Shift-Share Instrumental Variables (SSIV) are used to address endogeneity and selection challenges in many economic settings. This workshop will last two evenings and will introduce the basics of SSIV and cover the recent literature on its econometric foundations. Special focus will be paid on the different assumptions underlying the “exogenous shares” and “exogenous shocks” approaches to SSIV identification, and their practical implications. We will also cover a more general class of instrumental variable strategies combining exogenous shocks and non-random exposure. Group programming exercises will be used to illustrate various theoretical concepts in real-world applications.

This is one of our advanced courses. These courses are designed assuming a solid foundation in the basics of instrumental variables and will cover the frontiers of the topic. A good review is the intro course: https://github.com/Mixtape-Sessions/Instrumental-Variables.

This course provides an in-depth introduction to panel data approaches to causal inference. The first part of the course reviews how new "heterogeneity-robust" estimation strategies address some important limitations of traditional two-way fixed effects regressions in difference-in-differences applications, and then provides an in-depth discussion/comparison of many of these approaches. This part also includes a number of practical extensions such as how to include covariates in the parallel trends assumption and dealing with "bad controls". The second part of the course discusses how the insights of recent work on difference-in-differences can apply in a number of other settings that frequently arise in empirical work. And, in particular, this part of the course provides connections between the difference-in-differences literature and alternative identification strategies (conditioning on lagged outcomes, change-in-changes, and interactive fixed effects models) and also how to deal with more complicated treatment regimes (continuous treatments or treatments that can change value over time).

This is one of our advanced courses. These courses are designed assuming a solid foundation in the basics of the difference-in-differences methodology and will cover the frontiers of the topic. A good review is: https://github.com/Mixtape-Sessions/Causal-Inference-2.

The holy grail of causal inference is the individual-level treatment effect: how would a particular patient respond to a drug? Which users will respond most to a targeted ad? Would a given student be helped or harmed by a classroom intervention? This session introduces machine learning tools for estimating heterogeneous treatment effects like random causal forests. The course goes over the theory and concepts as well as the nitty-gritty of coding the methods up in python, R, and Stata using real-world examples. This course can be taken as a follow-up to the Machine Learning and Causal Inference mixtape session, or as a stand-alone course.

This is one of our advanced courses. These courses are designed assuming a solid foundation in the basics of machine learning and causal inference and will cover the frontiers of the topic. A good review is the intro course: https://github.com/Mixtape-Sessions/Machine-Learning.

This three-day workshop covers a wide range of practical results for regression and IV-based analyses of causal effects which leverage random or conditionally as-good-as-random shocks. Questions of particular focus include:

• "What controls do I need to include to avoid omitted variables bias?"
• "Do I need to worry about 'negative weighting' of heterogeneous effects?"
• "How should I be clustering my standard errors?"
• "What's the payoff to considering nonlinear/'structural' analyses?"

Results will be illustrated through several real-world applications.

This is one of our advanced courses. These courses are designed assuming a solid foundation in the basics of causal inference and will cover advanced methods. A solid understanding of the material covered in the material from Scott's courses (Part 1 and Part 2) will be assumed.

This one-day workshop will cover advanced topics from the recent difference-in-differences literature. One question of particular focus will be, "what should I do if I'm not 100% sure about the validity of the parallel trends assumption?" We will cover a variety of relaxations to the parallel trends assumption, and new tools for power calculations and sensitivity analysis. The workshop will focus not just on the theory, but also on practical implementation in statistical software such as R and Stata.

This is one of our advanced courses. These courses are designed assuming a solid foundation in the basics of the difference-in-differences methodology and will cover the frontiers of the topic. A good review is: https://github.com/Mixtape-Sessions/Causal-Inference-2.