Paper: arXiv 2604.09821

Authors: Oleg Roshka

Abstract

On a 93-actor quarterly panel mixing macro indicators, institutional data, and firm-level investment ratios, global factor augmentation degrades prediction for actor subgroups whose dynamics are misrepresented by the shared basis. A two-stage architecture – global pooled AR(1) for shared persistence, block-specific local models for residual dynamics – improves full-panel out-of-sample $R^2$ from 0.630 to 0.677 ($Δ= +0.047$, CI $[+0.036, +0.058]$, 10/10 windows, placebo $p \leq 0.001$). A held-out decade test (block partition frozen on 2005–2014 data, evaluated on unseen 2015–2024 windows) confirms the gain ($Δ= +0.050$, 10/10), and a stratified placebo that fixes the macro/firm data-type split and permutes only firm-sector assignments corroborates ($z = 7.25$, $p \leq 0.001$). Cross-regime replication on a 109-actor UK/EU heterogeneous panel ($Δ= +0.017$, 8/8 windows) and a combined US + UK/EU panel of 202 actors ($Δ= +0.030$, placebo $z = 9.68$ – exceeding the original US-only $z = 7.82$) confirms the architecture transfers across regimes. A 146-firm CapEx/Assets robustness check refines the scope condition: the gain depends on cross-sectional dispersion in autoregressive structure, which data-type heterogeneity reliably produces but which is also present in firm-only panels under suitable ratio choices.

Complexity vs Empirical Score

  • Math Complexity: 5.5/10
  • Empirical Rigor: 8.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper provides a highly rigorous empirical framework for forecasting heterogeneous panels using a two-stage global/local architecture. It features exceptional statistical testing across multiple geographic regimes, out-of-sample held-out decades, and stratified placebos.

Research Flowchart

  flowchart TD
    A["Research Goal: Forecast Heterogeneous Investment Panels<br/>(Addressing global factor mismatch)"] --> B["Data Inputs<br/>• US Panel: 93 actors (Macro, Inst., Firm)<br/>• UK/EU Panel: 109 actors<br/>• CapEx/Assets Panel: 146 firms"]
    
    B --> C["Two-Stage Architecture<br/>1. Global Pooled AR(1): Shared Persistence<br/>2. Block-Specific Local Models: Residual Dynamics"]
    
    C --> D["Validation & Stress Testing<br/>• Held-out decade test (2015–2024)<br/>• Stratified sector placebos<br/>• Cross-regime transfers (UK/EU & Combined)"]
    
    D --> E["Primary Outcome: Predictive Gain<br/>• Full-panel OOS R² increases from 0.630 to 0.677<br/>(Δ = +0.047, 10/10 windows, p ≤ 0.001)"]
    
    D --> F["Key Robustness & Scope Finding<br/>• Transferability confirmed across regimes (US + UK/EU z = 9.68)<br/>• Gains depend on cross-sectional dispersion in AR structure"]

    E --> G["Conclusion: Two-stage model outperforms global factor<br/>augmentation in heterogeneous panels"]
    F --> G