1. NOTE: To change the image on this slide, select the picture and delete it. Then click the Pictures icon in the placeholder to insert your own image. CHOOSING IN GROUPS MUNGER AND MUNGER Slides for Chapter 8 Uncertainty

2. Outline of Chapter 8  Sources of uncertainty  Perspective of the observer  Where the voters / candidates are  Uncertainty as a cause of divergence  Political competition with uncertain voters  Two results  Conclusion Slides Produced by Jeremy Spater, Duke University. All rights reserved.2

3. Sources of uncertainty  Strategic uncertainty  Optimal decision depends on others’ decisions, which are uncertain  Rational ignorance  Information is costly, so it is rational to be somewhat ignorant  Plays by nature under risk or uncertainty  Decision-makers may at least know the expectation value of the outcome (risk)  Decision-makers may not know anything about the probabilities of outcomes (uncertainty)  Perspective of observer  Observations and accounts carry some degree of uncertainty  Presence of observer can influence outcome 3Slides Produced by Jeremy Spater, Duke University. All rights reserved.

4. Perspective of the observer  Hawthorne effect  People act differently if they know they are being watched  Analogous to quantum uncertainty principle  Individual idiosyncrasies  People behave for reasons that are unpredictable and unknowable to observer  These sources of uncertainty can be captured with a random error term Slides Produced by Jeremy Spater, Duke University. All rights reserved.4

5. Where the voters are  According to MVT, candidates try to position themselves in middle of distribution  In practice, they are not quite sure where the middle is located  If politicians have different estimates of the middle, their platforms will differ Slides Produced by Jeremy Spater, Duke University. All rights reserved.5

7. Where the voters are (2)  Theorem 8.1:  Suppose exactly one median position is known to exist, but that its location is unknown. If the distribution of possible median positions is unimodal, symmetric, and shared, then candidates will adopt identical positions at the location of the mean of the distribution of medians.  Theorem 8.2:  If the distribution of possible median positions is unimodal and symmetric but candidates have different perceptions of this distribution, then candidates will initially adopt divergent positions at the location of the means of the perceived distributions of medians. These divergent positions are not an equilibrium, however. If candidates can move, they will adopt identical positions somewhere between the two divergent perceived means of the distributions of medians.  Theorem 8.3:  If candidates are uncertain about the location of voters and candidates seek some mix of policy and reelection, then candidate positions will diverge in equilibrium. In fact, no convergent equilibrium exists under these circumstances. However, the extent of the divergence depends on the extent of uncertainty and the mix of policy-reelection motivations by the candidates. Only if uncertainty is extreme and reelection motivations are trivial will the degree of divergence be substantively significant. Slides Produced by Jeremy Spater, Duke University. All rights reserved.7

8. Where the candidates are  Candidates might not keep their campaign promises  Voters observe a probability distribution of the candidate’s position  Based on their statements and past behavior  Voters respond to their expectations about the candidate’s likely behavior in office  In following example, Edward’s past flip-flopping makes voters uncertain Slides Produced by Jeremy Spater, Duke University. All rights reserved.8

11. Uncertainty as a cause of divergence Slides Produced by Jeremy Spater, Duke University. All rights reserved.11

12. Uncertainty as a cause of divergence (2)  Assume quadratic loss function in one-dimensional policy space  Assume voting rule: Slides Produced by Jeremy Spater, Duke University. All rights reserved.12

13. Uncertainty as a cause of divergence (3)  Setting cut point Q = x i, and setting utility functions equal:  Solving for Q:  This confirms the median voter theorem. Slides Produced by Jeremy Spater, Duke University. All rights reserved.13

14. Political competition with uncertain voters  In this case, voters are uncertain about candidates’ positions:  Because of the uncertainty, we express utility as an expectation: Slides Produced by Jeremy Spater, Duke University. All rights reserved.14

15. Political competition with uncertain voters (2)  Decision rule: Slides Produced by Jeremy Spater, Duke University. All rights reserved.15

16. Political competition with uncertain voters (3)  Solving for Q as before:  This differs from the equation on slide 13 by the bracketed term  The candidate with larger variance will lose votes to the more reliable candidate  Higher-variance candidate wants to move away from lower-variance candidate  This causes a surprising tendency toward divergence Slides Produced by Jeremy Spater, Duke University. All rights reserved.16

17. Two Results (1): No equilibrium Slides Produced by Jeremy Spater, Duke University. All rights reserved.17

18. Two Results (2): Surprising cut point Slides Produced by Jeremy Spater, Duke University. All rights reserved.18

19. Conclusions  Downsian model:  Voters uncertain of politicians’ actual policies  This leads to focus on ideology, not issues  Politicians uncertain of where median voter lies  They spend substantial resources finding out  Uncertainty is key to understanding real-world politics Slides Produced by Jeremy Spater, Duke University. All rights reserved.19