Nonfarm Payroll Data Is Not A Useful Indicator For Timing The Stock Market

|
 |  Includes: DIA, SPY
by: On Second Look

The Employment Situation report comes out monthly at 8:30am, EDT on the first Friday of each month. The release includes a significant amount of data. April's release on May 4th, 2012 included 38 pages of survey data. Probably the most watched (by investors) data item in this release is the change in nonfarm payrolls (NFP). April's release showed a gain of 115,000 jobs, well below the estimate of 160,000 and below what is considered necessary for a healthy U.S. economy. The S&P 500 closed down 1.6% the day of this announcement. (A link to the current release can be found here)

The release is a major news event each month. It typically captures a front page headline and gets considerable airtime on television. Many consider it a key barometer of economic health and therefore an important indicator for financial markets. The question I'm interested in discussing is whether or not this indicator is at all useful for timing the stock market.

The data is presented seasonally adjusted, as a month to month change in the actual number of jobs. The data include a "birth/death" adjustment which typically gets picked apart as it makes various assumptions and estimates about jobs from new and closing businesses. I have used the revised data in Bloomberg. Bloomberg presents the data monthly since 1939. The month to month data is extremely volatile and subject to big revisions. For those who are statistically inclined, the reported number often fails to meet statistical significance. To try to smooth this, I have focused on the 6 month moving average. Below I use a totally unscientific method whereby I have identified what I consider to be the big market moves since 1965 (in nominal terms). I chose 1965 since it captures the bear market of the 70's, the bull market of the 80's and 90's and the current malaise since 2000. Looking at the S&P 500 during this period there were a total of 6 major drops of 25%+ (I look at intraday lows) and 7 major rallies. I am interested to know if a change in the nonfarm payrolls can provide an accurate lead indicator to investors. We will see if it works at timing the market tops, the bottoms, and if it avoids false signals. One of the toughest determinations is deciding what constitutes a "signal" when looking at the data. I'm going to be completely unscientific and rely on simple observation. One could torture the data to make it confess - however a simple look is usually far more powerful.

Period 1 - November 1968 to May 1970 - a ~35% market drop ...

Click to enlarge
(Click to enlarge)

The NFP peak in March 1969 lagged the market peak in November 1968 by 4 months. The very slow topping out of the NFP index most likely would not have sounded any alarms until late 1969. By that time the market had fallen nearly 20% from peak.

Period 2 - May 1970 to January 1973 - a ~70%+ market rally ...

Click to enlarge
(Click to enlarge)

The NFP trough in October 1970 lagged the market trough in May 1970 by 5 months. The market had bounced considerably by this time.

Period 3 - January 1973 to October 1974 - a ~40% market drop …

Click to enlarge
(Click to enlarge)

Here again, the NFP peak in March 1973 lagged the S&P 500 peak in January of 1973 by 2 months. The move down was then very coincident. It may have been difficult to identify the "sell signal" as the NFP index did not obviously breakdown until late 1973.

Period 4 - October 1974 to September 1976 - a ~75% market rally

Click to enlarge
(Click to enlarge)

The NFP trough in April 1975 lagged the market trough in October 1974 by 7 months. The market had bounced considerably by this time.

Period 5 - March 1978 to November 1980 - a ~60% market rally

Click to enlarge
(Click to enlarge)

The NFP data actually seemed to have an inverse relationship to the market during this period. The NFP fell dramatically from its peak in July 1978 until the trough 2 years later in July 1980. During this time the S&P 500 actually rallied strongly. Relying on NFP during this rally may have caused an investor to miss it.

Period 6 - November 1980 to August 1982 - a ~25% market drop

Click to enlarge
(Click to enlarge)

This was similar to the relationship in Period 3 above. The NFP peaked in January of 1981, 3 months after the November 1980 peak in the S&P 500. Following the peaks, the two data series moved down together.

Period 7 - August 1982 to August 1987 - a ~230% market rally

Click to enlarge
(Click to enlarge)

The NFP troughed in November of 1982. This was 4 months after the S&P bottomed in August of 1982. The NFP then went on to give a pretty massive false signal in 1984 and actually fell for nearly 2.5 years while the market rallied into 1986.

Period 8 - August 1987 to December 1987 - a ~30% market drop

Click to enlarge
(Click to enlarge)

NFP was of no use in trying to predict the 1987 crash and subsequent rebound.

Period 9 - December 1987 to March 2000 - a 580% market rally

(see above - NFP failed to signal the beginning of this rally)

Period 10 - March 2000 to October 2002 - a ~50% market drop

Click to enlarge
(Click to enlarge)

NFP peaked at about the same time as the S&P 500 in March of 2000. It was solidly coincident.

Period 11 - October 2002 to October 2007 - a ~100% market rally

Click to enlarge
(Click to enlarge)

NFP bottomed in February of 2002 and led the October 2002 bottom in the S&P 500 by 8 months.

Period 12 - October 2007 to March 2009 - a ~55% market drop

Click to enlarge
(Click to enlarge)

NFP peaked in April 2006, 1.5 years ahead of the S&P 500 peak in October of 2007. Given the significant lead time and the magnitude of the move in the S&P 500 between the two peaks, it is not clear that this would have been particularly useful.

Period 13 - March 2009 to May 2012 (present) a ~110% market rally

Click to enlarge
(Click to enlarge)

NFP bottomed in April of 2009, 1 month after the S&P 500 bottom in March.

False Signals

There were definitely false signals. They include the late 1970's, 1984 to 1986, the period around the recession in 1990/1991, the slowdown during 1994 to 1996 and others. There were so many false signals that it would be unwise to make a market call based on NFP data.

Conclusion:

Nonfarm Payrolls may be an important measure for economic activity but it's historically been a poor lead indicator of the significant market turning points. On my quick and dirty analysis, the Nonfarm Payrolls measure only led 2 of the 13 major turning points. It was coincident in 1 and lagged or failed to provide any signal of the market turns in the remaining 10 instances. The lags ranged in time from 1 to 7 months. Moreover, there were a lot of false signals where the NFP moved in the exact opposite direction as the S&P 500. In my opinion, Nonfarm Payrolls is a terrible indicator of future market performance and I largely ignore it.

Nonfarm Payrolls, 6 Month Moving Average

# of instances it led the S&P500 turn

2

# of instances it was coincident with the S&P500 turn

1

# of instances it lagged the S&P500 turn

7

Provided a False Signal

At least 4

Failed to provide a signal at a major turning point

3

Click to enlarge

The full period …

Click to enlarge
(Click to enlarge)

Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.