The Role of Public Policy in Technology Diffusion: The case of Plug-in

Subscriber access provided by UNIV OF WESTERN ONTARIO

Policy Analysis

The Role of Public Policy in Technology Diffusion: The case of Plug-in Electric Vehicles Julio C. Zambrano-Gutiérrez, Sean Nicholson-Crotty, Sanya Carley, and Saba Siddiki Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b01149 • Publication Date (Web): 07 Sep 2018 Downloaded from http://pubs.acs.org on September 11, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 29

Environmental Science & Technology

1

1 2 3 4 5 6 7 8 9

The Role of Public Policy in Technology Diffusion: The case of Plug-in Electric Vehicles Julio C. Zambrano-Gutiérrez*, †, Sean Nicholson-Crotty†, Sanya Carley†, Saba Siddiki‡ †

School of Public and Environmental Affairs, Indiana University, Bloomington, IN 47405, USA Maxwell School of Citizenship and Public Affairs, Syracuse University, Syracuse, NY 13244, USA * Corresponding author’s email: [email protected]

‡

10 11 12

Abstract The plug-in electric vehicle (PEV) is regarded by many as a viable alternative to the

13

internal combustion engine, so long as the radical technology is able to overcome technical and

14

financial shortcomings that dictate consumer acceptance. States have instituted a variety of

15

policies aimed at mitigating these shortcomings and simultaneously increasing consumer demand

16

for PEV vehicles. Motivated by a limited body of literature on the effects of these policies, and a

17

significant need for information about policy efficacy, in this study we evaluate the relationship

18

between a suite of state-level policies and PEV registrations. Results reveal that tax credits for

19

individuals, grants programs for charging infrastructure and PEV purchases, and incentives for

20

state-owned PEVs fleets increase PEV registrations. The observed impact of grant incentives is

21

mediated by charging capacity or, alternative phrased, much of the influence of grants on

22

registrations is through the channel of first improving the charging infrastructure within a state. Abstract Art PEV registrations (thousand)

23 120 100 80 60 40 20 0 2010

2011

2012

2013

2014

2015

States with no incentives States with at least one incentive

24

ACS Paragon Plus Environment

2016


Page 2 of 29

2

Introduction

25

The plug-in electric vehicle (PEV) is a classic case of a radical technology. It is touted for

26 27

its potential to compete with, and potentially displace, the internal combustion engine (ICE) as a

28

dominant vehicle technology. As with other “radical” technologies (1), however, this

29

technological shift will require a confluence of policy, infrastructural, and behavioral

30

developments in its favor. Generally speaking, scholars interested in the adoption and diffusion

31

of product innovations have long recognized that government policies can encourage shifts in

32

consumer preferences toward them (2-4). Work on radical technologies suggest that these

33

“pushes” from government are particularly important when an innovation represents something

34

truly new or radical to the current technological regime with which consumers are familiar (5-6). The first mass-marketed PEVs hit the U.S. automobile market in 2010. As of 2016, the

35 36

stock of PEVs had grown to 563,710 (7), with uneven distributions of these vehicles across

37

states. A number of factors are known to dissuade potential consumers from adopting PEVs,

38

including cost, range, and battery recharging requirements (8-9). PEVs tend to have an upfront

39

cost that is 50 to 100% higher than similar internal combustion engine vehicles, though the level

40

of subsidies offered by government affects what the consumer pays (8). The range of most

41

current PEVs on a fully-charged battery without a hybrid engine is typically between 70 and 100

42

miles, significantly less than the 250- to 350-mile range of a conventional ICE. These distance

43

limitations contribute to “range anxiety” (8), a fear that one’s battery will die when on the road.

44

Finally, PEVs require specific charging infrastructure that must be installed in an owners’ home

45

or an accessible public facility. Previous work on the intent to purchase PEVs (10) find that

46

perceptions of these disadvantages damage consumer interest in adoption more than any other

47

factor.


Page 3 of 29


3

48

States and municipalities have adopted a number of policies designed to overcome these

49

barriers and facilitate PEV consumption, though the policy response varies considerably across

50

jurisdictions. Not surprisingly, there have been a number of studies of the relationship between

51

these policy incentives and the spread of PEVs; but these studies have reached mixed

52

conclusions about policy impacts (see 11-12 for a recent review of many of these studies). A

53

number of studies have also surveyed consumers and potential consumers in order to understand

54

the impact of policies (see for example 13-14) and this methodological approach has also

55

produced mixed conclusions. However, in the interest of brevity, we will only provide a partial

56

review of studies that examine policy impact using statistical analyses to predict actual PEV

57

purchases or registrations.

58

Several studies have focused exclusively on the most prominent policy incentives for

59

PEV purchase – direct financial incentives and charging infrastructure. Sierzchula et al. (15) find

60

that both are important, but that neither has a particularly large impact on PEV purchases. Lutsey

61

et al. (16) identify an interactive effect between these variables, suggesting that financial

62

incentives are positively associated with PEV sales, but only in states with sufficient charging

63

availability. A key drawback to both of these studies, however, is that they are estimated in a

64

single year of data and, as such, cannot include fixed effects to account for time or unmeasured

65

state-level characteristics, including other policies.

66

Li et al. (17) also focus on tax incentives, in the form of the federal income tax credit, and

67

municipal level charging capacity. They adopt a more sophisticated design, examining quarterly

68

sales over a two-year period and explicitly accounting for endogeneity between previous PEV

69

sales and investment in charging infrastructure. Because of the longer time frame, they are also

70

able to estimate a two-way fixed effects model, which helps to isolate within-jurisdiction effects.



Page 4 of 29

4

71

However, it is important to note that fixed effects cannot account for state-level policy incentives

72

that are time-variant but otherwise omitted from the analysis, such as grant programs and

73

incentives for state-owned PEVs fleets. Li et al. found that federal incentive does increase sales,

74

but that almost half of the effect is due to charging capacity, and that a direct incentive for

75

charging station deployment would have a larger overall effect on PEV consumption.

76

Some studies have also sought to analyze the impact of a larger suite of PEV related

77

policy incentives, but again, report varying results. Studies have found a positive association

78

between financial incentives, investment in charging infrastructure, high occupancy vehicle

79

(HOV) lane access, and, in one case, emissions testing exceptions (18-19). Vergis and Chen (20)

80

examine the impact of purchase incentives and a count of other “supportive” policies, but do not

81

distinguish the individual policies. They find a positive correlation between these variables and

82

market share for battery electric vehicles, but not for plug-in hybrid electrics. These results are

83

the opposite of those arrived at by Jin et al. (19). Santini and colleagues (21-22) find that DOE

84

grants affected state-level registrations in the absence of other policies, while activities by public

85

utilities to promote PEVs only worked in conjunction with other state-level incentives. This latter

86

result is consistent with Slowik and Lutsey (23), who suggest “a comprehensive package of

87

policy and promotion actions by state, local, utility, and other private stakeholders is a key for

88

developing the electric vehicle market.” Unfortunately, all of these studies employ a cross-

89

sectional research designs in a very limited study period and, thus, cannot isolate within-

90

jurisdiction effects.

91

Clinton et al. (24) analyze sales over a slightly longer time period—two years—and find

92

that tax credit and charging infrastructure are significant predictors, while direct rebates and

93

HOV lane access are not. These authors use an instrumental variables approach to deal with


Page 5 of 29


5

94

endogeneity between sales and charging infrastructure. Unfortunately, the financial incentives

95

analyzed by Clinton et al. do not vary over the short time span of their study, so they were unable

96

to estimate within-state effects.

97

Jenn et al. (25) confirmed the positive effect of financial incentives (tax credits and

98

rebates) on the number of PEVs registrations. They also find that consumer awareness of

99

incentives helps to explain heterogeneity in the response to them. These authors test the effects

100

of PEV incentives during a longer period (2010-2015) and adopt a sophisticated methodological

101

approach. Unfortunately, because of the use of monthly data, they cannot explicitly include

102

charging infrastructure and PEV model availability in their models.

103

We suggest that mixed findings in previous research regarding the impact of public

104

policies on PEV market share, along with uncertainty regarding the best way for governments to

105

incentivize the spread of this technology, are due to inconsistencies in the design of previous

106

studies. Studies to date tend to focus on sales in a single year or limited number of years,

107

evaluate a limited number of policy incentives, or employ methodological approaches that

108

compromise internal validity. In our analysis, we seek to build on these studies by analyzing a

109

larger number of policies, more PEV models, over a much longer period of time, and using an

110

identification strategy aimed at causal inference and the use of mediated or indirect impacts of

111

policy on the spread of PEVs.

112

More specifically, we analyze the effect of nine policy incentives on the spread of PEVs

113

between 2010 and 2016. We find that tax credits for individuals, grants programs targeting

114

charging infrastructure and the purchase of PEVs, and incentives for state-owned PEVs fleets

115

increase registrations even when controlling for other incentives. We also find that the observed

116

impact of grant incentives is mediated by charging capacity or, more specifically that much of



Page 6 of 29

6

117

their influence on registrations comes through their role in improving the charging infrastructure

118

within a state. The paper concludes with a discussion of the implications of these results for our

119

understanding of the diffusion of radical technologies and the ability for systems to overcome

120

technological “lock-in” (26). Methods

121

122

We examine PEV registrations in the American states between 2010 and 2016 because

123

this time frame spans from the introduction of modern PEVs through the last year for which data

124

are currently available. Our dataset is a longer time series than has been used in previous studies

125

and the set of PEV adoption policies—nine in total—is more extensive than those studied

126

previously. Following the precedent set by previous scholars (17,24), we employ a

127

methodological approach that accounts for endogeneity between charging infrastructure and PEV

128

sales. We also employ a research design that allows us to investigate the degree to which the

129

impact of policy on PEV diffusion actually operates through its impact on charging

130

infrastructure.

131

Dependent Variable: PEV Registrations

132

The data for new PEV registrations come from the private provider, IHS Markit (27).

133

They include information for battery electric vehicles (BEV) and plug-in hybrid electric vehicles

134

(PHEV) registrations for each year between 2010 and 2016 by state. This study includes 50 PEV

135

models but does not include data on low-speed electric vehicle registrations, since their

136

technology is not comparable with the mass marketed PEVs.

137

The primary dependent variable for this study is the number of annual new PEV

138

registrations by state. As displayed in Figure 1, the spread of PEVs has been unequal between

139

states. The state of California accounts for 48% of the total new PEVs registrations accumulated


Page 7 of 29


7

140

between 2010 and 2016, followed by Georgia with 5%; and Washington, New York, and Florida

141

each one with approximately 4% of the PEV registrations. The rest of states have a participation

142

between 0.02% (Wyoming) and 3.19% (Texas) of the new PEV registrations during our seven

143

years of analysis. [Figure 1 here]

144 145 146

Independent Variables: Public Policies Data on PEV related policies come from LexisNexis State Capital (28), which provides

147

access to legislative statutes from the fifty states. Using keywords “low emission vehicles OR

148

zero emission vehicles OR electric vehicles”, we found 150 out of 557 relevant results

149

containing information related to PEV incentives between 1990 and 2016. Search results that are

150

not part of this study only provide definitions about our keywords (70%) or do not contain new

151

information about policy incentives (30%). The history of the modifications in the statutory

152

codes are available and linked to a respective bill; the latter of which contains the year the policy

153

went into effect. In some occasions (30% of the useable cases), the bill only contains the

154

adoption year, in which case this study assumes that the following year is the effective year.

155

Our data extraction technique identified a total of 24 different types of PEV incentives at

156

the state level. Many of these policies, however, had diffused to only a single state or were

157

adopted long before modern PEVs hit the market, making assessments of their causal impacts

158

impossible. In our analyses, we chose to focus on nine public policies that diffused to two or

159

more states and were adopted by at least two states after 2010. Figure 1 reveals that, like PEVs,

160

the spread of these nine policies among the states has been uneven. The policies include financial

161

incentives, infrastructure related incentives, and symbolic policies as fully described in Table 1.

162

[Table 1 here]



Page 8 of 29

8

163

We use data on PEV incentives to create two different types of independent variables.

164

First, we create dichotomous indicators coded 1 if a state has a particular policy and 0 otherwise,

165

which allows us to estimate the impact of adopting an individual policy while controlling for the

166

adoption of other policies. Afterwards, we code the dollar amounts of grants and tax credits to

167

explore whether the size, rather than simply the presence, of a financial incentive influences the

168

number of PEV registrations within a state.

169

Control Variables

170

We control for internal state characteristics that may influence both PEV diffusion and

171

PEV policy adoption. The control variables represent obstacles and motivations to the adoption

172

of new technology by consumers within a state. All data sources are outlined in Table 2. One of

173

the primary impediments to the spread of PEVs identified in previous research is extent of the

174

charging infrastructure within a jurisdiction. In order to capture that capacity, we create a

175

measure of the number of charging stations in each state and year.

176

This study includes a measure of PEV-models available in each state in each year to

177

control for the influence of supply on PEV diffusion. Consistent with previous studies, we expect

178

higher number of PEV registrations in states with a larger number of PEV models available (16).

179

Because of their expense, both in terms of purchase price and investment required by

180

communities, there are also likely to be a number of socioeconomic obstacles to the spread of

181

PEVs. We capture these with measures of wealth, education and, employment. Specifically, we

182

measure gross state product (GSP) per capita, the proportion of the population with at least high

183

school education, and the unemployment rate.

184

Citizens may also purchase PEVs in an effort to reduce negative environmental

185

consequences of ICEs, such as carbon dioxide (CO2) emissions. This analysis includes the CO2


Page 9 of 29


9

186

emissions produced by the transportation sector as a proportion of the total CO2 emissions at

187

each state. States with higher proportion of CO2 emissions are expected to have a higher level of

188

PEV registrations. The data for CO2 emissions are only available until 2014 (38), therefore we

189

interpolate the values for 2015 and 2016 as a function of the GSP. One remark about PEVs is

190

that their combustion emission take place during the stage of electricity generation (29-30). For

191

that reason, we include renewable energy use within a state to capture the readiness of each state

192

to embrace less pollutant technologies.

193

Of course, the level of concern over issues such as CO2 emissions and the willingness to

194

embrace technologies such as PEVs more broadly is likely to be partially a function of the

195

environmental ideology of state residents. We capture this orientation with two political

196

measures. The first is the percent of Democratic legislators in the Senate and House of

197

Representatives. We also include the total number of victories of members from the Green Party

198

across different popularly elected positions in each state.

199

If one extends the literature on conventional hybrids to the context of the PEV, we could

200

hypothesize that consumers in states with higher gasoline prices are more likely motivated to

201

accept PEVs as an ICE alternative than those that live in states with lower gasoline prices (31-

202

33). We include annual data for average gasoline price per gallon, which incorporates federal and

203

state gasoline taxes, but excludes local taxes. In a related argument, previous work has assumed

204

that the price of electricity will influence the decisions of consumers considering PEVs.

205

Therefore, we control for the average price for electricity (cents/kilowatt-hour) in each state and

206

year. Finally, the number of licensed drivers operationalize the size of the potential market for

207

PEV at each state.

208

[Table 2 here]



Page 10 of 29

10

209 210

Design and Estimators For the primary analysis, this study exploits variation in the timing and extent of PEV

211

policy adoption in the American states in order to estimate a generalized difference-in-

212

differences model. As is standard in these estimators, the model includes two-way fixed effects

213

(FEs) which allows for the estimation of the within-state effects of a policy on PEV registrations

214

in the period following its adoption. Standard errors are clustered at the state level in this

215

specification. The basic estimator can be represented as such: = + .. + ′ + + + (1)

216 217 218 219 220 221 222

Where: Yit = New Electric Vehicle Registrations in Statei at Timet Tit = Electric Vehicle Promotion Policyn..k Adopt * Post in Statei at Timet X’it = a vector of time variant control variables in Statei at Timet ui = a state fixed effect vt = a year fixed effect

223

This specification allows us to compare PEV registrations within each state before and

224

after the adoption of each PEV related policy, while controlling the existence of other policies

225

and a host of controls.

226

As noted above, previous work suggests that charging infrastructure within a jurisdiction

227

not only incentivizes potential consumers of PEVs but may also be endogenous to previous sales.

228

Diagnostics reveal that this is also the case in our sample of states and years. In order to correct

229

for potential bias arising from this endogeneity, we implement an instrumental variables

230

approach with the differences-in-differences framework. Following the suggestion of Li et al.

231

(17), we instrument the number of charging stations with the interaction between the number of

232

stations in all states other than the state-year observation under analysis and the number of

233

grocery stores in each state (lagged one year). The first stage shows an F-test for the excluded


Page 11 of 29


11

234

instrument of 37.38, suggesting that the instrument is acceptable. So, in the equation discussed

235

above, X’it –the vector of time variant control variables in Statei at Timet –contains the

236

instrumented version of charging stations, rather than the actual number.

237

Finally, we run a mediating variables analysis in order to determine whether the observed

238

impact of PEV policies on sales is in part a function of their impact on charging infrastructure.

239

Evidence of mediation requires us to show that: 1) the number of charging stations influences

240

registrations; 2) policies influence the number of charging stations; 3) policies influence

241

registrations in the absence of a control for charging stations; and 4) the impact of policies on

242

registrations changes when we include the mediator in the model. The first of these requirements

243

can be met in our primary analysis presented in Table 3 (Column 1). We test for number 2 in a

244

two-way fixed effects model where number of charging stations is the dependent variable, which

245

is presented in the first column of Table 4. The last two requirements are tested with difference-

246

in-difference models of registrations presented in columns 2 and 3 of that table. We instrument

247

for charging stations in the last model because we still need to correct for diagnosed endogeneity.

248

Thus, the final analysis in Table 4 is identical to the main model presented in Table 3 (Column

249

2), but we present it again in order to facilitate the observation of any mediating effects.

250 251

Results The first column of Table 3 contains the difference-in-differences analysis of the impact

252

of nine policy incentives on PEV registrations between 2010 and 2016. Consistent with previous

253

work, instrumented charging infrastructure has a significant and positive impact on registrations.

254

The results suggest that a 1-standard deviation increase in charging stations is associated with a

255

5,360 or 0.82-standard deviation increase PEV registrations. As another way to think about this

256

result, it suggests that a 10% increase in available charging infrastructure (as measured by the



Page 12 of 29

12

257

instrumented charging stations variable) is associated with approximately 18 % more

258

registrations.

259

When we use a dichotomous indicator of policy, the results suggest that tax credits for

260

individuals have a positive impact on the number of new PEV registrations in each state.

261

Substantively, the findings suggest that the adoption of tax credits for individuals causes 1977

262

additional PEV registrations in adopting states, which represents an increase of more than 0.3 –

263

standard deviations. Interestingly, the findings also suggest that the presence of special financing

264

for PEV related equipment increases PEV registrations in a state and that the impact is larger

265

than that of individual tax credits. Specifically, we find that state guaranteed financing increases

266

registrations by 5,452 or 0.83-standard deviations compared with the level prior to adoption.

267

Finally, we find that the adoption of incentives for state-owned PEV fleets causes an increase of

268

2,115 PEV registrations, which represents an increase of 0.32 – standard deviations.

269 270

[Table 3 here] It is possible, of course, that a simple dichotomous indicator of polices obscures

271

important variation in their impact on the spread of PEVs. Therefore, in order to generate a more

272

precise estimate of the impact of grants and tax incentives, the second column of Table 3

273

presents models that uses the dollar amounts of these financial incentives, rather than simply an

274

indicator of their presence. When we take the dollar amounts into account, both monetary

275

incentives cause an increase in PEV registrations, and grants now appear to have a larger impact

276

than tax credits. Specifically, we find that the adoption of a $5 million program, which is the

277

median size among states with these programs, results in increase of 6,500 in PEV registrations

278

relative to the period before adoption, which represents almost a full standard deviation increase.


Page 13 of 29


13

279

Alternatively, adoption of the median tax credit of $6,000 increases registrations by 653, which

280

is or about 0.1 – standard deviations.

281

Before moving on, it is important to note that three policies had a negative and significant

282

impact on PEV sales. Specifically, models suggest that states that adopted regulations intended

283

to standardize public charging stations within a state, states in the planning stage, and those that

284

adopted policies reserving special parking spots for PEV drivers actually saw a decrease in new

285

registrations following those adoptions. We will return to these negative findings in the

286

discussion section, where we offer some potential explanations.

287

Exploring the mediating impact of charging infrastructure

288

The results thus far suggest that policies designed, at least in part, to contribute to a

289

state’s charging infrastructure may have an influence on PEV registrations. This raises the

290

possibility that this observed impact is in fact mediated by the degree to which these policies

291

actually increase the number of stations available to PEV owners.

292

The analyses presented in Tables 3 and 4 provide the information necessary to test that

293

possibility. First, the primary analysis presented in the first column of Table 3 confirms that

294

charging stations (instrumented) have a positive and significant impact on PEV registrations.

295

Turning now to Table 4, the two-way fixed effects model of charging stations in the first column

296

shows that some public policies do have an impact on charging infrastructure, which is the next

297

prerequisite for demonstrating a mediating effect. Specifically, the analysis suggests that the

298

presence of grants that can be used to enhance charging capacity are positively associated with

299

the number of charging stations within a state.

300

[Table 4 here]



Page 14 of 29

14

301

The final piece in the mediating variables analysis is to compare the impact of these

302

policies on registrations when the potential mediator—number of charging stations—is excluded

303

versus included in the models. Looking at the final two columns of Table 4 we can see that

304

grants increase registrations in both models, but that the impact shrinks significantly when the

305

instrumented charging station variable is included in the model. Specifically, the results suggest

306

that 46% of the total impact of PEV grant policies on new registrations actually comes through

307

the increase in available charging capacity that the policy stimulates. Discussion

308 309

There has been significant growth in the ownership of PEVs since the first mass-

310

marketed models became available in 2010. That diffusion has, however, been highly variable

311

across states. Incentives to facilitate the spread of PEVs have also varied dramatically across the

312

states and we test whether heterogeneity in the latter can explain the observed variation in the

313

former. This study represents a contribution to existing research on this subject because it

314

investigates a larger number of PEV related policies, over a longer time period, using causal

315

identification strategies that explores both direct and mediated effects of public policy.

316

Analyses of PEV ownership and nine PEV statutory policy-incentives across all 50 states

317

between 2010 and 2016 provide some evidence for the positive role of government incentives on

318

PEVs uptake. Specifically, they suggest that the adoption of grants targeting infrastructure and

319

ownership, as well as policies creating special financing for PEV-related equipment increase

320

PEV registrations within a state. These policies remain the significant predictors of registrations

321

even if we use the dollar amounts of financial incentives, such as grants and tax credits, as the

322

independent variables.


Page 15 of 29


15

323

These results offer an important confirmation of previous findings which suggest that

324

individual tax credits can spur PEV diffusion, even when we control for a much larger number of

325

policies and analyze registrations over a much longer time period. They also add significantly to

326

previous work by suggesting that other policies can also influence PEV registrations. Indeed, the

327

findings indicate that states may be able to overcome the observability problem inherent in new

328

technologies such as PEVs simply by putting these vehicles on the road in public fleets.

329

Interestingly, the influence of incentives for state-owned electric fleets on new registrations is

330

larger than the impact for individual tax incentives, which have been one of the most widely

331

touted mechanisms for incentivizing uptake of PEVs.

332

The findings also suggest the import of another heretofore unexplored state-level

333

incentive for PEV diffusion. Specifically, they suggest that grants targeted at improving charging

334

infrastructure and providing better financing for PEV purchases have a positive and significant

335

impact on new PEV registrations in the states that adopt them. Intuitively, we would expect that

336

part of this impact would occur because they actually do improve charging capacity and a

337

mediating variables analysis suggests that this is the case. The grants are positively associated

338

with the number of charging stations, which in turn is positively associated with PEV

339

registrations. The final part of the mediating variable analysis suggests that almost half of the

340

observed treatment effect of adopting a PEV grants policy is, in fact, working through the impact

341

of those policies on charging capacity.

342

Before moving on, it is important to note that the impacts of individual tax credits or

343

policies to grow state-owned PEV fleets are not moderated by charging infrastructure. This is

344

what we would expect, because these policies are not designed to overcome the barriers to PEV

345

ownership by improving charging capacity. The fact that they are not mediated by the number of



Page 16 of 29

16

346

stations provides a good falsification test and, therefore, increase our confidence in the validity

347

of the results of the mediating variables analysis.

348

Our results also suggest that numerous policy incentives do not appear to directly

349

increase PEV registrations. Of course, we exercise caution in interpreting these nonfindings, as

350

well as the positive results discussed above, because we are analyzing only six years of data at

351

the very outset of diffusion period for PEVs. Nonetheless, the failure of some of these policies,

352

like laws providing some funding to PEV manufacturers for R&D, to influence sales should not,

353

perhaps, be surprising. The mechanisms by which these policies might influence consumer

354

behavior in a meaningful way are difficult to imagine.

355

It is harder to understand why policies that seek to improve charging infrastructure by

356

standardizing charging stations within a state would not have a positive impact on PEV

357

registrations. However, these public charging station policies have actually increased the number

358

of charging stations (see Column 1, Table 4). As such, these policies help to reduce anxiety

359

among consumers about the range of PEVs. The fact that these policies appear to cause a

360

significant reduction in new registrations is more difficult to explain. It is possible that the debate

361

over these policies makes consumers aware that even if they find a charging station, they may

362

not be able to plug their new vehicle into it, which creates additional uncertainty and hampers the

363

diffusion of the technology. Obviously, however, more research is needed to understand the

364

mechanisms underlying this counter-intuitive result.

365

Previous research on PEVs suggests that price and range anxiety are the two major

366

factors that keep potential consumers from these vehicles. Older work also suggests that

367

observability is one of the key impediments to the spread of any truly innovative or novel

368

product (see Rogers (2)). Our findings indicate that policies that directly address these barriers


Page 17 of 29


17

369

are most effective at facilitating the spread of this radical technology. Tax incentives directly

370

reduce the higher price of PEVs; grants targeting charging infrastructure likely reduce range

371

anxiety because they actually increase the number of available charging stations; finally, policies

372

that grow state-owned PEV fleets allow consumers to see and become accustomed to the new

373

technology.

374 375

Acknowledgments

376

This article is part of a project funded by the National Science Foundation, under grant

377

award #1633082. Besides their appreciation for the financial support, the authors would also like

378

to acknowledge the helpful feedback offered by anonymous reviewers and other project team

379

members: Derek Ehrnschwender, Steven Kakovik, Amani Khoury, and Lucas Stegemiller.

380



Page 18 of 29

18

References

381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424

Christensen, C.M.; Overdorf, M. Meeting the challenge of disruptive change. Harvard business review. 2000, 78 (2), 66-77. (2) Rogers, E.M. Diffusion of innovations; The Free Press: New York, N.Y., 1962. (3) Kolodinsky, J.M.; Hogarth, J.M.; Hilgert, M.A. The adoption of electronic banking technologies by US consumers. International Journal of Bank Marketing. 2004, 22 (4), 238259. (4) Newell, R.G.; Jaffe, A.B.; Stavins, R.N. The effects of economic and policy incentives on carbon mitigation technologies. Energy Economics. 2006, 28 (5-6), 563-578. (5) Unruh, G.C. Escaping carbon lock-in. Energy policy. 2002, 30 (4), 317-325. (6) Kemp, R.P. Governance of environment-enhancing technical change: past experiences and suggestions for improvement. MERIT Research Memorandum 20-013, 2000. (7) Global EV Outlook 2017: Two million and counting; International Energy Agency: Paris, France, 2017; www.iea.org/publications/freepublications/publication/GlobalEVOutlook2017.pdf (8) Graham-Rowe, E.; Gardner, B.; Abraham, C.; Skippon, S.; Dittmar, H.; Hutchins, R.; Stannard, J. Mainstream consumers driving plug-in battery-electric and plug-in hybrid electric cars: A qualitative analysis of responses and evaluations. Transportation Research Part A: Policy and Practice. 2012, 46 (1), 140-153. (9) Sperling, D. Future drive: Electric vehicles and sustainable transportation; Island Press: Washington, DC, 1995. (10)Carley, S.; Krause, R.M.; Lane, B.W.; Graham, J.D. Intent to purchase a plug-in electric vehicle: A survey of early impressions in large US cites. Transportation Research Part D: Transport and Environment. 2013, 18, 39-45. (11)Zhou, Y.; Levin, T.; Plotkin, S.E. Plug-in Electric Vehicle Policy Effectiveness: Literature Review. United States, 2016; www.osti.gov/biblio/1255232 (12)Hardman, S.; Chandan, A.; Tal, G.; Turrentine, T. The effectiveness of financial purchase incentives for battery electric vehicles–A review of the evidence. Renewable and Sustainable Energy Reviews. 2017, 80, 1100-1111. (13)Hardman, S.; Tal, G. Exploring the Decision to Adopt a High-End Battery Electric Vehicle: Role of Financial and Nonfinancial Motivations. Transportation Research Record: Journal of the Transportation Research Board. 2016, (2572), 20-27. (14)Tal, G.; Nicholas, M. A. Exploring the Impact of High Occupancy Vehicle (HOV) Lane Access on Plug-in Vehicle Sales and Usage in California. Institute of Transportation Studies, University of California, Davis, 2014; http://docs.trb.org/prp/15-5180.pdf (15)Sierzchula, W.; Bakker, S.; Maat, K.; van Wee, B. The influence of financial incentives and other socio-economic factors on electric vehicle adoption. Energy Policy. 2014, 68, 183-194. (16)Lutsey, N.; Searle, S.; Chambliss, S.; Bandivadekar, A. Assessment of leading PEV promotion activities in United States cities. International Council on Clean Transportation, 2015; www.theicct.org/sites/default/files/publications/ICCT_EV-promotionUScities_20150729.pdf (17)Li, S; Tong, L; Xing, J; Zhou, Y. The market for electric vehicles: indirect network effects and policy design. Journal of the Association of Environmental and Resource Economists. 2017, 4 (1), 89-133. (1)


Page 19 of 29


19

425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470

(18)Narassimhan,

E.; Johnson, C. The Effect of State Incentives on Plug-in Electric Vehicle Purchases. Presentation, 2014; www.nrel.gov/docs/gen/fy15/62884.pdf (19)Jin, L.; Searle, S.; Lutsey, N. Evaluation of state-level U.S. PEV incentives. International Council on Clean Transportation, 2014; www.theicct.org/sites/default/files/publications/ICCT_state-EV-incentives_20141030.pdf (20)Vergis, S.; Chen, B. Comparison of plug-in electric vehicle adoption in the United States: A state by state approach. Research in Transportation Economics. 2015, 52, 56-64. (21)Santini, D., Zhou, Y., and Marcy, R. Electric drive technology market trends. Argonne National Laboratory. White paper for DOE Clean Cities Strategic Planning. 2015a; https://cleancities.energy.gov/files/u/news_events/document/document_url/95/2015_strategic _planning_electric_drive.pdf (22)Santini, D.; Zhou, Y.; Marcy, R. PEV adoption pattern and utility outreach study. Presentation to DOE. 2015b. (23)Slowik, P., & Lutsey, N. Expanding the electric vehicle market in us cities. International Council on Clean Transportation, 2017; www.theicct.org/sites/default/files/publications/USCities-EVs_ICCT-White-Paper_25072017_vF.pdf (24)Clinton, B.; Brown, A.; Davidson, C.; Steinberg, D. Impact of Direct Financial Incentives in the Emerging Battery Electric Vehicle Market: A Preliminary Analysis. Presentation, 2015; www.nrel.gov/docs/fy15osti/63263.pdf (25)Jenn, A.; Springel, K.; Gopal, A. R. Effectiveness of electric vehicle incentives in the United States. Energy Policy. 2018, 119, 349-356. (26)Unruh, G.C. Understanding carbon lock-in. Energy policy. 2000, 28 (12), 817-830. (27)IHS Markit. Electric Vehicles Registrations by state. Dataset, 2017. (28)LexisNexis State Capital. Statutes. State Statutes by Keyword. Dataset, 2017 (accessed October 2017). (29)Brinkman, G.L.; Denholm, P.; Hannigan, M.P.; Milford, J.B. Effects of plug-in hybrid electric vehicles on ozone concentrations in Colorado. Environmental science & technology. 2010, 44 (16), 6256-6262. (30)Ji, S.; Cherry, C.R.; J. Bechle, M.; Wu, Y.; Marshall, J.D. Electric vehicles in China: emissions and health impacts. Environmental science & technology. 2012, 46 (4), 2018-2024. (31)Diamond, D. The impact of government incentives for hybrid-electric vehicles: Evidence from US states. Energy Policy. 2009, 37 (3), 972-983. (32)Gallagher, K.S.; Muehlegger, E. Giving green to get green? Incentives and consumer adoption of hybrid vehicle technology. Journal of Environmental Economics and management. 2011, 61 (1), 1-5. (33)Jenn, A.; Azevedo, I. L.; Ferreira, P. The impact of federal incentives on the adoption of hybrid electric vehicles in the United States. Energy Economics. 2013, 40, 936-942. (34)United States Census Bureau. 2012 Retail Trade (NAICS Sector 44-45). 2018. https://www.census.gov/data/tables/2012/econ/census/retail-trade.html (accessed January 2018). (35)Alternative Fuels Data Center. Alternative Fuel Stations. Dataset. 2018. https://www.afdc.energy.gov/data_download (accessed February 2018). (36)Bureau of Economic Analysis. Regional Economic Accounts. Dataset. 2017. https://www.bea.gov/regional/index.htm (accessed October 2017). (37)American Community Survey. IPUMS. 2017. https://usa.ipums.org/usa/ (accessed October 2017).



Page 20 of 29

20

471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486

(38)Environmental

Protection Agency. State CO2 Emissions from Fossil Fuel Combustion, 1990-

2014. 2017. https://archive.epa.gov/epa/statelocalclimate/state-co2-emissions-fossil-fuelcombustion-1990-2014.html (accessed October 2017). (39)U.S. Energy Information Administration. Electricity. 2017. https://www.eia.gov/electricity/data.php#sales (accessed October 2017). (40)National Conference of State Legislatures. State Partisan Composition. 2017. Available at http://www.ncsl.org/research/about-state-legislatures/partisan-composition.aspx#2016. (41)Green Party U.S. GPUS Elections Database. 2017. https://www.gpelections.org/ (accessed October 2017). (42)U.S. Energy Information Administration. State Energy Data System (SEDS): 1960-2015. All prices and expenditures estimates. 2017. https://www.eia.gov/state/seds/seds-datacomplete.php?sid=US#CompleteDataFile (accessed October 2017). (43)Federal Highway Administration. Highways Statistics 2015. 2017. https://www.fhwa.dot.gov/policyinformation/statistics/2015/ (accessed October 2017). (44)United States Census Bureau. Cartographic Boundary Shapefiles – States. 2018. https://www.census.gov/geo/maps-data/data/cbf/cbf_state.html

487


Page 21 of 29


21

488 489

Figure 1. Accumulation of new Plug-in Electric Vehicles (PEVs) registrations (market share %) and PEV Policy-incentives (number of policies) between 2010 and 2016

490

491 492 493 494 495

Source: Upper panel: IHS Markit (27); lower panel: LexisNexis State Capital (28); this figure was created using ArcMap 10-6 (44).



Page 22 of 29

22

496

Table 1. Plug-in electric vehicle policy-incentives Category

Incentive

Description

Example

California with its Alternative and Renewable Fuel and Grant programs to finance the Vehicle Technology Program installation of electric vehicle and its Air Quality charging infrastructure and Improvement Program provides funding for electric the purchase of plug-in charging infrastructure and electric vehicles electric vehicle deployment up to $40 million per year

Grants

Tax credits for individuals to purchase or lease plug-in electric vehicles, and to Tax credits for individuals convert internal combustion engine vehicles into plug-in electric vehicles Financial

Electric vehicle supply equipment financing

Programs to issue bonds, notes and other types of financial instruments to finance the installation of electric vehicle charging station, electric vehicle conversions, and to support the deployment of plug-in electric vehicles

Colorado offers tax credits to individuals who own lowemission vehicles, including PEVs, for an amount up to $6,000 Alabama authorizes local governments to issue bonds, notes, or other type of financing methods to increase charging stations deployment

Financing research to South Carolina with its facilitate the development and Distributed Energy Resource commercialization of electric Program invests in vehicles and their parts, and technologies to improve the Research and development the investment in technologies load management of electric to improve load management vehicle charging of electric vehicle charging stations

Infrastructure

Public charging stations availability

Reduction of administrative Minnesota requires that any barriers to install electric installed electric vehicle vehicle charging stations, charging station must be requirements to have charging compatible for utilization with stations in interstate highway any type of electric vehicle rest areas, and compatibility (make, model) requirements to use charging stations for any type of electric vehicle


Page 23 of 29


23

Private charging stations availability

Parking availability

Planning

Symbolic

Reduction of administrative barriers to install electric vehicle charging stations for personal residential use

Oregon forbids homeowners associations to prohibit the installation of charging stations in a parking lot or in another area subject to the exclusive use of the owner

Hawaii mandates places with Assignation of parking spaces at least one hundred parking to plug-in electric vehicles spaces to designate no less and parking enforcement with than 1% of the parking spaces monetary penalties to electric vehicles Studies to evaluate the feasibility to deploy electric vehicles infrastructure, to purchase state-owned electric vehicle fleets, to reduce electricity rates to charge electric vehicles, and to use grid technology to facilitate the use of electric vehicles

Massachusetts requires an action plan to increase access to electric vehicle infrastructure, increment the purchase of PEV by reducing the cost of PEV purchase and identifying strategies to remove barriers to PEV deployment

Prioritize the purchase of new Connecticut requires that state-owned fleet to be plug-in alternative fuel cars must electric vehicles and to represent 50% of the lightIncentives for state-owned increase progressively the duty trucks purchase by the electric vehicles fleets proportion of plug-in electric state after 2008 and 100% vehicles in the state-owned after 2010 fleet 497 498



Page 24 of 29

24

499

Table 2. Descriptive Statistics (2010-2016) Variable

Mean

Std. Dev.

Min

Max

Source

1570.91

6577.5

0

74749

IHS Markit27

Dependent Variable PEV registrations

LexisNexis28

Independent Variables Financial incentives Grants

0.09

0.29

0

1

Grants (Million US$)

0.83

4.95

0

40.2

Tax credits for individuals

0.05

0.22

0

1

271.21

1425.14

0

15000


0.06

0.24

0

1

Research and development

0.03

0.17

0

1


0.07

0.25

0

1


0.01

0.11

0

1


0.06

0.24

0

1

Planning

0.17

0.37

0

1

Incentives for state-owned electric vehicles fleets

0.20

0.40

0

1

1322.86

1817.68

105

10073

US Census Bureau34

Charging Stations

55.05

103.73

0

1180

Alternative Fuels Data Center35

PEV models availability (logged)

2.05

0.86

0.00

3.53

IHS Markit27

GDP (2010 USD Millions per capita)

0.05

0.01

0.03

0.08

Bureau of Economic Analysis36

Tax credits for individuals (US$)

Infrastructure incentives

Symbolic incentives

Instrumental Variable Grocery stores and supermarkets

Control Variables


Page 25 of 29


25

3.9

1.08

1.36

7.07

American Community Survey37

High school education (%)

29.01

2.95

23

38.23

American Community Survey37

C02 Transportation sector (% total emission)

35.78

11.7

11.44

60.17

Environment al Protection Agency38

99.77

U.S. Energy Information Administrati on39

91.14

National Conference of State Legislature40

69

Green Party Elections Database41

4.57


34.04


25.53

Federal Highway Administrati on43

Unemployed (%)

30.27

Renewable energy (% total energy)

48.09

Democrats in the legislature (%)

4.13

Green party victories

3.21

Gasoline price (USD)

Price of electricity (USD Cents/kilowatthour)

Licensed drivers (Millions)

10.42

4.26

23.13

17.31

8.99

0.56

3.87

4.43

500


0.92

13.33

0

2.2

6.08

0.41


Page 26 of 29

26

501 502

Table 3. The effect of plug-in vehicle incentives on plug-in electric vehicle registrations (20102016) IV-FEs Grants

2543.3824 (2803.8188)

Tax credit for individuals

1976.8545*** (585.3514)

IV-FEs

Grants (US$)

0.0013*** (0.0004)

Tax credit for individuals (US$)

0.1089* (0.0464)


5451.7439+ (2862.3421)

6080.6510 (3713.9809)

Research and Development

-505.6160 (855.6333)

-77.3595 (838.0104)


-4231.1034*** (1041.4808)

-3448.6789*** (905.1473)


-2666.7845 (2041.3897)

-2103.7550 (1903.5174)


-3470.3898+ (1917.8229)

-2883.3600 (1838.9848)

Planning

-2532.7464+ (1505.6860)

-2001.4547+ (1090.5485)

Incentives for state-owned PEVs fleets

2115.1354+ (1233.3533)

2148.7102+ (1210.0167)

Charging Stations

58.5555*** (10.2670)

51.6742*** (9.4720)


638.3047 (575.1312)

485.0872 (476.9756)

-41059.2966 (83039.4611)

-49382.4971 (77673.3857)

Unemployed (%)

608.1368 (514.4131)

764.5352 (501.0593)


429.9259+ (241.6774)

326.6275+ (195.6089)



Page 27 of 29


27

CO2 Transportation sector (% total emissions)

-34.1718 (31.0155)

-39.9799 (32.2509)

Renewable energy (%total energy)

7.2845 (16.2786)

6.1597 (15.0851)


8.9964 (28.0325)

19.8077 (26.5373)

Green Party Victories

52.0547 (55.7435)

24.0641 (37.8413)


2889.9101 (3139.7872)

1654.2782 (2699.9509)

Price of electricity (USD Cents/kilowatt-hour)

303.3240 (329.4809)

161.7831 (223.1537)


1606.5935 (1271.1941)

476.9607 (691.7363)

-32150.5230** -20751.7210* (12297.1416) (9833.2524) Observations 300 300 Clusters 50 50 State and Year fixed-effects Yes Yes First-stage F statistic 62.02 37.38 R2 within 0.7543 0.8063 Intra-class correlation 0.9452 0.9462 Cluster robust standard errors in parentheses: + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. Note: all independent variables are lagged one period except tax credits for individuals Constant

503 504



Page 28 of 29

28

505

Table 4. Mediating analysis: How charging stations affect PEV registrations? Charging Stations FEs

Registrations

Registrations

FEs

FEs-IV

Grants (Million US$)

0.0000*** (0.0000)

0.0028*** (0.0004)

0.0013*** (0.0004)

Tax credits for individuals (US$)

-0.0012 (0.0011)

0.0470 (0.0617)

0.1089* (0.0464)

EVs supply equipment financing

82.0947 (70.4815)

10346.3421 (7956.6310)

6080.6510 (3713.9809)

Research and Development

-17.2048 (25.4381)

-731.6256 (748.7901)

-77.3595 (838.0104)


59.7012* (26.7834)

-167.9650 (764.7739)

-3448.6789*** (905.1473)


19.2522 (64.5896)

-1790.4073 (1861.6065)

-2103.7550 (1903.5174)


62.4284* (25.8884)

684.4644 (922.5250)

-2883.3600 (1838.9848)

Planning

-6.1189 (24.0510)

-2207.0213 (1597.9537)

-2001.4547+ (1090.5485)

Incentives for state-owned PEVs fleets

-36.7757 (28.0414)

417.1600 (811.4772)

2148.7102+ (1210.0167) 51.6742*** (9.4720)

Charging Stations -27.7733+ (13.9910)

-1775.7621* (882.9396)

485.0872 (476.9756)

2836.8092 (2586.2965)

86478.2534 (79655.3885)

-49382.4971 (77673.3857)

Unemployed (%)

-23.3429* (10.6120)

-493.8797 (315.6300)

764.5352 (501.0593)


-7.3525* (3.1919)

-9.7749 (154.1134)

326.6275+ (195.6089)

CO2 Transportation sector

-0.3410 (0.6426)

-30.3822 (43.2512)

-39.9799 (32.2509)

Renewable energy (%total energy)

-0.5519

-12.7263

6.1597




Page 29 of 29


29

(0.4662)

(15.5262)

(15.0851)


0.8073+ (0.4778)

49.1744+ (29.0196)

19.8077 (26.5373)

Green Party Victories

-1.0197+ (0.5697)

-44.1037* (17.8213)

24.0641 (37.8413)


184.3636 (171.3989)

9926.6873 (6927.6988)

1654.2782 (2699.9509)

Price of electricity

3.3564 (4.7780)

251.5893 (318.0826)

161.7831 (223.1537)


85.4523 (71.6702)

4255.1217 (3338.6606)

476.9607 (691.7363)

-20751.7210* (9833.2524) Observations 300 Clusters 50 State and Year fixed-effects Yes First-stage F statistic 37.38 R2 within 0.7423 0.7073 0.8063 Intra-class correlation 0.9909 0.9926 0.9462 Cluster robust standard errors in parentheses: + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. Note: all independent variables are lagged one period except tax credits for individuals Constant

506 507

-736.0855 (775.4530) 300 50 Yes

508 509


-52766.2221 (35431.9394) 300 50 Yes

The Role of Public Policy in Technology Diffusion: The case of Plug-in

Recommend Documents