Open Collections

Essays in political economy

University of British Columbia

Chapter 1 estimates how an individual's expressed sentiment responds to messages from their social network connections. I use machine learning to code messages for expressions of one type of sentiment: happiness. Because network link formation is not random, I use exogenous shifters to instrument for the message volume of each of a user’s neighboring nodes. Specifically, I interact neighbor daylight with average neighbor sentiment, and aggregate this across neighbors to construct an instrument for viewed messages. A user with neighbors in different places with different average sentiment receives a shock to their feed when light levels differ across those places. I find that a user's happiness increases by 3.4% when the happiness of incoming messages increases by 10%. Chapter 2 presents a general framework for estimating the causal effect of social interactions on online social networks. This context presents two challenges for causal estimation beyond the endogeneity problem discussed above. First, social networks are dynamic: users are affected not only by contemporaneous messages but also by past messages. Second, some data is missing. These networks have relatively low levels of clustering, which means that it is computationally infeasible to collect all of the neighbors of a sample of the network. I introduce an estimation strategy for addressing these two challenges. I also construct six new instruments within this framework and compare their strength. Chapter 3 develops a method for estimating the impact of voter demobilization efforts on voter turnout. We exploit two facts: a) demobilization is typically targeted to avoid the supporters of the intended beneficiary, and b) voting results are available at the sub-district (poll) level. Omitted variables will generally be constant across a district, while the impact of the violations will be decreasing in the level of support for the violating party. Our method is general in the sense that it does not require a natural experiment and is robust to countrywide shifts in voter support (”swing”). We apply this method to allegations of fraud in the 2011 Canadian federal election, and estimate that illegal demobilization efforts reduced turnout by 3.9% in affected districts.

Text

2018-10-18

Attribution-NonCommercial-ShareAlike 4.0 International

http://hdl.handle.net/2429/67600

Economics

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-sa/4.0/