Central banks in emerging economies routinely face a hard problem: how to forecast inflation when the underlying regime is shifting under their feet. Recurrent policy shocks, currency stress, and administered-price interventions make the parameters of any single model unstable, and the operational cost of getting the forecast wrong is high. This paper proposes an integrated, data-driven decision-support framework that combines three complementary techniques—structural break detection, Lasso regularization for feature selection, and ARIMAX modeling with identifiable policy dummies—to forecast monthly inflation in Bolivia. Using a monthly dataset that spans March 1980 to May 2026 (555 observations), we apply the Bai–Perron, CUSUM/CUSUMSQ, and Chow tests to identify structural regime changes; a Lasso model with time-based cross-validation to distill the most informative predictors from a candidate pool of 27 variables; and an ARIMAX(1,0,1) model with two policy dummies—one for the fuel subsidy period and one for the fixed-exchange-rate regime—to quantify how much specific policy interventions contribute to inflation dynamics. The results confirm that the 1985–1986 post-hyperinflation stabilization is the dominant structural break in the series (Bai–Perron BIC-selected break; Chow F = 212.5), while the August 2004 introduction of the fuel subsidy emerges as a statistically significant break in its own right (Chow F = 23.8, p < 0.0001), which empirically justifies modeling it explicitly. The Lasso singles out a monetary differential factor and international reserves as the most robust linear predictors, and the ARIMAX finds that the fuel subsidy period is associated with an additional 0.28 percentage points of monthly inflation (p = 0.023). It also delivers, to our knowledge, the first quantitative estimate of the parallel-market exchange-rate pass-through for Bolivia: 1.32 percentage points per boliviano of depreciation (p = 0.009). Together, these outputs provide a reproducible and interpretable pipeline that monetary authorities can use to track inflation drivers when regimes shift.

Keywords

Inflation forecasting; Structural breaks; Lasso regularization; ARIMAX; Central bank decision support.