John Irons's Blog


Economic News, Data and Analysis

The Economy and U.S. Presidential Elections: Part II

Predicting the
outcome of the election using the economy as a guide is a tricky business.
Many people have come up with different formulations, but there are some
similarities across the models – all show at least some effect of the economy
on presidential elections.  For some examples of the equations used
for predicting the election, see Part I of The Economy
and Presidential Elections


Listen to Audio
on Presidential Vote Predictions

(Audio from,
including some commentary from your humble guide.)

The Table below
shows the predictions from the 1996 race. When it comes to the 2000 election,
all of the models I have seen are currently predicting a solid Gore victory
unless there is a major economic downturn in the next 3 months.


Prediction Results
(% of 2-Party popular vote)
Fair 1996 Irons 1992 Tufte 1978 (96) Hibbs 1996
Final Predictions: 51.03% 57.25% 51.31% 53.51%
Actual: 54.63% 54.63% 54.63% 54.63%
Abs. Error: 3.60% 2.62% 3.32% 1.12%

How exactly
do these equations use the data to make predictions?

Measuring the

There are many
ways to measure the performance of the economy – so picking the right measure
is a first step in predicting the election outcome. The models tend to
use very broad measures of the overall performance of the macroeconomy
– per capita GDP growth or income growth are the typical ways to measure
the real output of the economy, while consumer price inflation is the most
common measure of prices. However other measures such as unemployment,
income distribution, stock market performance, government sending, have
all been used as well. Fair uses both an output measure as well as inflation,
while Hibbs and Tufte use only an output measure (per capita income and
GDP respectively).

A faster pace
of economic growth will lead to a greater vote share for the incumbent’s
party.  Among the 3 models listed in the table above, a 1% permanent
increase in growth will lead to anywhere between a 0.7 and a 4 percentage
point increase in the 2-party vote share for the incumbent party.


If it takes some
time for policies to be implemented and for those policies to have an impact
on the economy, it makes sense that the recent past should matter more
than the early part of a president’s term. This seems to be the case empirically
as well – the most recent past matters more than the early part of the
term; however, the models are split on exactly how to taper the past economic


Just about everyone
agrees that the incumbent has an advantage, and the empirical models bear
this out giving a substantial advantage to the party in power.  (Although
Fair’s model lowers the vote share the longer a party have been in office.)


No surprises
here. The Vietnam and Korean wars did a good job of hurting the incumbent
party’s chances come Election Day.  Taking into account these conflicts
allows us to better predict the outcome in peacetime.


Here’s where
you split the economists from the political scientists. While the Economists
seem primarily concerned with behavioral models of voting, the Political
Scientists may be more concerned with strait out prediction. For the later
purpose, many models use presidential approval ratings as an additional
piece of data. Edward Tufte, a political scientist (at the time), pioneered
the use of the popularity rating in 1978 and others have continued to use
this approach (including Iowa professor M. Lewis-Beck). Economists Ray
Fair and Douglas Hibbs prefer to use only economic data in their models.

An interesting
question is why the economy matters in addition to the approval rating.
I would have thought that the state of the economy would have been reflected
in the approval rating and would have no independent effect. (In fact,
Hibbs finds the opposite – adding the popularity data to his economy-only
model does not significantly improve his predictions).


Does this mean
that campaigns don’t matter?

One way of interpreting
the results is that it is only the economy that affect the outcome of the
election, and that all of the other things associated with the campaign
– debates, conventions, policies, ads, speeches, etc. – have no effect
on the election outcome.

However, there
is also another interpretation. All of these other factors are important
and both parties are able to use these tools effectively to influence the
outcome of the election. What happens is that the parties have similar
skills in using all of the tools of campaigns and they effectively cancel
each other out – leaving only the economy as an object that cannot be manipulated
by both parties.

it is difficult to use the available data to determine which interpretation
is correct.

Does this mean
that the incumbent party gets all the blame or credit for the economy?

Initially, it
would be hard to say that the president has that much control over the
economy. First, fiscal policy is determined jointly by congress and the
president, so some of the blame and credit should be given to the House
and Senate. In addition, most people think that fiscal policy is a rather
weak tool when it comes to steering a huge economy. Secondly, an important
tool in influencing the economy, monetary policy, is out of the direct
control of the president. Alan Greenspan does at least as much as Bill
Clinton in running the economy. Finally, there are many other factors that
impact the economy of which the president has little or no control, including
foreign economies, oil prices, technology breakthroughs, etc.

However, in this
case perception is more important than reality. The 1980’s had Ronald Reagan
as President and a Democratic Congress, and was characterized by an economic
policy dubbed “Reagonomics.” Most of the 1990’s had Bill Clinton as President
and a Republican Congress and most people will use the term Clintons’ economy.

The empirical
evidence currently supports the idea that the president gets some credit.
Certainly voters know that the president does not solely determine the
economy – the equations do not claim that they do, only that the president,
and his party, does get some, if not all, of the credit.

More readings

Links to vote prediction pages

gives Gore a lead: Pocketbook issues favor Democrats
– CBS MarketWatch.

the Economy, Stupid?
– Kiplinger’s

– Christian Science Monitor


Filed under: Economy



%d bloggers like this: