This dissertation collects three essays on two hotly debated policy questions during and after the COVID-19 pandemic. Coupled with elevated uncertainty and falling confidence, the pessimism about future productivity and growth had dramatically discouraged business investment before the Great Lockdown. It has been widely accepted that heterogeneity matters for understanding the propagation of business cycles and how to target any policy intervention, but little is known about how firm heterogeneity matters in the propagation of expectation driven business cycles, i.e., cycles driven by news about future fundamentals. In the first chapter, I use U.S. firm-level data from Compustat combined with identified news shocks from Structural VAR analysis to empirically study the heterogeneous responses of firms' capital accumulation following the sudden arrival of news about future technology. I find that non-dividend payers that are younger, smaller, with lower leverage, higher liquidity, or lower cash flow respond the most to news shocks over a five-year horizon. After taking into account the correlation among different characteristics, liquidity, age, and cash flow remain powerful in predicting the heterogeneity in firm investment. Besides, I also provide suggestive evidence on the transmission mechanism of technology news shocks to investment dynamics through the impact on firm finance. Contrary to the previous macro-level evidence, I find a very limited and short-lived increase in firms' market values and share prices following good news. In contrast, firms' book values respond more persistently and prominently. The borrowing of long-term debts responds significantly more for younger, smaller, less indebted, and more liquid firms. The cash flow channel is less important in explaining the heterogeneity in investment response as the response in sales or earnings is short-lived and more homogeneous across firms. Thanks to the massive policy measures put forward by both fiscal and monetary authorities in the last couple of years, the pandemic recession turned out to be short. But U.S. inflation soon hit a four-decade high record in June 2022 even though the Federal Reserves had taken increasingly bigger steps to raise interest rates. The effectiveness of monetary policy to stabilize inflation greatly depends on the trade-off between inflation and economic slack. While numerous scholars have documented a weaker trade-off, known as the flattening of the Phillips curve, the precise reasons behind the change are still under debate. In the second chapter, I provide a new explanation and study the state dependence of the slope of the Phillips curve on the trend productivity growth. By merging two longitudinal databases, I present estimates of the ``average'' New Keynesian Phillips curve (NKPC) for 17 advanced economies across TFP growth regimes since 1890. Following the state-of-the-art method, I estimate the NKPC using trilemma monetary shocks as instruments and find that the Phillips curve is steeper (flatter) in the high (low) growth regime. My empirical finding is consistent with the following mechanism: the structural changes that contribute to higher productivity growth could also result in a more competitive market, increasing the price elasticity of demand so that the pass-through of marginal costs from short-run demand changes is larger. This mechanism is qualitatively in tune with the recent trends of a flattening Phillips curve and productivity slowdown amid rising market concentration in major advanced economies. The policy implication is that structural reforms that can improve productivity and restore business dynamism help enhance the potency of monetary policy to stabilize inflation in the long run. Despite decades of research, the consistent estimation of the Phillips curve remains one of the most challenging empirical tasks in macroeconomic studies due to pervasive endogeneity issues. In Chapter 2, I adapted the two-step approach proposed by Barnichon and Mesters (2020) in the estimation of the NKPC and imposed useful restrictions to improve point estimates. In the final chapter, I provide my justification by comparing the simulation performance of different estimators using monetary shocks as instruments. I show that the flexibility of the two-step approach to allow for extra controls in estimating the impulse responses gains its additional advantage over the other IV estimators and imposing range inequality constraints along with long-run restriction brings the point estimates closer to their true values when the rank condition fails. Nevertheless, the state-of-the-art method might be confronted with other pitfalls in the estimation of the Phillips curve, such as the many weak instruments problems and small sample bias, which deserve further research.