The Effect of Congressional Majorities on Financial Variables
The estimated effect of a change in the majority party in the Senate on the S&P 500 was only 0.17 percent, ...much smaller than the corresponding estimate of 2 percent for the Presidency in 2004. The effect of partisan majorities in the Senate on other financial variables was essentially zero.
In Party Influence In Congress and the Economy (NBER Working Paper No. 12751), authors Erik Snowberg, Justin Wolfers, and Eric Zitzewitz measure financial market responses to changes in the majority party in the U.S. Congress. Their primary focus is the response of equity, currency, oil, and bond prices to changes in the probability that each party would gain a congressional majority in the 2006 midterm election. They then examine data from earlier midterm elections and compare their estimates of the effects of changes in congressional majorities to their own earlier estimates of the effect of a switch in the party of the president.
In the run-up to the 2006 midterm elections, "intrade.com" created two contracts tied to Republican majorities in Congress: one paid $10 if Republicans held onto their majority in the Senate, the other paid the same if the GOP continued its control of the House. The latter contract traded at $2 at the beginning of election night, suggesting that Republicans had a 20 percent chance of maintaining their majority in the House. At 5 p.m. EST, exit polls indicated a poor showing by Republicans. As vote tallies provided confirmation, the probability of Republican success sank to zero. By contrast, the probability of Republicans maintaining their majority in the Senate began at 70 percent on election night and fluctuated substantially thereafter, largely because of the extremely close tallies in Virginia and Missouri.
To quantify the economic effects of party majorities in Congress, the researchers paired prediction market data from intrade.com with prices for the December 2006 futures contract of various financial variables. They then regressed the changes in financial variables on the changes in the price of the contracts tracking the Republicans' chances of maintaining their majorities.
They found that the estimated effect of a change in the majority party in the Senate on the S&P 500 was only 0.17 percent, which while statistically significant was much smaller than the corresponding estimate of 2 percent for the Presidency in 2004. The effect of partisan majorities in the Senate on other financial variables was essentially zero. The relative unimportance of Congress for the economy relative to the Executive Branch is further underscored by the fact that the largest event in their financial data was a 0.6 percent rally in the S&P 500 and other indices following the post-election announcement of Donald Rumsfeld's resignation from the cabinet.
The 2002 election is the only other midterm vote for which high frequency data from a liquid prediction market exists. The Intrade data suggest that on election night 2002, Republicans had a 90 percent chance of maintaining their majority in the House and a 40 percent chance of gaining a majority in the Senate. Throughout the tally, these probabilities increased as Republicans won majorities in both houses. The outcome for the House was so close to expectations that there was no useful election-night variation in the contract to analyze. However, there was enough variation in the Senate contract to estimate an impact on financial markets. Once again, these effects were significantly smaller than the effect of the party of the president.
The 1998 midterm election was too predictable to be of any use to the researchers, but the 1994 midterms offered enough of a surprise that the authors could correlate shifting probabilities with shifts in various financial, oil, and foreign currency prices. Again, a change in the majority party in Congress had smaller effects on these prices than earlier or later changes in the party controlling the White House.
The authors conclude that the majority party in Congress has relatively little control over economic policy, at least as these economic levers affects equity, bond, oil, and currency prices. This is not to deny an important role for Congress, they say, but simply to note little evidence of influence on economic aggregates. The evidence from the various midterm elections further suggests that, contrary to the popular view, markets show no preference for "divided government" (when different parties control Congress and the White House), rather than a politically unified government. The researchers point out that equity prices and bond yields rose in response to news of Republican majorities in the House and Senate (1994, 2002) despite the fact that these elections created divided and unified governments, respectively. Further, these variables fell in response to the Democratic majorities gained in 2006, which created a divided government.
The authors end their paper with two qualifications about their results: "First, as with other applications of the event study method, our approach estimates market expectations about future policy, rather than actual differences in these policies. And second, the financial variables we analyze do not speak directly to economic welfare or yield immediate normative implications."
-- Matt Nesvisky