Money Supply (M2) and the Stock Market

http://www.cxoadvisory.com/economic-indicators/money-supply-m2-and-the-stock-market/

Money Supply (M2) and the Stock Market
Posted in Economic Indicators
July 5, 2010Share Post



Some investing experts track change in money supply as a potentially important indicator of future stock market behavior. When the money supply grows (shrinks), they theorize, asset prices go up (down). Or, money supply growth drives inflation, thereby elevating discount rates and depressing equity valuations. One measure of money supply is the M2 money stock, which consists of currency, checking accounts, saving accounts, small certificates of deposit and retail money market mutual funds. Is there a reliable relationship between historical variation in M2 and stock market returns? Using weekly data for seasonally adjusted M2 and the S&P 500 Index during November 1980 through June 2010 (1,546 weeks), we find that:

The following chart depicts M2 and the S&P 500 index over the entire sample period. Both series have generally risen as the economy grows (and inflates), but the stock market is far more volatile than M2. Visual inspection is not helpful in discovering any relationship between the two series.

To dig deeper, we relate changes in the two series.



The following scatter plot relates weekly change in M2 to same-week change in the S&P 500 Index over the entire sample period. One of the observations in September 2001 is a two-week relationship because U.S. stock markets were closed. The Pearson correlation between the two series is 0.01, and the R-squared statistic is 0.00, indicating that weekly variations in M2 explain nothing about concurrent weekly stock market movements.

Might the effect of money supply be evident only over longer intervals?

Relating the 4-week change in M2 and same-interval change in the S&P 500 Index (sample size 387 non-overlapping intervals), the Pearson correlation is -0.09 and the R-squared statistic 0.01.

Relating the 13-week change in M2 and same-interval change in the S&P 500 Index (sample size 122 non-overlapping intervals), the Pearson correlation is -0.14 and the R-squared statistic 0.02.

Relating the annual change in M2 (selecting the M2 reports nearest to year-ends) and same-year change in the S&P 500 Index (sample size 30 years), the Pearson correlation is -0.16 and the R-squared statistic 0.03.

It appears that strength of relationship may grow as measurement interval increases, but sample size decreases as measurement interval increases. The relationship becomes more negative, suggesting a fear of inflation/elevated discount rate effect. None of the relationships indicate a compelling material (tradable) relationship between changes in M2 and U.S. stock market returns.



A reader suggested looking at the relationship between real change in M2 (inflation-adjusted by subtracting the contemporaneous 12-month inflation rate derived from the all-items Consumer Price Index) versus stock market returns. The next scatter plot relates annual changes in M2, both nominal and real, to same-year change in the S&P 500 Index over the entire sample period. The relationship between real change in M2 and stock returns is weaker than that between nominal change in M2 and stock returns. A shared dependence of the two variables on the inflation rate or small sample size may explain the difference.

Might there be a non-linear but reliable effect of M2 variations on stock returns?



The next chart summarizes average S&P 500 Index returns across quintiles of ranked changes in M2 over intervals of one, four and 13 weeks over the entire sample period. At the longer measurement intervals, there may be some tendency for stronger returns when M2 growth is lowest (again suggesting the effect of change in M2 on discount rate). However, the progressions across ranges of M2 growth are not completely systematic, and there are only 24-25 observations per quintile for the 13-week interval.

Might changes in M2 lead changes in stock market returns?



The final three charts explore potential M2-stock market lead-lag relationships for non-overlapping intervals of one, four and 13 weeks over the entire sample period by offsetting changes in M2 relative to changes in the S&P 500 Index. Specifically,

The top chart relates the weekly change in M2 to the weekly change in the S&P 500 Index for offsets ranging from the stock market leads M2 by 13 weeks (-13) to M2 leads the stock market by 13 weeks (13).

The middle chart relates the 4-week change in M2 to the 4-week change in the S&P 500 Index for offsets ranging from the stock market leads M2 by about six months (-6) to M2 leads the stock market by about six months (6).

The bottom chart relates the 13-week change in M2 to the 13-week change in the S&P 500 Index for offsets ranging from the stock market leads M2 by about four quarters (-4) to M2 leads the stock market by about four quarters (4).

Correlations are generally small, and the lead-lag variations appear to be mostly noise. The third chart suggests the stock market may anticipate long-term increases in M2 (by declining), but the level of any such anticipation is economically weak.

In summary, evidence from simple tests does not support a belief that M2 money stock is a useful indicator for short-term or intermediate-term future stock market behavior.

The percentage of stocks in the S&P 500 now trading above their 50-day moving averages 7/2/2010

The percentage of stocks in the S&P 500 now trading above their 50-day moving averages is down to 4%. At the March 2009 lows, the reading only got down to 5%, so that gives investors a good idea of just how extreme this decline has gotten.

MA 50 and MA200 death cross

那，这个就是著名的所谓MA50和MA200的death cross。测试结果，其实不咋的，相信没人能等个几十年一直坚持到盈利。这是trend following system的通病。


The "Death Cross"

Sometimes a technical signal will occur that generates an unusual amount of questions, and over the past week that has been the case with an imminent crossover in some longer-term moving averages.

Last week, we looked at the 10-week/30-week crossover in the S&P as well as the slope of the 150-day moving average. Both had garnered media attention as imminent sell signals for stocks, though their records were spotty.

The current nail-biter is the horribly-named "death cross", when the 50-day moving average crosses below the 200-day moving average. We've looked at this in the past, along with its mirror-image "golden cross", and usually we come out un-impressed.



Like usual, let's forget the textbooks and opinions and instead look at history. Here are the results of going short on a bearish crossover and covering when they gave the opposite signal:

Net Profit/Loss: +646 points (versus +1,056 for buy-and-hold)
Winning %: 35%
Average Winner: 14.2%
Average Loser: -7.9%
Maximum Drawdown: -357 points (versus -889 for buy-and-hold)

Note that this strategy calls for going short, so a positive outcome means that stocks declined.

Most trend-following strategies, especially using moving averages, have a very low winning percentage, especially especially those that sell short, so a 35% winning rate isn't all that unusual. At least it made money.

But...

Out of the gross 646 points of profit, 589 of them came from one trade (the 2008 bear market). And 459 of them came from one other trade (the 2001-2002 bear market). So those two trade alone accounted for 1048 points...which means that the other 44 trades netted -402 points.

That is the great frustration with trend following - you'd better take every trade, and you'd better trade it for a long time (or trade a lot of markets), because you never know when that one trade is going to come along to finally make you profitable.

In fact, if you started trading this system back in 1928, you would have had to wait until 2002 to show a profit (there were some decades that were profitable, but overall the losses from the unprofitable ones ate them away).

Here is the breakdown by decade:

Average Return % Winning %
1930s -1.4% 44%
1940s +0.2% 50%
1950s -2.5% 0%
1960s -1.3% 50%
1970s +2.8% 40%
1980s -2.5% 25%
1990s -7.1% 0%
2000s +15.5% 50%

The strategy was piss-poor every decade, until we finally arrived in the 2000s with the wide, steady swings in prices. Up until the past 10 years, you wouldn't have done much at all by trying to time the market using these moving averages.

Instead of waiting for a bullish crossover, here are the results the given number of days after sell short after the "death cross":


1 Week
Later
2 Weeks
Later
1 Month
Later
3 Months
Later
6 Months
Later
1 Year
Later
Avg Return +0.4% +0.3% +1.2% -2.0% -3.0% -2.7%
% Positive 48% 43% 59% 52% 50% 43%

It really didn't make much different at all in the numbers if the 200-day average was sloping up or down at the time of the cross - except for one month later, it was a poor strategy for selling short.

Usually, stocks rebounded in the short- and long-term after these crossovers, though the signals were inconsistent. That's the problem with these things - they will keep you out of the worst bear markets, and if that's your biggest concern then by all means reduce risk when they occur. But you'd better have a strategy for getting back in, because more often than not they do nothing but whipsaw back and forth.

Head Shoulders Pattern 6/29/2010

Major Indices: “Looking UGLY, Billy Ray!”
By Barry Ritholtz - June 29th, 2010, 12:13PM

Ron Griess of The Chart Store recently mentioned to me that “one of the most bullish patterns around was a failed Head & Shoulders.”

Those comments are appropriate this morning, as we may soon find out if those H&S’s on the major stock indexes fail or manifest themselves as bearish signals.

Nasdaq, S&P, Dow Jones charts below.

ECRI Leading Index as an Investment Tool

Another interesting chart, via David Rosenberg

The ECRI not only leads but is also more timely than the ISM since the data are released weekly and the index covers the whole economy, not just manufacturing.

We divided the ECRI into four different phases:

1. From the trough to zero (coming out of recession).
2. From zero to the peak (“sweet spot” of the cycle — from the end of the recession to the cycle peak in growth).
3. From the peak back to zero (past the peak in growth; economy slows but not back in recession).
4. Zero back to the negative trough (heading back into recession).

>
ECRI Weekly Leading Indicator: Growth Rate
click for ginormous graphic

courtesy of Gluskin Sheff

>
More David Rosenberg after the jump…


From late 2008 to the fall of 2009, we were in Phase 2. Since last October, we have been in Phase 3 and it looks like we could be here for a while.

In Phase 3, historically, the S&P 500 has provided tiny positive returns (average price appreciation of +1.3%). Tech, industrials and energy are the top performing cyclicals, along with health care and staples in the more defensive area. This cyclical-defensive barbell works well — basic materials, consumer discretionary, financials and utilities tend to lag the most. In the credit market, this is a period to be focussing on reducing duration and scaling into quality: Baa spreads tighten, on average, by 11bps but widen in the high-yield space by an average of 13bps.

Nothing is to say that we will automatically revert to Phase 4 just because we are in Phase 3 right now, although we are only 5.1% away from zero, even with policy rates still close to 0%. Then again, this was a credit cycle, not a rates cycle. It was credit that created the 2003-07 boom, and it was credit that created the 2007-2008 bust. A 5.5% peak in the funds rate was hardly the culprit, and we know that it was not a 0% rate in late 2008 that triggered the 2009 renewal in economic activity and investor risk appetite but rather the Fed’s massive expansion of its balance sheet and the government’s willingness to push the fiscal deficit to record peace-time levels. In this sense, any analysis that relies on the classic post-WWII recession-recovery experience — even this one — has to be viewed in the context of a secular credit contraction, which began two years ago.

In Phase 4, the S&P 500 on average declines 6.3% with eight of the 10 sectors declining — a barbell of being long energy on the cyclical side and consumer staples on the defensive side has worked well.

Consumer cyclicals, technology, industrials and financials are crushed in this segment of the ECRI cycle; telecom, utilities and health care do not perform as well as staples but are areas where at least you don’t typically get beaten up (for relative-return folks). The CRB is down an average of 3% but gold and oil tend to be supported by a weaker U.S. dollar. The yield on the 10-year note rallies an average of almost 40bps; as with equities, corporate bonds are hurt in this quadrant — Baa spreads widen about 60bps and high-yield by close to 100bps. We have to be mindful that this can very well be the next phase of the cycle even without the Fed raising rates.

The ECRI bottomed this cycle a good four months before the equity market did and for those folks that paid attention, like Jim Grant, kudos to them.

Because from the trough to zero — Phase 1 — the equity market rallies on average by 12% with all 10 sectors in the green column, led by tech, consumer discretionary and basic materials. Energy, telecom and utilities tend to lag behind. Financials are basically market performers. The government bond market is still rallying in this segment and the curve is steepening — that along with a slight softening in the U.S. dollar provides a positive liquidity backdrop, which in turn is conducive to spread narrowing in the credit market (average tightening of around 40bps in investment-grade and 200bps in junk).

The TRIN index 6/10/2010

The TRIN index (often called the Arms index) is a market indicator developed by Richard Arms that measures market breadth and volume. A more detailed description of the indicator can be found at Investopedia, but to describe it quickly, the TRIN measures the ratio of up stocks to down stocks and then divides that number by the ratio of volume in up stocks to volume in down stocks. When the ratio is above one it indicates that volume in declining stocks is greater than the volume in rising stocks, while readings below one indicate that more volume is going into up stocks than down stocks.

In order to eliminate the noise in the daily readings of the TRIN, we often look at the ten-day average. This helps to give us a clearer picture of short-term sentiment. In the chart below, we highlight the 10-day average TRIN over the last several years. As shown, at a current level of 2.53 there has only been one other period since 2002 where the indicator was higher than it currently is. Surprisingly, that one occurrence was not during the heart of the bear market from October 2007 through March 2009. It actually came right before, in March 2007 when the S&P 500 saw a 5% correction as the first cracks of the sub-prime crisis began to emerge. Following that correction, the S&P 500 quickly recovered and hit new highs within a matter of weeks. It wasn't until seven months later, in October 2007 that the S&P 500 finally peaked at 1,565.15.

Some Indicators

NYADV = NYSE Advancing Issues.
NYDEC = NYSE Decline Issues.
NYAD = NYADV - NYDEC.
NYMO = NYAD EMA19 - NYAD EMA39
NYSI = NYMO + NYMO(1) + NYMO(2) + ... + NYMO(n)

NYUPV = NYSE Up Volume.
NYDNV = NYSE Down Volume.
NYUD = NYUPV - NYDNV

NYHGH = NYSE New 52 Week Highs.
NYLOW = NYSE New 52 Wekk Lows.
NYHL = NYHGH - NYLOW
RHNYA = NYHGH / ( NYHGH + NYLOW )
NYHLR = NYHGH / NYLOW
NYHILO = 10 Day MA of Record High Percent Index

NYA50 = NYSE Stocks Above 50 Day Moving Average
NYA200 = NYSE Stocks Above 200 Day Moving Average
NYA50R = NYSE Percent of Stocks Above 50 Day Moving Average
NYA200R = NYSE Percent of Stocks Above 200 Day Moving Average