Job Market Paper

This paper disentangles the roles of capital regulation and credit ratings in mitigating insolvency risk in the U.S. property insurance market. I first investigate the mechanism through which capital requirements affect the insurance market. Using an instrumental variable approach that exploits a 2017 policy change as a quasi-experiment, I find that a $1 million increase in required capital leads insurers to hold $3.34 million more in capital and to raise insurance prices by 0.218 percentage points. These results reveal a direct trade-off between financial stability and consumer affordability. To further explore the underlying mechanisms, I develop a structural model in which insurers make capital and pricing decisions in a competitive market with limited liability and exposure to catastrophic risks. Counterfactual analyses show that tightening capital requirements improves solvency but raises prices. In the absence of capital regulation, the model predicts that the insolvency rate would increase by 0.09 percentage points, while insurance prices would decline by about 5.1%, accompanied by greater risk-taking and market concentration. A third counterfactual scenario examines a market without capital regulation but with high credit rating salience. When consumers place greater emphasis on credit ratings, the insolvency rate decreases; however, intensified price competition reduces profitability and increases market concentration. Overall, the findings underscore that capital regulation remains crucial for sustaining market stability, as heightened rating salience alone cannot fully substitute for its stabilizing effects.

This paper studies the impacts of different types of post-secondary education on education and labor market outcomes, with a particular focus on training participation during unemployment under Canada’s Employment Insurance (EI) system. I first use variations of distances to institutions and exogenous variations of colleges that upgraded into universities to investigate the value added to university education. Then I separately estimate the impacts of the transformed universities and traditional colleges and universities. I compare results from OLS and IV regressions. To address treatment heterogeneity, I adapt the locally linear specification from Mountjoy (2022). Results suggest that the difference between university and college graduates is marginal. However, university entry improves labor market outcomes, such as employment and earnings, compared with cohorts without post-secondary education. Graduates from transformed universities obtain a higher probability of being employed and higher earnings compared with people without post-secondary education. Nonetheless, transformed university graduates may have worse performance in the labor market compared with graduates from colleges or traditional universities. In addition, transformed university graduates are more likely to register for EI-supported training programs during unemployment, compared with graduates from colleges, traditional universities, and individuals without post-secondary education, highlighting an important interaction between higher education pathways and training incentives embedded in the EI system.

The staggering computational requirements of artificial intelligence are creating an unprecedented terrestrial energy and thermal bottleneck. While mega-scale space solar energy remains a long-term goal, technology firms and sovereign actors are currently engaged in a pre-emptive orbital competition for AI data computation centers. Because near-Earth orbits are a finite, common-pool resource in a global game where international treaties are unenforceable, this paper develops a continuous-time stochastic differential game to analyze this astropolitical market failure caused by externality. I first prove why sovereign states will rationally defect from centralized space treaties, necessitating private, self-enforcing financial contracts. I then design a mechanism—the orbital-sustainability bond—and provide derivations for the firm's valuation under endogenous, state-dependent jump-to-default risk. By expanding the framework into structural corporate finance, I solve for staged compound real options and Power Purchase Agreement (PPA) derivative collars that mathematically cure the agency friction between founders and institutional bondholders. The optimally calibrated mechanism generates a double dividend: it lowers the Phase 1 entry threshold to P = 3.46, while heavily delaying secondary competitors and inducing a spatial pivot to deep space. Ultimately, this framework provides a highly actionable corporate finance architecture that halts the value-destroying pre-emptive race and yields a 21.7% recovery of deadweight loss.

Regulators in the U.S. property insurance market face a critical challenge: transitioning from static, formula-based capital requirements to dynamic, model-based regimes. While dynamic regulation offers the potential to improve social welfare and market resilience, it suffers from two major barriers: computational intractability under profound uncertainty and a lack of interpretability required for regulatory oversight. In this paper, I propose a novel framework for outcome-based regulatory design. I develop a Learn-Verify-Explain methodology that utilizes Deep Reinforcement Learning to discover optimal dynamic capital strategies. Unlike traditional black-box approaches, my framework integrates Formal Verification to mathematically guarantee compliance with safety constraints and Decision Tree Extraction to distill complex policies into transparent, implementable rules. Empirical results demonstrate that this hybrid approach outperforms traditional static benchmarks, increasing social welfare by approximately 35% while reducing insolvency rates to zero. Crucially, the distilled policy reveals a risk-sensitive stabilization strategy: the agent learns to prioritize market efficiency through deregulation during stable periods, while imposing immediate corrective tightening upon detecting early signs of distress. This study provides an experiment of AI-in-the-loop financial regulation.

I analyze the growing tension between insurer solvency and insurance affordability in property markets exposed to climate change. I develop a structural model where risk-averse households choose between purchasing insurance and investing in physical risk adaptation, while competitive insurers face regulatory capital requirements for systemic climate risk. My model demonstrates two findings: (1) high frictional cost of regulatory capital creates a high entry price for insurance that is relatively insensitive to individual adaptation, and (2) below a specific wealth threshold, both physical adaptation and insurance coverage collapse as households lack the liquidity to clear the solvency-loaded premium. Through policy counterfactuals, I find that premium subsidies are counterproductive; by making insurance artificially cheap, they cause households to increase financial coverage while decreasing physical adaptation effort. In contrast, adaptation grants that directly target the liquidity friction are highly effective, allowing households to jump to a higher state of resilience, which improves both household welfare and market-wide financial stability.

Modern industrial agriculture relies on monocultures to maximize economies of scale, yet this genetic homogenization introduces profound systemic risks to global food security. This paper investigates whether market competition naturally corrects this trend or exacerbates it, utilizing a two-stage structural model of endogenous cultivar choice and price competition. By estimating a random coefficients Logit model, I quantify the static welfare gains from cultivar variety and the dynamic insurance value of biodiversity against correlated supply shocks. Simulation results demonstrate that consumers derive significant utility from variety, with a transition to pure monoculture resulting in a 73% loss in static consumer surplus. Crucially, I find that market structure determines this vulnerability: a simulated competitive oligopoly sustains four distinct cultivars, whereas a monopoly restricts the market to just two, internalizing cannibalization. Furthermore, I identify a substantial market failure: while monoculture is statically efficient in normal states, it creates unpriced systemic tail risk. Under simulated shocks, the expected social welfare of a diverse biological portfolio is approximately 68% higher than that of a monoculture. These results suggest that profit-maximizing firms structurally under-provide biodiversity, providing a rigorous economic rationale for biodiversity subsidies and stringent antitrust scrutiny in agribusiness.

Emergency Department (ED) overcrowding presents a dual crisis: it compromises patient safety and destabilizes the medical liability insurance market. High-cost malpractice claims, particularly those stemming from failure to diagnose ambiguous cases, drive up insurance premiums, thereby increasing healthcare overhead and limiting patient access to affordable care. Traditional triage protocols and deterministic Machine Learning (ML) models optimize for operational efficiency but ignore the fat-tail liability risks associated with diagnostic uncertainty. This paper proposes a novel Defensive Triage mechanism that integrates Bayesian Uncertainty Quantification into queueing logic. By prioritizing patients based on epistemic uncertainty, we effectively hedge against catastrophic misdiagnosis. Using a discrete-event simulation benchmarked against standard protocols, I demonstrate that this approach significantly reduces aggregate liability exposure. I argue that by lowering the frequency of high-severity claims, Defensive Triage allows insurers to stabilize premiums. This creates a virtuous economic cycle: reduced insurer risk lowers provider costs, ultimately expanding the capacity of the healthcare system to provide accessible, insured care to a broader population.

As populations age, the primary economic threat to healthcare sustainability is the accelerated depreciation of health capital. While traditional triage focuses on short-term liability shielding, musculoskeletal (MSK) triage in elderly populations requires a long-term capital preservation approach. This research proposes three novel frameworks: The Functional Autonomy Hedge (FAH), which utilizes gradient-based symptom volatility; Expected Liability Calibration (ELC), which optimizes for insurance solvency by pricing the marginal cost of diagnostic delay; and the Frailty-Decay Shield (FDS), which incorporates a biological aging parameter. By identifying latent invisible risks near clinical cliffs, I employ actuarial policies to prioritize patients with high-risk profiles. Simulation results (N=5,000) demonstrate that while FAH offers a modest 5.6% liability reduction over standard AI, the optimized ELC framework achieves a 50.2% reduction. The FDS framework, by pricing frailty, provides the superior result with a 52.4% reduction, effectively serving as a mathematically rigorous hedge against the systemic risks of the Silver Tsunami.

(Draft available on request)

This paper studies how employer mandates and health insurance affect labor market outcomes and health. I use staggered difference-in-differences research design and variations in the Affordable Care Act to learn how employer mandate affects labor market outcomes. I use doubly robust difference-in-differences in my main specifications to reduce selection bias. Results in the full sample suggest that the employer mandates in the Affordable Care Act increased hourly wages and did not have significant impacts on employment and part-time employment. Employer mandates stimulate a larger increase in employer-sponsored health insurance coverage rates among low-income workers. However, low-income workers are more vulnerable to involuntary part-time employment if employers reduce work hours to circumvent employer mandates. Firms prefer to reduce work hours to circumvent employer mandates instead of firing workers. Using doubly robust estimators and staggered difference-in-differences research design, I find evidence that providing health insurance improves workers' health. The employer mandate may increase productivity by improving workers' health status. Still, it may widen income inequality in the long run because low-income workers are more vulnerable to work hours losses.

The rapid expansion of artificial intelligence is creating a terminal energy bottleneck that traditional terrestrial grids are ill-equipped to handle. While space-based solar power represents the ultimate solution for high-density compute clusters, the technology remains in its infancy, necessitating a multi-decade transition strategy involving terrestrial solar and Small Modular Reactors (SMRs). This paper develops a structural corporate finance model to analyze the optimal sequencing of these heterogeneous energy investments under conditions of technological uncertainty and capital irreversibility. Using a continuous-time real options framework, I examine the trade-offs between the lower capital intensity of terrestrial solar and the high-baseload reliability of nuclear SMRs as bridges to eventual orbital energy migration. The model characterizes the technological overhang created by long-lived nuclear assets and the resulting stranded asset risk if space-based breakthroughs occur prematurely. I solve the firm's intertemporal optimization problem to derive endogenous investment thresholds and optimal capital structure dynamics. The paper aims to propose a novel class of contingent-claim financing instruments. The resulting framework provides a rigorous valuation architecture for technology firms and infrastructure investors navigating the high-stakes frontier of the AI energy transition.