Presidential Elections and Market Returns

By David G. Booth

September 2004

Every four years or so, we are asked about the effects of a presidential election on stock and bond returns. The Fama/French factor returns provide good data for reviewing election history. Fama and French calculate the following five time series of returns as proxies for risk dimensions of stock and bond returns:

Market factor:   Stock market return above Treasury bills   Rm-Rf
Size factor:   Small cap minus large cap stock returns   SMB
Price factor:   Low-priced minus high-priced stock returns,
where price is scaled by book value		   HML
Term premium:
   Treasury bond minus Treasury bill return
   TERM
Credit premium:
   Corporate bond minus Treasury bond return
   DEFAULT

Table 1 displays the average factor returns from 1927 to 2003, both in total and by month of the year. The pattern of returns across calendar months is pretty random, except for January. Both SMB and HML have unusually large average returns in January.

Table 1
Fama/French Factors
Average Monthly Returns (%)
November 1927-October 2003
 
  Rm-Rf SMB HML TERM DEFAULT
All months 0.64 0.26 0.36 0.15 0.03
November 1.30 0.26 -0.36 0.45 -0.10
December 1.40 -0.11 -0.56 0.16 0.20
January 1.50 2.59 2.55 -0.01 0.60
February 0.30 0.69 0.29 0.02 -0.18
March 0.30 0.13 0.20 -0.12 -0.09
April 0.94 -0.08 0.98 0.01 -0.15
May 0.16 0.48 0.02 0.08 0.17
June 0.86 -0.24 0.06 0.56 -0.30
July 1.12 -0.19 1.04 -0.01 0.04
August 1.06 0.08 0.48 0.01 0.13
September -1.33 0.42 -0.34 0.09 0.15
October 0.10 -0.87 -0.08 0.57 -0.08
 
Data courtesy of Fama/French.

When presidential election years are considered separately, the pattern of returns is still fairly random. Table 2 displays results for the five months preceding an election and for the five months after an election. The entire month of November is treated as a post-election month. The market premium appears to be unusually high for July and August. However, it is not more than two standard deviations above the average premium for all months 1927-2003. The unusually large average return is attributable to the July-August return of over 100% in 1932.

Table 2
Fama/French Factors
Average Monthly Returns (%)
All 19 Presidential Election Years
1928-2000
 
Pre-Election Months Rm-Rf SMB HML TERM DEFAULT
June 1.72 -0.40 0.01 1.19 -0.42
July 1.85 -0.95 2.89 0.62 -0.10
August 3.48 1.28 1.29 -0.12 0.42
September -0.00 0.30 -0.40 0.16 0.29
October 0.20 -1.06 0.36 0.47 -0.11
 
Post-Election Months Rm-Rf SMB HML TERM DEFAULT
November 1.48 0.32 -0.36 0.45 -0.13
December 1.36 -0.16 0.17 0.17 0.22
January 1.21 1.71 2.64 -0.28 0.89
February -1.68 -0.25 1.68 -0.60 -0.02
March -0.12 0.11 1.41 -0.17 -0.08
 
Pre-Election Periods Rm-Rf SMB HML TERM DEFAULT
Last 5 months: Jun-Oct 1.45 -0.16 0.83 0.46 0.01
Last 4 months: Jul-Oct 1.38 -0.11 1.04 0.28 0.12
Last 3 months: Aug-Oct 1.23 0.17 0.42 0.17 0.20
Last 2 months: Sep-Oct 0.10 -0.38 -0.02 0.31 0.09
 
Post-Election Periods Rm-Rf SMB HML TERM DEFAULT
Next 2 months: Nov-Dec 1.42 0.08 -0.09 0.31 0.04
Next 3 months: Nov-Jan 1.35 0.62 0.82 0.11 0.32
Next 4 months: Nov-Feb 0.59 0.40 1.03 -0.07 0.24
Next 5 months: Nov-Mar 0.45 0.35 1.11 -0.09 0.17
Next year: Nov-Oct 0.42 0.26 0.53 0.19 0.01
 
Data courtesy of Fama/French.

The post-election results show strong returns for the first three months after an election, but the twelve-month results are strikingly similar to the overall averages displayed in Table 1.

An incumbent President is running for reelection this year. In 13 of the last 19 presidential elections, an incumbent stood for reelection. Of those, the incumbent won 9 times and lost 4. Table 3 displays the behavior of the Fama/French factor returns in the months around the election for those incumbents who won since 1928. It appears that the incumbent in those cases benefits from a strong stock and bond market, with large cap stocks outperforming small cap stocks. In the year after the election, the factor returns are about normal.

Table 3
Fama/French Factors
Average Monthly Returns (%)
All 9 Presidential Election Years
When Incumbent Is Reelected
1928-2000
 
Pre-Election Months Rm-Rf SMB HML TERM DEFAULT
June 2.01 -0.85 0.55 0.43 -0.25
July 0.03 -0.99 0.77 0.69 -0.01
August 1.89 0.30 0.33 -0.06 -0.01
September 0.13 0.36 -0.03 0.61 0.08
October 2.01 -1.34 1.56 1.12 -0.02
 
Post-Election Months Rm-Rf SMB HML TERM DEFAULT
November 0.46 0.56 0.58 1.06 -0.37
December 1.30 -0.01 0.30 -0.61 0.52
January 1.22 1.72 0.82 0.14 0.30
February -0.30 -0.66 1.81 -0.45 0.10
March -0.63 -0.52 2.35 -0.31 0.08
 
Pre-Election Periods Rm-Rf SMB HML TERM DEFAULT
Last 5 months: Jun-Oct 1.21 -0.51 0.64 0.56 -0.04
Last 4 months: Jul-Oct 1.01 -0.42 0.66 0.59 0.01
Last 3 months: Aug-Oct 1.34 -0.23 0.62 0.55 0.02
Last 2 months: Sep-Oct 1.07 -0.49 0.76 0.86 0.03
 
Post-Election Periods Rm-Rf SMB HML TERM DEFAULT
Next 2 months: Nov-Dec 0.88 0.28 0.44 0.23 0.08
Next 3 months: Nov-Jan 0.99 0.76 0.57 0.20 0.15
Next 4 months: Nov-Feb 0.67 0.41 0.88 0.03 0.14
Next 5 months: Nov-Mar 0.41 0.22 1.17 -0.03 0.13
Next year: Nov-Oct 0.36 0.07 0.47 0.23 0.00
 
Data courtesy of Fama/French.

Table 4 displays the results for the 10 elections since 1928 when the incumbent was not reelected. The factor returns are all negative for the month of October. Once again, though, the factor returns for the twelve months after the election are all about normal.

Table 4
Fama/French Factors
Average Monthly Returns (%)
All 10 Presidential Election Years
When Incumbent Is Not Reelected
 
Pre-Election Months Rm-Rf SMB HML TERM DEFAULT
June 1.46 0.02 -0.47 1.88 -0.58
July 3.50 -0.91 4.80 0.56 -0.19
August 4.92 2.17 2.16 -0.18 0.80
September -0.12 0.26 -0.72 -0.25 0.48
October -1.44 -0.81 -0.71 -0.11 -0.18
 
Post-Election Months Rm-Rf SMB HML TERM DEFAULT
November 2.40 0.10 -1.21 -0.11 0.07
December 1.41 -0.29 0.07 0.87 -0.06
January 1.21 1.70 4.28 -0.65 1.41
February -2.93 0.11 1.56 -0.74 -0.13
March 0.33 0.68 0.56 -0.04 -0.23
 
Pre-Election Periods Rm-Rf SMB HML TERM DEFAULT
Last 5 months: Jun-Oct 1.66 0.14 1.01 0.46 -0.03
Last 4 months: Jul-Oct 1.71 0.17 1.38 0.00 0.23
Last 3 months: Aug-Oct 1.12 0.54 0.24 -0.18 0.37
Last 2 months: Sep-Oct -0.78 -0.28 -0.72 -0.18 0.15
 
Post-Election Periods Rm-Rf SMB HML TERM DEFAULT
Next 2 months: Nov-Dec 1.91 -0.10 -0.57 0.38 0.01
Next 3 months: Nov-Jan 1.67 0.50 1.04 0.04 0.48
Next 4 months: Nov-Feb 0.52 0.40 1.17 -0.16 0.33
Next 5 months: Nov-Mar 0.49 0.46 1.05 -0.13 0.22
Next year: Nov-Oct 0.47 0.39 0.47 0.18 0.02
 
Data courtesy of Fama/French.

In summary, historical data suggests that the presidential election year results do not produce any significant findings that would suggest a change in an asset allocation strategy. Not surprisingly, factor returns are strong in the months before an election when an incumbent is reelected and weak when an incumbent is not reelected. Win or lose, the post-election factor returns are close to historical norms in the months after the election.

Even though election results don't seem to predict factor returns, factor returns in the months before an election may help predict the election winner. Table 5 displays the success of predicting election results from a derived stock variable. The derived variable, the "E" factor in Table 5, is the difference in returns between the Fama/French Large Cap Value Index and the Fama/French Small Cap Growth Index for the month of October in the election year. The "E" factor correctly predicts all 9 of the reelection victories for the incumbents and all 4 of the reelection defeats!

Table 5
Presidential Election Years
Using "E" Scores to Forecast Election Winners
"E" = Fama/French Large Cap Value Index return (%) minus Fama/French Small Cap Growth Index return (%) for the month of October in the election year.
 
YEAR   RESULT "E"
1928   DNP -4.20
1932   LOST -20.02
1936   WON 3.28
1940   WON 5.25
1944   WON 1.55
1948   WON 2.10
1952   DNP 0.69
1956   WON 0.22
1960   DNP 6.45
1964   WON 0.22
1968   DNP 3.34
1972   WON 4.39
1976   LOST -0.56
1980   LOST -7.61
1984   WON 2.80
1988   DNP 5.39
1992   LOST -3.99
1996   WON 7.62
2000   DNP 9.65
 
Average "E" Score by Election Outcome
Victory by Incumbent (9)   Won   3.05
Incumbent Does Not Participate (6)   DNP   3.57
Incumbent Loses (4)   Lost   -8.04

 

If an incumbent runs and "E" score is negative, then incumbent loses 4 times out of 4.

If an incumbent runs and "E" score is positive, then incumbent wins 9 times out of 9.
 
Data courtesy of Fama/French.

The "E" factor was developed through a process of egregious data mining, and there is no claim that it can successfully predict winners in the future. Nevertheless, the performance of the stock market between now and the election will probably influence election results, and I, for one, will be curious to see how well the "E" factor performs out of sample.

David G. Booth is the Chief Executive Officer and Chief Investment Officer of Dimensional Fund Advisors, an investment adviser registered with the Securities and Exchange Commission. This article may contain the opinions of the author but not necessarily the opinions of Dimensional Fund Advisors. All materials presented are compiled from sources believed to be reliable and current, but accuracy cannot be guaranteed. This article is provided for informational purposes only and should not be construed as an offer, solicitation, recommendation, endorsement of any of the products or services described in this website.

Back to Top