Bond Yields Foretell Polling Data

Since the 2000 election, I have been tracking the way that the presidential race polling data tend to lag the movements of the stock market by about a week.  I have been doing that this year for subscribers to my newsletter and Daily Edition, and I have been a little bit dissatisfied with the correlation this year.  But I recently found the key, which is shown in this week's chart.

It is not so much the stock market that the polling data are following this time, but rather it is bond yields.  This chart shows the Real Clear Polling (RCP) average daily spread, and it is compared to the Treasury Yield Index (TYX).  That measures the current yield to maturity on whichever 30-year T-Bond is the most recently issued series.  I am employing a 6 trading day offset to show how the polling data are following bond yields.

My hypothesis is that there really is not a lag in real life, and that changes in bond yields occur coincident with changes in public mood related to the election.  It is just that measuring public mood takes a while for the various polling firms to contact their poll respondents, gather the responses, tabulate and publish, and then have those get aggregated together by RCP.  I cannot prove that hypothesis about the coincident relationship.  But we can see that there indeed is a pretty good correlation. 

The latest data from RCP show Trump with a lead of just 0.3 percentage points.  That does not mean anything for the actual results, which will depend on state by state counts toward the electoral college totals, but it does indicate which way people are leaning.  The continued rise in the TYX implies that there should be a further drift in the polling data in Trump's favor, perhaps getting him up to about a 1 percentage point advantage in the popular vote (assuming that the polling data reflect the popular vote). 

One big difference this year is that a lot more of the voting is being done early, ahead of the Nov. 5 official election day.  So if the change in polling results are accurately reflecting shifting mood among voters (and yes, that is a BIG if), then any drift in the polling data which echoes the movements of bond yields will not affect votes that have already been cast.  But early voters also tend to be more firmly committed to their choice ahead of time, and thus they are not as likely to be voters whose decisions might change at the last minute.