Temporal fragmentation in the fiscal–monetary–inflation system: regime-dependent associations in Bangladesh, 1973–2025

The fiscal–monetary–inflation system does not operate as a coherent long-run structure in Bangladesh over 1973–2025. Using a five-link framework tested via autoregressive distributed lag (ARDL) bounds estimation, Bai–Perron structural break tests, regime-specific Granger causality, and a four-variable VAR, we report three findings. First, the system is temporally fragmented: the long-run association between money growth and inflation is 0.389 (p < .01) before the 1991 trade liberalisation and statistically indistinguishable from zero thereafter, the post-1991 effect being −0.010 with a standard error of 0.113. Second, the fiscal deficit–currency growth relationship becomes negative after the 2003 float, moving from +0.32 in 1973–1990 to −2.87 (p < .05) in 2004–2025; larger deficits are associated with slower currency growth, consistent with non-monetary deficit financing rather than textbook seigniorage. Regime-invariance tests do not reject parameter stability in this equation, so we describe the pattern as exploratory. Third, exchange-rate pass-through shifts significantly at the 2003 float (slope shift 1.090, p = .047), though post-float pass-through of 0.282 is imprecisely estimated. Bangladesh’s monetary aggregates ceased to co-move with prices before the recent inflation episode, which helps explain why monetary tightening has produced limited disinflationary effects.