Douglas Hibbs

October 27 2011

The Partisan Division of House Seats in 2012: Implications of the ‘Bread and Incumbency’ Model

For Democratic Party Prospects as of 2011:q3

The number of House seats won by the president’s party in presidential election years, whether it be in the majority or the minority, is well explained by just two fundamental pre-determined or exogenous variables: (1) the number of House seats won by the in-party at the previous midterm election, which registers the impact of institutional advantages enjoyed by incumbents in the US single-member district, constituency service-oriented legislative system, and (2) weighted-average growth of per capita real disposable personal income over the congressional term. No other factor, objectively measured ex-ante, systematically affects on-year House election outcomes.

Unlike votes for president, military Fatalities exert no systematic influence on the aggregate partisan division of House seats. Only presidential voting outcomes are affected by unprovoked, hostile deployments of American armed forces in foreign conflicts, unsanctioned by a congressional declaration of war. In present circumstances, political responsibility for American Fatalities in Afghanistan will be attributed to President Obama, not congress.

The Bread and Incumbency equation for the partisan division of House seats is written

where Seats_t denotes the number of House seats won by the president’s party at presidential election periods, Seats_t-8 is the number of won by the president’s party at the previous midterm election eight quarters ago, and ΔlnR is the quarter-on-quarter percentage rate of growth of per capita real disposable personal income, expressed at annual rates.

Estimating the Bread and Incumbency equation for the fifteen House elections in presidential election years spanning 1952-2008 yields coefficient values and related statistics

Coefficient estimate:					Adj. R^2= .89
(Std. error|p-value):	(20.4|.82)	(0.9|.00)	(1.8|.00)	(0.2|.00)	Root MSE = 13

The Democrats won 193 seats in the 2010 House mid-term election, a loss of 63 from their 2008 on-year showing of 256 seats, which put the president's party in the minority for the 112th Congress.
Over the first 3 quarters of the 112th Congress -- 2011:q1 to 2011:q3 -- weighted-average growth of per capita real income was an anemic -1.8%. According to the Bread and Incumbency model, an election held under such poor economic conditions would yield a notional number of seats going to the Democrats of just 169, as depicted in the graph below.

What of course matters for the Democrat’s House prospects are conditions at Election Day in 2012, not the notional outcome implied by the situation at the end of 2011:q3. The data in the table below indicate that according to the Bread and Incumbency model the prospect of the Democrats winning a bare majority of 218 House seats in 2012 is nil, even under the unlikely economic conditions that could yield a comfortable Obama victory in the presidential contest.

Expected Number of House Seats Going to the Democrats in 2012 at various real income growth rates 2011:q4 - 2012:q3
Hypothetical per capita real income growth rate 2011:q2 - 2012:q4:	-4	-2	0	+2	+4
=> Resulting weighted-average real income growth rate over the congressional term:	-3.5	-2	-.4	1.1	2.6
=> Expected number of seats going to the Democrats – the president’s party: (Change from 2010 mid-term election):	159 (-34)	68 (-25)	178 (-15)	188 (-5)	197 (+4)

Like my Bread and Peace model of votes for president, the Bread and Incumbency model paints a bleaker picture of the Democratic Party’s chances for a House majority in 2012 than current betting price data do. At the end of October 2011 Intrade prices (from rather thin trading) implied that the chances of the Democrats winning a House majority in 2012 was 28.7% (and their chances of retaining a majority in the Senate was 25%).

Both the Bread and Peace model and the Bread and Incumbency model aim to explain election outcomes in terms of objectively measured political-economic fundamentals, rather than to predict optimally voting results or to track them statistically after the fact. For those reasons neither model includes arbitrarily coded dummy, trend, or count variables or uses pre-election poll readings of voter sentiments, preferences and opinions. Trends and related time-coded variables are ad-hoc, statistical junk without scientific merit no matter how much they improve statistical fits or forecasting accuracy in various samples. Attitudinal-opinion poll variables are themselves affected by objective fundamentals, and consequently they supply no insight into the root causes of voting behavior, even though they may provide good predictions of election results.

Here are the Stata program (“do”) file and the Stata data (“dta”) file that generated results on this page.

Here you will find an analysis of the partisan division of seats in mid-term House elections based on my Bread, Incumbency and Balance model.