Date of Award

May 2021

Document Type


Degree Name

Doctor of Philosophy (PhD)



Committee Member

Patrick Warren

Committee Member

William Dougan

Committee Member

Chungsang Lam

Committee Member

Oriana Aragon


This dissertation is comprised of three essays in political economy. In the first chapter, I study the short-run political polarization between Republican and Democrat politicians in the House of Representatives before and after the November 2018 midterm election, using Twitter data. I compute various metrics of ideological polarization at weekly intervals using methods such as hashtag analysis, topic modelling ,Bayesian Ideal Point Estimation, mention and retweet network analysis. I empirically check for the patterns in political polarization during the election cycle at the level of discourse. Different measures of polarization signal different patterns in polarization. When polarization is measured by hashtag divergence or topical divergence, it seems to increase as the election approaches. However, when polarization is measured by divergence in word distribution, sentiment-augmented topic divergence, or cited-media ideology divergence, it seems to decrease as the election approaches. This pattern is consistent with a divergence in preferred electoral agenda but convergence in agenda-item-specific positioning.

In the second chapter, I extend the framework of analysis that I developed to the Indian context. I study the short run political polarization between the politicians of the two main national political parties in India contesting in the Lok Sabha, the lower house of the Indian parliament before and after the 2019 general elections, using data from their Twitter feed. I compute various measures of ideological polarization using the methods described in the first chapter, and empirically test the policy convergence hypothesis versus the policy divergence hypothesis discussed in the literature by analysis of these measures of ideological polarization. This chapter reiterates the findings of the previous chapter, which shows that the different measures of polarization signal different patterns in polarization. We find increase in polarization as measured through topical divergence and fall in polarization as measured through sentiment augmented topic analysis suggesting divergence in agenda setting behavior and convergence in agenda-item-specific positioning. This is similar to the pattern in the U.S data. However, in contrast to the US data, polarization as measured through hash-tag divergence decreases whereas polarization as measured by cited media ideology increases as we approach the election in India.

In the third chapter chapter, my coauthor Sagnik Das from City University of New York and myself study the effect of political business cycles on government expenditure in India as measured using data from the world’s most extensive public works programme (NREGA), new road constructed data as well as night light intensity data which is used as a proxy for development. Using panel data at the district level spanning from 2011 to 2020 for NREGA employment and expenditure, 2000 to 2014 for new road constructed under the PMGSY program and mean total calibrated nightlight intensity from 1994 to 2014, we can show the existence of political business cycles wherein politicians stimulate the economy before the election either to lure myopic voters or to signal their capability to forward-looking voters. We find the causal impact of political business cycle on expenditure undertaken under NREGA and on employment provided under NREGA at the intensive margin. We also find evidence of political business cycles impacting the length of new road constructed under PMGSY and money disbursed by the Government for new projects to be undertaken under PMGSY. For night light intensity, we do find some evidence of the causal impact of political business cycles. We also use high-frequency monthly nightlight intensity data spanning from 1993-2013 to investigate the political business cycle’s effect in the shorter run. We do see a statistically significant spike in night light intensity one month before the election. However, we are unable to find any conclusive trend with the approach of the election.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.