Monday, June 12, 2023

Policy evaluation of waste pricing programs using heterogeneous causal effect estimation

Using machine learning methods in a quasi-experimental setting, Marica Valente studies the heterogeneous effects of introducing waste prices – unit prices on household unsorted waste disposal – on waste demands and municipal costs. Using a unique panel of Italian municipalities with large variation in prices and observables, she shows that waste demands are nonlinear. The author finds evidence of constant elasticities at low prices, and increasing elasticities at high prices driven by income effects and waste habits before policy. The policy reduces waste management costs in all municipalities after three years of adoption, when prices cause significant reductions in total waste.
A growing number of municipalities have implemented Pay-As-You-Throw (PAYT) programs that require households to pay for each unit (per bag, can, or weight) of unsorted waste presented for collection.... Empirical estimates provide mixed evidence on the magnitude of PAYT average effect on unsorted waste as well as on possible indirect causal effects on recycling and total waste. Moreover, policy impacts may vary across municipalities depending, e.g., on the adopted price level and socio-economic characteristics.... In light of this, heterogeneity in causal effects plays an essential role in evaluating policy efficacy, allowing to ascertain subpopulations for which the policy is most beneficial and to generalize estimates to a new target population. On top of this, heterogeneous causal effect estimates can be combined with municipal waste management costs to deliver heterogeneous cost–benefit evaluations of waste policies.

In this paper, Valente examines heterogeneous demand responses to waste prices, and their impact on municipal waste management costs. The main challenge in the analysis is that determinants of waste generation and policy adoption are possibly many, and may confound the estimation of causal effects. This motivates the collection of a unique panel of municipalities with a large variation in prices and observables, and the estimation of municipal level causal effects of prices (continuous treatment) via machine learning methods....
The author estimates policy effects on municipal costs for each municipality by combining price causal effects on unsorted and recycling waste with their impacts on waste management costs. She finds that waste prices decrease municipal costs for all municipalities especially in the long-term (after three adoption years) when municipalities show the highest reductions of total waste.
Municipalities generate relatively more Recycling Waste (RW) than Unsorted Waste (UW) on average. Managing one unit of UW costs on average twice as much as RW. However, municipal unit costs are largely heterogeneous, and for some of these municipalities (40%) unit costs of RW are higher than those of UW. Table 1 also shows that average costs for UW are slightly higher in per capita terms. Per capita costs indicate that one individual on average spends about €100 per year for waste services, and pays the most for unsorted waste. This implies that UW reductions that do not translate into RW increases will potentially drive cost savings for both households and municipalities. Previous studies point to education and income levels as key determinants of policy adoption and effectiveness.... In her sample, PAYT is implemented in municipalities with on average slightly higher income and comparable education levels.
Using prices and UW post-policy, she calculates household variable costs in each municipality. In the third policy year, households in high-price municipalities (>13 cents) pay on average €176 per capita. In low-price municipalities (<3 cents), they pay on average €21 per capita. 
Policy effects on municipal costs are mostly positive. However, waste prices raise municipal costs in a fraction of municipalities where the reduction of unsorted waste is largely driven by an increase in recycling, and management costs of recycling are high compared to those of unsorted waste. In year 3, PAYT generates cost savings in all municipalities. Average savings are €20 per capita, which is about one fifth of what municipalities spend for waste management per person on average. As management unit costs are largely unaffected, savings come from UW reductions that do not translate into higher RW. In other words, municipal cost savings are largely driven by total waste reductions.
As higher prices promote increased recycling rather than waste avoidance, the largest cost savings occur in municipalities setting low prices after three years of adoption. ... Higher prices may be less desirable from a municipal cost perspective as they do not cause further total waste reductions, instead they rather reallocate waste to the recycling pile, and (ii) targeting total waste reductions in high-price municipalities could increase savings substantially.

The final implication of this paper may help local governments to overcome their initial concerns regarding the implementation of PAYT policies. This paper provides evidence that short-term costs may indeed increase due to higher recycling, but these costs are outrun by the long-term savings due to total waste reductions.

Compared to a counterfactual zero-price (no PAYT) scenario, UW reductions lead to a 24% decrease in municipal costs of waste management on average over the policy years. At the same time, increased RW raises municipal costs by on average 21%. In year 3, average per capita total cost savings amount to €24, however, there is large variation... Despite effect heterogeneity, waste prices cause cost savings for most municipalities.

by Marica Valente, University of Innsbruck, Austria
Journal of Environmental Economics and Management via Elsevier Science Direct
Volume 117; January, 2023; 102755

No comments:

Post a Comment