Applied Economics Research

Michela Faccioli (university of Trento, Italy) and Klaus Moeltner (VT)

We developed a new statistical framework to estimate societal benefits (in dollars) of environmental improvements using data from a common / popular type of survey approach. Our method avoids having to choose any functional relationships or statistical distributions. It is computationally fast and simple to implement with publicly accessible statistical software.

People’s willingness-to-pay (WTP) for environmental improvements is often captured through referendum-style survey questions, asking them to decide between the status quo and improved conditions, at additional cost. Traditionally, this “Contingent Valuation” data is estimated via structural  parametric models, at high risk of mis-specification. Nonparametric alternatives are available, but do not allow for individual-level WTP estimates. We developed a “best of both worlds" estimator that combines Random Forest with traditional distribution-free welfare estimators to derive individual-level choice probabilities and WTP estimates in a fully nonparametric fashion.

In simulations, we show that our model is robust to functional misspecifications, and generates accurate predictions of WTP. For an actual application to biodiversity enhancements in the UK we show that our framework produces reasonable WTP predictions with tight confidence intervals. It also highlights the difference across households in how they value these improvements. Given its robustness to mis-specification error and its user-friendliness, our method has the potential to become widely adopted within the environmental valuation community.

Robert J. Johnston (Clark U.) and Klaus Moeltner (VT)

We show how Machine Learning tools called Random Forests and Local Linear Forests can be a more accurate and efficient alternative to standard estimation techniques in analyzing meta-regression models (MRMs) to predict societal benefits related to enhanced environmental conditions.

The general framework of using existing meta-data to predict economic values is referred to as Benefit Transfer (BT). BT via MRM is the most common approach within federal agencies to determine the societal value (in dollars) of changes in environmental conditions and services. BT estimates are critical inputs to Benefit-Cost Analyses for environmental policies and rulemaking. The stakes for BT to generate accurate estimates are high, given the large amounts of both taxpayer dollars and industry profits affected by large-scale environmental policies.

Our approach improves model fit and reduces confidence intervals for out-of-sample predictions by an order of magnitude compared to existing methods.  The steep reduction in uncertainty for environmental benefits sharpens agencies’ decision tools and rules, reducing the risk of sub-optimal policy designs. We presented the new approach to the U.S. Environmental Protection Agency (EPA), who is already considering adopting it for future rulemaking.

Chunbei Wang and Le Wang

We examined the factors behind the surge in self-employment among Hispanic immigrants over the past two decades, focusing on the role of undocumented immigrants.

Although the surge in self-employment among Hispanic immigrants is often celebrated, there is limited understanding of the drivers behind this trend and its policy implications. Our paper addresses this gap by exploring the underlying causes.

Policymakers need to understand the true nature of Hispanic business growth in order to allocate resources effectively and address vulnerabilities in the informal economy. Additionally, this research underscores the importance of studying undocumented entrepreneurs to better assess their performance and growth potential.