Wilmott Forums - SP500 vs. GDP = "Great Depression 08-18"

Forum Navigation:

FORUMS > Economics Forum

< refresh >

Topic Title: SP500 vs. GDP = "Great Depression 08-18"
Created On Sat Feb 21, 09 02:34 PM

Topic View:


	Pages: [ 1 2 >> Next ]

BullBear
Senior Member

Posts: 853
Joined: Aug 2007

Sat Feb 21, 09 02:34 PM

SP500 Vs. GDP

-------------------------
The rise of the libertarians

Posts: 129
Joined: Oct 2005

Sat Feb 21, 09 04:16 PM

I didn't see anything conclusive in this analysis... correlation does not imply causation!

Edited: Sat Aug 22, 09 at 01:33 PM by seppar

Quote

Top

Bottom

Traden4Alpha
Senior Member

Posts: 14301
Joined: Sep 2002

Sat Feb 21, 09 08:37 PM

I applaud this attempt to examine the relationship between the GDP and the S&P 500, but it contains at least four serious methodological flaws:

1. This should be a log-log regression, not a linear one. The use of linear regression and plotting creates a biased view of the effects which makes the current excursions appear more extreme than they are. The linear regression also implies an absurdity -- that the "expected" value of the S&P 500 should have been less than zero when the GDP was smaller than 600 billion.

2. If we are believe this analysis, then the markets were as much as 50% under valued (relative to GDP) for a 15-year period of about 1980 to 1995. On the one hand. I doubt this was true (Greenspan warned of irrational exuberance in 1996!), which suggests that the natural trend line for S&P vs. GDP is quite a bit lower than the analysis suggests. On the other hand, if the market really was undervalued, then it suggests that there's nothing to stop the markets from being 50% undervalued for the next 15 years (i.e., we could have a "U" shaped recovery in GDP, but an "L" in equities). In either case, the market might drop further or remain low relative to GDP.

3. The conclusion of undervaluation rests, in part, on the assumption that GDP will continue to grow. Regressing S&P vs. GDP removes time as a variable of consideration. But as a prospective investor, I'm less concerned about S&P vs. GDP, and more concerned about S&P vs. time. Even if the regressed relation is true, a "buy" recommendation rests on a resumption of GDP growth. If GDP continues to decline or does not grow, then the S&P might further decline or fail to grow.

4. The biggest issue is that GDP and S&P 500 are extremely different measures of the economy. GDP is largely an historical, objective measure of revenue in the U.S. economy. S&P 500 is a future-focused, market measure of the supply and demand for earnings in a segment of the global economy. There are at least five categories of differences that would muddy any relationship between the two variables: 1) historical vs. future-focused; 2) objective measure vs. market measure; 3) revenue vs. earnings; 4) total economy vs. public equities; 5) U.S. vs. global economy. Each of these five categories of differences implies the S&P vs. GDP relationship can vary in time without implying an incorrect valuation of the instantaneous value of S&P 500 relative to the instantaneous value of the GDP.

-------------------------
"It often happens that a player carries out a deep and complicated calculation, but fails to spot something elementary right at the first move." -- grandmaster Alexander Kotov --inscribed on gift chess sets given by Amaranth hedge fund.

Edited: Sat Feb 21, 09 at 08:45 PM by Traden4Alpha

Quote

Top

Bottom

Anthis
Senior Member

Posts: 4073
Joined: Oct 2001

Sat Feb 21, 09 09:53 PM

Quote
On the other hand, if the market really was undervalued, then it suggests that there's nothing to stop the markets from being 50% undervalued for the next 15 years (i.e., we could have a "U" shaped recovery in GDP, but an "L" in equities). In either case, the market might drop further or remain low relative to GDP.

If this combination implies low P/Es and high dividend yields, i cant see anything wrong in this. An L shaped market implies that the market waits for the current 30-40 somethings to accumulate wealth before it "takes off", especially if we assume reinvestment of the high dividends.

Quote

Top

Bottom

BullBear
Senior Member

Posts: 853
Joined: Aug 2007

Sun Feb 22, 09 12:48 AM

Quote
Originally posted by: seppar
I didn't see anything conclusive in this analysis... correlation does not imply causation!

> Anything better than a "destructive depression" means the market is tremendously undervalued!

If you believe that the market is undervalued then buy it. I suggest you to start (I mean buying) with General Motors : @ $1.77 it looks like a real bargain!

This isn't supposed to address causality or spurious relations...

Already bought (but not GM).

-------------------------
The rise of the libertarians

Quote

Top

Bottom

BullBear
Senior Member

Posts: 853
Joined: Aug 2007

Sun Feb 22, 09 01:09 AM

Quote
Originally posted by: Traden4Alpha
I applaud this attempt to examine the relationship between the GDP and the S&P 500, but it contains at least four serious methodological flaws:

1. This should be a log-log regression, not a linear one. The use of linear regression and plotting creates a biased view of the effects which makes the current excursions appear more extreme than they are. The linear regression also implies an absurdity -- that the "expected" value of the S&P 500 should have been less than zero when the GDP was smaller than 600 billion.

2. If we are believe this analysis, then the markets were as much as 50% under valued (relative to GDP) for a 15-year period of about 1980 to 1995. On the one hand. I doubt this was true (Greenspan warned of irrational exuberance in 1996!), which suggests that the natural trend line for S&P vs. GDP is quite a bit lower than the analysis suggests. On the other hand, if the market really was undervalued, then it suggests that there's nothing to stop the markets from being 50% undervalued for the next 15 years (i.e., we could have a "U" shaped recovery in GDP, but an "L" in equities). In either case, the market might drop further or remain low relative to GDP.

3. The conclusion of undervaluation rests, in part, on the assumption that GDP will continue to grow. Regressing S&P vs. GDP removes time as a variable of consideration. But as a prospective investor, I'm less concerned about S&P vs. GDP, and more concerned about S&P vs. time. Even if the regressed relation is true, a "buy" recommendation rests on a resumption of GDP growth. If GDP continues to decline or does not grow, then the S&P might further decline or fail to grow.

4. The biggest issue is that GDP and S&P 500 are extremely different measures of the economy. GDP is largely an historical, objective measure of revenue in the U.S. economy. S&P 500 is a future-focused, market measure of the supply and demand for earnings in a segment of the global economy. There are at least five categories of differences that would muddy any relationship between the two variables: 1) historical vs. future-focused; 2) objective measure vs. market measure; 3) revenue vs. earnings; 4) total economy vs. public equities; 5) U.S. vs. global economy. Each of these five categories of differences implies the S&P vs. GDP relationship can vary in time without implying an incorrect valuation of the instantaneous value of S&P 500 relative to the instantaneous value of the GDP.

I agree T4A.
Lots of variables missing... Yes, log-log reg makes sense. Causality & Integration tests missing. This wasn't supposed to be a scientific analysis just a descriptive check on both variables.

I was just trying to confirm that some (firm-by-firm) valuations are discounting a really dark scenario (which is possible).

Bottom line, I was trying to show how some traders appear to over-react to macroeconomic news. Last quarter contraction seemed like doom's day was coming... While there's been a contraction in GDP last quarter it also increased 80% last decade and the SP500 is at the same level of the 90's.

Also, when we look at some individual valuations (which is my focus, but not mentioned in the post) one can notice the huge discount rate implied by the market. I'm not saying it is wrong (although I think we've fallen too far, too fast) just trying to show that it seems that investors have already priced in plenty of bad news.

If the economy begins to recover soon (and don't drop much before it resumes growth again) it appears that the SP500 would pop. But who knows... No causality analysis here although I think both series possess some degree of (non-spurious) correlation.

-------------------------
The rise of the libertarians

Quote

Top

Bottom

BullBear
Senior Member

Posts: 853
Joined: Aug 2007

Sun Feb 22, 09 01:36 PM

Just tried to improve the simplistic analysis with truncated samples...

zooming the sp500 vs. gdp #1

zooming the sp500 vs. gdp #2

-------------------------
The rise of the libertarians

Quote

Top

Bottom

Traden4Alpha
Senior Member

Posts: 14301
Joined: Sep 2002

Sun Feb 22, 09 02:07 PM

Quote
Originally posted by: BullBear
Quote

I agree T4A.
Lots of variables missing... Yes, log-log reg makes sense. Causality & Integration tests missing. This wasn't supposed to be a scientific analysis just a descriptive check on both variables.

I was just trying to confirm that some (firm-by-firm) valuations are discounting a really dark scenario (which is possible).

Bottom line, I was trying to show how some traders appear to over-react to macroeconomic news. Last quarter contraction seemed like doom's day was coming... While there's been a contraction in GDP last quarter it also increased 80% last decade and the SP500 is at the same level of the 90's.

Also, when we look at some individual valuations (which is my focus, but not mentioned in the post) one can notice the huge discount rate implied by the market. I'm not saying it is wrong (although I think we've fallen too far, too fast) just trying to show that it seems that investors have already priced in plenty of bad news.

If the economy begins to recover soon (and don't drop much before it resumes growth again) it appears that the SP500 would pop. But who knows... No causality analysis here although I think both series possess some degree of (non-spurious) correlation.
One of the challenges in assessing the valuation of the S&P 500 relative to recent valuations and to GDP is in the differences between a revenue number (GDP) and a value-of-earnings number such as the S&P. In particular the high levels of consumer debt and government debt create different effects on the GDP and the S&P. The recent accumulation of debt has boosted the recent historical revenue of the economy (GDP) but will be especially destructive of future earnings. To the extent that consumers are directing greater fractions of disposable income to repaying debt (either personal or governmental), then earnings will remain lackluster. To the extent that the government is forced to increase business taxes to repay debt, then corporate earnings cannot grow. Never forget that GDP counts all economic activity, but that shareholders are last in line for profits.

Second, the S&P 500 represents the market price of risky earnings. To the extent that prospective investors eschew risk (by psychology or regulation), then the S&P will track a lower-than-historical PE ratio. I suspect we will enter an era calling for greater reserves and less leverage. Consumers will be forced to maintain higher equity ratios in housing and less total consumer debt, which will lower their investment in stocks. Banks will be regulated to maintain higher reserves and a less risky portfolio. I doubt that HF will receive as much investment or be able to raise as much debt for leverage. With less demand for earnings, the S&P won't rise.

Third, other indicators suggest that we aren't at the bottom. By some measures, home prices are still significantly overvalued relative to long-term historical levels and especially overvalued in this environment of a long-term surplus of total housing and a long-term shortage of qualified buyers. Also, some analyses suggest that price-to-sale ratios of stocks haven't reached historical bottoms yet (although these analyses may not properly reflect changes in cost structure strategies). Finally, the situation in Europe looks really bad to me. The crisis with Eastern European debt looks comparable (or worse) than the subprime mortgage problem in the U.S.

We still have a ways to fall. (Of course, the Great Depression included 6 "bull markets" on the way to the bottom, so who knows what will happen)

-------------------------
"It often happens that a player carries out a deep and complicated calculation, but fails to spot something elementary right at the first move." -- grandmaster Alexander Kotov --inscribed on gift chess sets given by Amaranth hedge fund.

Edited: Sun Feb 22, 09 at 02:37 PM by Traden4Alpha

Quote

Top

Bottom

BullBear
Senior Member

Posts: 853
Joined: Aug 2007

Mon Feb 23, 09 08:13 PM

I'm not calling for a bottom. No one knows where the bottom is.

When the GDP grows 80% and a measure of mkt cap for the 500 strongest firms is halved it seems that the discount rate is already huge. It is discounting a huge depression, deleverage and shareholders' value destruction. This can happen! Not saying it's wrongly priced in nor that it can't fall even further.

Hey, do you have any idea on the price-to-book-value historical behavior?

I think P/E is not the right measure, now. I would take a look at price-to-sales, price-to-cashflow, and price-to-book-value. Including and excluding financials.

If you have any idea on these metrics (at the SP500 level) could you please post it?

-------------------------
The rise of the libertarians

Edited: Mon Feb 23, 09 at 08:14 PM by BullBear

Quote

Top

Bottom

Traden4Alpha
Senior Member

Posts: 14301
Joined: Sep 2002

Tue Feb 24, 09 12:24 AM

A couple of interesting articles with charts (but not data) on the ratios you speak of are:

http://www.hussmanfunds.com/rsi/medianvaluation.htm
http://www.safehaven.com/article-231.htm

Most of these charts seem to confirm that we are in a bubble and are not under-valued with respect to long-term levels and are probably still over-valued given where we are in the business cycle.

Price to sales is a reasonably useful ratio because sales figures are much less volatile than earnings. Sales also tends to track the GDP (modulated by the market share of the companies). The problem with P/S is that its one level removed from what investors really care about. Interpreting P/S means first guessing the long-term profit margin on sales and then guessing investors' preferences for those earnings.

Price-to-book is even more indirect and a lot messier due to leverage effects, asset write-downs, effects of intangibles, and the unknown future productivity of net assets.

Cashflow is nice (and very valuable during a down-turn) but it's also a very volatile number.

-------------------------
"It often happens that a player carries out a deep and complicated calculation, but fails to spot something elementary right at the first move." -- grandmaster Alexander Kotov --inscribed on gift chess sets given by Amaranth hedge fund.

Posts: 129
Joined: Oct 2005

Tue Feb 24, 09 12:37 AM

Quote

Top

Bottom

BullBear
Senior Member

Posts: 853
Joined: Aug 2007

Tue Feb 24, 09 01:19 PM

Quote
Originally posted by: Traden4Alpha
A couple of interesting articles with charts (but not data) on the ratios you speak of are:

http://www.hussmanfunds.com/rsi/medianvaluation.htm
http://www.safehaven.com/article-231.htm

Most of these charts seem to confirm that we are in a bubble and are not under-valued with respect to long-term levels and are probably still over-valued given where we are in the business cycle.

Price to sales is a reasonably useful ratio because sales figures are much less volatile than earnings. Sales also tends to track the GDP (modulated by the market share of the companies). The problem with P/S is that its one level removed from what investors really care about. Interpreting P/S means first guessing the long-term profit margin on sales and then guessing investors' preferences for those earnings.

Price-to-book is even more indirect and a lot messier due to leverage effects, asset write-downs, effects of intangibles, and the unknown future productivity of net assets.

Cashflow is nice (and very valuable during a down-turn) but it's also a very volatile number.

Thanks for the links.

Not going to argue on aggregate valuations.

just a remark:
Price-to-book... depends on the sector and accounting frauds. After 08 write-downs it should be a cleaner measure. If balance sheets are cleaner now then they were before I believe the proper way is to look at the absolute value of the ratio and not on a historical basis. Well, if there's more fraud now then before than it's the opposite.

-------------------------
The rise of the libertarians

Quote

Top

Bottom

BullBear
Senior Member

Posts: 853
Joined: Aug 2007

Tue Feb 24, 09 01:25 PM

Quote
Originally posted by: seppar
Also a related article from Bloomberg:

Dividends Falling Means S&P 500 Is Still Expensive

In general, the "buy and hold" investment strategy is broken. With HF, Leveraged Short/Long ETF, High-frequency trading, the equity is just a vehicle intended to make short-term profits. If you want to benefit from a dividend stream, then better buy bonds - the risk-adjusted return is higher and entry-exit decisions are easier to take.

Theoretically, dividend yields and dividend amounts should mean nothing to firms' valuations. The cash will get out of the firm so the mkt cap will be adjusted. In the current situation, when there's little risk taking then:
- firms that are able to keep paying dividends: will attract more investors since it will signal that the firm is robust and can release funds to shareholders
- firms that are needing more equity: slashing dividends increases equity but signals a tough environment for the firm. (typical case of Financials and highly-leveraged firms)

Anyone saying that dividends falling, per se, means S&P500 is expensive dunno what's saying!

-------------------------
The rise of the libertarians

Quote

Top

Bottom

Traden4Alpha
Senior Member

Posts: 14301
Joined: Sep 2002

Tue Feb 24, 09 03:08 PM

Quote
Originally posted by: BullBear
just a remark:
Price-to-book... depends on the sector and accounting frauds. After 08 write-downs it should be a cleaner measure. If balance sheets are cleaner now then they were before I believe the proper way is to look at the absolute value of the ratio and not on a historical basis. Well, if there's more fraud now then before than it's the opposite.

Good points. P/B varies by industry and the quality of accounting. In estimating the right P/B we need to consider the effects of the three main terms in book value on future earnings.

1) Total Assets: In general, assets, when utilized at some rate lead to revenues. Asset utilization and pricing power strongly influence the revenues of the company. The price factors of those asset-related inputs affect the earnings of the company. Asset utilization also comes with variable costs (the labor and materials consumed by each unit of asset utilization) that define part of the cost structure. Next, asset depreciation reduces earnings (but not cashflow). Finally, to the extent that a company with valuable assets will always have some market value, regardless of earnings, the assets of the company contribute directly to the market capitalization. But if the assets are illiquid, subject to deflation, or misvalued on the books, then those assets would be discounted.

2) Intangible Assets: The book value equation subtracts these, but that is probably not always right for purposes of valuation. I know that some think that the intangible assets category is just an accounting gimmick to justify overpaying for an acquisition, but sometimes a purchased entity really is worth more than it's book value. To the extent that a company buys valuable trademarks, patents, customer populations, innovative facilities, etc., the intangible assets should NOT be subtracted to estimate a book value for the purposes of P/B-based valuation. Again, we have the issue of asset utilization,pricing power (which may be higher than normal for good intangible assets), asset depreciation, and any resale value.

3) Liabilities: Obviously, liabilities subtract from earnings, but the key is the degree of this subtraction. The amortization schedule and interest rate structure of liabilities is a crucial variable in estimating future earnings. To the extent that companies are forced to accept onerous terms on debt, their earnings will be meager for the term of the loan (unless inflation subdues that debt). Moreover, to the extent that a liability has a specific due date and the firm is unable to cover that liability, the firm might declare bankruptcy despite maintaining a positive book value (i.e., a P/B of zero is valid in some situations).

My point is that P/B is a very indirect measure that fails to properly reflect a number of crucial earnings drivers. P/B requires numerous assumptions about the productivity of assets, has the wrong sign for the effects of intangible assets, and does not reflect the impact of servicing liabilities. Thus, I'd not use P/B on highly leveraged firms or those with substantive intangible assets.

-------------------------
"It often happens that a player carries out a deep and complicated calculation, but fails to spot something elementary right at the first move." -- grandmaster Alexander Kotov --inscribed on gift chess sets given by Amaranth hedge fund.

Edited: Tue Feb 24, 09 at 03:12 PM by Traden4Alpha

Quote

Top

Bottom

meteor
Senior Member

Posts: 319
Joined: Sep 2004

Tue Feb 24, 09 10:28 PM

Quote
Originally posted by: BullBear
Just tried to improve the simplistic analysis with truncated samples...

zooming the sp500 vs. gdp #1

zooming the sp500 vs. gdp #2

You got to be kidding me!
What do you mean by improving?
You are just fitting points, nothing else.
Your analysis is just pure garbage.

-------------------------
On m'a toujours dit qu'c'etait les meilleurs qui partaient en premiers, j'ai donc choisi d'etre pourri pour rester.

Quote

Top

Bottom

Traden4Alpha
Senior Member

Posts: 14301
Joined: Sep 2002

Wed Feb 25, 09 03:07 PM

Quote
Originally posted by: BullBear
Quote
Originally posted by: seppar
Also a related article from Bloomberg:

Dividends Falling Means S&P 500 Is Still Expensive

In general, the "buy and hold" investment strategy is broken. With HF, Leveraged Short/Long ETF, High-frequency trading, the equity is just a vehicle intended to make short-term profits. If you want to benefit from a dividend stream, then better buy bonds - the risk-adjusted return is higher and entry-exit decisions are easier to take.

Theoretically, dividend yields and dividend amounts should mean nothing to firms' valuations. The cash will get out of the firm so the mkt cap will be adjusted. In the current situation, when there's little risk taking then:
- firms that are able to keep paying dividends: will attract more investors since it will signal that the firm is robust and can release funds to shareholders
- firms that are needing more equity: slashing dividends increases equity but signals a tough environment for the firm. (typical case of Financials and highly-leveraged firms)

Anyone saying that dividends falling, per se, means S&P500 is expensive dunno what's saying!

Maybe, maybe not.

Four reasons for assigning a higher valuation to dividend-paying stocks and, by extension, judging stocks on a price-to-dividend ratio basis:

1) preferential taxation of qualified dividends
2) psychological preferences for dividends (dividends = steady repayment of invested capital)
3) amelioration of liquidity and valuation risks
4) management's signal of confidence in steady earnings

By that logic, a falling dividend yield implies that stocks are "more expensive."

Of course, the flip-side is that sticky dividends are quite dangerous. Some believe that the banks, in particular, have dramatically increased their risks of bankruptcy by maintaining generous dividend payouts as a show of confidence. Perhaps these firms have already distributed their book value of shareholders?

-------------------------
"It often happens that a player carries out a deep and complicated calculation, but fails to spot something elementary right at the first move." -- grandmaster Alexander Kotov --inscribed on gift chess sets given by Amaranth hedge fund.

Quote

Top

Bottom

BullBear
Senior Member

Posts: 853
Joined: Aug 2007

Thu Feb 26, 09 04:09 PM

Why I don't like to argue on market index valuation metrics:
The S&P Gets Its Earnings Wrong

S&P doesn't take weights into account!?

There's also a flaw when looking at historical market index metrics - the firms and index composition is always changing. Existent firms now are different from those on 1920... There's also a Darwinian evolution in the markets which preserve the stronger firms (Coca-Cola, Disney, McDonalds, Nike, Sony, Microsoft, Exxon Mobil, IBM) and get rid of weaker firms (lehman, bear, enron,...). Most robust and old firms are alive for decades... Their franchise value is unique so will be the franchise value of survivors.

For example, a Bank able to survive and quickly expand its activity will boost its franchise value locally and overseas - "Solid as rock" type of marketing.

-------------------------
The rise of the libertarians

Quote

Top

Bottom

Traden4Alpha
Senior Member

Posts: 14301
Joined: Sep 2002

Thu Feb 26, 09 05:12 PM

Quote
Originally posted by: BullBear
Why I don't like to argue on market index valuation metrics:
The S&P Gets Its Earnings Wrong

S&P doesn't take weights into account!?

The existing sum-of-dollar-earnings DOES take weighting into account. Jeremy Siegel's logic is so deeply flawed, that what he proposes would actually induce a weighting squared term in which money-making companies (whose share prices and market caps are, no doubt, higher) would effectively dominate his calculation of earnings. It's mathematical rose-coloured glasses.

The proper way to look at this is to realize that each unit of the S&P 500 buys a fixed fraction of each of the underlying companies. Buy the right amount of S&P 500 index shares, and you'd own 100% of every company on the index. The weighting means you'd own 100% of the small companies (in a small $-fraction of the investment) as well as 100% of the big ones (in a large $-fraction of the investment), too. It also means that you'd accrue 100% of the dollar earnings and losses of each company. Thus the earning on the S&P are the sum of the dollar value of earning of the components and each share in S&P earns a piece of that.

A second way to look at this is to imagine putting $100 into one share of a two-stock index in which one stock has 9X the market cap of the other. Imagine that both companies have a 1,000,000 shares and that the price of the big company is $90/share and the small company is $10 (i.e., market caps of $90 million and $10 million respectively). Then the big company reports earnings of $9 million and the little company reports losses of $9 million. Total earnings are zero. Your $100 investment just earned $9/sh from the big company, but lost $9/sh from the small one. The net gain to you to your share of the index is zero. Seigel's calculation would insist that you earned $7.2/share by double weighting the gains of the big company.

Quote
Originally posted by: BullBear
There's also a flaw when looking at historical market index metrics - the firms and index composition is always changing. Existent firms now are different from those on 1920... There's also a Darwinian evolution in the markets which preserve the stronger firms (Coca-Cola, Disney, McDonalds, Nike, Sony, Microsoft, Exxon Mobil, IBM) and get rid of weaker firms (lehman, bear, enron,...). Most robust and old firms are alive for decades... Their franchise value is unique so will be the franchise value of survivors.

For example, a Bank able to survive and quickly expand its activity will boost its franchise value locally and overseas - "Solid as rock" type of marketing.

The way to look at this is that the S&P 500 is really just a mutual fund with S&P being the fund manager in picking the investments. Yes, the fund does have turnover, but that does not mean that the fund/index doesn't have a consistent NAV that is defined over time (as long as the index turnover events are properly priced into the index).

-------------------------
"It often happens that a player carries out a deep and complicated calculation, but fails to spot something elementary right at the first move." -- grandmaster Alexander Kotov --inscribed on gift chess sets given by Amaranth hedge fund.