Prediction Markets and Election Polling: Granger Causality Tests Using InTrade and RealClearPolitics Data


This study tests for direct causality between RealClearPolitics (RCP) polling averages and InTrade (IT) share-prices by performing Granger causality tests. These tests are applied to the 2012 U.S. presidential election campaign between Barack Obama and Mitt Romney. Three time series are considered in this analysis of causal links between RCP and IT. In all of the estimations, the null hypothesis that IT does not Granger cause RCP cannot be rejected. On the other hand, the null hypothesis that RCP does not Granger cause IT is rejected in a majority of cases. Overall, these results imply that RCP Granger causes IT. While the behavior of participants in prediction markets such as InTrade is informed by political polling, these Granger causality test results suggest that prediction markets add little to what is forecasted by the polls, particularly aggregation polls such as that from RealClearPolitics, regarding the outcome of presidential elections.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2


  1. 1.

    See Cebula and Hulse (2007) for the implications of strategic behavior by poll respondents on voter turnout.

  2. 2.

    Since 1984, the aggregate female voter participation voter participation rate has exceeded the male voter participation rate in the U.S. (Cebula and Meads 2008).

  3. 3.

    InTrade closed its site to U.S. customers on December 31, 2012, due to regulatory pressures from the Commodities Futures Trading Commission, which charged InTrade with operating an unregulated options market. Just over two months later, on March 10, 2013, InTrade suspended operations worldwide.

  4. 4.

    Examples of events include outcomes of elections, Academy Awards winners, i.e., winners of the Oscars, and other scenarios having political-economic consequences (

  5. 5.

    BetFair ( and Pollster ( represent alternatives to InTrade and RealClearPolitics for testing the relationship between prediction markets and election polls. However, these alternatives are not as widely recognized as InTrade and RealClearPolitics, which, during the 2012 U.S. presidential election, were much more visible on cable television news networks, and were widely tracked by print media such as The Washington Post and The New York Times.

  6. 6.

    We note that both our share-price and polling data are all candidate-specific. In other words, there are no polls that were included in our RCP averages that were generic in scope (e.g., unnamed Republican vs. unnamed Democrat), whether before or after the conventions. As such, our comparison of the polling data with the share-price data by candidate over the full reporting period is therefore methodologically straightforward in the sense that there was no need to normalize different types of polls in order to construct the data set.

  7. 7.

    The other Republicans who campaigned actively for their party’s presidential nomination in 2012 were Michele Bachmann, Herman Cain, Newt Gingrich, Jon Huntsman, Ron Paul, and Rick Santorum.

  8. 8.

    It is also the case that national polling increases in frequency after the political conventions. This increased frequency is reflected in the freshness of the RCP averages used in this study.

  9. 9.

    The Republican Party Convention was held in Tampa, Florida, from August 27, 2012, through August 30, 2012. The Democratic Party Convention was held in Charlotte, North Carolina, from September 3, 2012, through September 6, 2012. What is traditionally referred to in the U.S. as the “Labor Day weekend” occurred in between these two political conventions.

  10. 10.

    In one of the two cases in which the null is not rejected (i.e., the RCP→IT test of the Obama post-conventions data), the Granger causality test only marginally fails to reject the null.

  11. 11.

    In recognizing that genuinely interesting hypotheses are neighborhoods, and not points (Leamer 1978, 1988; Attfield 1982), Kennedy (2008: 61) advises that researchers should allow for slightly larger significance levels when relatively small samples are employed.

  12. 12.

    In this case, major events (e.g., the killing of Osama bin Laden) are rare. As such, the fact that they cause asset prices to lead poll averages for a brief period of time does not upset the RCP-->IT Granger causality.

  13. 13.

    The level of InTrade trading activity for the two major-party candidates averaged roughly 10,000 trades per day during the summer of 2012 and skyrocketed to over 200,000 trades per day in the week before the election.

  14. 14.

    American intelligence agencies had pursued bin Laden for well over a decade prior to the U.S. military’s successful operation in early May of 2011. This pursuit involved a number of U.S. presidents. As such, Operation Neptune Spear, the codename given to the operation to capture or kill bin Laden (Bell 2013), was a political boon to Obama. With regard to the 2012 U.S. presidential race, most political pundits believe that Romney soundly defeated Obama in the first presidential debate in early October of 2012 (e.g., see Blake 2012). Given the importance of presidential debates to many voters, Romney’s victory in the debate was a political boon to his campaign for the presidency.


  1. Akaike, H. (1969). Fitting auto-regressions for predictions. Annals of the Institute of Statistical Mathematics, 21, 243–247.

    Article  Google Scholar 

  2. Akaike, H. (1970). Statistical predictor identification. Annals of the Institute of Statistical Mathematics, 22, 203–217.

    Article  Google Scholar 

  3. Attfield, C. (1982). An adjustment to the likelihood ratio statistic when testing hypotheses in the multivariate linear model using large samples. Economics Letters, 9, 345–348.

    Article  Google Scholar 

  4. Beaulieu, M. C., Cosset, J. C., & Essaddam, N. (2006). Political uncertainty and stock market returns: Evidence from the 1995 Quebec referendum. Canadian Journal of Economics, 39, 621–641.

    Article  Google Scholar 

  5. Bell, D. (2013). Osama bin Laden’s death, two years later, U.S. News and World Report, (2 May), available online at Accessed on 19 July 2013.

  6. Berg, J. E., Nelson, F., & Rietz, T. A. (2008). Prediction market accuracy in the long run. International Journal of Forecasting, 24, 285–300.

    Article  Google Scholar 

  7. Blake, (2012). Six reasons Mitt Romney won the first debate, The Washington Post (3 October), available online at Accessed on 19 July 2013.

  8. Brown, W. O., Jr. (1996). Friends in high places: The wealth effects of JFK’s assassination on the assets of LBJ’s supporters. Public Choice, 86, 247–256.

    Article  Google Scholar 

  9. Burke, J., & Taylor, C. (2008). What’s in a poll? Incentives for truthful reporting in pre-election opinion surveys. Public Choice, 137, 221–244.

    Article  Google Scholar 

  10. Cebula, R. J., & Hulse, D. (2007). The polls results hypothesis. Atlantic Economic Journal, 35, 33–41.

    Article  Google Scholar 

  11. Cebula, R. J., & Meads, H. (2008). An inquiry into the contemporary differential between female and male voter turnouts. Atlantic Economic Journal, 36, 301–314.

    Article  Google Scholar 

  12. Chatrath, A., Christie-David, R., & Moore, W. T. (2006). The macroeconomic news cycle and uncertainty resolution. Journal of Business, 79, 2,633–2,657.

    Article  Google Scholar 

  13. Christie-David, R., Chaudry, M., & Koch, T. W. (2000). Do macroeconomics news releases affect gold and silver prices? Journal of Economics and Business, 52, 405–421.

    Article  Google Scholar 

  14. Christie-David, R., Chaudry, M., & Lindley, J. T. (2003). The effects of unanticipated macroeconomic news on debt markets. Journal of Financial Research, 26, 319–339.

    Article  Google Scholar 

  15. Ejara, D. D., Nag, N., & Upadhyaya, K. (2012). Opinion polls and the stock market: Evidence from the 2008 U.S. presidential election. Applied Financial Economics, 22, 437–443.

    Article  Google Scholar 

  16. Erikson, R. S., & Wlezien, C. (2012). Markets vs. polls as election predictors: An historical assessment. Electoral Studies, 31, 532–539.

    Article  Google Scholar 

  17. Granger, C. W. J. (1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica, 37, 424–438.

    Article  Google Scholar 

  18. Gustafson, C. (2008). On web, political junkies make a Real Clear choice, New York Sun (10 March), available online

  19. Hsiao, C. (1981). Autoregressive modeling and money-income causality detection. Journal of Monetary Economics, 7, 85–106.

    Article  Google Scholar 

  20. Kennedy, P. (2008). A guide to econometrics. Malden: Blackwell Publishing.

    Google Scholar 

  21. Leamer, E. E. (1978). Specification searches: Ad hoc inference with non-experimental data. New York: John Wiley.

    Google Scholar 

  22. Leamer, E. E. (1988). Things that bother me. The Economic Record, 64, 331–335.

    Article  Google Scholar 

  23. Morgan, J., & Stocken, P. C. (2008). Information aggregation in polls. American Economic Review, 98, 864–896.

    Article  Google Scholar 

  24. Rhode, P. W., & Strumpf, K. S. (2004). Historic presidential betting markets. Journal of Economic Perspectives, 18, 127–142.

    Article  Google Scholar 

  25. Sinclair, B., & Plott, C. R. (2012). From uninformed to informed choices: Voters, pre-election polls and updating. Electoral Studies, 31, 83–95.

    Article  Google Scholar 

  26. Stout, C., & Kline, R. (2011). I’m not voting for her: Polling discrepancies and female candidates. Political Behavior, 33, 479–503.

    Article  Google Scholar 

  27. Zorn, E. (2006). “Political site polls well with election junkies,” Chicago Tribune (26 October), available online at:

Download references


Christopher Duquette’s affiliation with The MITRE Corporation is provided for information purposes only and is not intended to convey or imply MITRE’s concurrence with, or support for, the positions, opinions, or viewpoints expressed by the authors.

Author information



Corresponding author

Correspondence to Richard J. Cebula.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Duquette, C., Mixon, F.G., Cebula, R.J. et al. Prediction Markets and Election Polling: Granger Causality Tests Using InTrade and RealClearPolitics Data. Atl Econ J 42, 357–366 (2014).

Download citation


  • Prediction markets
  • Elections polling
  • Voter preferences
  • InTrade
  • Granger causality


  • D72
  • D83
  • G12
  • G14
  • D70
  • D80