A dynamic seat product model

The Seat Product Model uses just two institutional parameters (average district magnitude and assembly size) to predict legislative fragmentation with remarkable accuracy. However, it is best suited for simple, single-tier systems. Subsequent extensions to complex, two-tier systems rely on an upper-tier (multiplier-style) correction that accounts for cross-tier inflationary effects but estimate it as a single constant from pooled cross-national data. Applied to Mexico’s mixed-member lower house elections (1997–2021), both specifications systematically mispredict legislative fragmentation. This is because cross-tier inflationary effects are not a fixed system property but an evolving one, shaped by institutional features interacting with a changing party system. This study proposes a Dynamic Seat Product Model that retains the original institutional baseline while allowing the correction to update automatically via exponential smoothing. The DSPM tracks Mexico’s long-run rise in legislative fragmentation more closely than fixed alternatives, clarifies the mechanism behind it, and offers a principled path toward making the SPM applicable to virtually any single election.