Sterilized FX interventions may not be so sterilized

It is widely believed that sterilized interventions do not affect domestic currency interest rates. The reason is the word "sterilized". Yet we show in this paper that when collateral base for central bank operations isn't huge, sterilized interventions may still affect interest rates, loan extension and, hence, real economy (beyond the effects of altered exchange rate). The mechanism is simple: when banks make decisions about loan extension and, hence, deposit (money) creation, they take liquidity risk into account. When collateral base for central bank operations is not big enough, even if collateral constraint is not currently binding, banks may still fear (massive) withdrawals that, in principle, can get them to the constraint. This fear is reduced when they get permanent liquidity (from the central bank that buys FX) as opposed to getting the same amount of liquidity by borrowing from the central bank (that requires collateral). Reduction in this fear will then result in loan interest rate reduction and/or easier terms for loans. We demonstrate the importance of this mechanism through three different approaches: accounting, theoretical and empirical. The quantitative importance of this channel depends on the amount of unused collateral: the more the collateral the lower the liquidity risk and associated interest rate-effects of FX interventions. In addition, the framework provides other interesting insights about the relationship between liquidity risk and reserve requirements.