Paying for Knowledge
Teacher Certification, Pay, and Workforce Adjustment
Armenia's certification reform is not just a bonus program. It is an attempt to use a subject exam as a screening device, a salary rule, and eventually a workforce-management tool. The paper links first-attempt certification scores to teacher value added in a nationwide teacher–student panel and then asks what happens when a one-time knowledge screen becomes a permanent pay rule. It follows both classroom performance and workforce adjustment, including retesting, concurrent school assignments, staffing pressure in scarce subjects, and the possibility that later effects operate through hiring, attrition, hours, and entry rather than through incumbent effort alone.
The empirical results separate three objects that policy discussions often collapse. First, subject knowledge does predict classroom effectiveness, especially in mathematics: teachers with higher first-attempt scores have higher pre-exam value added. Second, the threshold-linked raise does not improve performance for teachers near the passing cutoff; by year three, the local estimates are negative. Third, average effects away from the cutoff are much closer to zero, with any gains concentrated among the strongest scorers. Read together, the paper suggests that the reform measures something real, but that the main policy action may lie in screening and workforce adjustment rather than in a broad effort response among marginal passers.
Administrative microdata from KTAK (NaCET), the National Center for Educational Technologies, linked to student achievement and school-level records. The panel covers the full population of public-school teachers across certification rounds.
Regression discontinuity at the certification exam threshold. Separates the informational content of subject knowledge from the incentive effect of the raise by comparing teachers just above and below the passing cutoff.
Bullets to Ballots
War, Bereavement, and Democratic Politics in Armenia
This paper is built around a micro–macro problem. At the household level, direct wartime loss can pull surviving family members out of politics. At the community level, concentrated casualties can instead intensify turnout and redirect votes away from the incumbent toward nationalist opposition. The paper brings those two margins into the same design, asking how a young democracy processes grief, blame, sacrifice, and political accountability after military defeat.
The identification uses an RD-IV strategy based on Armenia's mandatory conscription system and lottery-based draft allocation to generate plausibly exogenous variation in who faced elevated war exposure. On top of that administrative core, the project digitizes Armenian voter registries whose signature fields reveal turnout without revealing vote choice, and pairs those data with casualty records, election returns, and an original CATI survey. The survey is managed through both an in-house team and a contracted survey firm, and is not an ornamental add-on: it is designed around the same local exposure comparison, with a bereavement booster, so the paper can distinguish direct bereavement from broader conscription exposure and test mechanisms such as trauma, blame attribution, trust, and security preferences.
The substantive contribution is that the politics of wartime loss are scale-dependent. Preliminary results indicate that direct household bereavement reduces participation among surviving household members, while community-level responses are nonlinear: moderate losses are associated with demobilization and limited incumbent punishment, whereas severe losses produce higher turnout and sharper movement toward nationalist opposition. That combination is what makes the paper interesting. It explains how private withdrawal and public mobilization can coexist in the same postwar polity.
The Latency Tax
School Assignment, Congestion, and the Demand for Seats
This project studies school assignment in Armenia from two angles at once: the policy history of the shift from decentralized enrollment to a centralized online platform, and the inefficiencies that persist when the centralized platform still works as a live, priority-based FCFS system rather than as a true ranked assignment mechanism.
The point is not merely that some families can click faster. The current mechanism is not strategy-proof, accepts one school at a time, and does not aggregate preferences across schools. That means parents must act sequentially under uncertainty, often watching the system closely, reacting to rejections or waitlist movement, and sometimes canceling and re-applying. Those frictions are especially costly for families balancing sibling priorities or multiple school-age children. Empirically, the project studies congestion at opening phases, cancellation patterns, placement outcomes, commute implications, and sorting across schools.
The related policy work argues that a ranked offline assignment system would remove timing advantages, reduce overload, handle sibling priorities more cleanly, and let the planner use the full preference profile rather than a stream of timestamped moves.
Conflict, Son Preference, and Demographic Response
This project asks how external threat enters household demographic behavior. In a setting with low fertility, persistent son preference, and access to prenatal sex determination, conflict may affect not only politics but also sex ratios at birth, birth counts, and fertility timing.
Public School Choice and Racial Sorting in California
This paper uses California administrative data and a system-grade-year design to study how charter and magnet expansion reshape student sorting across traditional public schools. The question is deliberately narrow: not whether “school choice” is good or bad in the abstract, but how different institutional forms of public school choice change racial sorting inside the same public system. The current draft indicates that charter and magnet growth do not move segregation in the same way.
The Value of Information in School Choice
This project studies how information changes behavior inside New York City's Deferred Acceptance mechanism. It focuses on what families know — or fail to know — about priorities, tie-breakers, and risk, and how that information changes preference reporting, strategic behavior, trust, and perceived fairness. The central policy question is whether a formally strategy-proof mechanism remains behaviorally well-functioning when the information environment is opaque or uneven.
Brain Drain, Occupational Sorting, and Return Migration
This project studies how educational pathways in Armenia shape migration for higher education, subsequent occupational sorting abroad, and eventual return or repatriation. It combines Armenian educational histories with LinkedIn and other public professional sources to trace who leaves, where they study, what sectors they enter, and how those trajectories interact with home-country opportunity and talent misallocation.
Policy Writing
Short pieces for policymakers, practitioners, and nonacademic audiences.
Teacher Certification and Pay in Armenia
A policy brief on what Armenia's certification reform appears to measure well, where classroom effects are limited, and why staffing, retesting, and other workforce margins may matter more than a one-time threshold effect.
War, Bereavement, and Democratic Politics in Armenia
A short brief translating Bullets to Ballots for a broader audience, with emphasis on turnout, accountability, why private withdrawal and public mobilization can coexist, and what wartime loss means for democratic responsiveness.
School Assignment Reform in Armenia
A note explaining why live FCFS assignment creates congestion, sequential decision costs, and inequity, and how a ranked centralized assignment system changes the problem for families, schools, and the ministry.