In this dissertation, I use novel methodologies that incorporate machine learning into causal policy evaluation such as double robust machine learning to study some key issues in Productivity and Finance. In the first chapter, I evaluate the impacts of European public subsidies on innovation. I use double machine learning with deep neural networks to explore the effects of public subsidies on firms’ R&D input and output. I find that public subsidies increase both R&D intensity and R&D output and these results remain economically and statistically significant even after accounting for treatment endogeneity. In the second chapter, I evaluate the effects of public subsidies and collaboration agreements on innovation output. Many public schemes related to R&D have pushed towards collaborative agreements between firms/organizations and this chapter studies whether subsidies not promoting collaboration perform as well in terms of stimulating R&D output. Results show that subsidized noncollaborative firms would have gained in terms of R&D output had they collaborated. I also find that collaboration alone seems to generate significantly higher (double) R&D output compared to subsidies alone. In the third chapter, I analyze the impacts of offering non-core and non-financial ("plus") services in addition to core financial services on Microfinance Institutions' (MFIs) performance using a double machine learning model with random forests. The results indicate no differences in the performance of MFIs offering core financial and microfinance plus services, however, MFIs that offer non-core financial services together with non-financial services are serving less poor clients, suggesting a rather surprising "mission drift". In the fourth chapter, I analyze the impacts of regulation on MFIs' performance. I provide evidence of the impact of regulation on the double bottom line of the microfinance industry using double machine learning with neural networks. Results show that regulation does not affect financial results but affects the outreach of savings-and-loan MFIs. Regulation increases the depth of outreach of this group, indicating fewer poor clients, and suggesting a mission drift. In the fifth chapter, I investigate the link between the term structure of sovereign credit default swaps and the market efficiency of carry trades. I use Kneip et al. (2012) factor model to deal with large dimensions and unknown forms of unobservable heterogeneous effects. I document a divergent pattern of carry trade risk for developed and developing countries. In the sixth chapter, I use recurrent neural networks and feed forward deep networks, to predict NYSE, NASDAQ and AMEX stock prices from historical data. I experiment with different architectures and compare data normalization techniques. Then, I leverage those findings to question the efficient-market hypothesis through a formal statistical test and I find evidence of an inefficient stock market. Each of these studies requires the implementation of new methods of estimation and inference that have not been utilized to examine these important economic policy issues. My research points to many advantages of the approaches that I introduce in my dissertation. Robustness of inferences is a crucial dimension to acceptable policy recommendations and my development of semi/nonparametric estimators and their applications to crucial evaluations of public policy and regulatory oversight provides evidence that they are well-motivated theoretically, that they can be feasibly implemented in empirical applications, and they are in many cases, a dominant strategy in regard to model specification and estimation.