We estimate fiscal spending and tax multipliers for the US economy by employing di¤erent combinations of fiscal and non-fiscal instruments in a proxy-SVAR framework. We do so by working with a novel identification strategy which allows us to jointly estimate the multipliers and assess the overall validity of the SVAR specification. We provide robust evidence in favor of a fiscal spending multiplier larger than one. Turning to the tax multiplier, we show that the strikingly different estimates one finds in the literature may depend on the orthogonality assumption regarding the non-fiscal instrument (namely, total factor productivity shocks) and tax shocks. In particular, we show that assuming total factor productivity shocks to be orthogonal to tax shocks implies a tax multiplier close to one. However, if this orthogonality assumption is relaxed, the tax multiplier is estimated to be three times larger.