Eco-Friendly Car Policy: Targeted and Spillover Effects
Chenxi LI, Xueming LUO, Yiping SONG and Chee Wei PHANG
Version Sept 2014
Chenxi Li is PhD Candidate, Fudan University, chenxili13@fudan.edu.cn. Xueming Luo is the Charles Gilliland
Distinguished Professor of Marketing, Fox School of Business, Temple University, luoxm@temple.edu; Yiping
Song is Associate Professor, Fudan University, Marketing Department, ypsong@fudan.edu.cn; Chee Wei Phang is
Associate Professor, Fudan University, phangcw@fudan.edu.cn.

Developments in the automobile industry can be either a boon or a bane. On one hand, they can spur
economic growth and fuel employment (Miravete and Moral 2009). On the other hand, they can produce
environmental pollution from greenhouse gas emissions (IEA 2011). This double-edged sword may be
particularly sharp in developing countries such as China and India. In China, vehicular emission pollution is
the main source of dust-haze (Vehicle Pollution Prevention Annual Report 2013).1
Thus, to bolster the automobile industry while stemming environmental side-effects, governments
worldwide have developed various institutional policies. For instance, in 2009 the U.S. implemented the
“cash for clunkers” program, formally known as the Cars Allowance Rebate System. The program offered a
rebate of $3,500 - $4,500 for each old and low fuel-efficient vehicle that consumers traded in for a newer,
more fuel-efficient one (Knittel 2009). While intended to reduce pollution and stimulate sales of more
fuel-efficient cars, the policy garnered doubts about its effectiveness. Some researchers found that purchases
stimulated by the policy were offset by a dip in auto sales further down the road (Mian and Sufi 2012), and
others found the program to be too costly for tax payers (Knittel 2009). Thus, for institutional economies to
effectively achieve intended goals is a challenge.
To overcome this challenge, the present study evaluates the impact of an alternative institutional policy
launched in China. The Chinese automobile industry plays a critical role in the global economy as the
world’s largest automobile producer and consumption market. However, the pollution caused in China by
automobiles is enormous. To help curb this pollution, in 2009 China launched the Tax-reduction for
Low-emission Vehicles (TLV) policy, which reduced the purchase tax for cars with small engines for two
years, from 2009 through 2010. 2 The implementation of TLV was intended to bloom the market of

1

Source: http://www.vecc-mep.org.cn/news/e_home_gg/1390967956277.html; http://www.china-embassy.org/eng/xw/t650869.htm,
last accessed on June 17, 2014.
2
In China, three categories of automobile taxes, including the consumption tax, the value-added tax and the purchase tax, are
charged by the government. Among these three taxes, the purchase tax is the only tax paid by consumers directly, while the
consumption tax and value-added tax will be paid by manufacturers and retailers respectively. TLV reduced the purchase tax of
1

small-engine cars, of which the market share dramatically decreased from 2004 to 2007. The proposal of
TLV policy was passed on January 14th, 2009 and implemented on January 20th, 2009.
TLV differed from “cash for clunkers” in two key ways: first, in the “cash for clunkers” program the
federal government provided a direct incentive via monetary subsidy (up to 4,500 USD) for every qualified
replacement of old vehicles, implying a high pay-out for the policy-maker. As Tyrell and Dernbach (2010 p.
491) note, “the hard reality about programs that depend on direct public subsidies [such as “cash for
clunkers”] is that they are difficult to support over time.” In contrast, the TLV operated by allowing
consumers to pay a lower tax to the government if they purchased a small-engine car (i.e., 1.6-liter and
below). Thus, a large initial fund allocation was not required for this program, which also means that it may
be more financially feasible for developing countries. Second, TLV represented policy measures that
attempted to influence consumers’ choice and decision, i.e., on whether to buy a smaller-engine car to enjoy
a tax reduction. This implies that all consumers who intended to buy a new car would be affected by the
policy. In contrast, the “cash for clunkers” was only eligible for those who owned an old car that met the
program criteria (Busse et al. 2012). Thus, in the context of a developing country in China, where most new
car purchases are made by first-time buyers, TLV policy may have a larger impact.
Our identification hinges on the changes in car sales before, during, and after the TLV policy. In terms
of environmental aspects, we assess the change in sales of more environmental-friendly cars with
small-engines ( ≤ 1.6-liter) vis-à-vis less environmental-friendly cars with big-engine (>1.6-liter). Our
analyses reveal that the TLV produced three key results: 1) the policy indeed boosted sales of small-engine
cars (a positive direct effect or targeted effect); interestingly, it also simultaneously dampened sales of
big-engine cars (a negative spill-over effect or untargeted surprising effect). 2) Deeper analyses document
that the positive direct effect of the TLV policy was driven by sales in smaller rural cities with slower
economic and infrastructural developments. 3) The negative spillover effect of the TLV policy was attributed

to sales in larger metropolitan cities with faster economic and infrastructural developments. The dip in sales
occurred mostly for non-indigenous car brands, while indigenous car brands experienced increased sales.
Recently, research has addressed how car policies may curb pollution or “automobile externalities”
(Parry et al. 2007; Porter 1999). A variety of car policies have been considered to limit automobile ownership,
small-engine cars (1.6 liter and below) from 10% to 5% in 2009. The policy continued implementing in 2010 with a little revision,
the reduced tax rate changed to 7.5%, from January to December 2010. After the two-year TLV policy period, the purchase tax rate
for small-engine cars turned back to 10% in 2011. The purchase tax for big-engine (>1.6-liter) cars is 10% all the time.
2

reduce automobile use, and promote more eco-friendly vehicles. Policies limiting ownership involve bidding
and lottery mechanisms implemented in China (Chen and Zhao 2013; Hao et al. 2011).3 Policies that reduce
automobile use involve raising fuel taxes (Parry and Small 2005; Bento et al 2009) and prescribing the use
of odd- or even-numbered car plates on the road in some cities such as Beijing (Cai and Xie 2011). Policies
changing automobile use patterns involve charging higher toll fees during times and places when road
systems are congested (to encourage people to travel during off-peak hours or via less congested areas). Such
measures may reduce vehicular emissions because they help minimize trip durations and decrease speed
variations (Daniel and Bekka 2000). Though effective to varying extents, these policies often achieve their
environmental purposes at the expense of the automobile industry’s development, i.e., by discouraging car
purchases due to the high costs of purchase or usage. Some policies that promote eco-friendly vehicles
specifically target automobile makers or consumers’ cars. Examples include imposing requirements to meet
vehicle manufacturing standards such as the Corporate Average Fuel Economy (CAFE) program in the U.S.
and the system for controlling emissions of new vehicles in the E.U. Policies that target consumers include
the “cash for clunkers” program to address environmental problems and stimulate the automobile industry
(Mian and Sufi 2012).4
A series of studies have been conducted to assess the effectiveness of such programs with respect to
either automobile manufacturers or consumers, with mixed results. For instance, such programs have been
credited with “halo” sales of another quarter million cars (Maritz Research 2010).5 In addition, an NHTSA’s
(National Highway Traffic Safety Administration) Report to Congress indicated that 49% of the cars sold
under such a program were manufactured locally, thus creating over 60,000 jobs in automobile
manufacturing and sales and related industries (Tyrell and Dernbach 2010). However, a more recent study by
Mian and Sufi (2012) suggests programs simply shift purchases forward. Fewer autos were bought in the ten
months after the U.S. program expired in counties with a higher number of clunkers, thereby offsetting most
of the initial purchases made. They also found no evidence of the program’s effect on employment in cities
with higher exposure to the program. A study by Miravete and Moral (2009) on a comparable cash for
clunkers program in Spain reached similar conclusions – that these programs may simply accelerate
3

In the former (bidding-based mechanism), consumers who intend to purchase a car need to bid for a car license; in the latter
(lottery-based mechanism), non-transferrable car licenses are allocated through random draws among registered participants.
4
As stated by the U.S. Secretary of Transportation Ray LaHood on July 27, 2009, “with this program, we are giving the auto
industry a shot in the arm, and struggling consumers can get rid of their gas-guzzlers and buy a more reliable, fuel-efficient vehicle.
This is good news for our economy, the environment, and consumers' pocketbooks.”
5
Source: http://www.maritz.com/Press-Releases/2010/Maritz-Research-Findings-Cash-for-Clunkers-More-Successful-Than-Previously-Reported.aspx.
3

intentions to change cars. With respect to the program’s intended environmental effects, most studies have
provided affirmative evaluations (Huang 2010). For instance, the NHTSA’s Report to Congress estimated
that the program will lead to a reduction of greenhouse gas emissions by 9 million metric tons over 25 years.
Also by replacing older vehicles with new ones, the program may reduce the emissions of other criteria
pollutants. Yet, prior research points out that the costs of attaining these environmental benefits could be
staggeringly high (Knittel 2009; Li et al. 2011).
We contribute to the literature by showing that an institutional car policy may be effective in several
ways. At its fundamental level, the policy could indeed orient consumer purchases toward small-engine cars
as intended, while at the same inhibit the purchases of big-engine cars not targeted. In addition, automobile,
which belongs to transportation or manufacturing industry, is indeed a large source of air pollution(Muller et
al. 2011); our empirical evidence demonstrates the institutional policy could effectively reduce the air
pollution by decreasing the purchase of large-engine cars in large metropolitan, which are the area with the
most critical environment problems; on the other side, it could simultaneously increase the sales of
small-engine cars in middles size and small size cities, where the environmental problem is gradually
concerned, so as to pursue a long-term environmental improvement. Our results also show that
environmental policies could garner local economic benefits by limiting the sales of big-engine cars of
non-indigenous brands, while stimulating sales of indigenous car brands. In 2006, Djankov et al. claim that
the role of political institutions in fostering or restricting entrepreneurship is very vital. Our study expand
their study by improving the effectiveness of institutional policy, that TLV policy helped indigenous car
brands, especially for local entrepreneurs, who usually play a crucial role in the success of country’s
economic growth and innovation, fend off competition.
The rest of the paper is organized as follows. Section I briefly overviews the automobile industry in
China prior to the implementation of TLV. Section II reports the data and empirical results. Finally, Section
III discusses the study’s contributions and concludes with implications of the results.
I.

Background

The automobile industry in China has undergone rapid developments since the 1978 reform and opening up
policy. In 1980, the number of motor vehicles in China was only 1.8 million, but within 10 years the number
grew to 5.1 million with an annual growth rate of 11.1%. The period from 1990 to 1999 witnessed a
relatively stable growth of motor vehicles in the country, with the number reaching 14.5 million in 1999 (a
4

10.2% annual growth rate). More rapid growth subsequently followed from 2000 to 2009, with an annual
growth rate of 14.5%; by 2009, the number of motor vehicles in China had reached 62.1 million. As a
developing country, the automobile industry has played a key role in China’s economic development,
contributing to 28.6% of the nation’s total GDP (based on reported statistics from 1978 to 2006) while also
significantly supporting the developments of a wide range of other industrial segments, e.g., steel,
manufacturing, and service.6
China has been actively cultivating the automobile industry during the past 30 years. In the early days,
Chinese state-owned automobile manufacturers primarily focused on producing large commercial trucks, and
only a limited number of passenger cars (Baker and Hyvonen 2011). Starting from the early 1980s, joint
venture partnerships with major global automobile manufacturers were approved to develop China’s
domestic production capabilities through technology and knowledge transfer. The 1994 implementation of
the Automobile Industry Policy took a significant step in this direction. The policy allowed large global
manufacturers to operate in China under a joint venture with their ownership restricted to no more than half.
Major local participants in these joint ventures included Beijing-based First Automobile Works (FAW), the
Shanghai Automotive Industry Corporation (SAIC), and Hubei-based Dongfeng Motors, which collectively
were the so-called “big three.” Riding on the strong economic growth in China at the start of the 2000s,
automobile production picked up significantly, with new state-owned and private manufacturers (e.g., Geely,
Chery, and BYD) entering the market. In 2004 the Automobile Industry Policy was updated with a greater
emphasis on domestic research and development, as a key objective was to develop a few famous indigenous
brands and globally competitive automobile groups (top 500) by 2010 (Baker and Hyvonen 2011). Yet, given
the extremely competitive automobile market in China, indigenous brands occupy only a small percentage of
the market (Oliver et al. 2009).
It is important to recognize that due to the wide disparity in the economic development across the
country, China constitutes a multi-tiered socio-economic landscape. First-tier metro cities are the most
developed with the best infrastructures and typically host larger populations, e.g., Beijing and Shanghai. In
contrast, lower-tiered rural cities lag in infrastructural development, and have smaller populations with lower
average income levels. Traditionally the sales of private cars had been concentrated in higher-tiered cities,
but this trend is changing as economic growth now drives personal income. According to a Nielsen report
6

Source: http://www.cnautonews.com/pl/zl/201403/t20140321_298349.htm, last accessed on July 6, 2014.
5

(2013), more car purchases are increasingly coming from lower-tiered cities, with 56% of potential buyers
first-time ones.7 Coupled with the estimation that the penetration rate of the auto market is only 5% in China
overall, there is still much potential for the automobile market.
Concomitantly, the negative implications of automobile development such as environmental pollution
has also become increasingly striking. In 2009, vehicular emissions such as nitrogen oxide and hydrocarbon
accounted for over 25% of national total emissions in China. In large cities such as Beijing, Shanghai, and
Guangzhou, vehicular emissions have become a major source of air pollution (Wang and Hao 2012). It has
also been noted that car emissions are the leading contributor of summer problems in Beijing (Wang et al.,
2012). Against this backdrop, the TLV policy was initiated and evaluated in terms of its effectiveness.
II.

Empirical Analyses
A. Data

Our dataset contains the entire record of passenger car registrations in China. To study the impact of TLV, we
focus on the registration records from July 2008 through June 2011, covering the policy implementation
period and six months prior and six months after the policy implementation. During this three-year
observation window, 24,574,631 cars were purchased and registered in China. The registration records
contain the following information: time of purchase (month), brand, car model, engine type, the city in which
the car was purchased and car buyer’s demographic data. There are 67,283 missing values of demographic
variables (age and gender) in our data, which is about 1.13% of total 5,962,320 observations. We replace
them by mean values of car buyer’s age/gender proportion in specific month group, city-tier group and
price-tier group respectively. We also collected a number of variables as controls from the China Statistical
Yearbook8, including GDP, average amount of savings, unemployment rate, and population. These variables
have been considered in other studies investigating car policy stimuli (Mian and Sufi 2012; Miravete and
Moral 2009), given that they may influence consumer decisions to purchase cars vis-à-vis the cost of car
usage.
Figure 1 shows the sale changes of small- and big-engine cars in our study period. Table 1A presents the
summary statistics of our dependent variables. Apart from the statistics of total car sales during the study
period, we split the records into the sales of big-engine cars (>1.6-liter) and small-engine cars (≤1.6-liter),
7

Source: http://www.nielsen.com/us/en/insights/news/2013/in-china_s-auto-market--smaller-cities-are-in-the-drivers-seat.html, last
accessed on July 6, 2014.
8
Source: http://www.stats.gov.cn/tjsj/ndsj/, last accessed on July 9, 2014.
6

and separated the statistics into three time periods (before, during, and after the policy period), and by
different city tiers.
[Insert Figure 1 here.]
[Insert Table 1A here.]

Comparisons of monthly average sales six months prior to, during, and six months after the policy (TLV)
show significant differences in terms of the sales of all cars, big-engine cars, and small-engine cars (see
Table 1): sales were significantly higher during the policy implementation period, with a 124% increase in
sales of small-engine cars, and only a 52.3% increase for big-engine cars. Comparing with the increase of
total car sales (91.2%) from pre-policy period to policy period, it’s reasonable to state that the sales of
small-engine cars increased much more dramatically than those of big-engine cars in general. These provide
initial model-free evidence for the impact of the TLV on car sales.
Besides, Table 1B and 1C present the summary of car related variables and demographic data of our
study respectively.
[Insert Table 1B, Table 1C here.]

B. Empirical Analyses
We first examine the change in car sales with the following regression model:
𝑆𝑎𝑙𝑒𝑠𝑖𝑗𝑡 = 𝛼0 + 𝛼1 𝑃𝑜𝑙𝑖𝑐𝑦𝑡 + 𝛼2 𝑃𝑜𝑠𝑡𝑡 + 𝚫𝑪𝒐𝒏𝒕𝒓𝒐𝒍𝑽𝒂𝒓𝒔𝒊𝒕 + Ω𝑖 + 𝜖𝑖𝑗𝑡

(1)

where i indicates the city, j indicates the car type, and t indicates the month. The dependent variable, Salesijt,
is the natural logarithm of sales volume. As the data cover three periods, i.e., pre-policy, during-policy, and
post-policy, we use two dummies, Policyt and Postt, to identify the periods respectively. Policyt is equal to 1
if month t is during the implementation of the TLV policy and 0 if otherwise. Postt is equal to 1 if month t is
after the implementation of the TLV policy, and 0 if otherwise. Thus, when both Policyt and Postt are 0,
month t should be before the policy implementation. As previously mentioned, we also control the
macro-economic impacts, including time trend, regional GDP (100 billion in RMB), regional savings per
household (1,000 in RMB), unemployment rate (%), and regional population (10 million), as well as other
policies which were being implemented during the observation period of our dataset. Time trend is an
important control because it helps rule out alternative explanation of consumers’ ability to purchase cars in
rural lower-tier cities over time, beyond the income effects controlled for by other covariates. We separated
the analyses into the total car sales, the sales of big-engine cars, and the sales of small-engine cars, and
7

estimated the three models contemporarily by Seemingly Unrelated Regression (SUR) method. Table 2
presents the results.
[Insert Table 2 here.]

As the results show, the TLV policy had a positive effect on total car sales in general. However, when
we refine the analysis by car types, we see that the policy indeed has a positive effect on the sales of
small-engine cars (direct effects as intended by the policy). Interestingly, this policy also has a negative
effect on the sales of big-engine cars (a negative spill-over effect). These effects are robust to controlling for
the trends of growing car sales in China (denoted by the time trend variable), other policies that were
implemented during the study period (both of them, denoted by Scrappage and Trucks), and other variables
that may influence car sales: regional GDP, regional savings per household, unemployment rate, and regional
population.
The positive targeted effect is straightforward: the TLV policy decreased price of small-engine cars and
raised sales consequently. And about the negative spillover effect, it might be caused by the substitution
effect, especially for big-engine cars with emission closer to 1.6 L, which is the cut point. Our additional
study on more emission groups also provides an additional support to this explanation.
To gain deeper insights into these effects of the TLV policy, we segment the data analyses by city tiers
in China (see the Appendix). As previously mentioned, China has different city-tiers with considerable
variation in terms of development levels. First-tier cities consist of sixteen mega cities, including Beijing,
Shanghai, and main provincial capital cities (e.g., Hangzhou, Nanjing, and Chengdu). These cities possess
rich resources of economy, education, nature and have a relatively developed economy. The 154 cities in the
second-tier are usually the capital cities of small provinces (e.g., Taiyuan, Fuzhou, and Guiyang) or
important cities in medium-level provinces (e.g., Datong, Dalian, and Erdos). Most of the second-tier cities
have their own pillar industries. Beyond the second-tier, China has cities that do not have well-developed
economies. 194 cities belong to such third-tier in our dataset. To ensure the robustness of our study, we also
checked the result of first-tier cities without Beijing and Shanghai, where have additional limitation for car
plate application.9 Table 3 presents the regression analysis across the three city-tiers.
[Insert Table 3 here.]
The results show that the intended positive effect of the TLV on the sales of small-engine cars actually
9

Beijing has a lottery system for car plate application and Shanghai has an auction system for it.
8

occurred in the lower-tier cities (tiers 2 and 3), but not in the first-tier cities. In contrast, the negative
spillover effect of the policy on the sales of big-engine cars occurred in higher-tier cities (tier 1), but not in
the 2nd and 3rd tier cities at the bottom-end. As such, the direct, targeted positive main effects of promoting
smaller engine cars are driven by sales in lower-tier cities, whereas the indirect, untargeted negative
spill-over effects of limiting larger engine cars are driven by sales in higher-tier cities.
A possible explanation for the existence of positive targeted effect on smaller cities only is that citizens
of smaller cities usually have a lower income and are more price-sensitive. By and large, the public
transportation in smaller cities is comparatively inconvenient. Therefore, a lower price stimulates the sales of
small-engine cars in smaller cities. In contrast, citizens of mega cities are of less price-sensitivity, which
means the policy would not increase their willingness to buy small-engine cars to a great extent. And the
public transportation in larger cities is more convenient, so citizens of larger cities might change the means
of transportation to subway, bus and some other kinds of public transportation.
We next assess the policy effects across car brands. As show in Tables 4A, we find that the policy
stimulated the sales of small-engine cars of both indigenous and non-indigenous brands but only reduced the
sales of large-engine cars of non-indigenous brands. Specifically, we analyze twelve major brands, including
five indigenous brands and seven non-indigenous brands respectively. As shown in Table 4B and 4C, most
indigenous brands (FAW, BYD, Geely and JAC, except Chery) enjoyed a sales increase of small-engine cars
while only three of seven non-indigenous brands (Honda, Peugeot, Suzuki) had the same increase. And about
the large-engine cars, all indigenous brands enjoyed a sales increase; however, of the seven non-indigenous
brands, two (Toyota and Peugeot) suffered in sales, three experienced no significant change (KIA, Kyundai,
and Suzuki), and only two obtained a sales increase (Honda and Volkswagen). As such, this TLV policy
actually favors the growth of indigenous car brands vis-à-vis non-indigenous counterparts, an interesting
spillover effect not intended by the policy. The occurrence of negative spillover effect on non-indigenous
brands is possibly caused by the fact that prices of non-indigenous brand automobiles are generally higher
than indigenous brands. Consumers preferring to buy cars with non-indigenous brands might be influenced
more by the TLV policy, and the substitution effect thereby occurs more on non-indigenous brands.
[Insert Table 4A, Table 4B and Table 4C here.]

9

C. Additional Studies
1. More Emission Groups
To further explore the effect of TLV policy, we replicate our basic model with six different emission
groups.10
[Insert Table 5 here.]
The result shows that the effect of TLV policy on three emission groups of small-engine cars is
consistently positive, and the negative spillover effect also occurs on all emission groups of large-engine cars.
In addition, we find that the TLV policy has a comparatively larger positive targeted effect and negative
spillover effect on emission groups around 1.6 Liter. This finding, to some extent, supports our explanation
of substitution effect for the existence of negative spillover effect on big-engine cars.
2. Robustness Check
In order to ensure the reliability of our findings, we conduct a robustness check by adding individual control
variables and car related variables in other study to measure consumer heterogeneity:
𝑆𝑎𝑙𝑒𝑠𝑖𝑗𝑡 = 𝛼0 + 𝛼1 𝑃𝑜𝑙𝑖𝑐𝑦𝑡 + 𝛼2 𝑃𝑜𝑠𝑡𝑡 + 𝜞𝑰𝒏𝒅𝒊𝒗𝒊𝒅𝒖𝒂𝒍𝒊𝒋𝒕 + 𝜣𝑪𝒂𝒓𝒊𝒋𝒕 + 𝜟𝑪𝒐𝒏𝒕𝒓𝒐𝒍𝑽𝒂𝒓𝒔𝒊𝒕 + Ω𝑖
+ 𝜖𝑖𝑗𝑡

(2)

In model (2), Individualijt includes age and gender. Following Chandra and Kaiser (2014), we use the
Herfindahl Hirschman Indexes (HHIs) to measure the heterogeneity of car buyers’ demographic
characteristics. The value of a HHI closer to 0 indicates more diversity of characteristics, whereas it closer to
1 means greater homogeneity. Carsijt cover the distributions of several major car features, including car
model, ownership, usage, payment method of purchase, color, and fuel type. All these features are shown in
percentage terms. For instance, Blackijt indicating the percentage of black cars sold in a specific car type per
city per month. Other variables are the same as those in model (1).
We also do the robustness check in other two ways. As most Chinese retailers need to achieve the
settled sale targets by the end of year, there are usually plenty of promotions provided by different 4S shops
in the last quarter of each year. To eliminate the possible influence of those invisible promotions, we
replicating the results of model (2) with data dropped the records of last three months of each year (October
to December). Another robustness check of our study examines the model with the lag effect of sales. As

10

The emission of automobiles is from 0.8 to 5.7 Liter. The divided six emission groups are: L≤1.0, 1.1≤L≤1.5, L=1.6, 1.7≤L≤
2.5, 2.6≤L≤4.0, L> 4.0.
10

purchasing a car is, to some extent, a kind of conspicuous consumption, especially in China. The herding
effect may possibly exist (McCormick 1983), which means sales of a certain car model in current month
could be influenced by the previous sales. The results of the three robustness checks are shown in Table 5.
[Insert Table 6 here.]
We can see that the positive targeted effects on small-engine cars and the negative spillover effects on
large-engine cars are revealed in all models, which are consistent with the results of our basic model (1).
3. Car Buyer Targeting Analysis
To study the effect of TLV policy on heterogeneous car buyer groups, interaction terms of policy with car
buyer targeting variables (age, gender and car price) are added to the model as below:
𝑆𝑎𝑙𝑒𝑠𝑖𝑗𝑡 = 𝛼0 + 𝜝𝑷𝒐𝒍𝒊𝒄𝒚𝒕 ∗ 𝑪𝑻𝒊𝒋𝒕 + 𝛼1 𝑃𝑜𝑙𝑖𝑐𝑦𝑡 + 𝛼2 𝑃𝑜𝑠𝑡𝑡 + 𝜞𝑪𝑻𝒊𝒋𝒕 + 𝜟𝑪𝒐𝒏𝒕𝒓𝒐𝒍𝑽𝒂𝒓𝒔𝒊𝒕 + Ω𝑖
+ 𝜖𝑖𝑗𝑡

(3)
[Insert Table 7 here.]

Car buyer targeting variables include age, gender of car buyers and the price of purchased car. Three
age groups and gender, shown in percentage terms, are first used to study the effect of policy. We also use the
constructed variables of age and group to study the interaction effect of TLV policy and consumer
heterogeneity.
As shown in Table 6, the more heterogeneity of consumer in age, the more effective of TLV policy on
both small-engine cars and large-engine cars. On the contrary, the interaction of policy and gender reveals an
opposite phenomenon: the increasing gender heterogeneity of consumers reduces the positive targeted effect
as well as the negative spillover effect of TLV.
We then concentrate on the direct distinction of policy effect on different age group and gender. As
shown in Table 1C, the proportion of age group between 31 and 50 is the prime one of our data, which is
averagely 16.79% for each car type per city per month. Interaction between the TLV policy and the other two
non-prime age groups shows that comparing with the prime age group, the higher percentage of consumers
of non-prime age groups, the more effective of TLV policy on large-engine cars. However, the positive
targeting effect of TLV is different on these two age groups. The effect of TLV policy on small-engine cars is
strengthened by the increasing proportion of consumers under 30, while it’s eliminated by having more
consumers over 50. The result also reveals asymmetrical impact of male proportion on positive targeted
11

effect and negative spillover effect: the increasing male proportion of consumers enhances the positive
targeted effect, while reduces the negative spillover effect of TLV.
Besides, the interaction of policy and car price also reveals that the higher consumer paid for the
automobile, the less effectiveness of TLV policy as purchasers of more expensive cars are less price
sensitive.
III.

Concluding Remarks

The automobile industry may buttress a country’s economic development, but may also threaten the
environment. We demonstrate that a policy measure designed to influence consumer purchase of more
environmental-friendly cars with a tax-cut incentive could furnish both environmental and economical
benefits. Environmentally, such a policy could encourage ownership of low-emission, small-engine cars and
discourage ownership of high-emission, big-engine cars. Economically, the policy could favor the growth of
indigenous brands, which is desired in a developing country such as China that hopes to cultivate its own
brands to decrease dependence on non-indigenous companies.
Leveraging the uniquely huge and diverse markets of China, we also show that the policy is beneficial
for encouraging purchases of small-engine cars in lower-tier cities, which range from fairly well developed
to poorly developed areas. Furthermore, such a policy acted more to discourage the purchases of big-engine
cars in higher-tier, mega cities, where air pollution due to vehicular emissions is particularly severe.

11

The heterogeneous results across the different city-tiers highlights a caveat of the TLV policy – it
turned out to be ineffective in inducing the purchases of small-engine cars in first-tier mega cities in China to
pursue a long-term environmental improvement. While this does not negate the value of the policy given that
it did reduce large-emission big-engine cars in the first-tier cities, it does mean that prudent policy-makers
ought to ponder alternative policy designs if their objective is to promote a fleet of more
environmentally-friendly cars in their forerunner cities.
This study is limited because it does not use any environment index to directly measure the impact of
TLV on air pollution. However, environment is a long term goal, which is hard to quantify in couple of years,
especially by focusing on only several years of data and limited influential factors. Based on the main
finding of our paper, it’s reasonable to claim that the TLV policy will consequently reduce environmental
11

For instance, Beijing has become one of the cities with the most severe vehicular emission pollution. In Shanghai, although the
amount of cars is only 1/12 of that in Tokyo, the air pollutants due to vehicular emissions are about equal to that of the latter. In
Guangzhou, vehicular emissions constitute 22% of the pollution sources (the largest component), whereas the percentage even
reaches 70% in Shenzhen (Source: http://www.ytepb.gov.cn/html/ztbd/wqzl/768.html, last accessed on June 12, 2014).
12

pollution, to some extent, as it limits car sales of big engine and lifts up sales of small engine in the same
time period.

REFERENCES
Baker, Mark and Markus Hyvonen. 2011. “The emergence of the Chinese automobile sector.” RBA
Bulletin, March: 23-29.
Bento, Antonio M., Lawrence H. Goulder, Mark R. Jacobsen and Roger H. von Haefen. 2009.
“Distributional and Efficiency Impacts of Increased US Gasoline Taxes.” The American economic
review, 99(3): 667-699.
Busse, Meghan R., Christopher R. Knittel, Jorge Silva-Risso, Florian Zettelmeyer. 2012. “Did ‘cash for
clunkers’ deliver? The consumer effects of the car allowance rebate system.” November 2012. Working
Paper.
Cai, Hao, and Shaodong Xie. 2013. “Temporal and spatial variation in recent vehicular emission
inventories in China based on dynamic emission factors.” Journal of the Air & Waste Management
Association, 63(3): 310-326.
Chandra, Ambarish, and Ulrich Kaiser. 2014. "Targeted advertising in magazine markets and the advent
of the internet." Management Science, 60(7): 1829-1843.
Chen, Xiaojie, and Jinhua Zhao. 2013. “Bidding to drive: Car license auction policy in Shanghai and its
public acceptance.” Transport Policy, 27, 39–52.
Daniel, Joseph I., and Khalid Bekka. 2000. “The environmental impact of highway congestion pricing.”
Journal of Urban Economics, 47(2): 180-215.
Djankov, Simeon, Yingyi Qian, Gerard Roland and Ekaterina Zhuravskaya. 2006. “Who are China's
entrepreneurs?” The American economic review, 96(2): 348-352.
Hao, Han, Hewu Wang, and Minggao Ouyang. 2011. “Comparison policies on vehicle ownership and use
between Beijing and Shanghai and their impacts on fuel consumption by passenger vehicles.” Energy
Policy, 39(2): 1016–1021.
IEA 2011. World Energy Outlook 2011, IEA, Paris.
Knittel, Christopher R. 2009. “The Implied Cost of Carbon Dioxide under the Cash for Clunkers Program.”
University of California-Davis Working Paper.
Li, Shanjun, Joshua Linn, and Elisheba Spiller. 2012. “Evaluating ‘cash-for-clunkers’: Program effects on
auto sales and the environment.” Journal of Environmental Economics and Management, 65(2):
165-193.
McCormick, Ken.1983. "Duesenberry and Veblen: The demonstration effect revisited." Journal of
Economic Issues, 17(4): 1125-1129.
Mian, Atif R., and Amir Sufi. 2012. “What explains high unemployment? The aggregate demand channel.”
National Bureau of Economic Research, http://www.nber.org/papers/w17830, last accessed on July 14,
2014.
Miravete, Eugenio J. and Maria J. Moral. 2009. “Qualitative effects of cash-for-clunkers programs,
Centre for Economic Policy Research Discussion Paper No. 7517.
Muller, Nicholas Z., Robert Mendelsohn and William Nordhaus. 2011. “Environmental Accounting for
Pollution in the United States Economy.” The American Economic Review, 101(5): 1649-1675.
Oliver, Hongyan H., Kelly Sims Gallagher, Donglian Tian, and Jinhua Zhang. 2009. “China's fuel
economy standards for passenger vehicles: Rationale, policy process, and impacts.” Energy Policy,
37(11): 4720-4729.
Parry, Ian W. H. and Kenneth A. Small. 2005. “ Does Britain or the United States Have the Right
Gasoline Tax?” The American economic review, 95(4): 1276-1289.
Parry, Ian W. H., Margar Wall, and Winston Harrington. 2007. “Automobile externalities and policies.”
Journal of economic literature, 45(2): 373-399.
Tyrrell, Marianne, and John C. Dernbach. 2010. “Cash for Clunkers Program: A Sustainability
Evaluation.” The Univerity of Toledo Law Review 2010-2011. 42: 467.
Wallace, William. 1996. “Government without statehood: the unstable equilibrium.” In Policy-making in
the European Union, Wallace Helen and Wallace William (eds.), Oxford: Oxford Univeristy Press..
13

Wang, Shuxiao, and Jiming Hao. 2012. “Air quality management in China: Issues, challenges, and options.”
Journal of Environmental Sciences, 24(1): 2-13.
Wang, Shuxiao, Meng Zhao, Jia Xing, Ye Wu, Yu Zhou, Ye Lei, Kebin He, Lixin Fu, and Jiming Hao.
2012. “Quantifying the Air Pollutants Emission Reduction during the 2008 Olympic Games in Beijing.”
Environmental Science & Technology, 44(7): 2490-2496.
Xiao, Junji, and Heng Ju. 2014. "Market Equilibrium and the Environmental Effects of Tax Adjustments in
China's Automobile Industry." Review of Economics and Statistics, 96(2): 306-317.

14

Table 1, Panel A-Summary Statistics of Car Sales during the Study Period
Pre-policy

Policy

Post-policy

(Thousand)

Mean

SE

Mean

SE

Mean

SE

P-Value

Monthly Sales of Automobiles

380.79

12.36

728.75

41

799.98

79.32

＜0.0001

Monthly Sales of Automobiles

204.97

8.36

406.9

30.4

493.21

45.85

＜0.0001

175.81

6.11

267.85

12.02

306.77

33.93

＜0.0001

with Emission≤1.6L
Monthly Sales of Automobiles
with Emission＞1.6L

Table 1, Panel B- Summary Statistics on Car Data
Mean
Policy
0.667
Post
0.167
Time
18.5
Consumption Rate
0.944
Scrapped
0.528
Truck
0.611
GDP
15.39
Saving
10.20
Unemployment Rate 3.668
Population
5.228
Price (ten thousand) 14.85

SD
0.471
0.373
10.39
0.229
0.499
0.487
12.17
8.329
0.498
2.785
13.22

MIN
0
0
1
0
0
0
0.395
0.185
1.37
0.292
3.58

5%
0
0
2
0
0
0
1.654
1.058
2.6
0.633
4.48

95%
1
1
35
1
1
1
42.43
31.41
4.27
10.13
39.66

Max
1
1
36
1
1
1
53.21
40.41
4.57
10.50
96.2

Table 1, Panel C- Summary Statistics on Demographic Data

Male(%)
Age
Age<30(%)
31≤Age<50(%)
Age≥50(%)
Constructed HHIs
HHIage
HHIgender

Mean SD
MIN 5% 95%
18.45 34.93 0
0
100

Max
100

4.75 14.56 0
16.79 32.57 0
3.07 11.42 0

0
0
0

33.33 100
100
100
20
100

0.174 0.326 0
0.191 0.349 0

0
0

1
1

15

1
1

Dependent Variable
Model
Policy
Post
Time
Consumption Tax
Scrapped
Truck
GDP
Saving
Unemployment Rate
Population
Observations
R-Square

Table 2-Detecting the Effects of TLV
Total car sales Sales of small-engine cars Sales of big-engine cars
0.0183***
(0.002)
-0.1194***
(0.003)
0.0102***
(0.000)
-0.0196***
(0.002)
-0.0034a
(0.002)
-0.0124***
(0.002)
0.0178***
(0.000)
-0.0068***
(0.000)
0.0189***
(0.001)
-0.0079***
(0.000)
5962320
0.029

0.0216***
(0.002)
-0.0811***
(0.003)
0.0077***
(0.000)
-0.0096***
(0.002)
-0.0078***
(0.002)
-0.0069***
(0.002)
0.0100***
(0.000)
-0.0053***
(0.000)
0.0116***
(0.001)
0.0015***
(0.000)
5962320
0.017

-0.0037*
(0.001)
-0.0582***
(0.002)
0.0035***
(0.000)
-0.0116***
(0.001)
0.0001
(0.001)
-0.0084***
(0.002)
0.0095***
(0.000)
-0.0017***
(0.000)
0.0089***
(0.001)
-0.0122***
(0.000)
5962320
0.014

Note: Consumption tax indicates a change in the rate of consumption tax (in September 2008). Scrappage indicates a car scrappage
policy (from June 2009 to December 2010); Trucks indicates a policy to promote light trucks and vans in villages (from March 2009
to December 2010).
a
p<0.1, *p<0.05, ** p<0.01, ***p<0.001

16

Table 3-The Effects of TLV across City Tiers
1st city tier

1st city tier (Without
Beijing and Shanghai)
Panel A: Sales of small-engine cars ≤ 1.6 liter
Policy
-0.0085
-0.0103
(0.015)
(0.016)
***
Post
-0.2619***
-0.2231
(0.024)
(0.025)
***
Time
0.0210***
0.0203
(0.001)
(0.001)
*
*
Consumption Rate
-0.0340
-0.0328
(0.015)
(0.016)
Scrapped
-0.0210
-0.0142
(0.013)
(0.014)
Truck
-0.0171
-0.0179
(0.016)
(0.017)
GDP
-0.0037***
0.0006
(0.001)
(0.001)
*
Saving
-0.0003
-0.0026
(0.001)
(0.001)
Unemployment Rate
-0.0936***
0.0075
(0.005)
(0.010)
***
Population
0.0144***
0.0186
(0.002)
(0.002)
Observations
262080
229320
R-Squared
0.012
0.012
Panel B：Sales of big-engine cars＞ 1.6 liter
**
Policy
-0.0479***
-0.0392
(0.014)
(0.015)
***
Post
-0.1415***
-0.1098
(0.022)
(0.023)
***
Time
0.0098***
0.0098
(0.001)
(0.001)
Consumption Rate
-0.0126
-0.0081
(0.014)
(0.015)
Scrapped
0.0087
0.0117
(0.013)
(0.013)
Truck
-0.0152
-0.0211
(0.015)
(0.015)
***
GDP
-0.0043
0.0000
(0.001)
(0.001)
***
Saving
0.0118***
0.0049
(0.001)
(0.001)
a
Unemployment Rate
-0.0659***
-0.0179
(0.005)
(0.009)
**
Population
-0.0071***
0.0060
(0.002)
(0.002)
Observations
262080
229320
17

2nd city tier

3rd city tier

0.0282***
(0.003)
-0.1083***
(0.005)
0.0107***
(0.000)
-0.0153***
(0.003)
-0.0115***
(0.003)
-0.0113***
(0.003)
0.0094***
(0.000)
-0.0076***
(0.000)
0.0188***
(0.002)
-0.0014***
(0.000)
2522520
0.013

0.0244***
(0.002)
-0.0330***
(0.003)
0.0056***
(0.000)
-0.0072***
(0.002)
-0.0066***
(0.002)
-0.0061**
(0.002)
0.0040***
(0.000)
-0.0051***
(0.000)
0.0069***
(0.001)
0.0176***
(0.000)
3177720
0.014

0.0012
(0.003)
-0.0720***
(0.004)
0.0052***
(0.000)
-0.0179***
(0.003)
-0.0006
(0.002)
-0.0122***
(0.003)
0.0098***
(0.000)
-0.0050***
(0.000)
0.0063***
(0.002)
-0.0170***
(0.000)
2522520

0.0021
(0.001)
-0.0289***
(0.002)
0.0030***
(0.000)
-0.0103***
(0.001)
-0.0026*
(0.001)
-0.0081***
(0.001)
0.0043***
(0.000)
-0.0034***
(0.000)
0.0104***
(0.001)
0.0066***
(0.000)
3177720

R-Squared

0.008

0.007

0.009

0.010

Note: Consumption tax indicates a change in the rate of consumption tax (in September 2008). Scrappage indicates a car scrappage
policy (from June 2009 to December 2010); Trucks indicates a policy to promote light trucks and vans in villages (from March 2009
to December 2010).
a
p<0.1, *p<0.05, ** p<0.01, ***p<0.001

18

Table 4, Panel A - The Effects of TLV on Indigenous Brands and Non-indigenous Brands

Indigenous Brands Non-indigenous Brands
Panel A: Sales of small-engine cars ≤ 1.6 liter

0.0134***
(0.002)
Post
-0.0633***
(0.003)
Time
0.0075***
(0.000)
Consumption Rate
-0.0024
(0.002)
Scrapped
0.0007
(0.002)
Truck
-0.0032
(0.002)
GDP
0.0077***
(0.000)
Saving
-0.0057***
(0.000)
Unemployment Rate
0.0125***
(0.001)
Population
0.0045***
(0.000)
Observations
3313128
R-Squared
0.015
Panel B：Sales of big-engine cars＞ 1.6 liter
Policy
-0.0002
(0.001)
Post
-0.0174***
(0.002)
Time
0.0010***
(0.000)
Consumption Rate
0.0020
(0.001)
Scrapped
0.0053***
(0.001)
Truck
-0.0035*
(0.001)
GDP
0.0047***
(0.000)
Saving
-0.0020***
(0.000)
Unemployment Rate
0.0011a
(0.001)
Population
-0.0035***
(0.000)
Observations
3313128
R-Squared
0.006
Policy

0.0325***
(0.003)
-0.1029***
(0.005)
0.0080***
(0.000)
-0.0186***
(0.003)
-0.0184***
(0.003)
-0.0116***
(0.003)
0.0129***
(0.000)
-0.0048***
(0.000)
0.0105***
(0.001)
-0.0022***
(0.000)
2649192
0.020
-0.0072*
(0.003)
-0.1085***
(0.004)
0.0067***
(0.000)
-0.0285***
(0.003)
-0.0063*
(0.002)
-0.0146***
(0.003)
0.0154***
(0.000)
-0.0013***
(0.000)
0.0188***
(0.001)
-0.0232***
(0.000)
2649192
0.025

Note: Consumption tax indicates a change in the rate of consumption tax (in September 2008). Scrappage indicates a car scrappage
policy (from June 2009 to December 2010); Trucks indicates a policy to promote light trucks and vans in villages (from March 2009
to December 2010).
a
p<0.1, *p<0.05, ** p<0.01, ***p<0.001
19

Table 4, Panel B - The Effects of TLV across Non-indigenous Brands
Honda
Toyota
Non-indigenous brands:
Panel A: Sales of small-engine cars ≤ 1.6 liter

Volkswagen

0.0620***
0.0121
(0.015)
(0.010)
Post
-0.0679** -0.1155***
(0.023)
(0.015)
Time
-0.0015* 0.0016***
(0.001)
(0.000)
Consumption Rate
-0.0021
0.0081
(0.015)
(0.010)
Scrapped
0.0220a
0.0002
(0.013)
(0.009)
Truck
-0.0159
-0.0183a
(0.016)
(0.010)
***
GDP
0.0042
0.0064***
(0.001)
(0.001)
***
Saving
0.0065
0.0108***
(0.001)
(0.001)
***
Unemployment Rate
-0.0561
-0.0407***
(0.007)
(0.005)
***
Population
0.0099
-0.0127***
(0.002)
(0.001)
Observations
117936
222768
R-Squared
0.021
0.028
Panel B：Sales of big-engine cars＞ 1.6 liter
Policy
0.0491* -0.0747***
(0.021)
(0.013)
Post
-0.2557*** -0.0220
(0.032)
(0.019)
***
Time
0.0077
0.0059***
(0.001)
(0.001)
Consumption Rate
-0.0250
-0.0401**

-0.0008
(0.016)
-0.3655***
(0.025)
0.0240***
(0.001)
-0.0238
(0.016)
-0.1266***
(0.014)
-0.0146
(0.017)
0.0342***
(0.001)
-0.0144***
(0.001)
0.1213***
(0.008)
-0.0210***
(0.002)
196560
0.045

0.0167
0.0132
0.1496***
(0.011)
(0.015)
(0.025)
-0.0718*** -0.2839*** -0.3928***
(0.017)
(0.023)
(0.039)
0.0085*** 0.0171*** 0.0192***
(0.001)
(0.001)
(0.001)
-0.0350** -0.0266a
-0.0587*
(0.011)
(0.015)
(0.025)
0.0539*** -0.0385** -0.2454***
(0.010)
(0.013)
(0.022)
-0.0104
-0.0372*
0.0397
(0.011)
(0.016)
(0.026)
0.0135*** 0.0155*** 0.0456***
(0.001)
(0.001)
(0.001)
***
***
-0.0072
-0.0110
-0.0335***
(0.001)
(0.001)
(0.002)
***
-0.0060
0.0316
0.0815***
(0.005)
(0.007)
(0.012)
-0.0006
0.0006
-0.0009
(0.001)
(0.002)
(0.003)
196560
170352
52416
0.026
0.018
0.081

0.0302*
(0.014)
-0.3938***
(0.022)
0.0152***
(0.001)
-0.0017

-0.0029
(0.007)
0.0515***
(0.011)
0.0039***
(0.000)
-0.0200**

Policy

20

KIA

Hyundai

Peugeot

Suzuki

0.0566**
(0.018)
-0.4980***
(0.027)
0.0189***
(0.001)
-0.0660***
(0.018)
-0.1471***
(0.015)
-0.1119***
(0.018)
0.0257***
(0.001)
-0.0093***
(0.001)
0.0395***
(0.009)
0.0025
(0.002)
117936
0.049

-0.0150 -0.0755***
0.0007
(0.011)
(0.014)
(0.003)
-0.2697*** -0.3491*** 0.0557***
(0.017)
(0.021)
(0.005)
***
***
0.0184
0.0188
-0.0015***
(0.001)
(0.001)
(0.000)
***
***
-0.0858
-0.0702
0.0044

Scrapped
Truck
GDP
Saving
Unemployment Rate
Population
Observations
R-Squared

(0.021)
(0.013)
-0.0225
0.1057***
(0.018)
(0.011)
-0.0218
0.0087
(0.022)
(0.013)
0.0274*** 0.0143***
(0.001)
(0.001)
***
0.0099
0.0176***
(0.002)
(0.001)
0.0095
0.0268***
(0.010)
(0.006)
-0.0307*** -0.0415***
(0.003)
(0.002)
117936
222768
0.063
0.044

(0.014)
-0.1257***
(0.012)
-0.0550***
(0.015)
0.0379***
(0.001)
-0.0201***
(0.001)
0.0540***
(0.007)
-0.0339***
(0.002)
196560
0.040

(0.007)
-0.0144*
(0.006)
-0.0079
(0.007)
0.0093***
(0.000)
-0.0050***
(0.001)
0.0166***
(0.003)
-0.0106***
(0.001)
196560
0.019

(0.011)
(0.014)
(0.003)
***
***
-0.0630
-0.1331
0.0128***
(0.010)
(0.012)
(0.003)
***
***
-0.0401
-0.0686
0.0508***
(0.012)
(0.015)
(0.003)
***
***
0.0257
0.0141
-0.0002
(0.001)
(0.001)
(0.000)
***
*
-0.0225
-0.0026
0.0027***
(0.001)
(0.001)
(0.000)
0.0128*
0.0310***
0.0011
(0.006)
(0.007)
(0.002)
-0.0116*** -0.0211*** -0.0041***
(0.002)
(0.002)
(0.000)
170352
52416
117936
0.038
0.055
0.012

Note: Consumption tax indicates a change in the rate of consumption tax (in September 2008). Scrappage indicates a car scrappage policy (from June 2009 to December 2010); Trucks indicates a policy
to promote light trucks and vans in villages (from March 2009 to December 2010).
a
p<0.1, *p<0.05, ** p<0.01, ***p<0.001

21

Table 4, Panel C-The Effects of TLV across Indigenous Brands

Indigenous brands

FAW

BYD

Chery

Geely

JAC

Panel A: Sales of small-engine cars ≤ 1.6 liter

0.0162a
(0.009)
Post
0.0043
(0.013)
Time
0.0032***
(0.000)
Consumption Rate
-0.0025
(0.009)
Scrapped
0.0690***
(0.008)
Truck
-0.0068
(0.009)
GDP
0.0094***
(0.001)
Saving
-0.0077***
(0.001)
Unemployment Rate -0.0117**
(0.004)
Population
0.0017
(0.001)
Observations
183456
R-Squared
0.015
Panel B：Sales of big-engine cars＞
Policy
0.0108a
(0.006)
Post
-0.0220*
(0.009)
Time
-0.0009**
(0.000)
Consumption Rate
0.0544***
Policy

0.0605**
0.0116
0.0919*** 0.1340***
(0.020)
(0.008)
(0.012)
(0.015)
-0.1987*** -0.1024*** -0.0887*** -0.1558***
(0.031)
(0.012)
(0.019)
(0.023)
0.0200*** 0.0079*** 0.0059*** 0.0179***
(0.001)
(0.000)
(0.001)
(0.001)
**
0.0619
0.0104
0.0184
-0.0344*
(0.020)
(0.008)
(0.012)
(0.015)
*
***
***
-0.0422
0.0404
0.0520
-0.0667***
(0.017)
(0.007)
(0.011)
(0.013)
-0.0166
0.0007
0.0160
0.0251
(0.021)
(0.008)
(0.013)
(0.016)
0.0135*** 0.0106*** 0.0122*** 0.0195***
(0.001)
(0.000)
(0.001)
(0.001)
**
***
***
-0.0046
-0.0105
-0.0112
-0.0262***
(0.002)
(0.001)
(0.001)
(0.001)
0.0442*** 0.0153*** 0.0169** -0.0200**
(0.010)
(0.004)
(0.006)
(0.007)
0.0103*** 0.0174*** 0.0100*** 0.0185***
(0.003)
(0.001)
(0.002)
(0.002)
117936
353808
183456
78624
0.034
0.019
0.018
0.065
1.6 liter

0.0565***
(0.010)
0.0225
(0.015)
0.0021***
(0.000)
-0.0064
22

0.0136*** 0.0254***
0.0272a
(0.002)
(0.003)
(0.016)
***
***
-0.0300
-0.0383
-0.1291***
(0.004)
(0.005)
(0.025)
a
***
0.0002
-0.0020
0.0088***
(0.000)
(0.000)
(0.001)
*
**
-0.0064
0.0088
-0.0344*

(0.006)
Scrapped
-0.0066
(0.005)
Truck
-0.0023
(0.006)
GDP
0.0079***
(0.000)
Saving
-0.0079***
(0.000)
Unemployment Rate
0.0022
(0.003)
Population
-0.0026**
(0.001)
Observations
183456
R-Squared
0.005

(0.010)
0.0130
(0.008)
0.0176a
(0.010)
0.0047***
(0.001)
-0.0004
(0.001)
0.0079a
(0.005)
0.0009
(0.001)
117936
0.015

(0.002)
(0.003)
(0.016)
***
***
0.0092
-0.0334
0.0064
(0.002)
(0.003)
(0.014)
-0.0134*** -0.0174*** -0.0064
(0.003)
(0.003)
(0.017)
***
***
0.0015
0.0016
0.0353***
(0.000)
(0.000)
(0.001)
***
-0.0001 -0.0018
-0.0308***
(0.000)
(0.000)
(0.001)
-0.0163*** -0.0035* -0.0242**
(0.001)
(0.001)
(0.008)
-0.0004
0.0015*** -0.0055*
(0.000)
(0.000)
(0.002)
353808
183456
78624
0.007
0.017
0.059

Note: Consumption tax indicates a change in the rate of consumption tax (in September 2008). Scrappage indicates a car scrappage policy (from June 2009 to December 2010); Trucks indicates a policy
to promote light trucks and vans in villages (from March 2009 to December 2010).
a
p<0.1, *p<0.05, ** p<0.01, ***p<0.001

23

Dependent Variable
Model
Policy
Post
Time
Consumption Rate
Scrapped
Truck
GDP
Saving
Unemployment Rate
Population
Observations
R-Squared

Table 5-Detecting the Effects of TLV on Different Emission Groups
L=1.6
L≤1.0
1.1≤L≤1.5
1.7≤L≤2.5
2.6≤L≤4.0
0.0324*
(0.013)
-0.2075***
(0.021)
0.0119***
(0.001)
0.0002
(0.013)
-0.0380**
(0.012)
-0.0369**
(0.014)
0.0072***
(0.001)
-0.0001
(0.001)
0.0129a
(0.007)
-0.0135***
(0.002)
5962320
0.0003

0.1493***
(0.021)
-0.5749***
(0.033)
0.0378***
(0.001)
-0.0715***
(0.021)
-0.0634***
(0.019)
-0.0678**
(0.022)
0.0447***
(0.001)
-0.0058***
(0.002)
-0.2610***
(0.010)
-0.0532***
(0.003)
1.0871***
0.0036

0.1551***
(0.032)
-0.4493***
(0.050)
0.0135***
(0.002)
0.0353
(0.032)
0.0424
(0.028)
0.0188
(0.034)
0.0599***
(0.002)
0.0356***
(0.003)
-0.4732***
(0.016)
-0.1937***
(0.004)
5962320
0.0032

-0.1166***
(0.030)
-0.4318***
(0.046)
0.0055***
(0.001)
-0.0005
(0.030)
0.1802***
(0.026)
0.0049
(0.031)
0.0544***
(0.002)
0.1143***
(0.002)
-0.5331***
(0.015)
-0.3722***
(0.004)
5962320
0.0066

-0.0331***
(0.005)
-0.0360***
(0.007)
0.0012***
(0.000)
-0.0147**
(0.005)
0.0270***
(0.004)
-0.0027
(0.005)
-0.0004
(0.000)
0.0113***
(0.000)
-0.0313***
(0.002)
-0.0222***
(0.001)
5962320
0.0010

L> 4.0
-0.0009**
(0.000)
-0.0017***
(0.000)
0.0001***
(0.000)
-0.0030***
(0.000)
-0.0003
(0.000)
-0.0003
(0.000)
-0.0000
(0.000)
0.0001***
(0.000)
-0.0002a
(0.000)
-0.0005***
(0.000)
5962320
0.0001

Note: Consumption tax indicates a change in the rate of consumption tax (in September 2008). Scrappage indicates a car scrappage policy (from June 2009 to December 2010); Trucks indicates a policy
to promote light trucks and vans in villages (from March 2009 to December 2010).
a
p<0.1, *p<0.05, ** p<0.01, ***p<0.001

24

Table 6- The Robustness Checks

Model

Basic

Without Last Three Months Without Last Three Months
With Lag Effect

Panel A: Sales of small-engine cars ≤ 1.6 liter

0.0056***
(0.001)
Post
0.0436***
(0.002)
Time
-0.0017***
(0.000)
Consumption Rate
0.0199***
(0.001)
Scrapped
0.0116***
(0.001)
Truck
0.0094***
(0.001)
HHIage
-1.7712***
(0.002)
HHIgender
-0.4723***
(0.003)
NB
0.1287***
(0.001)
MPV
-0.3535***
(0.002)
Private
1.0532***
(0.002)
Non-operation
-0.2183***
(0.004)
Fullpaid
0.6351***
(0.003)
Black
-0.4681***
(0.002)
Gray
0.0497***
(0.002)
White
-0.0636***
(0.003)
Petrol
1.3676***
(0.004)
Price
-0.0076***
(0.000)
GDP
0.0025***
(0.000)
Saving
-0.0020***
(0.000)
Unemployment Rate -0.0095***
(0.001)
Population
-0.0011***
(0.000)
Lagged (Sales)
Policy

Observations
R-Squared

5962320
0.567

0.0120***
(0.002)
0.0712***
(0.003)
-0.0025***
(0.000)
0.0185***
(0.002)
0.0120***
(0.001)
0.0114***
(0.001)
-1.7260***
(0.002)
-0.4628***
(0.003)
0.1206***
(0.001)
-0.3430***
(0.002)
1.0675***
(0.002)
-0.2592***
(0.005)
0.6223***
(0.003)
-0.4272***
(0.002)
0.0495***
(0.002)
-0.0499***
(0.003)
1.3531***
(0.004)
-0.0076***
(0.000)
0.0024***
(0.000)
-0.0018***
(0.000)
-0.0096***
(0.001)
-0.0013***
(0.000)

4471740
0.562
25

0.0197***
(0.001)
-0.0234***
(0.002)
-0.0003***
(0.000)
0.0188***
(0.001)
-0.0023***
(0.001)
-0.0266***
(0.001)
-0.5958***
(0.001)
-0.1357***
(0.002)
0.0124***
(0.001)
-0.0948***
(0.001)
0.3811***
(0.001)
-0.1471***
(0.003)
0.2414***
(0.002)
-0.1529***
(0.001)
0.0008
(0.001)
-0.0324***
(0.002)
0.5249***
(0.003)
-0.0021***
(0.000)
-0.0001
(0.000)
-0.0001a
(0.000)
-0.0039***
(0.000)
-0.0001
(0.000)
0.7122***
(0.000)
4306120
0.854

Panel B：Sales of big-engine cars＞ 1.6 liter

-0.0096***
(0.001)
Post
-0.0398***
(0.002)
Time
-0.0002**
(0.000)
Consumption Rate
-0.0205***
(0.001)
Scrapped
0.0010
(0.001)
Truck
-0.0053***
(0.001)
HHIage
-0.9697***
(0.002)
HHIgender
-0.1160***
(0.003)
NB
0.1803***
(0.001)
MPV
0.3240***
(0.002)
Private
-0.6646***
(0.002)
Non-operation
1.4447***
(0.004)
Fullpaid
-0.0545***
(0.002)
Black
0.5703***
(0.002)
Gray
0.0622***
(0.002)
White
0.1957***
(0.003)
Petrol
0.3827***
(0.004)
Price
0.0075***
(0.000)
GDP
0.0042***
(0.000)
Saving
0.0012***
(0.000)
Unemployment Rate
0.0004
(0.001)
Population
-0.0148***
(0.000)
Lagged (Sales)
Policy

Observations
R-Squared

5962320
0.405

-0.0168***
(0.002)
-0.0597***
(0.003)
0.0004***
(0.000)
-0.0166***
(0.002)
-0.0021a
(0.001)
-0.0066***
(0.001)
-0.9840***
(0.002)
-0.1217***
(0.003)
0.1801***
(0.001)
0.3129***
(0.002)
-0.6857***
(0.002)
1.4631***
(0.004)
0.0003
(0.003)
0.5516***
(0.002)
0.0750***
(0.002)
0.1981***
(0.003)
0.3505***
(0.004)
0.0076***
(0.000)
0.0041***
(0.000)
0.0012***
(0.000)
0.0013*
(0.001)
-0.0148***
(0.000)

4471740
0.406

-0.0043***
(0.001)
-0.0282***
(0.002)
-0.0000
(0.000)
-0.0048***
(0.001)
0.0012a
(0.001)
-0.0054***
(0.001)
-0.3048***
(0.001)
-0.0046*
(0.002)
0.0548***
(0.001)
0.1487***
(0.001)
-0.3563***
(0.001)
0.6285***
(0.003)
0.0333***
(0.002)
0.2013***
(0.001)
0.0302***
(0.001)
0.0749***
(0.002)
0.0693***
(0.002)
0.0021***
(0.000)
0.0008***
(0.000)
0.0006***
(0.000)
0.0006
(0.000)
-0.0046***
(0.000)
0.7128***
(0.000)
4306120
0.800

Note: Consumption tax indicates a change in the rate of consumption tax (in September 2008). Scrappage indicates a car scrappage
policy (from June 2009 to December 2010); Trucks indicates a policy to promote light trucks and vans in villages (from March 2009
to December 2010). a p<0.1, *p<0.05, ** p<0.01, ***p<0.001
26

Table 7-Car Buyer Group Targeting

HHI

Share

Panel A: Sales of small-engine cars ≤ 1.6 liter

0.0790***
(0.002)
Post
0.0692***
(0.003)
Time
0.0024***
(0.000)
Consumption Rate
0.0069***
(0.002)
Scrapped
-0.0008
(0.001)
Truck
0.0031*
(0.002)
Price
-0.0053***
(0.000)
HHIage
-0.4541***
(0.005)
HHIgender
1.5575***
(0.005)
Policy*HHIage
-0.1245***
(0.006)
Policy*HHIgender
0.2261***
(0.006)
Policy*Price
-0.0048***
(0.000)
GDP
0.0075***
(0.000)
Saving
-0.0037***
(0.000)
Unemployment Rate 0.0038***
(0.001)
Population
-0.0044***
(0.000)
Age≤30
Policy

Age＞50
Male
Policy*Age≤30
Policy*Age＞50
Policy*Male

0.0801***
(0.002)
0.0806***
(0.002)
0.0017***
(0.000)
0.0032*
(0.001)
-0.0021
(0.001)
0.0040**
(0.002)
-0.0051***
(0.000)

-0.0048***
(0.000)
0.0062***
(0.000)
-0.0031***
(0.000)
0.0030***
(0.001)
-0.0039***
(0.000)
0.0074***
(0.000)
0.0045***
(0.000)
0.0105***
(0.000)
0.0018***
(0.000)
-0.0003***
(0.000)
0.0007***
(0.000)
5962320
0.359

Observations
5962320
R-Square
0.298
Panel B：Sales of big-engine cars＞ 1.6 liter

Policy

-0.0675*** -0.0666***
27

(0.001)
(0.001)
***
Post
-0.0854
-0.0685***
(0.002)
(0.002)
***
Time
0.0002
-0.0006***
(0.000)
(0.000)
Consumption Rate
0.0002
0.0002
(0.001)
(0.001)
Scrapped
0.0027*
0.0060***
(0.001)
(0.001)
Truck
-0.0030*
0.0004
(0.001)
(0.001)
Price
0.0049*** 0.0050***
(0.000)
(0.000)
HHIage
-0.0355***
(0.004)
HHIgender
0.9682***
(0.004)
Policy*HHIage
0.0442***
(0.005)
Policy*HHIgender
-0.0422***
(0.005)
Policy*Price
0.0043*** 0.0043***
(0.000)
(0.000)
***
GDP
0.0077
0.0067***
(0.000)
(0.000)
***
Saving
-0.0005
-0.0001
(0.000)
(0.000)
***
Unemployment Rate 0.0022
0.0017**
(0.001)
(0.001)
Population
-0.0170*** -0.0163***
(0.000)
(0.000)
***
0.0031
Age≤30
(0.000)
0.0024***
Age＞50
(0.000)
Male
0.0090***
(0.000)
-0.0007***
Policy*Age≤30
(0.000)
**
-0.0001
Policy*Age＞50
(0.000)
Policy*Male
0.0001***
(0.000)
Observations
5962320 5962320
R-Square
0.263
0.296
Note: Consumption tax indicates a change in the rate of consumption tax (in September 2008). Scrappage indicates a car scrappage
policy (from June 2009 to December 2010); Trucks indicates a policy to promote light trucks and vans in villages (from March 2009
to December 2010).
a
p<0.1, *p<0.05, ** p<0.01, ***p<0.001

28

Figure 1: Monthly Sales of Small- and Large-engine Cars
1000000
900000
800000
700000
600000
500000
400000
300000
200000
100000
0

small

29

large

Appendix: Cities in Different Tiers in China
City Tier

Names of the Cities

1st City Tier

Shanghai, Beijing, Guangzhou, Shenzhen, Tianjin, Hangzhou, Nanjing, Chengdu, Wuhan,
Xi'an, Chongqing, Shenyang, Harbin, Urumqi, Zhengzhou, Suzhou.

(Large
metropolitan,
16 in Total)
2nd City Tier
(Middle size,
154 in Total)

3rd City Tier
(Small size,
194 in Total)

Kunming, Changsha, Dalian, Jinan, Lanzhou, Qingdao, Wuxi, Taiyuan, Hefei, Changchun,
Fuzhou, Xiamen, Nanning, Nanchang, Baotou, Haikou, Ningbo, Foshan, Xining,
Dongguan, Shijiazhuang, Hohhot, Yinchuan, Wenzhou, Guiyang, Luoyang, Changzhou,
Xuzhou, Tangshan, Jilin, Liuzhou, Zhuzhou, Weifang, Huizhou, Yantai, Nantong, Daqing,
Erdos, Datong, Zhanjiang, Guilin, Qiqihar, Mianyang, Zhongshan, Zibo, Hengyang,
Changde, Anqing, Anyang, Anshan, Bengbu, Baoji, Baoding, Beihai, Benxi, Binzhou,
Cangzhou, Chaozhou, Binzhou, Chifeng, Chuzhou, Deyang, Dezhou, Dongying, Fushun,
Ganzhou, Handan, Hulunbeier, Huzhou, Huaian, Huainan, Huanggang, Huangshi, Jixi,
Jining, Jiyuan, Jiamusi, Jiaxing, Jiayuguan, Jiangmen, Jiaozuo, Jinhua, Jingmen, Jingzhou,
Jiujiang, Kaifeng, Kelamayi, Lasa, Langfang, Leshan, Lianyungang, Liangshan, Liaocheng,
Linfen, Linyi, Longyan, Loudi, Luzhou, Lvliang, Maanshan, Maoming, Nanchong,
Nanyang, Panjin, Pingdingshan, Putian, Qinhuangdao, Qingyuan, Quanzhou, Rizhao,
Sanya, Shantou, Shangrao, Shaoxing, Shihezi, Siping, Songyuan, Taizhou, Taian, Tongliao,
Tongling, Weihai, Weinan, Wuhai, Wuhu, Wuzhou, Xianyang, Xiangtan, Xiaogan,
Xinxiang, Xinyu, Xingtai, Xuchang, Yanan, Yancheng, Yangzhou, Yangquan, Yibin,
Yichang, Yichun, Yingkou, Yulin, Yuulin, Yueyang, Yuncheng.
Akesu, Ankang, Anshun, Bayanzhuoer, Bayinguole, Baicheng, Baishan, Baiyin, Baise,
Baoshan, Bijie, Haozhou, Changji, Chaohu, Chaoyang, Chengde, Chizhou, Chuxiong,
Dazhou, Dali, Dandong, Ezhou, Fangchenggang, Fuzhou, Fuxin, Fuyang, Guangan,
Guangyuan, Guigang, Hami, Hanzhong, Hetian, Heyuan, Heze, Hezhou, Hebi, Hegang,
Heihe, Hengshui, Huludao, Huaihua, Huaibei, Huangshan, Jian, Jieyang, Jinchang, Jinzhou,
Jincheng, Jinzhong, Jingdezhen, Jiuquan, Kashi, Laibin, Laiwu, Lijiang, Lishui, Liaoyang,
Liaoyuan, Lincang, Liuan, Liupanshui, Luohe, Meishan, Meizhou, Mudanjiang, Nanping,
Neijiang, Ningde, Panzhihua, Pingliang, Pingxiang, Puyang, Puer, Qitaihe, Qinzhou,
Qingyang, Qujing, Sanmenxia, Sanming, Shanwei, Shangluo, Shangqiu, Shaoguan,
Shaoyang, Shiyan, Shuangyashan, Suzhou, Suihua, Suizhou, Suining, Tianmen, Tieling,
Tonghua, Tongchuan, Tulufan, Wulanchabu, Wuzhong, Wuwei, Xiantao, Zhangjiakou,
Zhangzhou, Changzhi, Zhaoqing, Zhenjiang, Zhoushan, Zhuhai, Zigong, Xianning, Qizhou,
Xinyang, Xinganmeng, Suqian, Suzhou, Xuancheng, Yaan, Yangjiang, Yichun, Yiyang,
Yingtan, Yongzhou, Yuxi, Yunfu, Zaozhuang, Zhangjiajie, Zhangye, Zhongwei, Zhoukou,
Zhumadian, Ziyang, Zunyi, Aba,Alaer, Alashan, Aletai, Ali, Bazhong, Boertala, Changdu,
Chongzuo, Daxinganling, Danzhou, Dehong, Diqing, Dingxi, Dongfang, Enshi, Gannan,
Ganzi, Guyuan, Guoluo, Haibei, Haidong, Hainan, Haixi, Hechi, Kenqu, Hengyang,
Honghe, Huaiyin, Huangnan, Juzhou, Kezilesu, Kuitun, Linzhi, Linxia, Longnan, Naqu,
Nanhai, Nujiang, Qianjiang, Qiandongnan, Qiannan, Qianxinan, Qiongbei, Qionghai,
Qiongnan, Quzhou, Rikeze, Shannan, Shennongjia, Shizuishan, Shunde, Simao, Tacheng,
Tianshui, Tongren, Wanning, Wenchang, Wenshan, Wuzhishan, Xishuangbanna, Xilinguole,
Xiangxi, Xiangfan, Suxian, Yanbian, Yangling, Yangpu, Yikezhao, Yilihasake, Yushu,
Zhaotong, Zhuzhou.

30