll taxes. (Which they will pay back during retirement via the clawback provision-- an implicit stream of future revenues.)
There is nothing distorting about including the clawback provision. A retiree will absolutely, positively not get as large a guaranteed benefit if the worker took a private account than if the worker had not. That's the plan.
|# May 19th, 2005 1:35 PM kris|
|David & Brian,
Wouldn't you like to take a break and discuss American Idol instead ;-)
|# May 19th, 2005 2:34 PM BrianH|
|"Hahaha. No. See, today most OASDI payroll taxes go to pay current beneficiaries. If we instead divert 4 of 12.4 percentage points of those taxes into private accounts, that doesn't leave enough for current beneficiaries."
HA HA HA HA HA! Yourself.
I'm aware of how a PAYG system works.
I'm also aware that currently about 1/3 of the payroll taxes collected are diverted to the General Fund instead of paying for current benefits. Guess what, 1/3 of 12.4% happens to be a little more than the 4% that is being talked about.
What is doesn't leave room for is the extra $200 mill that Ted Kennedy and friends have been spending that is supposed to be paying for our retirement.
"You can't pay current retirees and yourself with the same dollar. So the plan is for you to make up for the money the government will have to borrow to make up for it. "
See figure 1 (old programmers will probably understand that)
"The clawback is broken out. It is clearly visible in the graph (as the difference in guaranteed benefits under the Bush Plan with and without private accounts.) It is also listed under "Loss due to benefit offset" in the table at the bottom. "
Sorry, the only way this could be possible is if you are assuming a PSA gain of less than the 3% rate. The reduced guaranteed benefit is calculated as:
PSA guaranteed benefit = Benefit without PSA (including Posen or any other changes) - (contributions + 3% compounding).
(This is not the clawback that is usually described which actually deducts this amount from your PSA account AND reduces the guaranteed portion of the benefit.)
If you are assuming even a 3.0001% growth in the PSA, your third column MUST be larger than your middle column and it's not in ANY of the cases I've run through it.
You're rigging the calculator.
Sorry Kris, I'm the odd ball of the group here. I don't follow American idol. American Choppers maybe...
|# May 19th, 2005 2:49 PM BrianH|
|- (contributions + 3% compounding).
- annuitized (contributions + 3% compounding)
|# May 19th, 2005 4:20 PM d_rosnick|
|Sorry, Brian, but you are wrong again.
The estimated 2005 Social security surplus is less than 12% of non-interest income, or less than 1.5% of taxable payroll. That means that if you diverted more than 1.5% of taxable payroll, either money would have to come from General Revenues or there would be a cut in benefits to current retirees.
Presumably, Congress would treat the OASDI Trust Fund, no differently than it treated the SMI (Medicare Part B) trust fund last year and make the transfer as mandated by law. However, this does leaves Treasury (as you said) about $190 billion shy of expectations ($118 billion in transfers, plus $69 billion in expected new loans from Social Security.)
So yes, this would make a bad general budget situation worse.
It's really hard to blame the Democrats for today's budget deficits. The Republicans control the White House and both Houses of Congress. So get a grip.
The growth in the private account would have to be over the clawback rate by enough to pay for annual and annuitization fees, which actually means something close to 3.9% (3.3% CBO rate + 0.3% annual fees + 5% one time annuitization fee, which is about equivalent to another 0.3% per year.) The reason there is little gain in total benefits for younger workers is the portfolio returns about 4%, barely more than the clawback and fees.
The reason older workers actually lose is that they don't have the the same time to let their savings compound and thus their accounts struggle to make enough to pay the annuitization fee. (So the equivalent annual rate for the annuitization, rather than 0.3%, is higher.)
Of course, they could come out ahead by annuitizing less than the full amount in the account.
Again, I'm not rigging the calculator. I've laid it all out there. If you can find an error, please please let me know. But you have to get your facts and your math straight first.
|# May 19th, 2005 4:23 PM d_rosnick|
|Brian, an item I missed above.
The clawback does not come out of the private account. Supposedly, it comes only out of guaranteed benefits. I don't know what you think you read where. I can only imagine you are confusing the clawback with the general benefit cuts that apply to everyone.
|# May 19th, 2005 4:32 PM BrianH|
I originally got the $200 billion number from a SSA spreadsheet of their data (must have been a year or 2 old).
It had been a while since I checked the numbers so I went back to the SSA for a source:
The 2005 Trustees Report
"Total benefits paid in 2004 were $493 billion. Income was $658 billion"
658 - 493 = 165 (yes a little less than the $200 billion I said.
165/658 = .25 or about 1/4 rather than 1/3
12.4 X .25 = 3.1% of the tax I pay is currently excess over benefits.
|# May 19th, 2005 4:41 PM BrianH|
|Sorry David your numbers don't add up.
The Bush plan you referenced sets the reduction at straight 3% the same as the rate paid by the Treasury to the SSA (not inflation adjusted). The CBO estimate for growth is inflation adjusted and after fees.
If you assume any rate above 3% and still get less than the reduction in benefits, your calculator is rigged.
|# May 19th, 2005 4:59 PM d_rosnick|
|That $658 billion includes interest on the Trust Fund which went right back to Treasury. The $493 did not include interest.
So if you want to know the effect on Treasury, you have to take out the interest income.
Brian, I do in fact assume a clawback at Treasury rates and I do not say otherwise. I'm trying to point out that the net effect of the clawback and fees is such that the return on the account before fees must be higher than the Treasury rate in order to break even.
I still compute everything at the correct rates.
|# May 19th, 2005 5:16 PM d_rosnick|
|Brian, you clearly have no idea what you're talking about.
3% is the long-term, inflation-adjusted rate on Treasuries assumed by SSA. CBO puts that rate at 3.3%.
Saying it is 3% unadjusted for inflation is just silly. SSA assumes 2.8% inflation in the long run. That would mean a real interest rate of only 0.2% on Treasuries.
(Admittedy, Treasuries are pretty amazingly low right now -- about 4.1% unadjusted, compared to 3.7% in the CPI-W over the last year. That's only 0.4% over inflation, and ought to tell you something about just how bad the economy really is today. But the fact is Social Security actually assumes a long term rate of about 5.8% unadjusted for inflation. See Table II.C1.)
In any case, if I assumed it to be 3% before inflation, I'd have to adjust the return on the accounts down as well.
Again, you really should get your facts straight.
|# May 19th, 2005 8:03 PM BrianH|
The bond they are talking about are the "trust fund" bonds. The treasury pays 3% on them.
Your calculator is still rigged.
|# May 19th, 2005 9:35 PM d_rosnick|
I'm sorry you are so confused. Feel free to email me if you decide you would prefer to learn about Social Security rather than make wild accusations you cannot back up.
|# May 19th, 2005 10:07 PM kris|
|Whoa. Clearly, everyone who has been posting on this thread has been providing a lot of sources to back up their opinions, so check the condescending attitude please. |
|# May 20th, 2005 2:26 AM BrianH|
In this case, I have to admit the error. I had read 3% interest on the special treasury bonds. It is actually 3% real (after inflation). The table in the reports he referenced clearly shows this.
However, the point still remains that David is assuming the PSA account will grow at less than this treasury rate. Otherwise, the far right column CAN'T be smaller than the center column. Admittedly, this variable interest rate means the PSA must generate a higher nominal return for break even than I had been using. But again stocks historically have returned 6.8% real so an expectation of better than 3% real is reasonable.
|# May 20th, 2005 9:48 AM d_rosnick|
Thank you for checking yourself on this matter. I must admit I get a little frustrated sometimes when I am asked to argue at length about facts which are clearly laid out.
Let's just do a little back-of-the-envelope calculation to see how the accounts might lose based on long-term numbers.
Gov't bonds: 3.3% over CPI-W
Corporate Bonds: 3.8% over CPI-W
Stocks: 4.35% over CPI-W
Annual Fees: 0.3% of annual assets
Annuitization Fee: 5% of returement assets
The CPI-W, bond rates, annual fees and the 50/30/20 stock/corp bond/govt bond mix all come from CBO. The stock return we've been over at length. A 5% fee on the annuity is low. For details, you might check out
Mitchell., O., J. Poterba, M. Warshawsky and J. Brown. "New Evidence on the Money's Worth of Individual Annuities." American Economic Review, December 1999.
The 50/30/20 leads to just under 4% real return before fees. (6.3% nominal) After annual fees, that's only 6%.
Now, a dollar put into a private account for 20 years turns into $3.21. $3.05 after annuitization. The clawback on the dollar is $2.96, leaving only nine cents. After inflation, just $0.06. A small gain, but better than nothing.
On the other hand, a dollar compounded for only 10 years turns into $1.79. $1.70 after annuitization. The clawback is $1.72, resulting in a net loss of total benefits.
Ironically, the worker would do much better if the clawback came out of the private account in a lump-sum rather than annuitized and taken out of defined benefits.
Over 20 years, the account less clawback is 25 cents, or 15 cents after annuitization and inflation.
Over 10 years, the account less clawback is 7 cents, or 5 cents after annuitization and inflation.
The reason for this is that the annuitized amount is much, much smaller (with a larger defined benefit from Social Security.) Unfortunately, the White House has reacted strongly to any suggestion that the plan is to claw back from the private account, rather than defined benefits.
In any case, if you still think that stocks will do better, you can always add in the 2.45% premium in the calculator and get that 6.8% return.
|# May 20th, 2005 11:28 AM BrianH|
When I make a mistake I try to admit it (and correct it). In this case I was wrong about the bonds being a fixed rate (I had missread that).
I still think you're low balling your estimate of stock returns, but as neither of us are going to back down on that one... I'll quit beating that drum for now.
I have a problem with you fee assumptions. The 5% annuity fee seems a bit high. In fact I did some random checking of fixed annuities being offered for sale and found no front end fees on any of them. I also missed any mention of annuity fees in any of the plan documents, summaries or discussions. I guess you're assuming these fees based on the article you referenced?
The .3% administration fee seems about right at face value. It's a little higher than some low cost funds and quite a lot lower than some high cost funds. I'll admit this annual fee would require a slightly higher return on the PSA to match the T-bond return. No real dispute here.
Finally, even if I were to receive a slightly lower return on a PSA (which I still dispute), I'd sign up for it on principal. I really don't think it's right to require a decreasing number of workers to pay my retirement benefits just so I can live a little better. I think it's an undue burden on the next few generations.