Research
Book Chapters
Abstract
This chapter provides a comprehensive overview of text-as-data methods in the social sciences, with a focus on political science applications. Beginning with foundational approaches such as lexicon-based analyses and statistical scaling methods and advancing to topic modeling frameworks, it then examines the emergence of word embeddings and transformer revolution. Applications include measuring ideological positions, sentiment analysis, and framing detection. The discussion highlights methodological innovations in zero-shot and few-shot classification, cross-lingual analysis, and causal inference, alongside considerations of validity, ethics, and interpretability. Finally, the chapter explores future directions including multimodal integration and responsible artificial intelligence (AI) practices.
Abstract
Panel data, also called time-series cross-sectional (TSCS) data, enable researchers to harness both cross-unit variation and within-unit dynamics to answer key questions in social sciences, from policy diffusion to regime change. This article surveys major panel data estimation techniques useful for social science researchers, highlighting their assumptions, strengths, and contexts of use. I begin by defining TSCS structures and illustrating common pitfalls of naïve pooling, before introducing two-way fixed-effects models for controlling unit- and time-invariant confounders. Next, I review conventional difference-in-differences designs and their extension to staggered adoption settings, emphasizing recent developments to correct bias when treatments roll out at different times. Building on these foundations, I delve into dynamic panel estimators, including the Arellano–Bond and system GMM frameworks, and the PanelMatch method, which matches on longitudinal histories to bolster causal identification. Throughout, research applications exemplify each method's potential and limitations.
Working Papers
Abstract
The WTO dispute settlement mechanism adjudicates trade conflicts through rule-based procedures ostensibly insulated from domestic politics, yet the pace at which disputes reach resolution varies widely. This study argues that the WTO enforcement space is constricted through the reactive mobilization of defendant-side firms whose anti-dumping protections are at risk. Using US lobbying disclosure data matched to firm-level dispute participation, the analysis exploits presidential party turnover as an exogenous shock to the political environment of defendant firms. A PanelMatch estimator with covariate balancing propensity scores reveals a sharp asymmetry. Defendant firms in non-prioritized industries significantly increase lobbying expenditure following partisan transitions, whereas firms in nationally prioritized sectors shielded by executive action under Section 232 or bipartisan legislation such as the CHIPS and Science Act show no response. A stratified Cox proportional hazards model demonstrates that this mobilization delays dispute resolution. Under partisan transition, increased lobbying reduces the hazard of reaching a panel report, prolonging the period during which contested duties remain in force. Without turnover, lobbying carries no meaningful effect on dispute duration. These findings reveal how domestic politics penetrates the procedural machinery of WTO dispute settlement, exposing a vulnerability in which political uncertainty activates defensive firm mobilization.
Abstract
The rise of superstar firms has accompanied the expansion of an international intellectual property (IP) regime that extends beyond the standards established in the WTO Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS). Despite sustained resistance from developing countries, the United States has successfully promoted TRIPS-plus provisions through preferential trade agreements, with other advanced economies following. This study attributes US success in diffusing TRIPS-plus standards to the interaction between corporate lobbying and development finance allocation. Using National Trade Estimate (NTE) reports from 1995 to 2022, published by the US Trade Representative, I construct a semantic proxy for US assessments of IP protection in partner countries using the DeBERTa-v3 language model. The analysis shows that lobbying by US corporate elites systematically shapes USTR evaluations of foreign IP regimes and that countries facing greater IP-related criticism are more likely to receive compensation for adopting TRIPS-plus commitments. Compensation varies by regime type: democracies receive increased foreign aid, while autocracies benefit more from International Finance Corporation lending to the private sector.
Abstract
Introduction of the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) marked the new era of intellectual property (IP) rights protection through preferential trade agreements (PTAs). TRIPS-plus IP rules impose substantial domestic costs on developing countries, yet many governments adopt them through preferential trade agreements with the United States and the European Union. This study explains TRIPS-plus adoption as a function of internal structural vulnerability and external political support. Developing countries whose labor-intensive exports rely heavily on discretionary preference schemes such as the Generalized System of Preferences face credible threats of suspension and therefore stronger incentives to formalize access through PTAs that include TRIPS-plus rules. A refined measure of political trade dependence that isolates these sensitive sectors captures this vulnerability more accurately than aggregate indicators, first introduced by Manger and Shadlen (2014). Foreign aid functions as a complementary mechanism that helps governments manage the domestic adjustment costs of strengthened IP protections, although compensation emerges only when obligations are substantively deep and donors can mobilize aid coherently. Empirically, TWFE OLS estimates show that higher PTD significantly increases the likelihood of adopting TRIPS-plus provisions. PanelMatch results confirm a short-lived rise in US aid after signature, while EU aid remains unresponsive due to weaker IP obligations and donor fragmentation.
Abstract
Lobbying disclosures often hide what firms lobby about, but we introduce new data and exploit timing to show that multinational firms lobby intensively about preferential trade agreements (PTAs). We argue that some firms lobby for ratification, which lowers tariffs on imported intermediate inputs, while others lobby to shape PTA terms to influence behindthe-border regulations. Our empirical analysis untangles these incentives by identifying groups of firms with different motivations and investigating which branches of government they lobby and when. We use PanelMatch (Imai, Kim and Wang 2023) to conduct staggered diff-in-diff analyses at the firm level, using Fortune Global 500 firms and a comprehensive set of US PTA negotiations. Pharmaceutical firms lobby for intellectual property rights protection; firms represented on the US Trade Representative’s general policy-making committee lobby to shape PTA provisions; other firms lobby for ratification of PTAs with countries where they have investments.
Abstract
US financial firms are widely believed to influence the politics of the International Monetary Fund, but the effects pose a puzzle: bank influence is believed to increase conditionality that favors banks, but to decrease conditionality for their biggest borrowers. We use LDA data on lobbying in the United States and IMF Monitor data on conditionality to test the hypothesis that lobbying influences conditionality. We find that lobbying by nonbankfinancialfirmsisassociatedwithincreasednumbersoffinancialconditions, quantitative performance criteria (QPCs), and labor-market conditions. Bank lobbying, in contrast, has both effects. When banks lobby about finance, borrowing countries receive more conditionality; but when banks lobby about banking, borrowers accept less financial conditionality and fewer QPCs. When banks are worried, they lobby about banking, and this undermines the IMF’s leverage to obtain the policy reforms that the banks most prefer. We go on to investigate firm-level responses to the initiation of IMF program negotiations and find that US financial firms shift their lobbying efforts when the Fund negotiates with countries where they have investments. Money-center banks and non-banking financial firms shift in opposite directions, which makes their lobbying portfolios converge: banks reduce their lobbying about banking and increase lobbying about finance, while non-bank financial firms decrease lobbying about finance and increase lobbying about banking.
Works in Progress
Research on development finance institutions (DFIs) has concentrated on state motives, governance, and project outcomes, with limited attention to how private firms use these instruments to pursue strategic objectives. This paper reframes projects of the US International Development Finance Corporation (DFC) as geoeconomic tools through which firms, especially in geopolitically sensitive sectors, integrate corporate political activity with development finance. By linking congressional and executive lobbying disclosures to DFC project records, the study analyzes how corporate sponsors recalibrate lobbying across venues, policy domains, and project phases, focusing on energy projects where development finance intersects with strategic competition, supply security, and geopolitical risk. PanelMatch estimates indicate a systematic temporal shift in lobbying venues: energy sector sponsors increase congressional lobbying during the first two years after project implementation, while lobbying directed at the Executive Office of the President and the Department of Energy peaks two to three years later, as congressional activity declines. Disaggregated evidence also reveals risk-contingent behavior: sponsors of projects in financially risky recipient countries sharply increase budget-related congressional lobbying immediately after implementation, but this surge fades within a few years, consistent with short-term political insurance rather than sustained influence investment.
Development finance institutions (DFIs) are central actors in closing infrastructure financing gaps in developing economies, yet we know little about when and how they mobilize private capital for infrastructure investment. This paper examines the conditions under which DFIs structure investments to attract private participation, using an original project-level dataset of infrastructure investments by twelve OECD-country DFIs. We combine large language model classification with human coding to identify DFI infrastructure investments, and link these to the World Bank's Private Participation in Infrastructure (PPI) database to analyze the composition of project consortia and financing structures. Our core argument is that private capital mobilization reflects a strategic tradeoff: it is more likely where projects are commercially viable and risks can be credibly priced, but less likely in low-return sectors and weaker institutional environments. We derive three expectations: mobilization is more likely in revenue-generating sectors (e.g., energy, transport) than in water and sanitation; in higher-income, institutionally stronger countries; and among DFIs with commercially oriented mandates. More public-service-oriented DFIs, by contrast, rely on non-conventional instruments such as guarantees and equity, and are more responsive to geostrategic priorities. We test these expectations using project-level models of both the likelihood and scale of private participation, exploiting within-DFI variation with sector and year fixed effects. The findings illuminate the organizational and political constraints shaping mobilization strategies, and advance understanding of how public development finance institutions structure the boundary between public and market-facing infrastructure provision.