2018 Conference - Session 4

2018 Conference - Session 4

SESSION 4 | Thursday, March 15 | 3:45-5:15PM

A. 4: Teaching BCA

Chair: Scott Farrow, University of Maryland, Baltimore County

This round table begins with a summary of each panel member's short teaching module in the edited volume titled "Teaching Benefit-Cost Analysis".   In addition, panel members are encouraged to  address higher level issues in teaching BCA such as: what are the core elements we want undergraduates to know or be able to do?  What about graduate students?  What are the hardest topics to teach?  Should we be only pro-BCA or have sections on critiques or where violation of assumptions causes BCA to break-down?  Is there a role for international examples and contexts in any given institutional setting? Questions will also be taken from the audience.


Susan Dudley, The George Washington University

David Weimer, University of Wisconsin, Madison

Stuart Shapiro, Rutgers University

Arnold Harberger, UCLA

Chiara Pancotti, Center for Industrial Studies (CSIL)

B. 4: Retrospective Costs and Benefits of Food Safety Regulations

Chair: Aliya Sassi, U. S. Food and Drug Administration

Recent developments in regulatory benefit-cost analysis reaffirm the need for retrospective analyses of existing rules and regulations. This panel provides a sampling of how United States regulatory agencies and scholars conduct retrospective analyses of costs and benefits of food safety and nutrition regulations. The aim is to discuss in detail the existing approaches, challenges faced, and methods used to overcome these challenges. In four presentations we 1) provide a retrospective review of the U. S. Food and Drug Administration's (FDA's) 2003 trans fat regulation and explore the factors that may have influenced producers to choose to reformulate their food products rather than relabel them as containing unhealthy ingredients; 2) re-examine the benefits of the United States Department of Agriculture's (USDA's) proposed and final rules aimed at reducing the incidence of Listeria monocytogenes in ready-to-eat meat and poultry products;  3) re-examine many aspects of the original regulatory impact analysis of the USDA's final Listeria monocytogenes rule and affirm that the rule has been cost-effective without imposing a disproportionate cost burden on small and very small establishments; and 4) retrospectively review the FDA's Prior Notice regulations for imported foods and discuss the impacts those new food security measures had on the food importation process and society.

Discussant: Elizabeth Ashley, OIRA/OMB


  1. Consumer and Producer Responses to the 2003 Trans Fat Labeling Regulation: A Retrospective View; Amber Jessup, U. S. Department of Health and Human Services (HHS/ASPE)
  1. Retrospective Benefit-Cost Analysis of the Regulation on “Control of Listeria monocytogenes in Ready-to-Eat Meat and Poultry Products; Flora Tsui, U. S. Department of Agriculture, Food Safety and Inspection Service
  1. Retrospective Analysis of the United States Food and Drug Administration’s Prior Notice Regulations for Imported Foods; Aliya Sassi, U. S. Food and Drug Administration

C. 4: Hedonic Price Studies and BCA

Chair: Dennis Guignet, U. S. Environmental Protection Agency

Discussant: Ben Witherell, New Jersey Department of Environmental Protection


  1. A Practical Guide to the Use of Hedonic Studies in Benefit-Cost Analysis; Alastair McFarlane, U. S. Department of Housing and Urban Development; Michael Hollar, U. S. Department of Housing and Urban Development

(1) The authors attempt to provide some guidelines concerning adapting hedonic estimates to benefit-cost analysis. (2) We review best practices and warn against common mistakes. (3) The research provides needed assistance in interpreting measures of willingness-to-pay taken from the real estate market. (4) We find that the ease in adapting results from other studies depends to a large extent on information concerning the cost of a regulatory action.

  1. The Impact of Air Pollution and Noise on Property Prices Over Time; Henrik Andersson*, Toulouse School of Economics; Peter Molnár, Department of Public Health and Community Medicine, University of Gothenburg, Sweden; Jan-Erik Swärdh, Department of Transport Economics, VTI, Sweden; Mikael Ögren, Sahlgrenska University Hospital, University of Gothenburg, Sweden

Revealed preference (RP) methods are preferred by many economists to stated preference methods to value non-market goods, since the former involves actual decision compared to the hypothetical decisions with the latter approach. However, one issue when using market data is that many of the variables may be correlated and/or confounded. This means that it may either be difficult, or even impossible, to examine how different non-market attributes influence individual's behavior, and hence they cannot be monetized with RP methods with good precision. In this study we examine the effect on property prices from air pollution and road noise using the hedonic regression technique (HP). There already exist a relatively large literature of HP studies on air pollution and transportation noise. One concern in these studies is that the estimated effect on the property prices from the environmental attribute, e. g. air pollution, also captures the effect from other attributes, e. g. noise (i. e. omitted variable bias). Hence, the concern is that the market data in the case of a high correlation between the air pollution and noise levels is not suitable for eliciting individuals' willingness to pay (WTP) for these attributes. The aim of this study is to examine the potential magnitude of this potential bias by running hedonic regression models on data from Gothenburg, Sweden. We have access to a very rich data set, that not only spans over a relatively large area, but also over a long time-period. The objectives are to estimate the effect on the property price from air pollution, in our case NOx, and road, and to estimate the effect if one of them is left out of the regression. Hence, we will obtain an empirical estimate of the potential size of the omitted variable bias in studies only examining the effect from one of the attributes. Our preliminary findings suggest that whereas the level of road noise has a statistically significant negative effect on the property price, we do not find such an effect from the level of air pollution. Regarding the size of the omitted variable bias our preliminary findings suggest that it is non-negligible, but relative small to other uncertainties in estimating HP models. The findings of the study is both of high research and policy relevance.

  1. Property Values and Water Quality: A Nationwide Meta-Analysis and the Implications for Benefit-Transfer; Dennis Guignet*, U. S. Environmental Protection Agency; Matthew Heberling; Michael Papenfus; Olivia Griot, Ben Holland

There is a well-established literature examining how water quality impacts home values, dating back over five decades. Despite this, hedonic property value studies have yet to be used in regulatory analyses of regional and nationwide water quality regulations enacted by the US Environmental Protection Agency. Heterogeneity in local housing markets, the types of waterbodies examined, and the water quality metrics used, are key reasons why the results of these local studies have not been applied to broader policies. Our objective is to synthesize this vast literature, and estimate unit values and value-transfer functions that can be used to assess broader water quality policies. We identify 38 studies in the published and grey literature that examine how home values vary with water quality. Our study is the first meta-analysis to aggregate this literature and systematically calculate comparable within- and cross-study elasticity estimates by accounting for differences in functional forms, assumed price-distance gradients, and baseline conditions. We convert the primary study coefficient estimates to common elasticity and semi-elasticity measures for both waterfront and near-waterfront homes, and then use Monte Carlo simulations to estimate the corresponding standard errors. Each study can yield numerous meta-observations due to multiple study areas, water quality metrics, and model specifications. Our preliminary meta-dataset contains n=686 unique observations, and for 524 of these observations sufficient information was reported to calculate the corresponding elasticity and semi-elasticity estimates. We find considerable heterogeneity in the meta-dataset in terms of the water quality metrics utilized, the type of waterbody studied, and the region of the US. The majority of estimated elasticities are with respect to water clarity (n=261). Among these studies, preliminary analysis suggests a mean price elasticity with respect to a one-percent increase in secchi depth of 0. 12% (median of 0. 04%), with a range from -0. 65% to 1. 72%. Other common metrics that are examined include dissolved oxygen, fecal coliform, E. coli, pH, chlorophyll, nitrogen, phosphorous, and total suspended solids. We find that the majority of estimates correspond to freshwater lakes (291), followed by estuaries (280), rivers (86) and small rivers and streams (29). Finally, we find reasonable spatial coverage, with studies covering 20 different states. The one exception, however, is the western US, where there are only four studies (providing just 75 meta-observations). A key contribution of this study is in highlighting key gaps in the literature regarding the types of waterbodies and regions covered, and the gap between the water quality metrics used by economists versus those currently examined by water quality modelers. We have just completed construction of the meta-dataset, and the next steps are to bin the meta-observations according to common water quality metrics, and estimate meta-analytic Random Effect Size (RES) means, RES meta-regressions, and Random Effects Panel models. We will then statistically test the appropriateness of pooling estimates from different regions and types of waterbodies, and discuss the implications for benefit-transfer.

D. 4: Applications of BCA Methods to Social Policies

Chair: Daniel Acland, University of California-Berkeley

Discussant: Lynn Karoly, RAND Corporation


  1. Conducting an Individual-Level Benefit-Cost Analysis of Family Self Sufficiency Programs; David Long, Princeton Policy Associates

I completed a benefit-cost analysis of a three-arm random control trial of FSS programs in New York City for MDRC earlier this year. I am giving a paper on the analysis next month at APPAM, where my focus will be on the substantive results of the analysis that is, the policy implications for HUD and local housing authorities operating FSS programs (the chair of the session is from HUD and the other papers address FSS programs). I propose to prepare a methodology-focused paper for SBCA. In this analysis, pertinent outcomes were monetized for individual sample members (2 treatment groups, 1 low-service control group) before they were used in impact estimation equations. Thus, costs, benefits, and net present value results in this study (from participant, government, and society perspectives) were all regression-adjusted impact estimates. This allowed significance tests to be conducted, confidence intervals to be calculated, and a subgroup analysis to be conducted.

  1. Does the Revocation of the Deferred Action for Childhood Arrivals (DACA) Survive an Accurate Cost and Benefit; Carolina Arlota, The University of Oklahoma, College of Law

In September 5th, 2017, Attorney General Jeff Sessions announced the rescission of the Deferred Action for Childhood Arrivals program “ DACA, hereinafter. DACA was created in 2012, and has been evaluated as an overall success. Its revocation has been criticized by members of both political parties. This article argues that the Administration's revocation does not pass the test of cost and benefit analysis (CBA) based on three accounts. First, from an administrative standpoint, the Executive determination is illegal under the Administrative Procedure Act of 1946. According to this Act, before creating, modifying, or repealing substantive rules, agencies must start a procedure known as "œnotice and comment." This procedure determines that agencies shall provide at least thirty days notice of its planned action and hear comments by affected/interested parties. The rationale is to assure that the agency's action is not arbitrary nor capricious. To the extent that neither notice or comment occurred, DACA's rescission will potentially increase litigation arising merely of procedural (and totally avoidable) claims. Importantly, until today, the Department of Homeland and Security was not capable of showing the benefits of its rescinding action. Therefore, this research discusses the costs of the current presidential immigration policy in light of this missing evidence and the violation of transparency requirements which have been obeyed previously by all administrations. Second, this article contends that DACA's revocation is inefficient if considered in substantive terms, because the children who were brought to the U. S. (and assuming they fulfilled the requirements of the program) no longer will be authorized to pursue an education “ high school and university included. Work and travel authorizations will be non existent once the two-years conditional approval expires. Consequently, instead of participating in the economy, paying taxes and tuition, and contributing to social programs, such individuals will become marginalized of a significant part of the economy today and, arguably, in the upcoming decades. In this scenario, DACA's revocation is likely to provide greater incentives for strategic behavior, such as marriage by convenience with U. S. citizens, for instance. The rescission has already escalated litigation between the federal government and the so-called "œsanctuary cities" and "œsafe heaven" states. Third, the analysis of the beneficiaries of DACA,who were brought to the U. S. in such an early age,is addressed under a CBA theoretical approach. Hence, this research contends that CBA shall not be blind-folded to moral considerations. In light of the three accounts above, the analysis presented in this article is unique, and focused on a current controversial topic. This framework contributes to the literature on CBA, because it addresses a contemporary example of a public policy enacted without cost considerations. It also advances a trending topic on CBA, namely, the inclusion ofmoral dimensions to costs considerations. The article concludes that DACA's revocation does not survive an accurate cost and benefit analysis.

  1. What Works to Increase Student Attendance Around the World? Recent Results from Comparative Cost-Effectiveness Analysis; Samantha Carter*, J-PAL Global; Radhika Bhula

The Abdul Latif Jameel Poverty Action Lab (J-PAL) recently released a major review of 58 randomized evaluations that tested programs designed to increase school enrollment and attendance, including a cost-effectiveness analysis comparing sixteen different programs in nine countries. This presentation will highlight the results from this recent publication, focusing specifically on the results of the cost-effectiveness analysis. The most cost-effective programs to improve student enrollment and attendance included in the analysis were those that addressed health problems and reduced the distance to school by creating low-cost schools in areas where few schools existed. On average, reducing the cost of school and providing other incentives were not as cost-effective as other approaches. Around the world, education is correlated with a range of positive outcomes including higher incomes, better health, and more active participation in politics and civil society. Encouragingly, there has been a dramatic rise in the number of children enrolled in school in recent decades, with 91 percent of primary school age children enrolled in school in 2015. Despite these gains, however, pockets of low enrollment remain. As of 2015, 61 million children of primary school age were out of school, along with an additional 202 million adolescents of secondary school age. In addition, millions of children who are enrolled in school do not attend regularly; in India, for instance, 29 percent of enrolled students were absent during unannounced visits to schools in 2016. Evidence from a substantial body of randomized evaluations suggests that the investments that parents and students make in education is sensitive to the costs and perceived benefits of schooling. And, while this literature suggests that certain approaches may be effective at increasing schooling, policymakers seeking to improve student attendance face resource constraints. The Abdul Latif Jameel Poverty Action Lab (J-PAL), a network of more than 140 affiliated professors who conduct randomized evaluations to test and improve the effectiveness of programs and policies aimed at reducing poverty, carries out comparative cost-effectiveness analyses to inform policy decisions. Comparative cost-effectiveness analysis can highlight the types of programs that tend to be most cost-effective, and can therefore serve as a useful starting point in the decision-making process.

E. 4: Investing in Infrastructure

Chair: Frits Bos, CPB Netherlands Bureau for Economic Policy Analysis

Discussant: Rick Geddes, Cornell University


  1. A Web-Based Land Use Benefit Cost Optimization Tool; Ken Acks, The Cost Benefit Gr, LLC, Env Valuation & Cost Benefit News

Currently, zoning and political power dictate land uses and density. This often results in suboptimal outcomes. Negative effects possibly include high housing costs arising from reduced supply and burdensome expenses, lower mobility, inequality, delays, inferior resilience to natural disasters, sprawl, pollution, and unjust dispersion of negative externalities. Zoning regulations can play a major role in where and how we live and work, and in the strength of economies. They can determine the size of our homes, what they look like, and where they can be located. Current land use regulations often are too static and difficult to change - many industrial and agricultural parcels are better suited to be improved with more dense residential structures, but can only be developed to their "œhighest and best use" with great difficulty. Most parcels within a particular zone may be well suited to the restriction placed upon uses in a zone, but some parcels may not be and thus remain vacant or used suboptimally. Zoning maps often force the same restrictions upon a large number of contiguous parcels with differing comparative advantages. Liberals, notably Jason Furman (2015), Orzag and Furman (2015) and Joseph Stiglitz (2015) and conservatives, including the Cato Institute, David Brooks (2017) and Edward Glaeser (2002, 2006 . . . ) alike have criticized these regulations. This paper will discuss the magnitudes of the welfare costs generated by flaws in current land use regulations, and then present a web-based model designed to begin providing superior alternatives by utilizing the tools of benefit-cost analysis. Advances in BCA and a plethora of new geo-locational data sources facilitated model development. The model compares the benefits and costs of eight alternative uses in four medium density urban/suburban locations. Four 60,000 square foot (SF) sites with 30,000 SF building footprints are analyzed for development with (1) 24 2-story 2,500 SF single family homes, (2) 20 3-story 4,500 SF 2-family homes, (3) a 10-story 250 unit apartment building containing 300,000 SF, (4) a 1-story neighborhood retail shopping center containing 30,000 SF, (5) 10-story 300,000 SF office building (6) a 10-story 300,000 SF mixed-use retail/office/apartment building, (7) a 1-story industrial building containing 30,000 SF and (8) a 50,000 SF park with several recreational options. Benefits are represented by estimated rents, revenues, consumer surpluses and shadow benefits generated. Costs include construction and other development costs and negative externalities. Benefits and costs are affected by neighboring uses and the proposed project. Environmental impacts of land use include habitat loss, reduction in biodiversity, flooding and reduced water quality from proliferation of impervious surfaces, air pollution from increased driving, and congestion. Health effects arising from increased driving attendant to sprawl include respiratory diseases and cancers, traffic fatalities, and obesity which increases disease risks. The algorithm maximizes the total value derived from a parcel, which primarily accrues to the owners and users (renters) (private value), but also is a function of how it influences surrounding parcels (social value). We find existing processes often don't maximize value.

  1. An Application of the Delphi Method to Benefit-Cost Analysis in Côte d’Ivoire; Timothy Breithbarth, Green Climate Fund; Djan Fanny, CNPC; Mamadou Yoda, CNPC, Government of Côte d’Ivoire; Alban Ahoure

In the field of international development, economists are often asked to estimate the benefits and costs of projects with limited data due to time, resource or capacity constraints. As a result, economists routinely search for low-cost methods of collecting or approximating data to inform their benefit-cost analysis. This presentation describes the Delphi Method used to obtain a consensus opinion on the economic value of government-owned land from a diverse group of experts in Côte d'Ivoire. During the development of the Cote d'Ivoire Compact with MCC, the Government of Côte d'Ivoire proposed a bridge crossing the Biétry bay in Abidjan. After an initial analysis deemed the traffic would be too low to justify the cost, an alternative of partially filling the lagoon to create land to support a road and additional lands for economic activities in the port area. The benefit-cost model needed estimates of the market value of this land, as rental rates for similar land was regulated by the Port of Abidjan. Due to a lack of time and available data, economists designed a workshop using the Delphi Method to obtain an estimate of land value from a group of experts. Through a three-hour workshop, experts were able to generate three increasingly precise and informed estimates of the land value per square meter, which could then be fed into the benefit-cost model. This experience showed the importance of timing, careful selection of participants, the brevity of questionnaires and the precise wording of questions to obtain a quality estimate from experts. Given its low cost, this method could be replicated in similar developing country contexts where data is scarce and time or resources are limited.

  1. Cost-Benefit Analysis and EU Cohesion Policy: Economic versus Financial Returns in Investment Project Appraisal; Massimo Florio, University of Milan; Valentina Morretta,   Università degli Studi di Milano, Department of Economics, Management and Quantitative Methods; Witold Willak, European Commission, DG Regional and Urban Policy

This paper investigates the role of Cost-Benefit Analysis (CBA) upon the evaluation of major infrastructure projects, in the context of the European Union Cohesion Policy. After presenting an overview of the changing landscape of the Cohesion Policy and the consequent development of the CBA practice over the last years, we use a dataset of around 1,000 major projects appraised during the period 2007-13, representing almost €180 billion of investment, and we study the expected difference between their Economic Rate of Return (ERR) and Financial Rate of Return (FRR). Such difference is a distinctive feature of the current CBA approach adopted by the European Commission and mainly depends on the use of shadow prices and the inclusion of externalities on the estimation of ERR, whilst FRR is based on market prices. Hence, we ask two simple questions: to what extent CBA leads to different expected returns compared with the financial analysis on a portfolio of major projects? What are the most important drivers of the difference between financial and economic analysis? We find that, on average, FRR is slightly negative (-2. 9%), while ERR is positive (16. 2%). The existing positive relation between ERR and FRR suggests that proposed projects are expected to be beneficial for the society and while they are not attractive for private investors, on average, they do not produce too large financial losses. Moreover, the analysis shows that the difference between ERR and FRR is wider in some sectors and we discuss possible interpretations of these findings.

G. 4: New Vehicles, Regulations, and Subsidies

Chair: Kerry Krutilla, Indiana University, Bloomington

Discussant: Ann Wolverton, U. S. Environmental Protection Agency


  1. Consumer Satisfaction with New Vehicles Subject to Greenhouse Gas and Fuel Economy Standards; Hsing-Hsiang Huang*, Oak Ridge Institute for Science and Technology; Gloria Helfand, U. S. Environmental Protection Agency; Kevin Bolon, U. S. Environmental Protection Agency

A variety of fuel-saving technologies have been implemented in light-duty vehicles since 2012 under the U. S. Environmental Protection Agency's and Department of Transportation's light-duty vehicle greenhouse gas emissions and fuel economy standards. Questions have arisen whether these new technologies create unforeseen problems -- hidden costs -- that have not been included in the net benefit calculations. We examine hidden costs associated with a vehicle's operational characteristics such as acceleration, handling, ride comfort, noise, and vibration; consumers' responses to these attributes are not easy to measure. To overcome the empirical challenge, a few recent studies used data coded from online professional auto reviews to explore potential hidden costs, and found little correlation of hidden costs with the technologies. However, there may be a gap of assessment of operational characteristics between professional auto reviewers and vehicle consumers. This study aims to fill this gap by using vehicle consumer satisfaction survey data gathered from 2014 to 2016 by Strategic Vision. In addition to a rich set of household characteristics, this survey provides consumers' overall experience with the purchased vehicle and assessment on a number of operational characteristics. Preliminary results indicate that, for each survey year, consumers' experiences with new vehicles are good; less than 3% of either passenger car or light-truck consumers gave an overall unsatisfactory review. For most of the operational characteristics, the unsatisfactory rates are also less than 3%; the unsatisfactory rates noise and vibration are higher but below 9%. Moreover, regression results suggest that more fuel-efficient vehicles are associated with lower probability of getting unsatisfactory ratings of operational characteristics, conditional on engine cylinders, brands, segments, and household characteristics. Additional work will explore the relationships between specific fuel-saving technologies and operational characteristics.

  1. Power and Fuel Economy Tradeoffs, and Implications for Benefits and Costs of Vehicle Greenhouse Gas Regulations; Gloria Helfand*, U. S. Environmental Protection Agency; Andrew Moskalik, Jeff Alson

As federal requirements for vehicle greenhouse gas (GHG) emissions and fuel economy (FE) have grown more stringent, concerns have been raised about possible opportunity costs of the standards. Some research argues that improvements in FE come at the expense of vehicle performance. If so, then losses in performance may be an opportunity cost of the standards. There are at least two counters to this hypothesis. First, the standards appear to have increased innovation and adoption of new technologies in the auto industry. If regulation caused innovation that can improve power, FE or both, then standards may also be responsible for a performance benefit for consumers that would not have happened in their absence. Second, the opportunity cost argument raises the question of the relationship between fuel economy and performance. Previous analyses, based on regressions of the relationship between power, fuel economy, and weight for vehicles produced, may not address some key concerns. First, vehicles produced do not provide a random sample of all possible technological relationships; estimates based on a non-random sample may be biased. Next, results are sensitive to measures used for power. Third, measures of innovation are sensitive to specification. Thus, tradeoff measures from these analyses may not accurately reflect underlying technological relationships. This paper provides evidence from an engineering simulation model of the changing relationship between power and fuel economy, and assesses its implications for benefit-cost analysis. We examine the tradeoffs for vehicles representing model years 1980, 2008, and 2016, and a simulated advanced technology. For newer powertrains, for constant fuel consumption, vehicles can have different acceleration rates; fuel consumption appears unaffected by power. These results come from powertrain innovations. If these innovations are attributable to regulations, then there may be ancillary benefits, rather than solely opportunity costs, to be considered in regulatory analysis.

  1. Network Externality and Subsidy Structure in Two-Sided Markets: Evidence from Electric Vehicle Incentives; Katalin Springel, Resources for the Future

In an effort to combat global warming and reduce emissions, governments across the world are implementing increasingly diverse incentives to expand the proportion of electric vehicles (EVs) on the roads. Many of these policies provide financial support to lower the high upfront costs consumers face and build up the infrastructure of charging stations. There is little guidance theoretically and empirically on which governmental efforts work best to advance electric vehicle sales. My paper empirically investigates the impact different incentives have on electric vehicle adoption using a two-sided market framework. The electric vehicle market is inherently two-sided, with one side being the electric vehicle drivers, and the other side being battery charging stations and between them are the platforms, the car manufacturers. Particularly, the focus is on whether it is preferable to subsidize consumers by lowering the high upfront costs associated with EV purchases, or charging stations, by lowering their sunk entry costs with a one-time subsidy. To motivate my research question, I began by developing a stylistic two-sided market model to show that subsidies to the different sides of the EV market are non-neutral, in the sense that one dollar spent on subsidy given to the charging side has a different economic impact as the same amount spent on subsidy given to consumers purchasing EVs. This result is driven by a key feature of two-sided markets, namely the positive network externalities between the two sides of the market. This clearly shows that if we really want to learn where to give subsidies to achieve the policy goal of increased EV sales we need to empirically estimate the impact of price subsidies versus station subsidies using a two-sided market framework. In my analysis, I use new large-scale vehicle registry data from Norway to empirically estimate the impact different subsidies have on electric vehicle adoption where network externalities are present. I present descriptive evidence to show that electric vehicle purchases are positively related to both consumer price and charging station subsidies. I estimate a structural model of consumer vehicle choice and charging station entry, which incorporates flexible substitution patterns and allows me to analyze out-of-sample predictions of electric vehicle sales. In particular, the counterfactuals compare the impact of direct purchasing price subsidies to the impact of charging station subsidies. I find that between 2010 and 2015 every 100 million Norwegian kroner spent on station subsidies alone resulted in 835 additional electric vehicle purchases compared to a counterfactual in which there are no subsidies on either side of the market. The same amount spent on price subsidies led to only an additional 387 electric vehicles being sold compared to a simulated scenario where there are no EV incentives. However, the relation inverts with increased spending, as the impact of station subsidies on electric vehicle purchases tapers off faster.