Abstract For two decades Bangladesh ran monetary policy through a reserve money programme, with broad money as the intermediate target for price stability. The programme assumed that changes in the base pass durably into the wider aggregates, yet this link has seldom been tested directly. This paper tests it, asking whether reserve money innovations reach narrow and broad money and, if so, whether the effect lasts. Monthly data for January 2003 to April 2025 (268 observations) are taken from Bangladesh Bank's Economic Trends, and observations from February 2023 onward are corrected for a unit inconsistency. The money multiplier enters only as a derived accounting ratio, because including it beside its own components would make the system singular. Unit root tests point to integration of order one and Johansen tests to cointegration, so the model is a vector autoregression in log levels with two lags and monthly seasonal dummies. Innovations are orthogonalised by a Cholesky decomposition, and confidence bands come from a moving-block bootstrap. A reserve money innovation of 3.25 per cent lifts broad money by 0.43 per cent on impact. The response is insignificant after eleven months, and most of the injection itself is reversed within a year. The derived multiplier drops by 2.83 per cent on impact in every specification tried, and in the baseline the fall loses significance after nine months. How narrow money responds depends on the identifying order. The evidence supports measurable short-run transmission alongside limited persistence, indicating that base money is a serviceable liquidity instrument but not a sufficient instrument of monetary control. The analysis identifies transmission within the money supply only, not effects on prices, output, credit or the exchange rate, and the evidence is consistent with, rather than proof of the superiority of, the move to an interest rate corridor in July 2023.