This thesis is comprised of three essays on the role of financial intermediaries in the transmission of various macroeconomic shocks to the real economy. The first essay explores why currency depreciations do not always have an expansionary impact on the economy. I posit a new channel of exchange rate transmission that works by affecting the lending capacity of banks, and show that this channel can offset the trade channel, explaining the muted response of economic activity to exchange rate shocks. To circumvent endogeneity concerns, I exploit a large and unanticipated currency appreciation shock from Switzerland in January 2015 when the Swiss National Bank surprised the markets by abandoning the lower bound on the chf/eur exchange rate. Using a novel hand-collected dataset on foreign currency exposure of Swiss banks and bank-firm relationships, I show that the currency appreciation shock enabled banks with a net foreign currency liability exposure to increase credit supply. I find that non-financial firms that had a pre-shock relationship with positively affected banks were able to invest more, partially offsetting the negative impact of currency appreciation on exporters. The second essay is co-authored with Rupa Duttagupta and Andrea Presbitero. This essay examines whether the banking sector amplifies the negative effects of a commodity price decline on real activity in commodity exporting countries. Using bank-level data for low-income developing countries, we find that banks with a relatively high exposure to commodity prices reduce lending in periods of commodity price busts. To disentangle the credit demand from the credit supply channel, we use loan-level data from a credit register in Uganda and show that a fall in commodity prices reduces loan growth of more exposed banks, even after controlling for borrowers' demand for bank credit. This finding has implication for macroprudential policies in low-income developing countries that are particularly dependent on commodity exports. The third essay, co-authored with Matthew Baron, documents a new channel through which unexpected inflation leads to adverse short-run effects on the macroeconomy. We hypothesize that unexpected inflation shocks weaken the banking sector (mainly due to asset-liability mismatch), which leads to credit contraction, which, in turn, transmits the shock to the real economy. We test this hypothesis in two settings. In the first setting, which looks at a sudden and unexpected inflation shock in the U.S. in mid-1976, we exploit across-state differences in reserve requirements for state-chartered non-member banks and within state differences between state- and nationally-chartered banks. These differences substantially affect bank cash holdings, and thus banks' inflation exposure. Through the affected banks, the inflation shock is then transmitted to the real economy through a lending contraction, with small bank-dependent non-financial firms most affected. In a second setting, we use newly-uncovered historical data on individual banks' financial statements to explore prominent high inflation episodes of the past, from France in the 1920s to Argentina, Brazil, Turkey, and Venezuela in recent decades. We exploit banks' cross-sectional heterogeneity in exposure to large, unexpected inflations to show the importance of the banking channel in these prominent historical inflation episodes.