In this thesis, I investigate effective predictors for corporate defaults and measurement of economic recessions. I use corporate default probabilities in US industrial firms from April 1980 to December 2014 and effective predictors extracted from 92 macroeconomic and financial variables and I propose a framework to determine whether there is a dynamic in effective predictors for US corporate defaults. I apply LASSO, an advanced variable selection technique, and I find that there are no macro factors that can consistently explain default risk over time, suggesting that default risk has been dynamic during the 35-year period. These dynamics persist even during non-recession periods. Interestingly, the strong predictive powers of macro factors over shorter periods are related to monetary policy indicators and tools, which provides empirical support for prior theoretical theories that emphasize the unique role of monetary policies in corporate defaults. Another interesting finding is that institutional market funding, as the liquidity source provided by non-bank groups, has gradually impacted on corporate defaults since the 2001 recession. In the financial crisis occuring in 2007 this money market funds (MMFs) have strongest impact on corproate defaults. This implies that MMFs paly a crucial role of destabilizing credit markets in recent years, which is consistent with current studies of whether MMFs can result in financial stability. I study US economic recessions by introducing a cutting-edged technique from the Natural Sciences in order to capture the critical transitions in the macroeconomic system, using hundreds of macroeconomic and financial variables covering the period January 1980 to December 2014. Based on this method, I construct macro indicators to measure the interactions among these variables in order to capture the critical transitions in the macroeconomic system. Then I employ a standard logit model to study whether these proposed indicators offer a prediction one-month ahead of US economic recessions from September 1980 to December 2014. I find that the interactions among macroeconomic and financial variables measured by covariance among these variables can provide one-month ahead prediction for US economic recessions. In particular, the best predictive macro indicators are constructed by employing procyclical factors and factors from six economic groups based on results from both in-sample and out-of-sample analysis. Regarding the standard ideal indicator defined by Shiskin and Moore (1967), the indicator constructed by procyclical factors is preferable over that constructed by the factors from six economic groups since the former is smoother. I also find that the threshold based on 25% for classifying the recessions can generate better estimation results than using 50%, consistent with prior studies. In forecasting economic recessions in the US, the indicator generated by using broad macro factors is able to provide predictive power. The implication of these results is to provide a new quantitative approach for central bankers and policy makers to predict economic recessions one month ahead.