Unpleasant Facts and Other Musings

Michael Pettis recently posted an interesting piece on the Shanghai market.  He makes the following points.

1. In markets, there are fundamental or value investors, arbitrage traders and speculators. Speculators react to information about supply and demand, as well as the expected short term impact of fundamental information. Value investors buy assets in order to capture the profits generated by these assets over a longer time frame.
2. In the Shanghai market, transaction costs are too high for more relative value traders.  Trustworthy fundamental information is too scarce, predicting government action is too important [This isn’t too different from the US market over the past crisis] and valuations are usually too high for fundamental or value investors to participate. This means that speculators dominate the market.
3. Without value investors buying when the market is cheap and selling when the market is expensive, the market will be more volatile.

After giving many examples of how the Shanghai market trades on non-economic information, he concludes:

“A market driven almost exclusively by speculators, and with little to no participation by fundamental or value investors, is not a market that pays much attention to long-term growth prospects.  It is driven largely by fads, technical factors, liquidity shifts, and government signaling.

So what does this year’s crash in the Shanghai stock market tell us?  It might be saying something about the impact of the European crisis on export earnings.  It might suggest that liquidity in the system is being driven into real estate rather than into stocks.  It may reflect contagion and nervousness about the fall of stock markets abroad.

But we should be cautious about reading too much into it.  In fact attempts by Beijing to hammer down real estate bubbles in the primary cities without addressing underlying liquidity expansion may simply push asset price bubbles elsewhere, and this could easily cause a surge in the Shanghai stock markets.  But this should not then be interpreted as signaling a surge in the economy.

Shanghai’s markets will go up and down, but they are not driven by investor evaluation of long-term growth prospects.  China does not yet posses the tools to make such evaluation useful, so be careful about reading too much into the stock market numbers.”

This does not mean that the Shanghai composite is useless. It means that it is a combined measure of liquidity and government intervention.  If the fund managers who speculate on government policy or liquidity are successful in making money, then the market is giving useful information. The information may be slightly different from other markets (although even developed world markets are more focused on the next few quarterly earnings than on growth prospects 5 to 10 ears out), but it is still important.  It is particularly important for those who think that China’s recent growth during the downturn is unsustainable because it is built on real estate speculation and government spending without a clear way to transition back to a more normal convergence style growth. A sharp fall in equity prices is one of the indicators of a tighter monetary policy (or that a tighter monetary policy is finally having an impact), and further extreme moves could be signs of an actual deflationary unwind. Any stock market move has many causes, but it definitely gives us useful information.

A – H share premium on top, Shanghai A Shares normalized to Jan 3^rd, 2010 on bottom. The premium of the Shanghai A shares over the foreign traded H shares is another measure of the relative excess liquidity in China’s market compared to the rest of the world. Source: Palantir.

After Katrina and 9/11, some people thought that higher GDP from rebuilding what was broken could be a silver lining. Since the accident and subsequent spill obviously had a negative effect on the wealth of our economy, it will be interesting to see if anyone is insane enough to suggest that the GDP impact of cleanup and containment spending (already totally around a billion dollars) is a silver lining to the spill. Broken window fallacies are popular, but in this case the negative impact of the spill on other industries such as fishing, tourism and other deepwater oil drilling projects might obviously outweigh the supposed benefits.

Analyzing the market capitalization movements of a certain company (which will remain nameless for legal reasons) suspected to be liable for clean up costs and economic claims, adjusting for the market capitalization change in their entire sector (which shall also remain nameless, I’m counting on my readers to figure out which company and which sector I am talking about here), suggests that the liability and clean up costs may total about $40 billion dollars.  Of course, there are a few caveats to this calculation.  For one thing, a lot of the market cap drop is probably related to potential sanctions or damage to their brand image and investors not wanting to hold a stock that is no longer a simple {insert sector} company. For another, the market might be pricing in a high chance of relatively contained costs and a small chance of very large clean up costs.   Still, 40 billion dollars is a reasonable cap so the “positive” GDP impact is at most 0.3% of GDP.

A lot of the basic market indicators have been flashing panic signs recently. One example of this panic is the VIX, a measure of expected S&P 500 volatility over the next month, has spiked up from just over 15 to over 35 today (It hit 45 last week) in a little over a month. When the VIX is purified for recent price action in the S&P 500, the recent volatility spike suggests that people should be buying the market. However, there is increasing pressure in the financial system and buying against a panicky market while the financial system’s pressure is accelerating can be very risky.  

There are a few measures that can be used to measure financial system pressure. Equity prices and volatility, the former going down and the latter going up, are one sign. Another is credit default swaps (CDS). CDS for the average major American banks rose between 50 to 100 basis points over the past month.  A less volatile measure of market stability is the LIBOR-OIS spread. This is the spread between the rate banks lend to each other on the interbank market and the effective federal funds rate. Not only does this spread indicate stress, but according to research published by the St. Louis Fed “the LIBOR-OIS spread has been the summary indicator showing the “illiquidity waves” that severely impaired money markets in 2007 and 2008.” This time the illiquidity stress is coming from European banks exposed to sovereign risk, as well as other financial institutions that are exposed to distressed European banks (The flight to quality towards Switzerland banks caused short term interest rates there to briefly go negative, hitting -0.25% on May 25^th). 

The chart bellow shows the LIBOR-OIS, which fell through most of 2009, has rose over 20 basis points in the past month.

 

 

Greece has been in the news again recently. The 5 year credit default swaps have jumped from 344 basis points last Friday to 444 basis points today, surpassing Iceland as the most worrisome sovereign debt in Western Europe. One hundred basis points in one week is a relatively large move, so it is expected that the impact of a worsening situation in Greece should impact both currency and equity markets. 

The more trouble Greece is in, the more people worry about the future of the Euro. However, it doesn’t take a probability of a currency crisis to drive the currency lower; it only takes an expectation of a relatively weak monetary policy in Europe to support Greece.

The worry that the Greek crisis will spill over into the financial system, along with its effect on the currency, has made the S&P outperform the Euro Stoxx 50 in dollar terms whenever Greece’s CDS spikes.

The rolling two month differences correlations have been relatively steady at 0.2 and 0.3 for the past 4 months (0.4 and 0.5 when looking at financial stocks). There are obviously other factors affecting currency and relative equity market performance, but the change in Greek CDS has been having a definitive impact. Those who don’t like trading the Greek CDS directly might consider a currency hedged long short equity approach.

The aggregate amount of consumer credit in the economy dropped by 11.5 billion dollars in February, the economists surveyed expected only a 0.7 billion dollar drop.  For those people thinking that the credit correction is completely over, the chart below puts the amount of consumer credit relative to GDP in historical context.  There is still further room for a decline in consumer credit.

I came across two interesting papers on capital gains taxes and their effect on the stock market.

The first paper, “Capitalization of capital gains taxes: evidence from stock price reactions to the 1997 rate reduction” was published in 1999 and was written by Mark H. Lang and Douglas A. Shackelford. 

This paper uses data from 1997 to try to get to the bottom of a few different views on capital gains taxes. The three theories are as follows:

1. When capital gains taxes are reduced, shares are more valuable to individuals and so share prices rise.  Theoretically, this effect should be larger when share repurchases or liquidation of assets are expected. It should also be larger if stock holders are likely to be individuals who are subject to capital gains taxes.
2. Capital gains taxes only tax assets that are sold, so they create a lock in effect on current owners. A reduction of the capital gains tax diminishes this lock in effect by reducing the tax on sellers, so a reduction of the tax should drive the value of stocks down.
3. If stocks are valued as the sum of their future dividends, or if the marginal investors are unaffected by capital gains tax rates, then changes in the capital gains tax rate shouldn’t have any effect at all.

The study found that during the week of the announced 1997 capital gains but not dividends tax cut, non-dividend paying stocks outperformed dividend paying stocks. This result is consistent with the first theory, where lower capital gains taxes increase the value of assets and vice versa.  The paper also failed to find evidence of a lock-in effect, as stocks with prior gains didn’t decline after the new capital gains tax rate went into effect.

The second paper was published in 2001 by James M. Poterba and Scott J. Weisbenner wrote “Capital Gains Tax Rules, Tax-loss Trading, and Turn-of-the-year Returns” 

This paper combines an analysis of end of the year trading anomalies with tax-loss trading. One theory is that losing trades will be sold at the end of the year in order to counter the tax implications of realized gains in other parts of the portfolio. Unfortunately, this same behavior can also be explained by “window dressing”, in which firms sell losers so their investors don’t see bad stock picks in their year end statements.  The paper attempts to distinguish the two by looking at how end of the year effects react to changes in tax laws.  A capital gains tax change would only affect individual investors while it wouldn’t affect window dressing at all, as window dressing may be done by managers for untaxed institutions as well managers for individual investors.  

The study looked at three different regimes.

Regime 1: Six month short term holding period, long term losses 100 deducible against adjusted gross income

Regime 2: Six month short term holding period, long term losses 50 deductible against adjusted gross income

Regime 3: 12 month holding period.

Stocks with losses at the start of the year exhibited less year end anomalies when 6 months was the demarcation before short term and long term holding. The paper doesn't disprove window dressing, but it does show that tax laws impact year end trading.

Data from the Federal Reserve shows that debt service ratios have declined into the end of 2009. 

Debt service ratio looks at debt payments to disposable personal income while the financial obligation ratio includes other obligatory payments such as property tax, automobile leases, rental payments and homeowner’s insurance. These ratios peaked in the first quarter of 2008 and have each dropped about 1.3%.

Looking specifically at homeowners, the drop from 2008 is seen to have been driven equally by both consumer and household related payments.  However, it can be argued that mortgage debt has more to fall before it reaches historical averages.

Interestingly enough, the financial obligation ratio of renters is lower than its historical averages. This is probably due to selection bias, as many people with a predisposition for borrowing money figured out that becoming homeowners was the easiest way to access more debt. The higher average financial obligation ratio of renters can be explained by renters earning less because they are younger than homeowners and because a sizable fraction of homeowners don’t have any mortgage to pay while all renters have to pay rent.

As I mentioned in my introductory post, the basic pessimistic chart of complete white male median income stagnation is very misleading.  However, the picture is bad enough that it does bear an explanation.

This analysis will focus on white male personal income in order to avoid complications from the change in household size and the impact of the changing gender and racial components of the workforce. Real median income growth from 1973 was negative, while real mean income growth occurred at only a rate of 0.6% a year. This means that mean income growth doesn’t come close to explaining the increasing gap to between per capita GDP growth and real median income growth. Unless GDP per capita growth has been completely disconnected from the well being of the average person in the economy, there should be some way to reconcile the two.

Some people think that the explanation is that the rich have rigged the economy in their favor so all of the improvements have gone to them. If this explanation was true, we would expect to see mean income diverge much from median income than it actually does.  If the rich getting richer is the reason for flat wages and it is not fully captured by looking at mean income instead of median income, then corporate profits are the next logical place to look. The below chart shows that there has been no drastic change in corporate profits/GDP since 1973.

While looking at GDP per capita we should acknowledge that there is a problem when comparing real GDP and real personal income from the census; these time series are deflated by different indices. The GDP deflator has not risen nearly as much as the CPI index, as the CPI Index counts the oil increase while the GDP deflator only looks at domestic goods.

Deflating the mean male income by the GDP deflator instead of CPI gets us a little bit closer to the increase in GDP per capita, but there is still a 55% difference that has yet to be explained.

A large part of the gap is explained by the changing nature of pay over time. In 1973, total compensation consisted of 73% of personal income and this dropped to 65% in 2008.  Wages and salaries consisted of 87.4% of total compensation, dropping to 81.4% in 2008.

If the wages had remained constant as a percentage of personal income, the median income would be 22% higher today. This approach yields a lower value to benefits that I found when looking at the employment cost index (Which by some measures has increased by almost 250% since 1981, explaining all the difference between median income and per capita GDP growth).  When the changing nature of pay is combined with the other factors, this helps close the gap between per capita GDP growth and the change in while male income since 1973.

Much of the rest of the difference can be explained by the changing workforce participation numbers.

When adjusting for this additional factor, the difference between median income and per capita GDP is almost fully explained.

At times, the adjusted factor is higher than the per capita GDP, but since white males earn on average more than females and other ethnic groups this is to be expected. The below table explains why median income differs from per capita GDP, but it also explains why real income has been flat in the census data since 1973.

From this table, we see that when looking at why median income and per capita GDP diverge, the changing nature of pay from total income to benefits is the most important variable, followed by changes in the distribution of income, differences in inflation calculation and the changing nature of the workforce. If the changing nature of pay is broken up into wages vs. total compensation and other forms of income, then benefits would account for 11% of the change with transfer payments and asset income explaining the rest.  Understanding why the real median income has been stagnant is interesting, but the situation is much more complicated than the simple graph would lead naïve readers to believe.

"The greatest shortcoming of the human race is our inability to understand the exponential function." 

-Albert A. Bartlett

Exponential growth is often widely underestimated. Here are a few issues where many people make mistakes because they don’t understand the simple math of exponential growth.

1. Economic Growth: When there are trade offs between efficiency and equality, many people today think that the taxes, regulation and redistribution are worth a slightly lower growth rate.  However, when this trade off is applied over a long time period, the results can be staggering. If the choice was made in 1870 to have more equality at a cost of 1 percentage point of growth a year, America in 1990 would be no richer than Mexico.
2. Entitlement Spending and National Debt: As I have pointed out previously, the United States is headed for very high debt levels if entitlement spending is not reformed.  One very simple way to fix this is to index entitlement benefits to inflation and not income. The growth of the economy would make it easy to pay for a safety net at today’s living standards. Unfortunately, this would only work for Social Security and not Medicare as the medical system is structured in a way that leads to health care inflation greater than that of the real economy.  Additionally, there is another problem when the net national debt reaches 100% of GDP. If the market perception of the debt turns negative and nominal interest rates remain higher than nominal GDP growth, then there is no way for the economy to grow itself out of debt.  This is the current situation with Greece, and Japan isn’t doing too much better.
3. Personal Finance and Pension Plans: If a prudent investor can make 10% real returns in a year, then they can turn 50 thousand dollars into over 1.6 million dollars after 35 years. This simple math explains how many of the rich people today consist of those who have saved and invested prudently. On the other hand, a supposedly fully funded pension fund planning on a world of 8% real returns that finds itself in a world of 4% real returns will find itself underfunded by over 75% 35 years later (In this case, the people making pension return assumptions are underestimating how much they matter, they just know that their books look better if they assume a higher return). Robin Hanson has been proposing that people don't give to the future because they don't care about it, but it may also be that they do not fully understand the impact of exponential growth*.
4. Overpopulation and increasing Resource Consumption: Overpopulation does not seem to be the exponential problem that we once thought it was. Once become rich enough, their population growth rate slows down. The UK’s Ministry of Defense 2008 Strategic Trends report expects the population to level out at around 9 billion people between 2050 and 2100 (page 25).  While overpopulation is itself not a problem, the exponential economic growth of these emerging economies are coincident with an exponential increase in demand for resources and these limited resources present constraints on growth.

Having established that exponential growth rates are important, here is a handy rule of thumb that will give an intuitive understanding of exponential growth. To calculate the doubling time of an exponentially growing series, take 70 (or 69.3 to be exact) and divide it by the growth rate. This means that a 10% growth rate leads to a doubling every 7 years, a 7% growth rate is a doubling every 10 years and a 3.5% growth rate is a doubling every 20 years.

*It is also possible that someone who both cares about the future and understands exponential growth might think that there were existential problems for the current society that are significant enough to reduce the probability of a far future donation from ever paying off.

While thinking about the CDS model in general and the Greek situation in particular, I realized that I left out a key variable from my previous analysis: Net Debt Interest Payments.

This variable should be very useful, as it implicitly measures net debt as well as the part of the deficit that isn’t changed by fiscal policy.  It also measures interest rates in the previous time period, so the information content contains last year’s CDS prices as well. While using data on which countries were considered at risk last year’s will give us a good idea of which countries are at risk this year, it only helps so much because it doesn’t explain why the country was in trouble in the first place.

However, this is a good example of Soros’s theory of economic reflexivity.  Countries that the markets see as having problems have to pay higher interest payments, making their fundamentals deteriorate further, which in turn drives interest rates even higher.  It will be interesting to see how Greece breaks out of this vicious cycle.