PAI 777 -- Spring 2016
Economics of Environmental Policy
Lecture # 16 -- Estimating Costs
I. Estimating Costs
Establishing the baseline
We want to compare costs
Even without regulation, we expect some things to change over time.
Note that costs will often be focused on a few individuals (e.g. affected firms or communities).
Benefits are more likely to affect a wider range of people.
Thus, equity concerns will be an issue.
Types of costs
Direct costs – purchases of tangible units such as equipment, labor, and land
Opportunity costs – the value of the best forgone opportunity.
It is what we give up by using a resource for this use, rather than the next best alternative use.
Example: an opportunity cost of going to school is foregone salary.
Important to distinguish between costs and transfers.
Environmental costs – because most regulations focus on a single pollutant, regulating one pollutant may increase the use of another pollutant.
Example: using scrubbers to clean SO2 emissions leaves behind a sludge that must be disposed of.
Abatement cost data
Until 1994, the US did an annual survey of pollution abatement expenditures by industry
The US spent $121.8 billion on Pollution Abatement and Control Expenditures (PACE) in 1994
About 1.7% of GDP
Within other OECD countries, typical numbers range from 0.5% - 2% of GDP
Changes in output
If regulations lead to changes in output, we need to adjust cost estimates
We also need to account for lost welfare from lower output
Note that, even if costs are passed on to consumers, they are still costs.
Also note that some costs domaynot occur because of change in quantity.
Figure 8.1 in the text provides an example
II. Cost of Job Losses
Do conventional cost-benefit analyses underestimate employment effects?
Typically assume full employment
That is, workers displaced in one sector will be re-employed elsewhere.
The Department of Labor surveys firms for the reasons for mass layoffs.
Only 1% attributed to any type of government regulation.
What are the net effects of new environmental rules on employment?
Job losses in one location may be offset by new jobs elsewhere
In addition, new equipment must be produced to reduce emissions.
This may create jobs
Morgenstern et al. (2002)
Studied effect of regulation in pulp & paper, plastics, petroleum, and steel
Overall effect of added pollution abatement expenditures on employment insignificant
How does non-attainment status of a county affect regulation?
Only non-attainment status for carbon monoxide has a negative effect on employment
Also studied ozone, PM, and SO2
Kahn and Mansur (2013)
Compared employment at county level for adjoining counties with different attainment status
Using neighbors helps control for other factors likely to affect employment
NA status does lead to job losses in specific industries that are intensive in electricity, labor, and pollution
Examples include petroleum products, paper, primary metals, and transportation equipment
Effect is equivalent to job losses that would result from a 33% increase in electricity prices in attainment counties
Also uses NA status
Uses plant-level data: is a plant in a NA county?
Analyzes effects using sectoral employment by county, sector, and year
Finds employment falls by 10% below 1990 levels after a change in NA status
This is 15% less than counterfactual trends
Summary: note that more disaggregated studies more likely to find employment effects of regulation
But, is on/off measure of regulatory status sufficient? (Sheriff, Ferris, and Shadbegian 2015)?
Sheriff, Ferris, and Shadbegian note that regulations interact
Attainment counties in the northeast still face regulation from the Ozone Transport Region rules of the CAA
In addition, even if people are re-hired, there may be adjustment costs that matter.
Resource re-allocations that arise from unanticipated shocks
E.g. new equipment purchased to comply with regulation
Layoff that leads to a spell of unemployment and lost earnings for a worker
Measuring adjustment costs of job losses
Easier to measure individual components of job losses than the adjustment costs themselves
Components of adjustment costs:
Job losses from employer initiated: establishment closes or downsizes
US layoff rate between 1990-20111 averaged 7% per quarter
But, these are desired adjustments in the economy as resources are reallocated to more productive uses
Adjustment costs related to these layoffs depends on the state of the economy: how hard is it to be re-employed
Lost earnings both between jobs and after obtaining new job
Depends on duration of employment spells
Davis and von Wachter (2011): when unemployment rises from 6 to 8 percent, adjustment costs rise from 1.4 years of predisplacement earnings to 2.8 years.
Walker (20123) finds higher earning losses for workers displaced by plant-specific regulatory changes under the Clean Air Act
After nine years, average present discounted value of lost earnings was about 20% of preregulatory earnings
Social costs of job losses
Lost earnings are not welfare losses.
What else matters?
Does a person need to relocate?
Are they able to relocate, or must they look for a new job in the same location?
Economics typically considers these choices separately.
If combined into a single choice, the margins of adjustment expand.
How can these be measured?
Consider a job where risk of unemployment is known
What wage premium do workers require to take that job?
That is, wages vary depending on job security
Kuminoff et al. (2015) consider the costs of relocation
Estimate that earnings losses are about 75% of the welfare losses from unemployment
However, relocation may allow workers to find a higher paying job
What we currently know about job losses from environmental regulation comes from studies that focus on net effects.
As these studies become more disaggregated, the effect of regulation becomes stronger
It is possible to calculate adjustment costs to job losses by studying how an individual’s employment history and wages change
To study the whole economy, need to develop models that incorporate these adjustment costs
Current models simply show results from pre-and post-regulatory equilibria.
Smith: “focus less on the “origin” and “destination” of each “trip” and instead pay more attention to the path that evolves along the way.” (p. 177)
III. How Accurate are Cost Estimates?
Sources of error
One difficulty is that errors can come in many ways.
The most obvious is incorrectly estimating the costs of control.
However, even if the cost of control is estimated correctly, predictions about emissions levels, number of plants, etc. can also be wrong.
See examples on pages 303-304 of Harrington
Evaluation of estimates
. study 28 estimates of the cost of regulation.
They compare pre-regulation estimates to actual costs after the regulation is in place.
Label an estimate as “accurate” if it within 25% higher or lower than the actual costs.
Unable to Determine
Unit Pollution Reduction Cost
Costs more likely to be overestimated
Discussion of results:
The three underestimates were for rather “minor” regulations: EPA aldicarb and CDEC bans and OSHA’s powdered platform regulation.
EPA and OSHA tend to overestimate reductions, but not per unit costs. As a result, total costs are overestimated.
State and foreign agencies were more likely to overestimate per unit costs.
For market-based polices, seven of the eight estimates overestimated costs!
Note that this is where we would expect technological innovation to be most important.
Why do errors occur?
Many estimates ignore the possibility of technological innovation.
Regulators have an obligation to identify a means of complying with the regulation, which usually means considering current technologies.
Future technologies are much harder to predict.
Quantity errors: misestimating baseline emissions
However, keep in mind that this also means that benefits are wrong.
For example, overestimating emissions reductions overestimates costs, but also overstates the benefits.
Regulations may change during the public comment period.
Thus, cost estimates aren’t for the final regulation.
Estimates may focus on maximum values, rather than means.
Especially if rely on industry for data.
Asymmetric correction of errors.
Firms are likely to bring underestimates to the attention of regulators. There is no similar group with strong incentives to bring overestimates to the attention of regulators.
IV. Introduction to Benefit-Cost Analysis
: Maximize total net benefits (= total benefits - total costs).
Benefit-cost analysis calculates the costs and benefits of a project and finds the total net benefits.
Note that some costs and benefits can be observed directly from market data. Others will need to be inferred from data.
Steps to benefit-cost analysis
Specify clearly the project or program.
For environmental economics, this is usually a physical project such as a dam or wastewater treatment plant, or a regulatory program, such as pollution control standards.
Determine quantitatively the inputs and outputs of the program.
Can be difficult – for example, general equilibrium effects.
Also, it is important to distinguish between
of resources due to substitution and the creation of new resources.
For example, jobs created by a project should normally not be included as a benefit.
Jobs created are a transfer of resources. If the project wasn't done, the workers could have been used elsewhere.
Estimate the social costs and benefits of these inputs and outputs.
Compare these costs and benefits.
Here, one can also include other considerations, such as equity.
For students who would like a review of the nuts and bolts of benefit-cost analysis, you can view my lecture notes from the topic in my policy analysis class at:
. The last set of lectures cover benefit-cost analysis.
V. Dealing with Uncertainty
The first step in dealing with uncertainty is risk assessment.
Risk has two components:
– depends on chance
– depends on circumstances (e.g. a smoker is more likely to get cancer)
In addition, assessing risk involves two concerns:
the probability of an event occurring
how serious the event will be
First, we focus on finding the probability of an event occurring.
Risk can be determined by looking at past records.
However, it is important to be aware of changes that occur over time. For example, increased safety features reduce the risk of death from auto accidents. This is a change in
Component analysis is often used to assess the risks of new technologies.
Problem: components may be related.
Risk by analogy
Often, time lags make perceiving risk difficult.
For example, cancer may be caused after exposure to a toxin, but only after many years.
As a result, studies on animals are often used to extrapolate human risks.
Animals are exposed to unrealistically high doses of toxins in the laboratory.
Need to extrapolate risk of humans from low exposure from calculated risk based on high exposure.
Physiology of animals and humans may be different.
The risk may be different for different people.