Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Assessing the Heterogeneous Impact of Economy-Wide Shocks: A Machine Learning Approach Applied to Colombian Firms

Marco Dueñas, Federico Nutarelli, Víctor Ortiz, Massimo Riccaboni, Francesco Serti

TL;DR

This study investigates the effectiveness of various Machine Learning techniques in predicting firms' trade and uses these predictions to reconstruct the counterfactual distribution of firms' trade under different COVID-19 scenarios and to study treatment effect heterogeneity.

Abstract

Our paper presents a methodology to study the heterogeneous effects of economy-wide shocks and applies it to the case of the impact of the COVID-19 crisis on exports. This methodology is applicable in scenarios where the pervasive nature of the shock hinders the identification of a control group unaffected by the shock, as well as the ex-ante definition of the intensity of the shock's exposure of each unit. In particular, our study investigates the effectiveness of various Machine Learning (ML) techniques in predicting firms' trade and, by building on recent developments in causal ML, uses these predictions to reconstruct the counterfactual distribution of firms' trade under different COVID-19 scenarios and to study treatment effect heterogeneity. Specifically, we focus on the probability of Colombian firms surviving in the export market under two different scenarios: a COVID-19 setting and a non-COVID-19 counterfactual situation. On average, we find that the COVID-19 shock decreased a firm's probability of surviving in the export market by about 20 percentage points in April 2020. We study the treatment effect heterogeneity by employing a classification analysis that compares the characteristics of the firms on the tails of the estimated distribution of the individual treatment effects.

Assessing the Heterogeneous Impact of Economy-Wide Shocks: A Machine Learning Approach Applied to Colombian Firms

TL;DR

This study investigates the effectiveness of various Machine Learning techniques in predicting firms' trade and uses these predictions to reconstruct the counterfactual distribution of firms' trade under different COVID-19 scenarios and to study treatment effect heterogeneity.

Abstract

Our paper presents a methodology to study the heterogeneous effects of economy-wide shocks and applies it to the case of the impact of the COVID-19 crisis on exports. This methodology is applicable in scenarios where the pervasive nature of the shock hinders the identification of a control group unaffected by the shock, as well as the ex-ante definition of the intensity of the shock's exposure of each unit. In particular, our study investigates the effectiveness of various Machine Learning (ML) techniques in predicting firms' trade and, by building on recent developments in causal ML, uses these predictions to reconstruct the counterfactual distribution of firms' trade under different COVID-19 scenarios and to study treatment effect heterogeneity. Specifically, we focus on the probability of Colombian firms surviving in the export market under two different scenarios: a COVID-19 setting and a non-COVID-19 counterfactual situation. On average, we find that the COVID-19 shock decreased a firm's probability of surviving in the export market by about 20 percentage points in April 2020. We study the treatment effect heterogeneity by employing a classification analysis that compares the characteristics of the firms on the tails of the estimated distribution of the individual treatment effects.

Paper Structure

This paper contains 15 sections, 10 equations, 13 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Methodological framework
  3. Data
  4. Control Variables
  5. Results
  6. Selection of the machine learning algorithm
  7. Evaluation of the COVID-19 effect
  8. Heterogeneity of the COVID-19 effect on Colombian exporters
  9. Estimations based on $Y-SUM$
  10. Concluding discussion
  11. Appendix - The Colombian economy amidst the COVID-19 crisis
  12. Appendix - Data
  13. Appendix - Descriptive Statistics
  14. Appendix - Statistical Significance of the $CADiff$
  15. Appendix - Comparison with chernozhukov2020generic

Figures (13)

  • Figure 1: Average Individual Treatment Effect, by months, comparing Logit-LASSO and RF. Standard errors were obtained with 100 bootstrap replications. Confidence intervals for a $5\%$ significance level.
  • Figure 2: The quarterly mean difference in the predicted probability of success (SAM vs. SUM) by industry, using the Logit-LASSO predictions. Standard errors were obtained with 100 bootstrap replications. Confidence intervals for a $5\%$ significance level.
  • Figure 3: Annual Sorted Partial Effects (SPE) and Average Partial Effects (APE) of COVID-19 on Colombian firm export's status. The treatment effect is calculated as a difference between SAM and SUM predictions. Standard errors were obtained with 100 bootstrap replications. Confidence intervals for a $5\%$ significance level.
  • Figure 4: Monthly Sorted Partial Effects (SPE) and Average Partial Effects (APE) of COVID-19 on Colombian firm export status. The treatment effect is calculated as a difference between SAM and SUM predictions. Standard errors were obtained with 100 bootstrap replications. Confidence intervals for a $5\%$ significance level.
  • Figure 5: Mean difference in the predicted probability of success (SAM vs. SUM / $Y$ vs. SUM) by month, using Logit-LASSO predictions and (SAM vs. SUM). Standard errors were obtained with 100 bootstrap replications. Confidence intervals for a $5\%$ significance level.
  • ...and 8 more figures