A Novel Grouped-Gram-Based Algorithm for Fast and Memory-Efficient Fixed Effects Estimation
High-Performance Variance-Covariance Matrix Construction Using an Uncentered Gram Formulation
Shows that Cov(X) can be computed as (XᵀX − (1/n)ssᵀ)/(n−1) with s=Xᵀ1, algebraically identical to the pairwise-difference form but avoiding explicit centering. Benchmarks indicate clear speedups over numpy.cov, with additional gains from faster Gram routines.
🧃 Political Interventions to Reduce Single-Use Plastics (SUPs) and Price Effects: A TWFE-Bias Corrected Event Study for Austria and Germany
High-frequency retail panel + fixed-effects event study. Clear price pass-through in Austria: pooled treated items +13.01 index points within 12 months (p<0.001) and +19.42 over the full post period (p<0.001). By product, balloons show strong and lasting effects (12m: 13.43, p=0.007; full: 19.96, p<0.001).
🚴♂️ Two-Stage Asymmetric Tullock Contests with Cost Shifters and Endogenous Continuation Decision
Sequential two-stage model of triathlon: Stage-1 continuation; Stage-2 Tullock contest with drafting as a multiplicative shifter of quadratic effort costs. Closed-form two-player solutions; existence/uniqueness and comparative statics for asymmetric n-player games.
📐 On Bessel’s Correction: Unbiased Sample Variance, the Bariance, and a Novel Runtime-Optimized Estimator
Derives a fast unbiased estimator for sample variance and introduces a new concept, the “Bariance.”
🏊♂️ Using COVID-19 Response Policy to Estimate Open Water and Swim Drafting Effects in Triathlon
Uses staggered triathlon starts during COVID to estimate causal drafting effects in open water swimming.
📊 Strategic Effort and Bandwagon Effects in Finite Multi-Stage Games with Non-Linear Externalities: Evidence from Triathlon using a Leave-One-Out IV Strategy
Analyzes group size and bandwagon effects in triathlons using a leave-one-out IV strategy.
🛒 Price-Endings and Seller Signals in Online Second-Hand Retail: Evidence From a Large Consumer Marketplace
Examines strategic .99 pricing and spatial clustering using scraped data from online bike markets.
🧪 alpha-miner 0.1.3 (WIP)
Implements the α-algorithm for automated process discovery using event logs. Lightweight, handy for high-level process mining in industrial settings.
austrian-cpi
Zero-dependency helper for Austrian CPI tables from Statistik Austria — perfect for dashboards, notebooks, web apps, or quick inflation adjustments.
Statistically Significant Linear Regression Coefficients Solely Driven by Outliers in Finite-Sample Inference
Shows how a single outlier can make a coefficient look significant in small samples; compares OLS to robust alternatives and emphasizes diagnostics.