Collateral-based monetary policy tools have been used extensively by major central banks. Lack of proper policy counterfactuals, however, makes it difficult to empirically identify their causal effects on the financial market and the real economy. We exploit a quasi-natural ex-periment in China, where dual-listed bonds are traded in two mostly segmented markets: the interbank market regulated by the Central Bank, and the exchange market regulated by the securities regulator. During a policy shift in our study period, China’s Central Bank included a class of previously ineligible bonds in the interbank market to become eligible collateral for financial institutions to borrow money from its Medium-Term Lending Facility (MLF). This policy shift allows us to implement a triple-difference strategy to estimate the causal impact of the collateral-based unconventional monetary policy. We find that in the secondary market the policy reduced the spreads of the newly collateralizable bonds in the treatment market (the interbank market) by 42-62 basis points. We also find that there is a pass-through effect from the secondary market to the primary market: the spreads of the treated bonds newly issued in the interbank market were reduced by 54 basis points.