Content uploaded by Brett David Hishigata Williams
Author content
All content in this area was uploaded by Brett David Hishigata Williams on Dec 03, 2022
Content may be subject to copyright.
EVS35 International Electric Vehicle Symposium 1
35th International Electric Vehicle Symposium and Exhibition (EVS35)
Oslo, Norway, June 11-15, 2022
Targeting Incentives Cost Effectively: “Rebate Essential”
Consumers in the New York State Electric Vehicle Rebate
Program
Brett D.H. Williams1
1Center for Sustainable Energy; 3980 Sherman Street, Suite 170, San Diego CA 92110, USA; brett.williams@energycenter.org,
Summary
To increase the cost-effectiveness of electric vehicle (EV) incentives and outreach, this research examined
consumers who would not have purchased/leased their EV without New York State’s Drive Clean Rebate—or
“Rebate Essentials.” Using survey responses from 5,191 participants rebated for 2017–2019 adoption, it analyzed
consumers of plug-in hybrid EVs (PHEVs), Tesla battery EVs (BEVs), and non-Tesla BEVs separately. Weighted
descriptive statistics and logistic regressions identified factors that increase the odds of a consumer being Rebate
Essential, and dominance analysis rank-ordered factors for prioritization. Profiles generated for each vehicle
category summarize characteristics and describe top opportunities for reinforcing Rebate Essential adoption
through incentive design and outreach. Recommendations are provided. Among the factors discussed are: 1)
interest in EVs at the beginning of the car search, 2) rebate awareness before visiting the dealership, 3) other
perks for EVs, 4) having lower income, and 5) giving relatively lower importance to environmental impacts.
Keywords: incentive, market development, marketing, policy, state government
1 Background
The New York State Drive Clean Rebate Program (NY DCRP) offers up to $2,000 as a point-of-sale rebate for
the purchase or lease of a new EV [1]. This paper seeks to help target outreach and incentive design away from
free riders (who would buy an EV without the incentive) and toward consumers most highly influenced by
financial incentives to become “true additions” to the EV market. It does so by identifying and rank-ordering
characteristics that statistically distinguish past NY DCRP rebate recipients who would not have acquired their
vehicle without the rebate, or “Rebate Essential” program participants.
In contrast to prior examinations of EV purchase incentives in the literature that establish their importance to, or
magnitude of effect on, vehicle markets (e.g., [2,3]), this work characterizes incentive recipients. In contrast to
prior research characterizing recipients of EV purchase incentives generally (e.g., [4–7]), this work adds to the
more modest body of research characterizing specifically those who were most highly influenced to buy an EV
(e.g., [5,7]). Compared to [5]—which examined the influence of the U.S. federal tax credit on consumers, the
EVS35 International Electric Vehicle Symposium 2
majority of which acquired a Tesla Model S or Nissan LEAF in 2013—this work examines the influence of the
New York State rebate on a wide variety of EV consumers in 2017–2019. Compared to [7]—which analyzed
characteristics associated with data-determined clusters produced using latent class analysis—this work examines
a consumer segment of predetermined interest (Rebate Essentials), with the hope that amplifying such adoption
will increase incentive cost-effectiveness. Further discussion of the incentive literature and its findings relative
to another, similar examination by the authors of the federal tax credit is available in another EVS35 paper [8].
Prior study of Rebate Essentials specifically [9–11] examined participants in a post-purchase rebate program in
California who purchased or leased their EV in 2013–2017. This work is the first that examines Rebate Essential
consumers 1) outside of California, 2) who participated in a dealer-based, point-of-sale rebate program, and/or
3) that purchased or leased their EVs in 2017–2019. It also further develops the methodology and pushes the
results farther toward actionability than prior analyses.
2 Data and Representativeness
The analysis primarily utilized NY DCRP program data, described next. National Household Travel Survey data
[12] and vehicle registration data [13] were also used to provide context and baseline metrics. The program data
analyzed included application and survey data. Application data provided the basic characteristics of 21,843
rebated EVs purchased or leased from late March 2017 (the launch of NY DCRP) through the end of 2019. Data
from the program’s Adoption Survey—which all rebate recipients are invited to take within a few months of their
application approval— included a total of 5,474 survey responses from those 21,842 rebated adopters (25%).
Weights were generated using iterative proportional fitting (raking) to make the survey responses more precisely
represent the program population. After data processing for logistic regression, 5,191 of those survey responses
were used in vehicle-category-specific modelling (see Section 3).
Compared to 39,029 EVs sold in the state from April 2017 through the end of 2019 [14], these rebate recipients
amount to roughly 56% of the total New York EV market during the period examined. The top rebates by model
were for the Toyota Prius Prime (25%), Tesla Model 3 (18%), Honda Clarity Plug-in Hybrid (10%), Chevrolet
Volt (7%), Ford Fusion Energi (7%), and the Chevrolet Bolt (6%). It is notable that 61% of the DCRP rebates
studied were for plug-in hybrid electric vehicles (PHEVs). The high proportion of PHEVs draws a sharp contrast
to other markets like California where all-battery electric vehicles (BEVs) make up the majority of EV sales and
incentives issued. For example, from 24 March 2017 through 31 December 2019, only 32% of applications
received and approved for rebates by California’s Clean Vehicle Rebate Project (CVRP) were for PHEVs [15].
Further details about the data used can be found in a related project report [16].
3 Methodology
3.1 Separate Modelling for Each Vehicle Category
Following previous analysis of Rebate Essentials in California [9–11], consumers of PHEVs and BEVs were
examined separately to account for their unique qualities, including differing program participation rates [17,18],
demographic and housing characteristics [19,20], purchase motivations [19], and so forth. Indeed, rebated
consumers of these two distinct product types differ in their responses to a wide variety of the survey questions
that form the basis of characterizing them in California [21,22].
Similarly, Tesla products and consumers have been found to be distinct from other BEV products and consumers,
for example in their purchase motivations [19], vehicle pricing [23,24], charging requirements, etc. Further,
interpretation of previous demographic [9] and consumer-segmentation [25] analysis that examined all BEVs as
a single group has been complicated by “counter-currents” running beneath the surface of BEV averages and
program averages. These and other muddied findings are caused by conflating conflicting trends that differ
between Tesla and non-Tesla BEV consumers. Additionally, after years of disproportionate headlines, Tesla came
EVS35 International Electric Vehicle Symposium 3
to truly dominate the EV market in mid-2018 with the rollout of the Model 3, resulting in similar dominance of
rebate program funding [23]. For these and other reasons, Tesla is also treated as its own vehicle category in this
analysis, alongside non-Tesla BEVs (which need to be collectively grouped for sample size reasons) and PHEVs.
3.2 The Grouping/Dependent Variable: Rebate Essentiality
Survey respondents were asked, “Would you have purchased/leased your electric car without the State car rebate
(Drive Clean Rebate)?” Respondents who selected “No” are considered Rebate Essential.[9–11] This
dichotomous variable will serve as both a grouping variable in the descriptive results and as the dependent
variable in the binary logistic regressions. Data were filtered to remove cases where no response was received
for the dependent variable question (14 PHEV cases, 2 Tesla cases, 0 non-Tesla BEV cases). Eleven cases were
also removed where the respondent indicated they were from 16 to 20 years of age.
3.3 Independent Variable Selection and Preparation
Program data fields (e.g., survey questions) were selected for use as independent variables based upon theoretical
relevance and for anticipated “actionability” of the results. The total number of independent variables for each
vehicle category was capped based upon sample size (typically < 35 variables). Further details are available in a
related program report [16]. When less than 50 responses were available for a survey response option, or when
needed to reduce interpretation complexity, adjacent or conceptually similar response bins were combined.
3.4 Analysis
Following data cleaning, weighted descriptive statistics by vehicle category and Rebate Essential status were
obtained. Weighted descriptive statistics were used to summarize the data, characterize participants, test for
significant differences across vehicle categories and consumer segments, and compare groups to metrics
characterizing new-vehicle buyers in New York State.
Missing data are problematic for logistic regression, for which cases missing data in variables of interest are often
deleted. To keep case-wise deletion losses limited to less than 5% of the sample available for each vehicle-
category-specific model, multiple imputation was used for variables where more than 0.8% of cases were missing
data [26]. Twenty datasets were created with 20 iterations for each vehicle-category’s modelling.
After preparing the data for analysis (filtering, combining bins, and creating datasets with imputed missing
scores), a total of 5,191 survey respondents were analyzed. “Full” binary logistic regression models were
specified (using unweighted data) for each dataset. Full models utilized all independent variables to identify
factors that significantly contribute to predicting Rebate Essentiality while controlling for other variables. Full
models were then reduced to “parsimonious” models consisting of only significant factors. To facilitate
prioritization and comparison, the relative importance of factors was determined using dominance analysis.
Factors were rank-ordered by the average of their average explanatory contributions to the modelling for each
dataset (using Estrella’s pseudo-R2). Further methodological details can be found in an open-source journal article
that took a similar approach to analyzing a different consumer segment with a different dependent variable [25].
Significance and notable non-significance are discussed and summarized along with descriptive findings into
profiles specific to each of PHEV, Tesla, and non-Tesla BEV consumers. Caveats, recommendations, and other
concluding thoughts are provided.
4 Descriptive Findings About Rebate Essentials
Among the rebated consumers represented by the weighted survey results, 52% of PHEV consumers were found
to be Rebate Essential, as were 40% of Tesla consumers and 60% of non-Tesla BEV consumers (Figure 1). These
percentages are discussed further in Section 6.
EVS35 International Electric Vehicle Symposium 4
Comparing Rebate Essentials to their counterparts within their vehicle category (Table 1), descriptive statistics
highlight several differences: 1) Rebate Essential PHEV and non-Tesla BEV consumers are male more frequently
than their non-Rebate Essential counterparts, 2) PHEV and Tesla Rebate Essential participants tend to be younger,
and 3) PHEV Rebate Essentials more frequently rent and less frequently identify solely as white/Caucasian.
Overall, each Rebate Essential group tends to more closely resemble New York new-vehicle buyers than
non-Rebate Essentials do, based on metrics developed to characterize mainstream markets using NHTS data.
Notable exceptions include: Rebate Essentials are more frequently male, Tesla Rebate Essentials more frequently
own homes, and Tesla consumers as a whole already less frequently identify as solely white/Caucasian.
Figure 1: Percent of weighted responses indicating they would not have acquired their vehicle without the rebate.
Table 1: Weighted Descriptive Statistics Summary †
Characteristic
PHEV
(n = 2,766)
Tesla
(n = 1,430)
Non-Tesla BEV
(n = 995)
New-Vehicle
Buyers
Not Rebate
Essential
Rebate
Essential
(52%)
Not Rebate
Essential
Rebate
Essential
(40%)
Not Rebate
Essential
Rebate
Essential
(60%)
(NY
Responses,
2017 NHTS‡)
Selected only
white
88%**
83%**
71%
68%
87%
84%
74%
Greater than 40
years old
82%**
77%**
72%**
65%**
74%
71%
69%
Bachelor's
degree or more
75%
75%
84%
84%
82%*
77%*
64%
Own home
92%**
88%**
85%
88%
91%
88%
73%
≥ $100k HH
income
61%
61%
85%
84%
66%
60%
53%
Selected female
37%**
30%**
18%
15%
30%**
24%**
51%
† Percentages are weighted to represent the program population along the dimensions of technology type (PHEV vs. BEV), model,
purchase vs. lease, and residence county. ‡ NHTS 2017 is weighted to represent population, not new-vehicle subset. New-vehicle
buyers identified based on within-100-mile match between odometer and miles driven while owned.
* p < 0.10, ** p < 0.05: two-sample test (with continuity correction) for equality of proportions between Rebate Essential and Not
Rebate Essential segments.
EVS35 International Electric Vehicle Symposium 5
5 Logistic Modelling Findings About Rebate Essentials
Table 2 provides the factors found to significantly increase the odds of being Rebate Essential for each vehicle
category. To facilitate prioritization of strategic efforts, results are rank-ordered by importance using dominance
analysis (see Section 3) and sorted into “high,” “medium,” and “low” importance groups. To facilitate comparison,
factors common across vehicle categories are color-coded.
The results indicate that factors such as consumer awareness of the rebate before the first dealership visit and
having some initial interest in an EV significantly increased the odds of being Rebate Essential across vehicle
categories. Rating other incentives important, both financial and convenience-related, also appears to be highly
related to Rebate Essentiality. PHEV and non-Tesla BEV Rebate Essentials also shared as significant factors
dealer awareness of the rebate on the first visit, the importance of free charging away from home, lower income,
and male gender. Giving lower importance to environmental impacts was found to increase the odds of being
Rebate Essential for Tesla consumers, as well as for PHEV consumers in alternative PHEV modelling.
Table 2: Summary of Rank-Ordered Factors that Increase the Odds of Being Rebate Essential
PHEV
Tesla
Non-Tesla BEV
“High-ranked” [> 0.01]
1. Initial interest in an EV is some or
very interested (vs. no knowledge or
interest) [0.037]
1. Green Pass or similar toll/E-ZPass
discounts are more important (vs. not
important) [0.04]
1. Consumer aware of the rebate
before visiting a dealership (vs.
not aware) [0.029]
2. Consumer aware of the rebate
before visiting the dealership [0.026]
2. Consumer aware of the rebate before
visiting a dealership (vs. not aware) [0.038]
2. Free charging away from
home is very or extremely
important (vs. not important)
[0.027]
3. Age 40–49 (vs. 21–29) [0.014]
3. Special EV electricity rates are
extremely important or not applicable (vs.
not important) [0.023]
3. Initial interest in an EV is
some interest or very interested
(vs. no knowledge or no
interest) [0.027]
4. Special EV electricity rates are
extremely important or n.a. (vs. not
important) [0.0119]
4. Saving money on fuel is more important
(vs. not important) [0.012]
4. Lower household income (vs.
higher incomes) [0.023]
5. Free charging away from home is
extremely important or n.a. (vs. not at
all important) [0.0118]
5. Vehicle make is not
Chevrolet (Nissan or other
makes) [0.012]
6. Slightly satisfied with the rebate
amount (vs. not at all satisfied) [0.011]
6. Dealer aware of rebate on
first visit (vs. not aware or don't
know) [0.01]
“Medium-ranked” [> 0.005]
7. Lower household income (vs. higher
income) [0.008]
5 (tied). Race/ethnicity is white/Caucasian,
relative to non-Latinx, non-Asian other
selections (individual or multiple) [0.009]
7 (tied). Male (vs. female)
[0.006]
8. Dealer is extremely knowledgeable
about home charging (vs. very
knowledgeable) [0.006]
5 (tied). Initial interest in an EV is some
or very interested (vs. no knowledge or
interest) [0.009]
7 (tied). Number of cars in
household - multiple cars (vs. 1
car) [0.006]
9. EV is an additional vehicle (vs.
replacement) [0.005]
7. Reducing environ. impact is slightly or
not important (vs. more important) [0.006]
EVS35 International Electric Vehicle Symposium 6
8. Access to the latest technology is
slightly or not important (vs. very or
extremely ) [0.005]
“Low-ranked” [< 0.005]
10. Dealer aware of rebate on first visit
(vs. not aware or don't know) [0.0048]
9. Vehicle performance is moderately,
slightly, or not important (vs. very or
extremely important) [0.003]
11. Male (vs. female) [0.0044]
12. HOV lane access is slightly
important (vs. not at all ) [0.0038]
13. Bachelor's degree or postgraduate
(vs. high school or other) [0.0024]
14. Rent residence (vs. own ) [0.0022]
Notably not significant were residence type (e.g., multi-unit vs. single family) and access to charging. If
rebates help compensate for the complexities of charging in multi-unit dwellings (MUDs), residence type might
be expected to be significant. It was not for any consumer group. Similarly, access to charging at home or near
work was not found to be significantly associated with Rebate Essentiality. This could be partially explained by
the significance of rent-vs.-own status for PHEV consumers, and the possibility that rebated EVs have not
penetrated deeply enough into MUDs in general for sufficient variation to exist yet in the data.
More broadly, few factors categorized as “household” factors were significant. This puts a greater emphasis
on demographics (particularly for PHEV consumers) and motivational findings (particularly for Tesla
consumers). On the other hand, most demographic factors were not significant for Tesla consumers, nor
were many motivational factors significant for non-Tesla BEV consumers.
6 Summary of Findings into Vehicle-Category Profiles
Profiles specific to each vehicle category (PHEV, Tesla, and non-Tesla BEV consumers) were generated. Each
begin by setting the stage with a summary of additional information characterizing the consumers in the program
dataset as a whole, for example the portion of rebates received by the category, their vehicle replacement
behaviours, and which adoption-enabling factors they most frequently rated extremely important. This is
followed by vehicle-category-specific breakdowns of the highly ranked predictors of Rebate Essentiality from
Table 2. With this information, the cost-effectiveness of supporting each vehicle category can then be reinforced
by either targeting potential consumers with similar characteristics or increasing the prevalence of certain
characteristics (e.g., awareness of the rebate) in order to “unlock” additional similar consumers who might
otherwise be highly influenced by the rebate. Descriptive characteristics help us better understand what type of
consumer was most highly influenced by the Drive Clean Rebate to purchase or lease an electric vehicle. The
results from Table 2, in turn, help us rank-order the most robustly significant distinguishing factors and tell us
where to focus first and most. Cumulatively, these findings inform the best guess of how to avoid program free
riders and target additional consumers that the rebate has the greatest chance of influencing, thereby increasing
program cost-effectiveness.
6.1 PHEV Rebate Essentials Summary Profile
Rebates received for PHEVs constitute 61% of the rebated vehicles analyzed, nearly double the share received
in California over a similar time frame (Section 2), making PHEV rebates an important consideration. Nearly 90%
of rebated PHEVs replaced a household car, the highest replacement rate among vehicle categories, making
EVS35 International Electric Vehicle Symposium 7
PHEV rebates impactful (e.g., from the perspective of greenhouse gases avoided). Over half (52%) of PHEV
rebates were given to PHEV Rebate Essential consumers in New York State (Figure 1) higher than the 40% of
rebates given for Tesla vehicles and lower than the 60% of rebates for non-Tesla BEVs.
Compared to other vehicle categories, PHEV consumers most frequently rated carpool-lane access and the Green
Pass or similar toll discounts as extremely important, tend to be older, more frequently female, and/or less
frequently have a college degree. The frequency of rating environmental impacts as extremely important is the
lowest for PHEV consumers (58%), as is the presence of residential solar, initial interest in an EV at the beginning
of the car search, and awareness of the rebate prior to visiting a dealership.
Based on Table 2, the most impactful and straightforward ways to reinforce the cost-effectiveness of PHEV
rebates include targeting potential PHEV consumers who already have one or more of the following
characteristics (in order of decreasing importance). Alternatively, measures that increase the prevalence of some
of the following characteristics (e.g., rebate awareness) may “unlock” consumers who would otherwise be highly
influenced by the rebate. PHEV Rebate Essentials tend to:
1. Have some (32%) or a lot (56%) of interest in EVs at the beginning of their car search.
2. Be aware of the rebate before visiting a dealer (55%).
3. Be in their 40s (22%) rather than their 20s (6%).
4. Rate electricity rates for charging either extremely important (22%) or not applicable (21%), rather than
not at all important (18%).
5. Rate free charging away from home either extremely important (28%) or not applicable (9%), rather
than not at all important (13%).
6. Be satisfied with the rebate amount (that ranged $500–1,700), rather than not (2%), particularly those
slightly satisfied (10%).
Significant distinguishing factors for PHEV Rebate Essentials tend to be related to demographics and not
household characteristics (e.g., housing type and access to charging at home). Unlike in California, the
importance of saving money on fuel and the absence of residential solar were not distinguishing factors for New
York PHEV Rebate Essentials.
Several other factors with modest contributions are listed detailed in Table 2. These may reinforce, if somewhat
weakly, the pertinence of, for example, targeting lower-income renters with college degrees.
Overall, rebated PHEV consumers have the lowest relative levels of initial interest in EVs and pre-dealership
awareness of the rebate among the vehicle categories. However, these two factors rank at the top of the list of
factors associated with increased odds of being a Rebate Essential PHEV consumer. A substantial opportunity
would appear to exist to increase PHEV cost-effectiveness and market adoption through education and outreach
about the benefits of PHEVs and availability of the rebate.
6.2 Tesla Rebate Essentials Summary Profile
Rebates received for Teslas constitute 24% of the rebated vehicles analyzed, making Tesla products the most
prominent brand rebated. Eighty-one percent of rebated Teslas replaced a household car, midway between PHEVs
(87%) and non-Tesla BEVs (75%), perhaps indicating the greater confidence in, and impact of, long-range BEVs.
Further, Teslas replaced gasoline vehicles the most frequently (85% vs. 83% for PHEVs and 74% for non-Tesla
BEVs). On the flip side, Teslas more frequently replace newer vehicles, and their confidence-inspiring high-
performance and less-compromised products may be reflected in Tesla consumers reporting being Rebate
Essential less frequently than the other vehicle categories (40% vs. 52% for PHEV consumers and 60% for non-
Tesla consumers, Figure 1).
Overall, in comparison to PHEV and non-Tesla consumers, Tesla consumers most frequently were very interested
in EVs at the start of their car search (80%) and their Tesla was their first EV (81%), indicating the brand may be
pre-converting consumers to EV interest and adoption. Tesla consumers also most frequently rated extremely
EVS35 International Electric Vehicle Symposium 8
important: the convenience of charging, vehicle performance, vehicle styling and comfort, the desire for new
technology, parking incentives, and special electricity rates for charging. They also most frequently use Level 2
charging at home (73%), but least frequently live in a single-family home (74%) and least frequently own their
residence (86%). Notably, Tesla Rebate Essentials more frequently are homeowners, making them less like new-
car buyers on this one dimension, but bringing them to equivalence with the frequency of home ownership among
PHEV and non-Tesla BEV Rebate Essentials (88%, see Table 1). They most frequently identify as male (83%)
and have annual gross household incomes greater than $150,000 (66%) but are least frequently white (70%—
lower than even new-vehicle buyers in general, per Table 1).
Based on Table 2, the most impactful and straightforward ways to reinforce the cost-effectiveness of Tesla rebates
include targeting potential Tesla consumers who already have one or more of the following characteristics (in
order of decreasing importance). Alternatively, measures that increase the prevalence of some of the following
characteristics (e.g., rebate awareness) may “unlock” consumers who would otherwise be highly influenced by
the rebate. Tesla Rebate Essentials tend to:
1. Rate the Green Pass or similar toll/E-ZPass discounts important (79%) vs. not at all important (13%).
2. Be aware of the rebate before visiting a dealer (78%).
3. Rate electricity rates for charging either extremely important (28%) or not applicable (18%) rather than
not at all important (13%).
4. Rate saving money on fuel costs moderately to extremely important (90%) rather than slightly or not
important (10%).
Significant distinguishing factors for Tesla Rebate Essentials tend to be related to motivations and not
demographics or household characteristics (e.g., housing type and access to charging at home). Notably, income
was not a distinguishing factor for New York Tesla Rebate Essentials.
Several other factors with modest contributions are detailed in Table 2. These may reinforce, if somewhat weakly,
the pertinence of targeting potential Tesla consumers placing lower importance on reducing environmental
impacts, the latest technology, and vehicle performance.
Although not a significant predictor of segment status, Tesla Rebate Essentials are typically the youngest of the
Rebate Essentials among the vehicle categories. For example, they had the highest percentage under 50 years old
and the highest percentage in each of the 20-, 30-, and 40-year-old buckets.
6.3 Non-Tesla Rebate Essentials Summary Profile
Rebates received for non-Tesla BEVs constitute the smallest portion of the rebated vehicles analyzed (15%).
Non-Tesla BEV consumers are most-frequently Rebate Essential (60%, Figure 1) and aware of the rebate before
visiting a dealership (70%). Perhaps because of historically more-compromised/lower-performance vehicles,
non-Tesla vehicles least frequently: replaced a household car (75%) and replaced gasoline vehicles (74%).
Overall, in comparison to PHEV and Tesla consumers, non-Tesla consumers most frequently rated extremely
important reducing environmental impacts (65%), the state rebate (53%), and the federal tax credit (56%). They
least frequently rated extremely important carpool-lane access (7%), vehicle performance (26%), vehicle styling
and comfort (14%), and the Green Pass or similar toll/E-ZPass discounts (13%).
They also most frequently had workplace charging (28%) and residential solar (22%).
Based on Table 2, the most impactful and straightforward ways to reinforce the cost-effectiveness of non-Tesla
BEV rebates include targeting potential consumers who already have one or more of the following characteristics
(in order of decreasing importance). Alternatively, measures that increase the prevalence of some of the following
characteristics (e.g., rebate awareness) may “unlock” consumers who would otherwise be highly influenced by
the rebate. Non-Tesla BEV Rebate Essentials tend to:
1. Be aware of the rebate before visiting a dealer (76%).
EVS35 International Electric Vehicle Symposium 9
2. Rate free charging away from home either extremely important (29%) or very important (22%) rather
than not at all important (9%).
3. Have some (22%) or a lot (66%) of interest in EVs at the beginning of their car search.
4. Have lower annual gross household income.
5. Acquire non-Chevrolet makes (67%).
6. Find their dealer was aware of the rebate on their first visit (83%).
Significant distinguishing factors for this group tend not to be related to household factors (e.g., housing type
and access to charging at home) or motivations. Unlike in California, the importance of environmental impacts
was not a distinguishing factor for New York non-Tesla BEV Rebate Essentials.
Two other factors with modest contributions are listed in Table 2. These may reinforce, if somewhat weakly, the
pertinence of targeting potential BEV consumers that are male or have multiple cars in the household.
7 Recommendations & Concluding Thoughts
To supplement the summary profiles in Section 6, and in an effort to push the findings of this research toward
“actionable” recommendations for outreach and incentive design, six additional recommendations are provided,
each after a brief description of the findings from which they stem.
Findings Summary: The odds of being Rebate Essential was not found to be associated with whether the rebated
EV was the consumer’s first EV or not.
Conclusion/Recommendation 1: No evidence was found to support limiting the number of rebates per
individual on the basis of rebate influence. Reassess over time as the market matures.
Findings Summary: Rebate Essentiality was associated with importance given to a variety of other financial
and convenience-based incentives.
Recommendation 2: Support or advertise other incentive programs (e.g., free charging, toll discounts, EV
charging rates) that reinforce the influence of the rebate.
Findings Summary: Descriptively, Rebate Essentials trend relatively younger and lower-income and rent
housing somewhat more frequently than non-Rebate Essentials (Table 1). PHEV Rebate Essentials specifically
identify less frequently as white than their counterparts. Predictively, Rebate Essentials are most highly
distinguished by having initial interest in EVs at the start of the new-car search and by consumer and dealer
awareness of the rebate before their first dealership visit (Table 2).
Recommendation 3: Use these characteristics and others provided in the detailed summary profiles in
Section 6 to target outreach and incentive design toward consumers with Rebate Essential characteristics.
This will increase the odds of reinforcing and amplifying adoption by those most highly influenced by
supportive resources to adopt an EV and minimize free ridership.
Recommendation: Identify ways to increase the prevalence of one or more characteristics associated with
Rebate Essentiality (see Section 6), in order to unlock more Rebate Essentials. For example, findings support
the need for rebate awareness campaigns (see next).
Findings Summary: To date, the rebate has been most influential as a tool for bringing people into the EV
market when presented to consumers during the information gathering phase prior to a dealership visit, rather
than as a sales tool at the dealership.
Recommendation 4: Support consumer awareness of the rebate during the information gathering phase
(especially for potential consumers of PHEV products, who have the lowest awareness).
Findings Summary: However, dealer awareness is also a (lower-ranked) significant predictor.
EVS35 International Electric Vehicle Symposium 10
Recommendation 5: Support rebate awareness among dealers, who may act as a “backstop” and either
reinforce consumer awareness or use the incentive to convert non-aware consumers into EV adopters.
Findings Summary: The goal of increasing cost-effectiveness both overlaps with, and has trade-offs with, the
goal of expanding EV markets more toward the mainstream. As described in a related project report [16], a
disproportionate amount of Rebate Essentials are also EV Converts—consumers with low or no initial interest at
the beginning of their car search [9,25]. There is particular overlap among those consumers with “some” initial
interest in EVs, which comprise 76% of EV Converts and 28% of Rebate Essentials, compared to 23% of the
program overall. (Those “very interested,” by definition, are not EV Converts, and those having no initial interest
were found to have lower odds of being Rebate Essential.) An even larger percentage of EV Converts are also
Rebate Essentials: 61% (vs. 51% for the program overall).
Recommendation 6: Findings indicate an opportunity to improve program cost-effectiveness by seeking out
consumers with lower initial interest in EVs and “converting” them to EV interest with the incentive and other
means.
Further discussion of the evidence that rebates are converting consumers into EV adopters, as well as other ways
to expand EV market frontiers outside of the enthusiastic core of early adopters, is currently in the process of
publication. In the meantime, additional details are available in a related project report [16]. It discusses the
integration of various consumer segments, including Rebate Essentials and EV Converts, into a roadmap of sorts
aimed at progressing EVs toward the mainstream and beyond to equitable access to transformative transportation
electrification.
8 Caveats & Next Steps
This work is centred on consumers who overcame their barriers to adoption, purchased/leased an EV, and
participated in the NY DCRP. Extrapolating these findings should be done with caution. Additional research is
required to understand consumers highly influenced by the incentive who did not overcome other barriers to
acquiring an EV or chose not to acquire one.
Even within the focus of the research, the range of topics explored is limited by sample size, which effects the
number of independent variables that can be effectively explored per vehicle category, and the availability of data
characterizing any given topic of interest. Although the NY DCRP Adoption Survey, which is summarized in
program reporting on the program website [1], is an extremely rich source of options, additional topics can of
course be of interest and relevance. One example that proved significant in related work in California was the
association found between Rebate Essentiality and lower vehicle price. This finding is particularly relevant in
the context of equity and incentive design features—such as the NY DCRP’s MSRP-based rebate amount or other
programs’ MSRP-based rebate eligibility (aka “MSRP caps”). These features cost-effectively support the volume
production of affordable new EVs—and, subsequently, affordable used EVs. Data characterizing the MSRP of
rebated vehicles was not readily available for this analysis but could be generated for follow-on analysis.
From a more technical modelling perspective, additional analysis could further examine the relationship between
PHEV Rebate Essentiality and the importance of reducing environmental impacts, which was problematic due
to multicollinearity. For example, additional modelling could use moderator variables or a variable that combines
a variety of social factors (e.g., environmental impacts and energy independence) into a single factor.
Acknowledgements
Particular thanks and a deep debt of gratitude are due to John Anderson for his indefatigable efforts in support of
this work and as a core author on nearly all the related research and project reports upon which it is built. His
contributions—to methodology, software, validation, formal analysis, investigation, data curation, visualization,
etc.—more than just laid the foundations for this paper, they are integral to it and the overall effort.
EVS35 International Electric Vehicle Symposium 11
Thanks are also due to CSE’s Eric Fullenkamp for analytical and codebase development support, Francis Alverez
for quality control, James Tamerius for methodological consultation, Keir Havel for programming support, and
John Gartner for editorial review and managerial support. We also thank Adam Ruder and David McCabe of
NYSERDA for their oversight of the research contract (66267) that funded the work, and for their tremendous
flexibility. We are thankful for the exciting opportunity to explore and learn from data generated by the Drive
Clean Rebate Program. However, any opinions expressed or mistakes remaining herein are those of the author.
References
[1] NYSERDA, Drive Clean Rebate for Electric Cars, (n.d.). https://www.nyserda.ny.gov/All Programs/Programs/Drive
Clean Rebate (accessed April 5, 2021).
[2] A. Jenn, K. Springel, A.R. Gopal, Effectiveness of electric vehicle incentives in the United States, Energy Poli cy.
119 (2018) 349–356. https://doi.org/10.1016/J.ENPOL.2018.04.065.
[3] E. Narassimhan, C. Johnson, The role of demand-side incentives and charging infrastructure on plug-in electric
vehicle adoption: analysis of US States, Environmental Research Letters. 13 (2018) 074032.
https://doi.org/10.1088/1748-9326/AAD0F8.
[4] S. Borenstein, L.W. Davis, The distributional effects of US clean energy tax credits, Tax Policy and the Economy.
30 (2016) 191–234. https://doi.org/10.1086/685597.
[5] G. Tal, M. Nicholas, Exploring the impact of the federal tax credit on the plug-in vehicle market, Transportation
Research Record. 2572 (2016) 95–102. https://doi.org/10.3141/2572-11.
[6] M.F. Sherlock, The Plug-In Electric Vehicle Tax Credit, Congressional Research Service, 2019.
https://crsreports.congress.gov (accessed February 28, 2021).
[7] A. Jenn, J.H. Lee, S. Hardman, G. Tal, An in-depth examination of electric vehicle incentives: Consumer
heterogeneity and changing response over time, Transportation Research Part A: Policy and Practice. 132 (2020)
97–109. https://doi.org/10.1016/j.tra.2019.11.004.
[8] B.D.H. Williams, J.B. Anderson, Lessons Learned About Electric Vehicle Consumers Who Found the U.S. Federal
Tax Credit Extremely Important in Enabling Their Purchase, in: 35th International Electric Vehicle Symposium and
Exhibition (EVS35), AVERE, Oslo, Norway, 2022.
[9] B.D. Williams, C. Johnson, Presentation: “Characterizing California Electric Vehicle Consumer Segments,” in:
BECC Conference, ACEEE, UC Berkeley CIEE, and SEEPAC, Baltimore, 2016.
https://doi.org/10.13140/RG.2.2.29388.13444.
[10] C. Johnson, B.D. Williams, Characterizing Plug-In Hybrid Electric Vehicle Consumers Most Influenced by
California’s Electric Vehicle Rebate, Transportation Research Record. 2628 (2017) 23–31.
https://doi.org/10.3141/2628-03.
[11] B.D. Williams, J.B. Anderson, Strategically Targeting Plug-in Electric Vehicle Rebates and Outreach Using
Characteristics of 'Rebate-Essential” Consumers in 2016–2017, in: The 31st International Electric Vehicle
Symposium, Society of Automotive Engineers of Japan, Inc., Kobe, Japan, 2018.
https://energycenter.org/sites/default/files/docs/nav/resources/EVS31_TargetingRebateEssentialConsumers_revise
d.pdf.
[12] FHWA, 2017 National Household Travel Survey, (2017). https://nhts.ornl.gov/ (accessed April 15, 2021).
[13] IHS Markit, Data License: New York State New Vehicle Registrations January 2016 – January 2021, (2021).
[14] CSE, AAI, Electric Vehicle Sales Dashboard, Alliance for Automotive Innovation, 2021.
https://www.autosinnovate.org/resources/electric-vehicle-sales-dashboard (accessed April 4, 2021).
[15] CSE, CVRP Rebate Statistics | Clean Vehicle Rebate Project, California Air Resources Board (CARB), 2021.
https://cleanvehiclerebate.org/eng/rebate-statistics (accessed April 4, 2021).
EVS35 International Electric Vehicle Symposium 12
[16] B.D.H. Williams, An Electric-Vehicle Consumer Segmentation Roadmap: Strategically Amplifying Participation in
the New York Drive Clean Rebate Program, 2021. https://www.nyserda.ny.gov/About/Publications/Research-and-
Development-Technical-Reports/Transportation-Reports (accessed March 16, 2022).
[17] B.D. Williams, J.B. Anderson, C. Santulli, G. Arreola, Clean Vehicle Rebate Project Participation Rates: The First
Five Years, Clean Vehicle Rebate Project (CVRP), San Diego CA, 2015.
[18] J. Tamerius, Clean Vehicle Rebate Project: Percent of Eligible-Model Vehicles Rebated Before and After Income-
Based Consumer Eligibility, Clean Vehicle Rebate Project (CVRP), San Diego CA, 2020.
[19] C. Santulli, B.D. Williams, Presentation: “Implementation Status Update” | Clean Vehicle Rebate Project, in: CVRP
Long-Term Planning Workshop, California Air Resources Board (CARB), Sacramento CA, 2015.
https://cleanvehiclerebate.org/en/content/implementation-update-dec-2015
[20] CSE, CVRP EV Consumer Survey Dashboard | Clean Vehicle Rebate Project, 2019.
https://cleanvehiclerebate.org/eng/survey-dashboard/ev (accessed February 4, 2020).
[21] C. Johnson, B.D. Williams, C. Hsu, J.B. Anderson, Summary Documentation of the Electric Vehicle Consumer
Survey, 2013–2015 Edition | Clean Vehicle Rebate Project, Center for Sustainable Energy (CSE), San Diego CA,
2017. https://cleanvehiclerebate.org/eng/content/summary-documentation-electric-vehicle-consumer-survey-2013-
2015-edition (accessed February 4, 2020).
[22] B.D. Williams, J. Orose, M. Jones, J.B. Anderson, Summary of Disadvantaged Community Responses to the Electric
Vehicle Consumer Survey, 2013–2015 Edition | Clean Vehicle Rebate Project, Center for Sustainable Energy (CSE),
San Diego CA, 2018. https://cleanvehiclerebate.org/eng/content/summary-disadvantaged-community-responses-
electric-vehicle-consumer-survey-2013–2015-edition (accessed June 30, 2020).
[23] J.B. Anderson, B.D.H. Williams, Presentation: “Proposed FY 2019–20 Funding Plan: Final CVRP Supporting
Analysis,” Clean Vehicle Rebate Project. (2019). https://cleanvehiclerebate.org/eng/content/proposed-fy-2019–20-
funding-plan-final-cvrp-supporting-analysis (accessed February 5, 2020).
[24] B.D.H. Williams, Presentation: “EV Purchase Incentives: Program Design, Outputs, and Outcomes of Four
Statewide Programs with a Focus on Massachusetts,” in: Behavior, Energy, and Climate Change Conference (BECC),
ACEEE, UC Berkeley CIEE, and SEEPAC, 2020. https://beccconference.org/wp-content/uploads/2020/12/Multi-
state-EV-rebate-Impacts-Brett-Williams_2.pdf (accessed February 24, 2021).
[25] B.D.H. Williams, J.B. Anderson, Strategically Targeting Plug-In Electric Vehicle Rebates and Outreach Using “EV
Convert” Characteristics, Energies (Basel). 14 (2021) 1899. https://doi.org/10.3390/en14071899.
[26] S. van Buuren, K. Groothuis-Oudshoorn, mice: Multivariate imputation by chained equations in R, Journal of
Statistical Software. 45 (2011) 1–67. https://doi.org/10.18637/jss.v045.i03.
Author
Brett Williams, MPhil (cantab), Ph.D., is Principal Advisor for Electric Vehicle Programs at CSE. He is a
point person for EV market and policy analysis, stakeholder engagement, and program design, strategy,
and evaluation. Previously, he was Assistant Adjunct Professor of Public Policy at UCLA, postdoctoral
scholar at UC Berkeley, and a researcher for Amory Lovins at RMI. Brett has a Ph.D. in Transportation
Technology & Policy from UC Davis, a master’s in Environment & Development from Cambridge
University (UK), and an undergraduate degree in Physics/Public Policy Analysis from Pomona College.
With particular thanks to:
John Anderson is a Senior Manager on the Transparency & Insights team at the Center for Sustainable
Energy. He has more than nine years’ experience working in EV markets and in myriad roles for
California’s Clean Vehicle Rebate Project. John’s role includes incentive program design and planning
activities, market and program projections, and analyses to inform implementation and outreach strategy.
John has a B.A. in International Security and Conflict Resolution from San Diego State University.
