Statistical Model Checking of the Keynes+Schumpeter Model: A Transient Sensitivity Analysis of a Macroeconomic ABM

Battle Count: 50

This paper is fundamentally about methodological rigor in macroeconomic ABM analysis. It does not address trading strategies, asset pricing, portfolio construction, or market microstructure. The macroeconomic observables (unemployment, GDP growth) are policy-relevant but not directly used in quantitative trading. The SMC methodology could theoretically be applied to financial ABMs, but the paper itself has no trading application.

The integration requires: (1) a C++ macroeconomic ABM simulator (K+S model, non-trivial with heterogeneous firms, banks, households, two sectors); (2) a thin command-protocol bridge (main_mv entry point with reset/step/eval operations); (3) MultiQuaTEx query specification; (4) MultiVeStA configuration with precision targets and parallel workers; (5) 12 one-parameter sweeps × 6 values × 3 observables = 216 experiment-observable-parameter combinations; (6) 40 parallel workers with block size 30. The bridge itself is minimal (~20 lines of C++), but the underlying model is complex and the full campaign requires substantial computational resources (1450-14000 simulations per configuration).

The paper provides a detailed experimental design with explicit parameter grids (Tables 1-2), precision targets (Table 3), query templates (Listing 1.3), simulator interface code (Listings 1.1-1.2), and statistical settings (α=0.05, 40 workers, block size 30, seed-of-seeds=1). The C++ simulator is the well-known K+S model from Dosi et al. The MultiQuaTEx query pattern is fully specified. However, the exact simulator code and MultiVeStA configuration files are not explicitly linked in the extract. The methodology is described as a reusable workflow.