Political Portfolio

Thesis

Campaign finance is usually studied as a causal question: does spending affect elections?

In practice, campaign committees face a different problem: given a fixed budget, where does the next dollar generate the highest expected increase in seats — and is now even the right moment to spend it?

This reframes campaign spending as a capital allocation problem under uncertainty, not a simple outcome regression — and, because committees don't spend once but continuously over a two-year cycle, extends into a sequential decision problem about when that capital should move.

Approach

I. Causal Spending → Vote Margin → Win Probability
Using Levitt's (1994) repeat-challenger design, I isolate the causal effect of spending on vote margin within matched incumbent–challenger pairs (2012–2024). Vote margin is then mapped nonlinearly into win probability, reflecting that dollars are more valuable in competitive races and marginal effects are convex near 50/50 contests.

I. Portfolio Allocation Across Races
House races are not independent. National swings introduce correlated risk through the generic ballot. I model this explicitly using a shared factor structure across races, constructing a covariance-aware allocation model that treats the House map as a portfolio rather than isolated bets. A persuasion ceiling regularizes the model against inferring unbounded returns at near-zero spending floors.

II. A Sequential Architecture for Deploying Capital
A committee's budget splits into capital already irreversibly committed and capital still deployable. Paper II reframes the static optimizer as a rollout policy — re-solving the allocation each reporting period over only what remains deployable — and shows this is precisely not full model-predictive control, since it prices no explicit value for preserving future flexibility.

III. Pricing the Value of Waiting
Capital held in reserve isn't idle — it's an option. Paper III specifies the stochastic process campaign state actually follows (opponent reaction to spending, national-environment drift, race-level noise) and prices Θ, the value of not yet committing capital, via Longstaff–Schwartz regression-based Monte Carlo — the same machinery used to price early exercise in American options.

Key Insight (Open Seats)

Open seats behave differently from incumbency races. They resemble higher-variance assets:

• Wider outcome uncertainty increases baseline win probability
• But reduces marginal return to additional spending

This creates a risk/return tradeoff analogous to volatility exposure in options pricing.

Empirical Test

If spending is efficient, observed allocations should be positively correlated with estimated marginal seat gain. I test this among competitively matched races controlling for partisan lean and incumbency — and separately test whether a patience-blind allocation policy matches how committees actually pace spending over a cycle.

Results

• ρ = −0.809 (p < 0.001) across 53 competitive 2024 races → spending is systematically misallocated relative to marginal seat gain
• +2.83 expected seats from reallocating the same budget using model ranking
• Out-of-sample validation (trained on 2012–2020, tested on 2022): ρ = −0.847, +3.22 seats improvement
• Calibration benchmark — Brier score: 0.031 (model) vs. 0.036 (Cook PVI baseline), 2024
• DCCC's actual spending doesn't reach 25% of its eventual cycle total until the final four weeks of a ten-month cycle, in both 2022 and 2024 — a patience-blind policy would front-load nearly all deployable capital instead
• Θ(0) = +4.5 to +5.1 expected seats — the value of holding a deployable reserve one more period, at a 98-day horizon, across three independent calibration scenarios; every scenario recommends holding rather than deploying
• A mechanism decomposition attributes roughly 70% of that value to candidates' own predictable organic spending growth, not to resolving genuine electoral uncertainty

Contribution

• Reframes campaign finance as a constrained optimization problem over expected seat gain
• Builds a causal estimate of spending → vote margin using repeat-challenger identification
• Extends inference into a risk-aware allocation model across correlated races
• Formalizes the sequential capital-commitment problem committees actually face, distinguishing a rollout policy precisely from full model-predictive control
• Prices the option value of an uncommitted reserve using a real-options / regression Monte Carlo framework, calibrated to public FEC and polling data
• Produces falsifiable, out-of-sample claims about allocation inefficiency and about the value of patience

Methods

Repeat-challenger causal design, Bayesian shrinkage, heteroskedastic uncertainty modeling with Duan smearing, factor-based covariance modeling, nonlinear portfolio optimization (LP + SLSQP), Longstaff–Schwartz regression-based Monte Carlo, random-effects panel estimation, bootstrap inference.

Data: MIT Election Lab, FEC filings, OpenSecrets, Cook PVI, Census ACS (all public sources, one proprietary: Cook)