This dissertation explores the relationship between financial intermediaries and the macroeconomy, both internationally and domestically. The dissertation is composed of two chapters, which study the role of financial sectors in international finance and the US economy, respectively.

In the first chapter, I build a two-country two-goods open economy model with asymmetric financial development levels to jointly match two puzzling observations during the 2008 global financial crisis (GFC): (1) the real consumption growth declined more in foreign countries than in the US although the crisis broke out in the US; (2) the US dollar appreciated against foreign currencies despite the sharp net foreign asset position (NFA) deterioration in the US. The key innovation is to introduce a milder financial constraint in the US within an international context. In tranquil times, the US holds more risky assets due to its more advanced financial market and consequently is more exposed to risks. When a crisis unfolds, the US financial intermediaries suffer heavier capital losses and are forced to liquidate risky asset holdings to deleverage. This deleveraging process leads to a capital retrenchment in the US, which will smooth the US consumption, raise the relative demand of US goods, push up the domestic price index, and lead to a US dollar appreciation. Meanwhile, the NFA deterioration in the US is mainly due to the valuation effect rather than the transaction effect. I show that the valuation effect does not affect the contemporaneous relative consumption between countries, a fact that has been overlooked in the literature. I then verify the model predictions in the data covering major countries from 2003 to 2018. The empirical evidence echoes the theoretical predictions: (1) The US financial sector net worth plunged more than its foreign counterparts and there was a large capital retrenchment in the US during the GFC. (2) The US capital retrenchment had a significant negative impact on foreign countries’ consumption and currency value.

In the second chapter, I identify a series of credit supply shocks in the US since the 1970s using sign restriction VAR and apply the identified shock to firm-level data using the panel local projection method. I document two main findings. First, the identified credit supply shock has a significant and persistent impact on firm investment. One unit of a credit supply shock (0.39% movement of credit volume) induces over 1% movement of firm investment on average at the peak. Different from monetary policy shocks, the credit supply shock has medium- and long-term effects on real variables. Second, the heterogeneous responses of firm investment to the credit supply shock conditional on firm default risk are state-dependent. Firms with lower default risk will increase investment more relative to average firms during credit expansions, and contract investment less during credit contractions. Low default risk firms have consistently lower borrowing costs and better access to external financing over credit cycles. This state-dependence of heterogeneous effects also holds for firms that are large, mature, and less volatile. The second finding has important implications for both empirical and theoretical studies in the future. On the empirical side, it suggests a possibility of misspecification when assuming the effects of firm characteristics on firm activity to exogenous shocks are invariant in expansionary and contractionary phases. This paper highlights that the effects of firm characteristics on firm investment to credit supply shocks have flipped signs and commensurate magnitudes during credit expansions and contractions. On the theoretical side, the evidence is consistent with the insights of financial friction literature arguing that financially constrained firms will be more negatively impacted during credit contractions. Nevertheless, it also suggests that this effect is nonlinear and state-dependent, i.e., financial frictions will amplify the effects of negative shocks, but dampen the effects of positive shocks. A future model featuring credit cycles and heterogeneous financial frictions should take this observation into account.