\begin{table}[htbp]
\centering\scriptsize
\caption{Simulation — MAE of 3-month index-change forecasts relative to the no-change forecast at the province level, by scenario (Monte Carlo s.e. over replications)}\label{tab:sim_province}
\begin{threeparttable}
\begin{tabular}{lrrrrr}
\toprule
Model & rw & momentum & regimes & spillover & sparse \\
\midrule
No change & 1.00 (0.00) & 1.00 (0.00) & 1.00 (0.00) & 1.00 (0.00) & 1.00 (0.00) \\
Mean growth & 0.68 (0.07) & 1.12 (0.09) & 0.79 (0.11) & 0.94 (0.13) & 1.12 (0.09) \\
Momentum & 0.96 (0.01) & 0.98 (0.01) & 0.82 (0.04) & 0.89 (0.02) & 1.00 (0.01) \\
Random walk with drift & 0.76 (0.06) & 1.05 (0.08) & 0.66 (0.07) & 0.84 (0.09) & 1.05 (0.08) \\
Seasonal naive & 1.33 (0.12) & 1.65 (0.19) & 1.05 (0.14) & 1.32 (0.16) & 2.02 (0.23) \\
AR(2), cell & 0.68 (0.08) & 1.10 (0.08) & 0.74 (0.10) & 0.89 (0.11) & 1.11 (0.09) \\
Local linear trend & 1.37 (0.09) & 1.35 (0.13) & 0.70 (0.07) & 1.11 (0.11) & 1.89 (0.23) \\
Ridge (panel) & 1.18 (0.16) & 1.87 (0.15) & 1.22 (0.22) & 1.59 (0.26) & 1.98 (0.20) \\
LightGBM (panel) & 0.83 (0.10) & 1.27 (0.13) & 0.75 (0.08) & 0.92 (0.11) & 1.26 (0.12) \\
MLP (panel) & 0.80 (0.11) & 1.36 (0.13) & 0.79 (0.11) & 1.19 (0.27) & 1.55 (0.28) \\
LSTM & 0.87 (0.14) & 1.29 (0.12) & 0.87 (0.15) & 1.10 (0.24) & 1.48 (0.25) \\
TCN & 0.82 (0.13) & 1.31 (0.17) & 0.85 (0.15) & 1.08 (0.26) & 1.38 (0.22) \\
TFT-lite & 0.71 (0.11) & 1.19 (0.13) & 0.79 (0.15) & 0.87 (0.12) & 1.22 (0.17) \\
BHPF & 1.19 (0.17) & 2.02 (0.17) & 1.24 (0.22) & 1.63 (0.26) & 2.04 (0.20) \\
RAREF & 1.24 (0.18) & 1.96 (0.20) & 1.26 (0.21) & 1.66 (0.29) & 2.01 (0.17) \\
QANSS & 1.90 (0.24) & 1.58 (0.15) & 1.24 (0.28) & 1.38 (0.27) & 1.68 (0.16) \\
DSGFN & 0.99 (0.12) & 1.46 (0.18) & 0.90 (0.13) & 1.33 (0.21) & 1.47 (0.25) \\
\bottomrule
\end{tabular}
\begin{tablenotes}\footnotesize\item rw: random walk with a constant drift; momentum: AR(1) drifts at every level; regimes: Markov-switching provincial drift; spillover: municipal drift follows the lagged regional drift; sparse: momentum with 35\% of the sales. Values below 1 beat the no-change forecast. Classic and tree models: 10 replications; neural models: 6.\end{tablenotes}
\end{threeparttable}
\end{table}