Working papers

• The Market and the Jungle: Dual Economies in Emerging Countries
  Juan Carlos Carbajal, Marcelo Gallardo and Cesar Martinelli.
  Abstract | Draft coming soon

  We combine Piccione and Rubinstein's (2007) jungle economy with the trading framework of Makowski and Ostroy (1995). We establish existence and efficiency of a jungle equilibrium for a fixed occupational profile, and existence in lotteries when occupations are chosen. We also define an equilibrium for a dual economy in which a competitive market operates side by side with a jungle. In the lotteries case the power relation is drawn after occupations are chosen, so the expected payoff averages over finitely many power relations and is continuous. Strategies are Borel measures over each menu, a convex set that is weak*-compact by Banach–Alaoglu. Berge's maximum theorem makes the best-reply correspondences upper hemicontinuous with non-empty compact values, convex because payoffs are affine, and the Kakutani–Fan–Glicksberg fixed point theorem then yields the equilibrium — Kakutani's own theorem does not suffice, the strategy sets being infinite-dimensional.
• Information and voting: Evidence from Peru’s 2026 presidential election - Accepted for presentation at LACEA LAMES 2026 (Lima, Peru)
  Marcelo Gallardo, Nicolás Velarde and Cristina Gutarra.
  Abstract | Preprint (very preliminary)

  We study how election-night flash estimates shape voting in Peru's fragmented 2026 presidential election. We exploit a natural experiment: on April 12, 2026, 187 polling tables across 13 voting centers failed to install, and the Jurado Nacional de Elecciones (JNE) extended voting for the affected ≈55000 electors to Monday, April 13. These voters cast ballots after observing the Ipsos and Datum flash estimates; otherwise comparable Sunday voters did not. A Bayesian-updating model of multi-candidate plurality voting frames the analysis, yielding predictions about vote reallocation toward the three candidates the estimates rendered viable -- López Aliaga, Sánchez, and Nieto. We estimate treatment effects on candidate vote shares at both the acta level and the acta-weighted polling-station level, comparing treated and control locales de votación matched on pre-treatment covariates. How flash estimates reshape voting is of first-order importance for Peru, given its institutional instability and high political volatility over the past decade.
• Economic and Political Uncertainty Index for Peru Using X and DeepSeek-V3
  Marcelo Gallardo, Manuel Loaiza and Gabriel Rodriguez.
  Abstract | Preprint

  We propose the first daily index of economic and political uncertainty (EPU) for Peru, and its defining novelty is the source: rather than newspapers, we read the real-time discourse on X (formerly Twitter). Dominant policy-uncertainty indices count newspaper articles and are typically monthly and lagged; X instead records, at daily frequency, how the country's most influential voices in politics, economics, journalism, and business react to events as they happen. We classify their messages along economic, political, and uncertainty dimensions with a large language model (LLM), DeepSeek-V3, and recombine these dimensions into a family of complementary daily indices, among them a broader political-uncertainty (PU) index. We then ask how closely the index behaves like two established market-based measures of uncertainty, the Chicago Board Options Exchange Volatility Index (VIX) and the volatility of the Bolsa de Valores de Lima (BVL). It co-moves with both at a moderate level, and more strongly within documented crises. This is what one expects of measures that share a common uncertainty component but differ in what else they capture: the index reads domestic political and economic uncertainty, the VIX global financial risk, and BVL volatility the Lima equity market. The moderate co-movement is therefore not a shortcoming but evidence that the index is not redundant with the benchmarks.
• Field Experiments on Seasonal Products and Markdown Pricing
  Marcelo Gallardo, Carlos Noton and Marcelo Olivares.
  Abstract | Draft coming soon

Completed papers

• Congestion and Penalization in Optimal Transport, 2025 - R&R at Decisions in Economics and Finance (Springer)
  Marcelo Gallardo, Manuel Loaiza and Jorge Chávez.
  Abstract | Preprint

  We propose a new model that transforms the classical discrete optimal transport framework by incorporating heterogeneous congestion costs and replacing traditional equality constraints with weighted penalization terms. The resulting formulation is a strictly convex optimization problem that better captures demand–supply imbalances in economic matching contexts and the congestion phenomenon. We first introduce the model and establish existence and uniqueness of the optimal transport plan under general conditions. For interior solutions, we present two analytical methods—based on the Neumann series expansion and the Sherman–Morrison formula—and develop a practical $O((N+L)N^2L^2)$ algorithm for computing the optimum. We then address the case of infinitely many types, corresponding to optimal transport on measure spaces, absolutely continuous with respect to Lebesgue, and prove existence and uniqueness under reasonable assumptions via infinite-dimensional optimization methods. Finally, we illustrate the applicability of our framework with examples from Peru’s health and education sectors, showing how it yields allocation patterns that differ from classical approaches and provide more accurate predictions. Pre-print in arXiv differs from the last version.
• Heterogenous Quadratic Regularization in Optimal Transport, 2025 - Submitted to RAIRO - Operations Research
  Marcelo Gallardo, Manuel Loaiza and Jorge Chávez.
  Abstract | Preprint

  In this paper, we build upon the optimal transport quadratic regularization model to develop a framework that incorporates congestion costs, particularly in matching within the healthcare and education sectors. Specifically, we introduce a model with heterogeneous quadratic costs. We analyze the model's properties under specific cases, extending the existing literature. Furthermore, we explore key structural characteristics of the model and provide numerical examples illustrating why this formulation more accurately captures real-world phenomena, particularly in the Peruvian context. The main result consists of identifying a specific type of corner solution when matching the same number of clusters, i.e., N=L.
• Irregular wave dynamics driven by a random force within the Burgers equation, 2024 - Submitted to International Journal of Applied and Computational Mathematics
  Marcelo Gallardo and Marcelo Flamarion.
  Abstract | Preprint

  In this article, we study the classical Burgers equation as a model for random fields. First, we consider initial data defined as a sum of harmonics with random phases and compute the blow-up time. Several simulations are performed, revealing that, while the critical blow-up time is approximately distributed according to a Gaussian law, the statistical tests reject the normality hypothesis. For the viscous case, we analyze waves driven by a random force. Using the Cole-Hopf transformation, the averaged wave field is computed numerically. Through a change of variables, we demonstrate that randomness primarily affects the phase of the wave field. Assuming the phase follows a uniform distribution, we show that the averaged field spreads and diminishes over time.

  ---