Content uploaded by Li Tang
Author content
All content in this area was uploaded by Li Tang on Feb 07, 2024
Content may be subject to copyright.
e China Review, Vol. 23, No. 4 (November 2023), 159–195
Zheng Wang is Associate Professor and Reader in Economics at the School of
Accounting, Finance and Economics, De Montfort University.
Li Tang is Full Professor at the School of International Relations and Public Aairs,
Fudan University. Correspondence should be sent to litang@fudan.edu.cn.
Cong Cao is Full Professor at the Nottingham University Business School China,
University of Nottingham Ningbo China.
Zhuo Zhou is Associate Professor at the Institute of World Economy, Shanghai
Academy of Social Sciences.
* e authors would like to thank two anonymous reviewers for their construc-
tive comments on the earlier dras of the article. Any mistakes are the author’s
alone.
e Impact of U.S.-China Tensions on
People Mobility
Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Abstra ct
Using novel monthly air passenger trac data, we assess the impact of
U.S.-China tensions on people inows from China to the U.S. We nd
that there was a 6 percent decline in air passenger ows from China to
the U.S. compared to other source countries during the period between
2017 and 2019. When dierentiated by geographical locations, relative
to other U.S. airports, U.S. airports near universities with a signicant
presence of Chinese students are found to have experienced a more
than 10 percent annual drop in passengers originating from China. A
further investigation reveals that the decline in people inows is mainly
attributed to the loss of passenger arrivals in August and that this
decline is consistently more signicant than the decrease experienced
by airports near tourist destinations during the same period. ese
ndings provide updated evidence of the detrimental eect a hostile
political climate could have on international people mobility between
two major scientic powers.
160 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
International scientific and technological exchanges are inextricably
intertwined with the economy, politics, and diplomacy. Looking back
into history, from the Coordinating Committee for Multilateral Export
Controls to its post-Cold War successor, the Wassenaar Agreement, and
from the 40-year Japan-U.S. trade war (1950s–1990s) within the
Western bloc to the mutual withdrawal of overseas students from China
and the former Soviet Union in the 1960s, scientic cooperation and
academic exchanges between nation states not only reect their bilat-
eral diplomacy, which is influenced by the political and economic
development environment, but also have a signicant impact on the
development of international relations.
China and the United States are the two most important nation states
today. Exchanges in science and education between the two countries
have a long history and can be traced back at least to the late Qing
dynasty. Aer the founding of the People’s Republic of China in 1949,
science and technology and higher education collaboration with the U.S.
was disrupted until in the early 1970s aer Nixon’s visit to China. In fact,
such collaboration became the earliest arena of Sino-U.S. cooperation.
e U.S.-China Agreement on Cooperation in Science and Technology (1979),
the very rst formal government agreement between the two countries,
launched a new era of top-down governmental collaboration. Since then,
despite the ups and downs in their bilateral diplomacy, overall U.S.-China
scientic and technological exchanges have sustained the momentum of
growth. Yet starting from 2011, and especially since 2018, the growing
tensions between the United States and China have become arguably the
most dominant events in international politics. As the effects of the
conict start to unfold, the damages caused by the changing political
climate between the two countries have also reached education and
academic research arenas.1
Despite accumulating anecdotal evidence, no study to our knowledge
has systematically investigated the impact of U.S.-China tension on bilat-
eral people ow and scientic and educational exchange. Using a newly
constructed dataset, we study the recent political tensions between two
economic and scientic powers and estimate the eect of the deterio-
rating relationship of the two countries on people inows from China to
the United States.
Our ndings enrich the literature on the consequences of political
tensions. Existing studies detect negative eects of political conicts on
bilateral trade,2 as well as on financial market performances.3 Our
e Impact of U.S.-China Tensions on People Mobility 161
research discloses that the damages caused by a worsening bilateral rela-
tionship also extend to people flows between countries. This study
contributes by providing evidence of how a turbulent political climate
between two countries aects international travels, particularly in the
context of knowledge-intensive activities. It refreshes and adds to the
accounts of the impact, at least indirect one, of skilled immigrants on
knowledge production, drawing from historical political shocks, such as
the diaspora of Soviet scientists,4 as well as German Jewish émigrés in the
United States.5
With another distinctive feature, this study combines highly disag-
gregated air trac data with geographical feature of airports and univer-
sities, which enables explorations of useful variations at a granular level
for a credible estimation of the short-term impact of political climate on
people ows and education exchange. is supplements the commonly
used bilateral migration data that is typically drawn from decennial
censuses of national governments and thus is limited in both geograph-
ical coverage at the country-pair level, and the time frequency and time-
liness of data.6
e rest of the paper proceeds as follows. e next section reviews
three lines of related literature and proposes research questions for
empirical investigations. Section 2 describes our research design and
data. The empirical results are presented and analyzed in section 3.
Section 4 concludes the article with a discussion on the limitations of this
research, further directions, and policy remarks.
1. Literature Review and Research Questions
We identify three strands of literature relevant to our research: the drivers
of the U.S.-China tension since 2018, its multi-dimensional impacts, and
determinants of international travels.
a. Causes of Post-2018 U.S.-China Tension
Since the diplomatic relationship of the United States and China was estab-
lished in 1979, the frictions and conicts between the two counties have
waxed and waned: the South China Sea disputes, arms sales to Taiwan, and
the Trans-Pacic Partnership Agreement are just a few of these issues. But
the ongoing U.S.-China trade war, beginning in 2018, is unique not only
for its wide scope and scale but also the positions of involved parties. It
162 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
started with tari battle and soon escalated and embroiled bilateral scien-
tic cooperation, technology markets, and talent mobility.
Scholars from dierent backgrounds have investigated the causes of
this trade war. Some believe that the trigger is trade imbalance, and former
U.S. President Donald Trump’s intensifying disputes with China were
sought to reduce the United States trade and scal decits.7 Some macro-
economists disagree. Stiglitz argued that the United States’ low saving and
tax cuts caused the Trump administration’s alarming fiscal deficit.8 As
noted by Lai, reducing the trade decit with China does not improve the
United States’ overall current account decit in the era of globalization as
other developing countries will sell similar goods to America.9
Another dominating view is that the then-Trump administration
aimed to halt China’s high-tech advancement and limit Chinese direct
overseas investment for national security reasons. International relations
scholars second this opinion. ey posit that the United States’ concern
about its declining supremacy and China’s rapid emergence as a chal-
lenger of U.S. hegemony catalyzed the U.S.-launched trade war.10 is line
of thinking is also popular among Chinese scholars.11 The term
“Thucydides Trap,”12 indicating the inevitability of war between a
declining superpower and a rising one, is adopted.13
Aligned with this international relation perspective, some scholars
believe that Trump’s goal of presidential reelection in 2020 was the
reason for the trade war. Autor et al. found that Chinese exports led to
unemployment and hardship for manufacturing workers in U.S. rural
areas,14 which mattered for Trump who had considerable support from
the Midwest and motivated him to take a harder stance against China.15
b. Consequences of Escalating Tension between the United States and
China
Undoubtedly, the escalating frictions between the two largest economies
have far-reaching eects. Much research has examined the impact of the
deteriorating relationship on U.S. prices, new automobile sales, welfare,
foreign direct investment in both China and the United States, and the
spillover eect on their trading partners, especially Asian economies.16 e
ndings are rather consistent: the conict would lead to a loss-loss situa-
tion for both sides. For instance, the Financial Times reports that the U.S.
tariff battle with China cost American colleges considerable revenue.17
According to the simulation by Itakura, the U.S.-China trade war would
e Impact of U.S.-China Tensions on People Mobility 163
bring about a reduction of nearly all sectoral imports and outputs in both
countries, and a 1.41 percent and 1.35 percent drop of the gross domestic
product (GDP) in China and the United States, respectively.18 He also
asserts that the spillover eect of the U.S.-China trade war would account
for a loss of 450 billion U.S. dollars in global trade.19
Not until recently has research explored if the U.S.-China scientic
collaboration can weather the tempestuous political fallout. For example,
based on evidence from joint publications and interviews, Woolston
noted no inuence of the U.S.-China tension on international scientic
collaboration, while acknowledging the increasing diculties of getting
nancial support for China-related work from the U.S. government.20
Recent studies on a possible U.S.-China decoupling argued that, without
intervention, the deteriorating relationship would wreak havoc on
commercial and scientic bonds.21 e China Initiative launched by the
U.S. Department of Justice to prosecute certain U.S.-based ethnic Chinese
science researchers and academics with links with Chinese research
institutions,22 the arrest of Meng Wanzhou, the then chief nancial ocer
of Chinese telecommunications giant Huawei, the U.S.’s tightened visa
scrutiny on Chinese students in science, technology, engineering, and
mathematics (STEM), and its chokepoint strategy of sanctioning Chinese
high-tech rms, to name just a few, are all expected to adversely aect
the ows between the U.S. and China.23 ough the aggregated statistics
of people inows and academic yearly enrolment data of international
students and visiting scholars are available from the U.S. Department of
Commerce and Institute for International Education (IIE) respectively, no
study to our knowledge has rigorously examined whether and, if any, to
what extent the intensifying U.S.-China conict aected international
travels, especially people inows, nor has its impact on educational and
academic exchange, controlling for other confounding factors, been
established.
c. Determinants of International Travels
A variety of factors aect international travel and people ows. Most of
this line of research is positioned in the eld of tourism management and
regional development. It has been largely accepted that supply and
demand jointly contribute to the dynamics of international travel.24
Demand factors of inbound tourism oen include population, income,
preferences, and expectations of inbound travelers and other features of
164 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
their country of origin.25 By contrast, supply factors of foreign ow oen
consist of characteristics of destination and arrival airports, such as
cultural and natural capital, income per capita, hotel capacity, flight
supply, environmental quality, agglomeration economies, and others.26
Regarding the effect of one-off events on international tourism,
existing studies largely focus on global pandemics and disasters, such as
the impact of avian flu, pandemic influenza, severe acute respiratory
syndrome (SARS), and Covid-19 Pandemic on international tourism and
Asian economies,27 as well as the enduring deterrence of the Chernobyl
nuclear accident on people inflows to Sweden.28 Yet no research has
empirically investigated whether a deteriorating bilateral relationship can
signicantly aect exchange activities between countries, while the post-
2018 U.S.-China tensions, as a quasi-natural experiment, offer rare
opportunities for an examination.
d. Research Questions
Integrating these three lines of literature, our study evaluates the impact
of U.S.-China political climate on U.S.-bound people ows from China in
general and on academic exchange in particular. Our empirical investiga-
tions can be summarized as being centered on two related research ques-
tions. e rst one is about the impact on international travels in general:
Question 1: How do the U.S.-China tensions aect people inows to
the U.S. dierently between those from China and those from other origin
countries?
We expect to see that, other factors (i.e., demanding factors associated
with country-of-origin and supply factors associated with destination and
arrival airports) held constant, the U.S.-China political tensions more
adversely aect inbound travelers from China than from other countries.
People travel internationally for a variety of reasons, including
business, education or leisure. e U.S. has long been the most favorable
destination for Chinese students and scholars to pursue their studies and
research careers abroad. Considering the ongoing U.S.-China decoupling
in science and technology, we propose the second research question that
explores the destination heterogeneity associated with the expected
impact in Question 1. It can be formulated as:
Question 2: Is the negative eect of U.S.-China tensions on the people
inows from China stronger for knowledge-intensive destinations than for
other destinations in the U.S.?
e Impact of U.S.-China Tensions on People Mobility 165
In our empirical analysis we use the number of passengers destined
for airports near university towns which are not popular tourist destina-
tions as a broad measure or proxy of travels for knowledge-intensive
activities. If there is an overall negative impact of the tensions as expected
following Question 1, this second question takes one step further by
asking how this impact varies with respect to the purpose of the visit
(knowledge-intensive activities versus others).
2. Data and Methods
a. Data Description and Measurement
Our primary source of data is OAG, a globally leading air trac database
that covers more than 99 percent of scheduled ights worldwide. We
retrieved the information on the monthly number of inbound passenger
arrivals by each origin country and each destination airport in the United
States.29 e original data spans 84 months from January 2013 through to
December 2019 (prior to the COVID-19 outbreak),30 and covers interna-
tional passenger ows from 228 countries or regions ending at 401 U.S.
airports. e total number of observations is 2,145,262, where the unit of
analysis represents a unique combination of an origin country (or
region), a U.S. airport, and a year-month. For quality check, we bench-
marked data against the ocial travel gures from the U.S. Department
of Transportation.31 Our original data is highly representative, accounting
for 96 percent of total international air passenger arrivals in the U.S. for
the period under study. Table 1 gives aggregate statistics of OAG data for
international passenger arrivals in the U.S., distinguishing between those
from China and from other countries, and also singling out those
arriving at university-town airports. It can be seen that in contrast to the
continued growth in passengers from other countries, passengers from
China declined in 2019. is unique drop in Chinese travelers is even
more prominent for passengers destined for university-town airports
(dened as airports located within a 100-mile radius of a university with
a signicant presence of Chinese students in the U.S.), and the decrease
started to happen in 2018.
166 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Table 1: Number of International Passenger Arrivals in the U.S. (ousand)
From
China
From all
countries
except
China
From all
countries
except China,
average
From China
at U.S.
university-
town airports
From all countries
except China at
U.S. university-
town airports
From all countries
except China at U.S.
university-town
airports, average
2013 2,772 80,080 359 267 7,458 34
2014 3,180 89,795 399 307 8,632 40
2015 3,254 95,226 419 325 8,901 41
2016 3,696 101,542 453 375 9,497 43
2017 3,869 106,988 476 389 9,911 46
2018 3,954 110,894 486 377 10,390 48
2019 3,894 113,746 501 353 10,691 50
Source: Authors’ calculation based on OAG.
One caveat with the OAG global air trac data, as with all other inter-
national air trac databases such as the International Civil Aviation Orga-
nization Aviation Data, is that it has no information about the travel
purposes of passengers. Given this data limitation, to estimate the potential
heterogeneous eects of U.S.-China tensions on people mobility, we adopt
two surrogates to exploit systematic variations in destination features and
seasonal travel patterns of these international travelers. The first one
approximates the purpose of passenger inows based on the distance of
their destination airport from nearest universities and tourist cities, and the
other one explores the differential impact on August, the peak arrival
month for international students and scholars, relative to other months, to
estimate the impact of U.S.-China tensions relevant to international educa-
tion and academic exchange. It is reasonable to believe that if the changed
U.S.-China relationship had a more signicant implication on education
and academic exchanges, the eect should be more pronounced for passen-
gers with destination airports nearer to knowledge-intensive areas and
arrivals in the months of August than the rest of the year.
e geolocation data of U.S. airports and their distances from the
nearest universities and tourist cities are obtained from Google Maps.
Table A1 (in appendices aer the main body of this article) provides
the names of U.S. universities with a signicant number of Chinese
students and Table A2 (in appendices) lists top U.S. tourist cities for
Chinese visitors. e three categories of nal destination airports are
dened as follows:
e Impact of U.S.-China Tensions on People Mobility 167
University-town airports (treated group 1): Airports located within a
100-mile radius of a university with a signicant presence of Chinese
students but outside the 100-mile radius of any major tourist cities.
Group size: 36 airports.
Tourist-city airports (treated group 2): Airports located within a
100-mile radius of a major tourist city but outside the 100-mile radius of
any universities with a signicant presence of Chinese students. Group
size: 37 airports.
Airports that are neither of the above (reference group): airports located
outside a 100-mile radius of any universities with a signicant number of
Chinese students and any major tourist cities. Group size: 285 airports.32
b. Descriptive Statistics
e rst glance of the data reveals that though the United States has been
one of the Chinese most favorite destinations for international travels
despite the long distance, its popularity is declining.33 As shown in Figure 1,
the United States is losing its ground not only to other traditionally popular
English-speaking countries outside Asia, but also to other destinations with
a reputation in scientific research and education. As noted in existing
studies, international travels, including education and research-oriented
ones, are inuenced by a variety of factors. Next, we adopt both dierence-
in-differences and difference-in-difference-in-differences estimation
approaches to investigate whether or not and to what extent political
tensions impacted people mobility from China to the U.S.
Table 2 describes the size and structure of the analytical samples, in
which a unit of observation is the combination of an origin country, a U.S.
airport, and a month in a year. In our baseline sample (sample A) where
the 100-mile radius is used to dene a university-town airport, we have
around 23 thousand observations, only moderately less than the number
of observations in the category of tourist-city airports, and about one
sixth of those observations in neither of these two categories. Table 3
reports the key summary statistics of our baseline analytical sample. Here
the number of passenger arrivals are converted to natural logarithms and
the timespan is split into two periods: pre-2018 (i.e. 2016–2017) and
post–2018 (i.e. 2018–2019). Comparing 2018–2019 with 2016–2017, it
can be seen that passenger ow from China declined, which is in contrast
to the growth of passengers from other countries. is pattern holds for
all types of international arrivals.
168 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
0 20 40 60
%
2013-Jan
2013-Aug
2014-Jan
2014-Aug
2015-Jan
2015-Aug
2016-Jan
2016-Aug
2017-Jan
2017-Aug
2018-Jan
2018-Aug
2019-Jan
2019-Aug
% China-originated air traffic to US among main English-speaking study destinations outside Asia
% China-originated air traffic to US among main study destinations
% China-originated air traffic to US among all destinations
Figure 1: Percentages of China-Originated Air Passengers Traveling to the United States.
Notes: “Main English-speaking destinations outside Asia” include the United States, the United
Kingdom, Canada, Australia, and New Zealand. “Main study destinations” include the
above countries plus Germany, Japan, and Singapore. “All destinations” include all travel
destinations (223 countries or regions) with ights from China.
Source: Passengers data from authors’ calculation based on OAG. e outbound ights from
China in this study comprise departure ights from China excluding Taiwan and the two
special administrative regions of Hong Kong and Macao.
168 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
0 20 40 60
%
2013-Jan
2013-Aug
2014-Jan
2014-Aug
2015-Jan
2015-Aug
2016-Jan
2016-Aug
2017-Jan
2017-Aug
2018-Jan
2018-Aug
2019-Jan
2019-Aug
% China-originated air traffic to US among main English-speaking study destinations outside Asi
a
% China-originated air traffic to US among main study destinations
% China-originated air traffic to US among all destinations
Figure 1: Percentages of China-Originated Air Passengers Traveling to the United States.
Notes: “Main English-speaking destinations outside Asia” include the United States, the United
Kingdom, Canada, Australia, and New Zealand. “Main study destinations” include the
above countries plus Germany, Japan, and Singapore. “All destinations” include all travel
destinations (223 countries or regions) with ights from China.
Source: Passengers data from authors’ calculation based on OAG. e outbound ights from
China in this study comprise departure ights from China excluding Taiwan and the two
special administrative regions of Hong Kong and Macao.
e Impact of U.S.-China Tensions on People Mobility 169
Table 2: Size and Structure of Analytical Samples
Sample A (distance criterion: 100-mile radius)
University-town
airports Tourist-city airports
Non-university-town &
non-tourist-city airports University &
airports
#
Obs
229,059 272,281 1,322,618 321,304
#
U.S. airports
36 37 285 43
#
Origin countries
225 227 227 227
#
Year-months
84 84 84 84
Sample B (distance criterion: 50-mile radius)
University-town
Tourist-city
Non-university-town &
non-tourist-city University &
airports
airports airports airports
#
Obs
138,260 219,823 1,579,052 208,127
#
U.S. airports
16 25 339 21
#
Origin countries
225 227 227 227
#
Year-months
84 84 84 84
Sample C (distance criterion: 150-mile radius)
University-town
Tourist-city
Non-university-town &
non-tourist-city University &
airports
airports airports airports
#
Obs
269,373 322,820 1,112,481 440,588
#
U.S. airports
48 43 240 70
#
Origin countries
222 227 227 227
#
Year-months
84 84 84 84
Notes: e unit of observation is a country-airport-year-month cell, where the country is the origin country and the airport is a U.S. destination airport.
Source: Authors’ calculation based on OAG.
170 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Table 3: Summary Statistics
China as origin country Other countries as origin country
2016-2017
Mean(SD)[N]
2018-2019
Mean(SD)[N]
2016-2017
Mean(SD)[N]
2018-2019
Mean(SD)[N]
Ln # passenger
arrivals
4.04 4.00 2.83 2.89
(2.34) (2.35) (2.04) (2.05)
[6,414] [6,516] [605,415] [606,617]
Ln # passenger
arrivals at university-
town airports
5.16 5.10 3.06 3.13
(1.94) (2.06) (2.04) (2.07)
[633] [625] [64,592] [64,069]
Ln # passenger
arrivals at tourist-city
airports
5.26 5.17 3.45 3.50
(2.55) (2.61) (2.23) (2.26)
[636] [645] [77,305] [77,393]
Ln # passenger
arrivals at non-
university-town &
non-tourist-city
airports
3.75 3.72 2.70 2.76
(2.27) (2.26) (1.98) (1.99)
[5,145] [5,246] [463,518] [465,155]
Notes: The whole data set consists of 228 origin countries, 401 U.S. arrival airports, and 84
months (January 2013 to December 2019). e rst number reported in each cell is the
mean, the second (in parentheses) is the standard deviation, and the third (in brackets) is
the number of observations.
Source: Authors’ calculation based on OAG.
Figure 2 plots the percentages of incoming passengers from China by
the above categories of airports. Compared with the reference group (i.e.,
airports that are neither near a university nor a tourist destination
popular with the Chinese) which shows a stable trend for the entire
sample period, the decrease in the share of passengers from China
becomes prominent after 2018, and is more so for university-town
airports than for tourist-destination airports. Furthermore, these
observed contrasts are starker for the peak travel month (August) than
for other times of the year.
e Impact of U.S.-China Tensions on People Mobility 171
Figure 2: Percentages of China-Originated Passengers in All International Air Passenger
Arrivals in U.S. Airports
Source: Authors’ calculation based on OAG.
c. Estimation Strategy
Dierence-in-Dierences (DD) Approach
To control for the inuence of other factors on people mobility, we start
the estimation with a dierence-in-dierences (DD) framework, which
can be expressed as:
where ln Pijym is the log number of air passenger arrivals from country i
in U.S. airport j in year y and month m; Ty denotes the year dummies
with 2017 as the reference year; Chinai is the dummy for China as the
origin country; δij captures all inuencing factors that are specic to a
given pair of origin country and arrival airport such as geographical
distance and number of universities near the destination; ωym captures
time trends that are common to all observations, such as changes in U.S.
or industry-level policies and seasonality in air trac;34 α is the intercept
and ∑ijy m the estimation residual; and βy, the associated year-specic coef-
cient of the interaction term Ty * Chinai, embodies the eect of U.S.-China
tensions (when y ≥ 2018) that is to be identied.
1 2 3 4 5 6 7
%
2013-Jan
2013-Aug
2014-Jan
2014-Aug
2015-Jan
2015-Aug
2016-Jan
2016-Aug
2017-Jan
2017-Aug
2018-Jan
2018-Aug
2019-Jan
2019-Aug
University-town airports
Tourist-city airports
Non-university town & non-tourist-city airports
e Impact of U.S.-China Tensions on People Mobility 171
Figure 2: Percentages of China-Originated Passengers in All International Air Passenger
Arrivals in U.S. Airports
Source: Authors’ calculation based on OAG.
c. Estimation Strategy
Di erence-in-Di erences (DD) Approach
To control for the in uence of other factors on people mobility, we start
the estimation with a di erence-in-di erences (DD) framework, which
can be expressed as:
where ln Pijym is the log number of air passenger arrivals from country i
in U.S. airport j in year y and month m; Ty denotes the year dummies
with 2017 as the reference year; Chinai is the dummy for China as the
origin country; δij captures all in uencing factors that are speci c to a
given pair of origin country and arrival airport such as geographical
distance and number of universities near the destination; ωym captures
time trends that are common to all observations, such as changes in U.S.
or industry-level policies and seasonality in air tra c;34 α is the intercept
and ∑ijy m the estimation residual; and βy, the associated year-speci c coef-
cient of the interaction term Ty * Chinai, embodies the e ect of U.S.-China
tensions (when y ≥ 2018) that is to be identi ed.
1 2 3 4 5 6 7
%
2013-Jan
2013-Aug
2014-Jan
2014-Aug
2015-Jan
2015-Aug
2016-Jan
2016-Aug
2017-Jan
2017-Aug
2018-Jan
2018-Aug
2019-Jan
2019-Aug
University-town airports
Tourist-city airports
Non-university town & non-tourist-city airports
172 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Intuitively, the DD design exploits variations within airport-time and
within country-airport pairs, and estimates (i) the difference in the
number of U.S. airport arrivals between passengers arriving from China
(the treated group) and other countries (the reference group) in the base
year (2017), and (ii) how the observed dierence in (i) (if any) changes
aer 2018. A negative sign of βy for y ≥ 2018 indicates a sharper drop in
the passenger trac from China than that from other countries.
Dierence-in-Dierence-in-Dierences (DDD) Approach
As our main target of quest is the impact on knowledge-intensive desti-
nations, a concern with βy is that a similar pattern could exist for other
destinations that are neither close to a university nor a popular tourist
city, in which case βy could be capturing an eect that was common to
all airports. To improve the credibility of the estimate, we therefore use a
dierence-in-dierence-in-dierences (DDD) approach by adding a new
comparison to Equation (1) to check the dierential impact on univer-
sity-town (or tourist-city) airports relative to airports that are distant
from university and tourist destinations:
where the new comparison comes from the triple interaction term Ty *
Chinai * Treatedj, in which Treatedj is an added dummy for university-
town (or tourist-city) airports; θiy is additional xed eects controlling for
country-year-specic confounding factors;35 and λy is the key parameter
to be estimated, allowing βy in Equation (1) to dier between the treated
and reference airports.36 In this DDD design, a negative sign of λy for y ≥
2018 would indicate how, relative to the passenger arrivals at the refer-
ence airports and from origins other than China, the passenger arrivals
at the treated airports from China are more adversely aected by the
U.S.-China frictions. Compared to the DD design, this specification
further explores variations across destinations with dierent levels of
concentrations of knowledge creation activities, thus getting closer to the
eect of interest.
e Impact of U.S.-China Tensions on People Mobility 173
3. Estimation Results
a. DD Estimation Results
e key estimated parameter βy and its 95 percent condence intervals
for dierent specications and subsets of airport are displayed in Figure 3
(also see columns (a) to (c) of Table A3 in appendices for the full estima-
tion results). e result on all U.S. airports suggests a 6 percent drop in
air passenger ows from China into the U.S. between 2017 and 2019,
relative to other source countries in the same period as shown in the
subgure (a). is deeper drop in passengers from China combined with
its timing is indicative of the impact of the tensions on people ows from
China in general, thus consistent with our expected impact in Question 1.
To put the estimated size of the impact in perspective, the number of
passenger arrivals from China in the U.S. in 2017 was 3.9 million. e
estimated eect of 6 percent drop is then equivalent to a loss of 234
thousand (i.e. 3,900,000*0.06) visits. Considering that the average number
of arrivals in the U.S. from countries other than China stands at 475
thousand in 2017, the above drop amounts to a reduction of nearly half
(234/475) of the number of international trips of an average country to
the U.S.
Subgure (b) shows that the estimated eect of U.S.-China tensions
on passenger arrivals in U.S. university-town airports is -11 percent in
2018 and dropped further down to -18 percent in 2019 at the signicance
level of 0.05. Again, to put the size of this eect in context, 389,000 trips
were made from China to university-town airports in the U.S. in 2017. So
the estimated eect amounts to a loss of 70 thousand (i.e. 389,000*0.18)
trips from China to these knowledge-intensive destinations, which is
about 1.5 (70/46) times the number of an average country’s trips to the
same destinations. In comparison, in subgure (c) a negative eect is also
found for passenger arrivals in tourist-city airports in 2019, but the size (12
percent) is smaller than that for university-town airports. is accords
with our prediction regarding the eect on travels for knowledge-inten-
sive activities in Question 2.
b. DDD Estimation Results
We add a third comparison to check the dierential impact of the DD
estimate for university-town or tourist-city airports relative to reference
airports, following the strategy in Equation (2). We have two treated
174 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
groups and one reference group of airports as dened previously: univer-
sity-town airports (treated group 1), tourist-city airports (treated group 2),
and airports that are neither a university-town nor a tourist-city airport
(reference group). e subgures (d) and (e) of Figure 3 visually display
the DDD estimation results for λy of Equation (2) (also see columns [d]
and [e] of Table A3 for the full estimation results). Indeed, we nd a
statistically differential effect for the two groups of treated airports
relative to the reference group, with the magnitude in 2019 being -11
percent for university-town airport arrivals (subgure [d] of Table A3).
Benchmarked against the total size of the ow from China into the U.S.
university-town airports in 2017, the magnitude of this eect amounts to
a reduction of approximately 43 thousand (i.e. 389,000*0.11) visits from
the country in a year. In contrast, the eect on tourist-city airport arrivals
is not statistically dierent from zero. Again, the expectation following
Question 2 is supported in this DDD estimation.
e fall semester of many U.S. universities starts in mid- or late-
August, while international travel for business or sightseeing is oen
spread throughout the year. So we re-estimate Equation (2) with
aviation data for August and for other months separately. is exercise
is conceptually equivalent to a quadruple-dierence design but para-
metrically less cumbersome. e results shown in Figure 4 support our
expected impact in Question 2 (also see Table A4 in appendices for the
full estimation results). For university-town airports, the estimated
eects for 2018 and 2019 are -21 percent and -28 percent respectively
(subgure [a], statistically signicant at the 5 percent level) for August
arrivals, whereas that for other months is statistically insignificant
(subgure [b]). is nearly 20-plus-percentage-point dierence suggests
that for university-town airports August arrivals of passengers from
China are indeed more adversely hit by the tensions than those in other
months, providing further evidence about the scale of the negative
impact of the U.S.-China tensions on academic inows to the United
States. For tourist-city airports, though the effect is -2 percent for
August arrivals (subgure [c]) and -7 percent for other months’ arrivals
(subgure [d]) in 2019, and neither turns out to be statistically signi-
cant at conventional levels.
e Impact of U.S.-China Tensions on People Mobility 175
0.03 -0.00 0.01 0.05 0.00 -0.01 -0.06
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(a) DD estimates
Sample: all airports
0.02 -0.01 -0.01 0.01 0.00
-0.11 -0.18
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(b) DD estimates
Sample: university-town airports
-0.04 -0.03 -0.03
0.06 0.00 0.00
-0.12
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(c) DD estimates
Sample: tourist-city airports
-0.08 -0.07 -0.08 -0.08
0.00
-0.10 -0.11
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(d) DDD estimates
Treated: university-town airports
-0.11 -0.08 -0.09 -0.01 0.00 -0.00 -0.07
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(e) DDD estimates
Treated: tourist-city airports
Figure 3: DD and DDD Estimates of the Eect of U.S.-China Tensions on Chinese Air
Passenger Arrivals in All 401 Airports, 36 University-Town Airports, and 37 Tourist-
City Airports in the U.S.
Notes: Estimated eect is extracted as the parameter βy, estimated from Equation (1), where 2017
is the reference year and countries other than China is the reference country group.
Subgures (d) and (e) display the DDD estimates of the eect of U.S.-China tensions on
Chinese air passenger arrivals in 36 university-town airports and 37 tourist-city airports in
the United States. Estimated effect is extracted as the parameter λy estimated from
Equation (2), where 2017 is the reference year and countries other than China is the
reference country group. In subgures (d) and (e), the reference airports (285 airports) are
those that are neither university-town nor tourist-city airports.
Source: Authors’ calculation based on OAG.
e Impact of U.S.-China Tensions on People Mobility 175
0.03 -0.00 0.01 0.05 0.00 -0.01 -0.06
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(a) DD estimates
Sample: all airports
0.02 -0.01 -0.01 0.01 0.00
-0.11 -0.18
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(b) DD estimates
Sample: university-town airports
-0.04 -0.03 -0.03
0.06 0.00 0.00
-0.12
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(c) DD estimates
Sample: tourist-city airports
-0.08 -0.07 -0.08 -0.08
0.00
-0.10 -0.11
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(d) DDD estimates
Treated: university-town airports
-0.11 -0.08 -0.09 -0.01 0.00 -0.00 -0.07
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(e) DDD estimates
Treated: tourist-city airports
Figure 3: DD and DDD Estimates of the E ect of U.S.-China Tensions on Chinese Air
Passenger Arrivals in All 401 Airports, 36 University-Town Airports, and 37 Tourist-
City Airports in the U.S.
Notes: Estimated e ect is extracted as the parameter βy, estimated from Equation (1), where 2017
is the reference year and countries other than China is the reference country group.
Sub gures (d) and (e) display the DDD estimates of the e ect of U.S.-China tensions on
Chinese air passenger arrivals in 36 university-town airports and 37 tourist-city airports in
the United States. Estimated effect is extracted as the parameter λy esti mated from
Equation (2), where 2017 is the reference year and countries other than China is the
reference country group. In sub gures (d) and (e), the reference airports (285 airports) are
those that are neither university-town nor tourist-city airports.
Source: Authors’ calculation based on OAG.
176 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Figure 4: DDD Estimated Eects of U.S.-China Tensions on Chinese Air Passenger Arrivals in
University-Town and Tourist-City Airports in the U.S., August versus Other Months
Notes: Estimated eect is extracted as the parameter λy estimated from Equation (2) using a DDD
strategy where 2017 is the reference year and countries other than China is in the reference
country group. e treated airports are university-town airports (36 airports) in subgures
(a) and (b), and are tourist-city airports (37 airports) in subfigures (c) and (d). The
reference airports (285 airports) are neither university-town nor tourist-city airports.
Source: Authors’ calculation based on OAG.
c. Robustness Checks
For robustness checks, we adopt alternative denitions of a university-
town or tourist-city airport in our estimate models. e distance criterion
is changed from a 100-mile radius to a 50-mile or 150-mile radius. For
university-town airports (Figure A2 in appendices), as expected, when
the distance criterion becomes more relaxed (i.e. the distance cuto is
higher), the size of the estimated effect gets smaller as more distant
airports are now included in the treated group that previously would
have been in the reference group. August is invariably the worst aected
month and is largely responsible for the overall negative impact esti-
mated. A generally similar pattern exists for tourist-city airports (Figure
A3 in appendices) when the distance cuto is reduced to 50 miles, but
the estimated eect for a 150-mile radius becomes statistically indierent
from zero. Table A5 (in appendices) contains the full estimation results.
-0.16 -0.20 -0.20 -0.17
0.00
-0.21 -0.28
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(a) Treated: university-town airports
August
-0.08 -0.06 -0.07 -0.07 0.00
-0.09 -0.09
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(b) Treated: university-town airports
Other months
-0.12 -0.04 0.02 0.01 0.00 0.07
-0.02
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(c) Treated: tourist-city airports
August
-0.11 -0.08 -0.09 -0.01 0.00 -0.01 -0.07
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(d) Treated: tourist-city airports
Other months
176 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Figure 4: DDD Estimated E ects of U.S.-China Tensions on Chinese Air Passenger Arrivals in
University-Town and Tourist-City Airports in the U.S., August versus Other Months
Notes: Estimated e ect is extracted as the parameter λy estimated from Equation (2) using a DDD
strategy where 2017 is the reference year and countries other than China is in the reference
country group. e treated airports are university-town airports (36 airports) in sub gures
(a) and (b), and are tourist-city airports (37 airports) in subfigures (c) and (d). The
reference airports (285 airports) are neither university-town nor tourist-city airports.
Source: Authors’ calculation based on OAG.
c. Robustness Checks
For robustness checks, we adopt alternative de nitions of a university-
town or tourist-city airport in our estimate models. e distance criterion
is changed from a 100-mile radius to a 50-mile or 150-mile radius. For
university-town airports (Figure A2 in appendices), as expected, when
the distance criterion becomes more relaxed (i.e. the distance cuto is
higher), the size of the estimated effect gets smaller as more distant
airports are now included in the treated group that previously would
have been in the reference group. August is invariably the worst a ected
month and is largely responsible for the overall negative impact esti-
mated. A generally similar pattern exists for tourist-city airports (Figure
A3 in appendices) when the distance cuto is reduced to 50 miles, but
the estimated e ect for a 150-mile radius becomes statistically indi erent
from zero. Table A5 (in appendices) contains the full estimation results.
-0.16 -0.20 -0.20 -0.17
0.00
-0.21 -0.28
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(a) Treated: university-town airports
August
-0.08 -0.06 -0.07 -0.07 0.00
-0.09 -0.09
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(b) Treated: university-town airports
Other months
-0.12 -0.04 0.02 0.01 0.00 0.07
-0.02
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(c) Treated: tourist-city airports
August
-0.11 -0.08 -0.09 -0.01 0.00 -0.01 -0.07
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(d) Treated: tourist-city airports
Other months
e Impact of U.S.-China Tensions on People Mobility 177
In Figure A4 (in appendices), we conduct a before-aer comparison
in a DDD setting by clustering years into two periods: before or aer
2018. e results oer a simple before-aer contrast and are consistent
with what has been estimated from the full-edged model (see Table A6
in appendices for the full estimation results). Overall, relative to the
pre-2018 period, Chinese passenger arrivals in U.S. university-town
airports dropped by 11 percent aer 2018, which is primarily driven by a
sharp decline (25 percent) in August arrivals. For tourist-city airports,
again, the negative eect is statistically and economically insignicant.
4. Conclusions and Discussions
a. Main Findings and Policy Implications
Using disaggregated passenger trac data, this study reveals an uncom-
fortable and inconvenient truth about the impact of international politics
on people mobility and knowledge ows across countries. It shows that
the post-2018 U.S.-China tensions have already led to an alarmingly
signicant drop in the number of passengers traveling from China to
knowledge-intensive destinations in the U.S., more than that is observed
for air trac from other countries and that for other U.S. destinations.
Such a drop is found to exist robustly even aer accounting for system-
atic dierences across airports and sending countries, seasonality, and all
possible interactions between factors along these dimensions. Our
ndings corroborate with the patterns from the recently released IIE data
which show a similarly gloomy but less nuanced picture of the situation.
e sharp drop of China-originated people inows is a reasonably
good surrogate of the loss of Chinese students and scholars in the U.S.
universities, which has implications far beyond nancial hardship for
some universities. Historically, the scientic supremacy of the United
States has been deeply rooted in its capacity of tapping into the pool of
global talent allowing the country to retain a signicant number of
U.S.-trained students, especially those at the highest-end who complete
a PhD degree. On the one hand, in light of the high research produc-
tivity of Chinese PhD graduates in U.S. universities,37 we expect that
the reconsideration by young Chinese scientists of study and career
locations away from the United States would lead to the decline of
research productivity in U.S. institutions and enterprises where there
has been a signicant dependency on these academics for day-to-day
178 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
research activities. 38 On the other hand, China’s scientic leap forward
has beneted from skilled returnees from the United States.39 It will be
an interesting avenue for future research to evaluate the incurred
damage to knowledge production on both sides of the Pacic as well as
beyond. Moreover, the less mobility of talent from China to the U.S.,
among other adverse factors, has led to the sharply decrease in the
collaboration of scientists between both countries measured by the
number of joint publications.40
b. Limitations and Future Research
Admittedly, this research is not without limitations. First, our identica-
tion comes from geographical and time variations, and thus the accuracy
of our estimate depends partly on the measurement accuracy and how
the reference groups are aected by the politico-economic tension. For
instance, our measure for academic exchange assumes most enrolled
student or visiting scholars y to the nearest airport to their campus in
August. Yet it is possible that they arrive at non-university-town airports,
or academic personnel exchanges start in months other than August.
Inbound passengers can also be returning Americans residing abroad and
other non-Chinese nationals. For the U.S. universities with multiple
campuses, we use the longitude-latitude information of their main
campuses, possibly miscoding the airport for branch campuses and
leading to an inaccurate estimate of impact.
Secondly, students and scholars from countries other than China
may also have been affected, directly or indirectly, by U.S.-China
tensions or other country-level factors correlated with them in two
directions. On the one hand, as the U.S. domestic immigration policy
and its relationships with the European Union and the Middle East also
tightened or worsened during the same period as the U.S.-China trade
war, a negative eect could also exist for passenger ows from these
countries aer 2018. However, as these origin countries are part of the
reference group and are thus dierenced out in our estimations, our
estimated eect on China could be interpreted as a lower bound of the
true effect. On the other hand, the fact that U.S. universities may
expand their admissions of students and receive more researchers and
academic visitors from other parts of the world to compensate for the
loss of talent from China could lead to an overestimation of the impact
on the inows of Chinese passengers in the U.S.
e Impact of U.S.-China Tensions on People Mobility 179
irdly, this study only covers a relatively short time aer the trade
war. Although no evidence suggests that the tensions in the Sino-U.S.
relations have substantially eased during the Biden area—in fact the
rivalry has even worsened in certain areas including matters related to
Taiwan and sanctions on advanced technology and parts—it would be
useful to have an updated assessment of the impacts to incorporate more
recent events.
In light of these limitations, several questions are worthy of further
investigation. To begin with, it would be interesting to explore the spill-
over eect of escalating U.S.-China tensions on China and other coun-
tries’ international people ow. In this study, we only consider unilateral
inbound flows from China and other countries to the United States.
Future work could extend to explore how political tensions impact U.S.-
origin passengers travelling to China, or multilateral business and
tourism ows among dierent origin and destination countries.
Secondly, in addition to tertiary education, there has been an
increasing number of Chinese students attending U.S. high schools. It
would be interesting to check if they are less inuenced by the tightened
U.S. visa policy on China.
irdly, in this research the possible collateral damage of political
tensions on tourist receipts are only checked very broadly in the back-
ground by looking at passenger arrivals at tourist cities. Subject to avail-
ability on more precise and geographically granular tourist data, future
research could explore this impact more explicitly.41 e year 2023 is the
sixth of the simmering political tensions between the world’s two largest
economies. Because of the outbreak and rapid spread of COVID-19,
suspension of ights between the U.S. and mainland China and travel
warnings have delivered a heavier blow on top of U.S.-China tourism,
academic exchanges and knowledge coproduction.42 Their impacts,
combined with the escalating Sino–U.S. rivalry, the ongoing war in
Ukraine, the Chinese balloon incident and among others cast more
turbulence and uncertainties on the global landscape of innovation and
politics which could go far beyond what we expect.
180 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Appendices
Table A1: List of top 25 U.S. Universities Recruiting Chinese Students
Rank (by # of F1
visas issued) Name of university
1 University of Illinois Urbana-Champaign
2 University of Southern California
3 Purdue University
4 Northeastern University
5 Columbia University
6 Michigan State University
7 Ohio State University
8 University of California, Los Angeles
9 Indiana University
10 University of California, Berkeley
11 New York University
12 Pennsylvania State University
13 University of Minnesota
14 University of Washington Seattle
15 Arizona State University
16 University of Michigan Ann Arbor
17 Boston University
18 Illinois Institute of Technology
19 Rutgers, e State University of New Jersey
20 University of Texas at Dallas
21 University of Wisconsin-Madison
22 University of California, San Diego
23 Carnegie Mellon University
24 State University of New York at Stony Brook
Syracuse University
Source: https://foreignpolicy.com/2016/01/04/the-most-chinese-schools-in-america-rankings-
data-education-china-u/.
e Impact of U.S.-China Tensions on People Mobility 181
Table A2: List of U.S. Tourist Cities Most Popular with Chinese Visitors
Name of tourist city Chinese characters
Atlanta 亞特蘭大
Baltimore 巴爾的摩
Boston 波士頓
Bualo 水牛城
Chicago 芝加哥
Dallas 達拉斯
Detroit 底特律
Guam 關島
Hawaii 夏威夷
Honolulu 檀香山
Las Vegas 拉斯維加斯
Los Angeles 洛杉磯
Miami 邁阿密
Monterey 加州蒙特雷
New Orleans 新奧爾良
New Yor k 紐約
Orlando 奧蘭多
Philadelphia 費城
Portland 波特蘭
Saipan 塞班島
Salt Lake City 鹽湖城
San Diego 加州聖地亞哥
San Francisco 舊金山
San Jose 加州聖荷西
Santa Barbara 聖巴巴拉
Seattle 西雅圖
Washington D.C. 華盛頓
Source: e list is based on compiled information from multiple leading Chinese providers of
travel services including Ctrip, Qiongyou, and Mafengwo.
182 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Table A3: Baseline DD and DDD Estimates, Dep. Var.: Log U.S. Airport International Air Passenger Arrivals
(Unit of Observation: Country-Airport-Year-Month)
DD estimates DDD estimate
Sample: all airports Subsample: university-
town airports
Subsample: tourist-
city airports
Treated airports:
university-town
airports
Treated airports:
tourist-city airports
(a) (b) (c) (d) (e)
T2013*China 0.027 0.025 -0.044
(0.020) (0.024) (0.024)
T2014*China -0.001 -0.005 -0.034
(0.016) (0.019) (0.020)
T2015*China 0.013 -0.009 -0.025
(0.013) (0.016) (0.018)
T2016*China 0.047 0.007 0.061
(0.008) (0.011) (0.011)
T2018*China -0.011 -0.108 0.000
(0.007) (0.009) (0.009)
T2019*China -0.063 -0.183 -0.119
(0.009) (0.011) (0.010)
T2013*China*TreatedAirports -0.083 -0.106
(0.052) (0.052)
T2014*China*TreatedAirports -0.075 -0.076
(0.052) (0.052)
(Continued on next page)
e Impact of U.S.-China Tensions on People Mobility 183
T2015*China*TreatedAirports -0.081 -0.086
(0.052) (0.052)
T2016*China*TreatedAirports -0.078 -0.014
(0.052) (0.052)
T2018*China*TreatedAirports -0.102 -0.005
(0.053) (0.052)
T2019*China*TreatedAirports -0.108 -0.072
(0.053) (0.052)
Orig-dest FE Yes Ye s Ye s Ye s Ye s
Year-month FE Yes Ye s Ye s No No
Orig-year FE No No No Yes Ye s
Dest-year-month FE No No No Ye s Ye s
Obs 2,139,185 228,551 271,965 1,545,406 1,588,680
Adj. R20.905 0.913 0.925 0.886 0.894
Notes: Time period: January 2013 to December 2019. Reference year (omitted): 2017. In the indicators of xed eects (FE) used, “Orig” means departure
country, and “dest” means destination airport in the U.S. Standard errors reported in parentheses are clustered by country (columns (a)–(c)) or
country-airport (columns (d)–(g)). *p < .05. **p < .01. ***p < .005.
184 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Table A 4: DDD Estimates, August versus Other Months, Dep. Var.: Log U.S. Airport International
Air Passenger Arrivals (Unit of Observation: Country-Airport-Year-Month)
Treated airports:
university-town airports
Treated airports:
tourist-city airports
August Other months August Other months
(a) (b) (c) (d)
T2013*China*TreatedAirports -0.164 -0.078 -0.116 -0.106
(0.112) (0.071) (0.135) (0.115)
T2014*China*TreatedAirports -0.202 -0.061 -0.040 -0.079
(0.103) (0.063) (0.118) (0.079)
T2015*China*TreatedAirports -0.200 -0.068 0.015 -0.092
(0.102) (0.055) (0.086) (0.055)
T2016*China*TreatedAirports -0.170 -0.070 0.009 -0.014
(0.090) (0.040) (0.086) (0.041)
T2018*China*TreatedAirports -0.211 -0.090 0.067 -0.013
(0.092) (0.062) (0.091) (0.054)
T2019*China*TreatedAirports -0.279 -0.087 -0.016 -0.074
(0.111) (0.064) (0.122) (0.052)
Orig-dest FE Yes Ye s Ye s Yes
Orig-year FE Yes Ye s Ye s Ye s
Dest-year-month FE No Yes No Ye s
Obs 132,685 1,405,930 135,784 1,446,226
Adj. R20.902 0.887 0.908 0.895
Notes: Time period: from January 2013 to December 2019. Reference year (omitted): 2017. Standard
errors reported in parentheses are clustered by country-airport. *p < .05. **p < .01.
***p < .005.
e Impact of U.S.-China Tensions on People Mobility 185
Table A5: Robustness Checks I—Alternative Denitions of University-Town and Tourist-
City Airports, Dep. Var.: Log U.S. Airport International Air Passenger Arrivals
(Unit of Observation: Country-Airport-Year-Month)
(A) Treated airports: university-town airports
Radius: 50 miles Radius: 150miles
(a) (b) (c) (d) (e) (f)
T2013*China*TreatedAirports -0.030 -0.028 -0.031 0.029 -0.078 0.035
(0.074) (0.129) (0.074) (0.084) (0.136) (0.083)
T2014*China*TreatedAirports -0.108 -0.186 -0.101 -0.014 -0.123 -0.002
(0.066) (0.121) (0.069) (0.078) (0.126) (0.078)
T2015*China*TreatedAirports -0.088 -0.196 -0.079 0.008 -0.126 0.022
(0.053) (0.112) (0.052) (0.057) (0.112) (0.057)
T2016*China*TreatedAirports -0.070 -0.100 -0.069 -0.012 -0.099 -0.005
(0.033) (0.100) (0.032) (0.044) (0.098) (0.043)
T2018*China*TreatedAirports -0.097 -0.178 -0.091 -0.079 -0.184 -0.071
(0.035) (0.157) (0.033) (0.057) (0.109) (0.057)
T2019*China*TreatedAirports -0.129 -0.247 -0.118 -0.075 -0.168 -0.065
(0.052) (0.089) (0.054) (0.058) (0.113) (0.057)
Orig-dest FE Yes Ye s Ye s Ye s Yes Ye s
Orig-year FE Yes Ye s Ye s Yes Ye s Ye s
Dest-year-month FE Yes No Ye s Ye s No Yes
Obs 1,710,478 146,918 1,556,091 1,376,156 118,163 1,251,919
Adj. R20.886 0.901 0.887 0.884 0.901 0.885
(B) Treated airports: tourist city airport
Radius: 50 miles Radius: 150miles
(a) (b) (c) (d) (e) (f)
T2013*China*TreatedAirports -0.108 -0.061 -0.111 -0.033 -0.036 -0.033
(0.151) (0.167) (0.151) (0.103) (0.130) (0.105)
T2014*China*TreatedAirports -0.124 -0.126 -0.121 0.005 -0.011 0.008
(0.094) (0.120) (0.095) (0.079) (0.107) (0.080)
T2015*China*TreatedAirports -0.122 -0.091 -0.122 -0.039 0.006 -0.041
(0.065) (0.074) (0.068) (0.052) (0.086) (0.053)
T2016*China*TreatedAirports -0.059 0.058 -0.068 0.022 0.090 0.017
(0.039) (0.069) (0.043) (0.040) (0.093) (0.042)
T2018*China*TreatedAirports -0.084 -0.002 -0.091 0.039 0.116 0.028
(0.027) (0.079) (0.026) (0.047) (0.112) (0.048)
T2019*China*TreatedAirports -0.095 -0.143 -0.089 0.108 0.206 0.100
(0.055) (0.100) (0.054) (0.071) (0.131) (0.068)
Orig-dest FE Ye s Ye s Ye s Yes Ye s Yes
Orig-year FE Yes Ye s Yes Ye s Yes Ye s
Dest-year-month FE Ye s No Ye s Ye s No Yes
Obs 1,791,903 153,511 1,630,856 1,429,801 122,009 1,301,904
Adj. R20.893 0.907 0.894 0.896 0.910 0.896
Notes: Time period: January 2013 to December 2019. Reference year (omitted): 2017. Standard errors
reported in parentheses are clustered by country-airport. *p < .05. **p < .01. ***p < .005.
186 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Table A6: Robustness Checks II—Clustering Years, Dep. Var.: Log U.S. Airport International
Air Passenger Arrivals (Unit of Observation: Country-Airport-Year-Month)
Treated: university-town
airports Treated: tourist-city airports
(a) (b) (c) (d) (e) (f)
Post2018*China*Treated Airports -0.113 -0.253 -0.096 -0.030 0.015 -0.033
(0.030) (0.096) (0.063) (0.029) (0.093) (0.030)
Orig-dest FE Yes Ye s Ye s Ye s Yes Ye s
Orig-year FE Ye s Ye s Yes Ye s Ye s Yes
Dest-year-month FE Yes Ye s Ye s Ye s Yes Ye s
Obs 656,962 52,438 597,652 676,881 53,936 616,089
Adj. R20.895 0.914 0.896 0.904 0.922 0.905
Notes: Time period: January 2013 to December 2019. Reference year (omitted): 2017. Standard errors
reported in parentheses are clustered by country-airport. *p < .05. **p < .01. ***p < .005.
Figure A1: Location of the ree Categories of Airports in the Research
Notes: University-town airports (36 airports) are dened as airports located within a 100-mile
radius of a university with a significant presence of Chinese students but outside the
100-mile radius of any major tourist cities. Tourist city airports (37 airports) are dened as
airports located within a 100-mile radius of a major tourist-city but outside the 100-mile
radius of any universities with a signicant presence of Chinese students. Non-university-
town and non-tourist-city airports (285 airports) are dened as airports located outside
the 100-mile radius of any universities with a signicant number of Chinese students and
any major tourist cities.
-0.16 -0.20 -0.20 -0.17
0.00
-0.21 -0.28
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(a) Treated: university-town airports
August
-0.08 -0.06 -0.07 -0.07 0.00
-0.09 -0.09
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(b) Treated: university-town airports
Other months
-0.12 -0.04 0.02 0.01 0.00 0.07
-0.02
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(c) Treated: tourist-city airports
August
-0.11 -0.08 -0.09 -0.01 0.00 -0.01 -0.07
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(d) Treated: tourist-city airports
Other months
186 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Table A6: Robustness Checks II—Clustering Years, Dep. Var.: Log U.S. Airport International
Air Passenger Arrivals (Unit of Observation: Country-Airport-Year-Month)
Treated: university-town
airports Treated: tourist-city airports
(a) (b) (c) (d) (e) (f)
Post2018*China*Treated Airports -0.113 -0.253 -0.096 -0.030 0.015 -0.033
(0.030) (0.096) (0.063) (0.029) (0.093) (0.030)
Orig-dest FE Yes Yes Yes Yes Yes Yes
Orig-year FE Yes Yes Yes Yes Yes Yes
Dest-year-month FE Yes Yes Yes Yes Yes Yes
Obs 656,962 52,438 597,652 676,881 53,936 616,089
Adj. R20.895 0.914 0.896 0.904 0.922 0.905
Notes: Time period: January 2013 to December 2019. Reference year (omitted): 2017. Standard errors
reported in parentheses are clustered by country-airport. *p < .05. **p < .01. ***p < .005.
Figure A1: Location of the ree Categories of Airports in the Research
N otes: University-town airports (36 airports) are de ned as airports located within a 100-mile
radius of a university with a significant presence of Chinese students but outside the
100-mile radius of any major tourist cities. Tourist city airports (37 airports) are de ned as
airports located within a 100-mile radius of a major tourist-city but outside the 100-mile
radius of any universities with a signi cant presence of Chinese students. Non-university-
town and non-tourist-city airports (285 airports) are de ned as airports located outside
the 100-mile radius of any universities with a signi cant number of Chinese students and
any major tourist cities.
-0.16 -0.20 -0.20 -0.17
0.00
-0.21 -0.28
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(a) Treated: university-town airports
August
-0.08 -0.06 -0.07 -0.07 0.00
-0.09 -0.09
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(b) Treated: university-town airports
Other months
-0.12 -0.04 0.02 0.01 0.00 0.07
-0.02
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(c) Treated: tourist-city airports
August
-0.11 -0.08 -0.09 -0.01 0.00 -0.01 -0.07
-.5 -.25 0 .25 .5
Estimated effect
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(d) Treated: tourist-city airports
Other months
e Impact of U.S.-China Tensions on People Mobility 187
Figure A2: Robustness Checks I (a)—Alternative Denitions of University-Town Airports.
Notes: Estimated eect is extracted as the parameter λy estimated from Equation (2) using a DDD
strategy where 2017 is the reference year and other countries than China are the reference
country group. e treated airports are university-town airports (16 airports in subgures
(a), (b), and (c); 48 airports in subgures (d), (e), and (f)). e reference airports are those
that are neither university-town nor tourist-city airports (339 airports in subgures (a), (b),
and (c); 240 airports in subgures (d), (e), and (f)). Denitions of these airports follow the
text or Figure 3 except that the distance criterion is now 50-mile radius in subgures (a), (b),
and (c), and 150-mile radius in subfigures (d), (e), and (f). See Table A5 for the full
estimation results.
Source: Authors’ calculation based on OAG.
e Impact of U.S.-China Tensions on People Mobility 187
Figure A2: Robustness Checks I (a)—Alternative De nitions of University-Town Airports.
-0.03
-0.11 -0.09 -0.07
0.00
-0.10 -0.13
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(a) Treated: university-town airports
50 miles radius, all months
-0.03
-0.19 -0.20
-0.10
0.00
-0.18
-0.25
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(b) Treated: university-town airports
50 miles radius, August
-0.03
-0.10 -0.08 -0.07
0.00
-0.09 -0.12
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(c) Treated: university-town airports
50 miles radius, other months
0.03 -0.01 0.01 -0.01 0.00
-0.08 -0.08
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(d) Treated: university-town airports
150 miles radius, all months
-0.08 -0.12 -0.13 -0.10
0.00
-0.18 -0.17
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(e) Treated: university-town airports
150 miles radius, August
0.03 -0.00 0.02 -0.01 0.00
-0.07 -0.07
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(f) Treated: university-town airports
150 miles radius, other months
(a) Treated: university-town airports
50-mile radius, all months
(b) Treated: university-town airports
50-mile radius, August
(c) Treated: university-town airports
50-mile radius, other months
(d) Treated: university-town airports
150-mile radius, all months
(e) Treated: university-town airports
150-mile radius, August
(f) Treated: university-town airports
150-mile radius, other months
No tes: Estimated e ect is extracted as the parameter λy estimated from Equation (2) using a DDD
strategy where 2017 is the reference year and other countries than China are the reference
country group. e treated airports are university-town airports (16 airports in sub gures
(a), (b), and (c); 48 airports in sub gures (d), (e), and (f )). e reference airports are those
that are neither university-town nor tourist-city airports (339 airports in sub gures (a), (b),
and (c); 240 airports in sub gures (d), (e), and (f)). De nitions of these airports follow the
text or Figure 3 except that the distance criterion is now 50-mile radius in sub gures (a), (b),
and (c), and 150-mile radius in subfigures (d), (e), and (f). See Table A5 for the full
estimation results.
Source: Authors’ calculation based on OAG.
Estimated effect
Estimated effect
Estimated effect
Estimated effect
Estimated effect
Estimated effect
188 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Figure A3: Robustness Checks I(b)—Alternative Denitions of Tourist City Airports
Notes: Estimated eect is extracted as the parameter λy estimated from Equation (2) using a DDD
strategy where 2017 is the reference year and other countries than China are the reference
country group. e treated airports are tourist city airports (25 airports in subgures (a), (b),
and (c); 43 airports in subgures (d), (e), and (f)). e reference airports are those that are
neither university-town nor tourist-city airports (339 airports in subgures (a), (b), and (c);
240 airports in subgures (d), (e), and (f)). Denitions of these airports follow the text and
Figure 3 except that the distance criterion is now 50-mile radius in subgures (a), (b), and
(c), and 150-mile radius in subgures (d), (e), and (f). See Table A6 for the full estimation
results.
Source: Authors’ calculation based on OAG.
188 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Figure A3: Robustness Checks I(b)—Alternative De nitions of Tourist City Airports
-0.11 -0.12 -0.12 -0.06 0.00
-0.08 -0.09
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(a) Treated: tourist-city airports
50 miles radius, all months
-0.06
-0.13 -0.09
0.06 0.00 -0.00
-0.14
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(b) Treated: tourist-city airports
50 miles radius, August
-0.11 -0.12 -0.12 -0.07
0.00
-0.09 -0.09
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(c) Treated: tourist-city airports
50 miles radius, other months
-0.03 0.00 -0.04 0.02 0.00 0.04
0.11
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(d) Treated: tourist-city airports
150 miles radius, all months
-0.04 -0.01 0.01
0.09
0.00
0.12
0.21
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(e) Treated: tourist-city airports
150 miles radius, August
-0.03 0.01 -0.04 0.02 0.00 0.03
0.10
-.5 -.25 0 .25 .5
2013 2014 2015 2016 2017 2018 2019
95% CI
Estimated effect
(f) Treated: tourist-city airports
150 miles radius, other months
(a) Treated: tourist-city airports
50-mile radius, all months
(b) Treated: tourist-city airports
50-mile radius, August
(c) Treated: tourist-city airports
50-mile radius, other months
(d) Treated: tourist-city airports
150-mile radius, all months
(e) Treated: tourist-city airports
150-mile radius, August
(f) Treated: tourist-city airports
150-mile radius, other months
Not es: Estimated e ect is extracted as the parameter λyestimated from Equation (2) using a DDD
strategy where 2017 is the reference year and other countries than China are the reference
country group. e treated airports are tourist city airports (25 airports in sub gures (a), (b),
and (c); 43 airports in sub gures (d), (e), and (f )). e reference airports are those that are
neither university-town nor tourist-city airports (339 airports in sub gures (a), (b), and (c);
240 airports in sub gures (d), (e), and (f )). De nitions of these airports follow the text and
Figure 3 except that the distance criterion is now 50-mile radius in sub gures (a), (b), and
(c), and 150-mile radius in sub gures (d), (e), and (f ). See Table A6 for the full estimation
results.
Source: Authors’ calculation based on OAG.
Estimated effect
Estimated effect
Estimated effect
Estimated effect
Estimated effect
Estimated effect
e Impact of U.S.-China Tensions on People Mobility 189
Figure A4: Robustness Checks II—Clustering Years for a Before-Aer Comparison
Notes: Estimated eect is extracted as the parameter λy estimated from a modied version of
Equation (2) using a DDD strategy where Ty is replaced by a period dummy, post2018,
which takes on the value of one when y2018 and zero otherwise. Countries other than
China are the reference country group. e treated airports are university-town airports (36
airports) in subfigures (a), (b), and (c), and are tourist city airports (37 airports) in
subgures (d), (e) and (f). In all subgures, the reference airports (285 airports) are those
that are neither university-town nor tourist-city airports. See Section 2-b for the exact
denitions of these airports and Table A6 for the full estimation results.
Source: Authors’ calculation based on OAG.
e Impact of U.S.-China Tensions on People Mobility 189
Figure A4: Robustness Checks II— er Comparison
0.00
-0.11
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
(a) Treated: iversity-town airports
All months
0.00
-0.25
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
August
0.00
-0.10
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
Other months
0.00 -0.03
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(d) Treated: tourist-city airports
All months
0.00 0.01
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(e) Treated: tourist-city airports
August
0.00 -0.03
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(f) Treated: tourist-city airports
Other months
(a) Treated: university-town airports
All months
(b) Treated: university-town airports
August
(c) Treated: university-town airports
Other months
(d) Treated: tourist-city airports
All months
(e) Treated: tourist-city airports
August
(f) Treated: tourist-city airports
Other months
ect is extracted as the parameter λy ed version of
Equation (2) using a DDD strategy where Tyis replaced by a period dummy, post2018,
which takes on the value of one when y2018 and zero otherwise. Countries other than
e treated airports are university-town airports (36
airports) in subgures (a), (b), and (c), and are tourist city airports (37 airports) in
gures, the reference airports (285 airports) are those
that are neither university-town nor tourist-city airports. See Section 2-b for the exact
nitions of these airports and Table A6 for the full estimation results.
Source: Authors’ calculation based on OAG.
e Impact of U.S.-China Tensions on People Mobility 189
Figure A4: Robustness Checks II— er Comparison
0.00
-0.11
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
(a) Treated: iversity-town airports
All months
0.00
-0.25
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
August
0.00
-0.10
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
Other months
0.00 -0.03
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(d) Treated: tourist-city airports
All months
0.00 0.01
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(e) Treated: tourist-city airports
August
0.00 -0.03
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(f) Treated: tourist-city airports
Other months
(a) Treated: university-town airports
All months
(b) Treated: university-town airports
August
(c) Treated: university-town airports
Other months
(d) Treated: tourist-city airports
All months
(e) Treated: tourist-city airports
August
(f) Treated: tourist-city airports
Other months
ect is extracted as the parameter λy ed version of
Equation (2) using a DDD strategy where Tyis replaced by a period dummy, post2018,
which takes on the value of one when y2018 and zero otherwise. Countries other than
e treated airports are university-town airports (36
airports) in subgures (a), (b), and (c), and are tourist city airports (37 airports) in
gures, the reference airports (285 airports) are those
that are neither university-town nor tourist-city airports. See Section 2-b for the exact
nitions of these airports and Table A6 for the full estimation results.
Source: Authors’ calculation based on OAG.
Estimated effect
Estimated effect
Estimated effect
Estimated effect
Estimated effect
Estimated effect
e Impact of U.S.-China Tensions on People Mobility 189
Figure A4: Robustness Checks II— er Comparison
0.00
-0.11
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
(a) Treated: iversity-town airports
All months
0.00
-0.25
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
August
0.00
-0.10
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
Other months
0.00 -0.03
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(d) Treated: tourist-city airports
All months
0.00 0.01
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(e) Treated: tourist-city airports
August
0.00 -0.03
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(f) Treated: tourist-city airports
Other months
(a) Treated: university-town airports
All months
(b) Treated: university-town airports
August
(c) Treated: university-town airports
Other months
(d) Treated: tourist-city airports
All months
(e) Treated: tourist-city airports
August
(f) Treated: tourist-city airports
Other months
ect is extracted as the parameter λy ed version of
Equation (2) using a DDD strategy where Tyis replaced by a period dummy, post2018,
which takes on the value of one when y2018 and zero otherwise. Countries other than
e treated airports are university-town airports (36
airports) in subgures (a), (b), and (c), and are tourist city airports (37 airports) in
gures, the reference airports (285 airports) are those
that are neither university-town nor tourist-city airports. See Section 2-b for the exact
nitions of these airports and Table A6 for the full estimation results.
Source: Authors’ calculation based on OAG.
e Impact of U.S.-China Tensions on People Mobility 189
Figure A4: Robustness Checks II— er Comparison
0.00
-0.11
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
(a) Treated: iversity-town airports
All months
0.00
-0.25
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
August
0.00
-0.10
-.5 -.25 0 .25 .5
Estimated effect
Pre-2018 Post-2018
95% CI
Estimated effect
Other months
0.00 -0.03
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(d) Treated: tourist-city airports
All months
0.00 0.01
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(e) Treated: tourist-city airports
August
0.00 -0.03
-.5 -.25 0 .25 .5
Pre-2018 Post-2018
95% CI
Estimated effect
(f) Treated: tourist-city airports
Other months
(a) Treated: university-town airports
All months
(b) Treated: university-town airports
August
(c) Treated: university-town airports
Other months
(d) Treated: tourist-city airports
All months
(e) Treated: tourist-city airports
August
(f) Treated: tourist-city airports
Other months
ect is extracted as the parameter λy ed version of
Equation (2) using a DDD strategy where Tyis replaced by a period dummy, post2018,
which takes on the value of one when y2018 and zero otherwise. Countries other than
e treated airports are university-town airports (36
airports) in subgures (a), (b), and (c), and are tourist city airports (37 airports) in
gures, the reference airports (285 airports) are those
that are neither university-town nor tourist-city airports. See Section 2-b for the exact
nitions of these airports and Table A6 for the full estimation results.
Source: Authors’ calculation based on OAG.
190 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Notes
1 Li Tang, Cong Cao, Zheng Wang, and Zhuo Zhou, “Decoupling in science
and education: A collateral damage beyond deteriorating US–China rela-
tions,” Science and Public Policy, Vol. 48, No. 5 (2021), pp. 630–634.
2 Christina L. Davis and Sophie Meunier, “Business as Usual? Economic
Responses to Political Tensions,” American Journal of Political Science, Vol.
55, No. 3 (2011), pp. 628–646; Andreas Fuchs and Nils-Hendrik Klann,
“Paying a Visit: e Dalai Lama Eect on International Trade,” Journal of
International Economics, Vol. 91, No. 1 (2013), pp. 164–177; Guy Michaels
and Xiaojia Zhi, “Freedom fries,” American Economic Journal: Applied
Economics, Vol. 2, No. 3 (2010), pp. 256–281.
3 Raymond Fisman, Yasushi Hamao, and Yongxiang Wang, “Nationalism and
Economic Exchange: Evidence from Shocks to Sino-Japanese Relations,”
e Review of Financial Studies, Vol. 27, No. 9 (2014), pp. 2626–60.
4 George J. Borjas and Kirk B. Doran, “e Collapse of the Soviet Union and
the Productivity of American Mathematicians,” e Quarterly Journal of
Economics, Vol. 127, No. 3 (2012), pp. 1143–1203; Ina Ganguli, “Immigration
and Ideas: What Did Russian Scientists ‘Bring” to the United States?, “
Journal of Labor Economics, Vol. 33, No. S1 (2015), pp. S257–S288.
5 Petra Moser, Alessandra Voena, and Fabian Waldinger, “German Jewish
émigrés and US Invention,” American Economic Review, Vol. 104, No. 10
(2014), pp. 3222–55.
6 Sari Pekkala Kerr, William Kerr, Çağlar Özden, and Christopher Parsons,
“High-Skilled Migration and Agglomeration,” Annual Review of Economics,
Vol. 9 (2017), pp. 201–234.
7 Dongsheng Di, Gal Lu, and Dian Zhong, “Why Did Trump Launch a
Trade War? A Political Economy Explanation from the Perspective of
Financial Constraints,” Economic and Political Studies, Vol. 7, No. 2 (2019),
pp. 203–216; Xiao Li, “Zhongmei maoyi shiheng yu Telangpu fadong maoy-
izhan de mudi” (e Trade Imbalance between China and the United States
and the Purpose of Trump’s Trade War), Nankai Xuebao (Nankai Journal),
No. 3 (2018), pp. 5–8.
8 Joseph Stiglitz, “e US is At Risk of Losing a Trade War with China,”
Project Syndicate, Vol. 30 (2018), pp. 24–35, https://www.project-syndicate.
org/commentary/trump-loses-trade-war-with-china-by-joseph-e--stiglitz-
2018-07?barrier=accesspaylog.
9 Edwin L-C Lai, “e US–China Trade War, the American Public Opinions
and Its Eects on China,” Economic and Political Studies, Vol. 7, No. 2 (2019),
pp. 169–184.
10 Steve Chan, China, the US and the Power-Transition eory: A Critique
(London: Routledge, 2008); Woosang Kim and Scott Gates, “Power
e Impact of U.S.-China Tensions on People Mobility 191
Transition eory and the Rise of China,” International Area Studies Review,
Vol. 18, No. 3 (2015), pp. 219–226; Richard Ned Lebow and Benjamin
Valentino, “Lost in Transition: A Critical Analysis of Power Transition
e or y, ” International Relations, Vol. 23, No. 3 (2009), pp. 389–410; Min-
hyung Kim, “A Real Driver of US–China Trade Conict: e Sino–US
Competition for Global Hegemony and Its Implications for the Future,”
International Trade, Politics and Development, Vol. 3, No. 1 (2019), pp.
30–40; David Zweig, “Is Sino–American Scientic Collaboration a ing of
the Past?,” International Higher Education, No. 108 (2021), pp. 5–7.
11 Yuhan Zhang, “e US–China Trade War,” Indian Journal of Asian Aairs,
Vol. 31, No. 1/2 (2018), pp. 53–74; Guangyuan Ma, “Cong xiuxidide xianjin
kan zhongmei maoyizhan bukebimian” (The Unavoidable Sino-China
Trade War: From the Perspective of the ucydides Trap), Zhongguo Waizi
(Foreign Investment in China), No. 4 (2018), pp. 10; Edwin L-C Lai, “e
US–China Trade War, the American Public Opinions and Its Eects on
China,” Economic and Political Studies, Vol. 7, No. 2 (2019), pp. 169–184.
12 Graham Allison, “e ucydides Trap: Are the US and China Headed for
War?,” e Atlantic, Vol. 24, No. 9, 24 September 2015.
13 Fred C. Bergsten, “China and the United States: Trade Conict and Systemic
Competition,” Peterson Institute for International Economics, October 2018,
https://piie.com/publications/policy- briefs/china-and-united-states-trade-
conict-and-systemic-competition; Imad A. Moosa, “e ucydides Trap
as An Alternative Explanation for the US–China Trade War,” Global Journal
of Emerging Market Economies, Vol. 12, No. 1 (2020), pp. 42–55.
14 David H. Autor, David Dorn, and Gordon H. Hanson, “The China
Syndrome: Local Labor Market Eects of Import Competition in the United
States,” American Economic Review, Vol. 103, No. 6 (2013), pp. 2121–68.
15 Lai, “e US–China Trade War”; David Frum, “Trump Repeats Nixon’s
Folly,” e Atlantic, 2 March 2018.
16 Michael E. Waugh, “e Consumption Response to Trade Shocks: Evidence
from the US-China Trade War,” National Bureau of Economic Research, No.
w26353 (2019); Riad A. Ajami, “US-China Trade War: The Spillover
Eect,” Journal of Asia-Pacic Business, Vol. 21, No. 1 (2020), pp. 1–3; Mary
Amiti, Stephen J. Redding, and David E. Weinstein, “e Impact of the 2018
Taris on Prices and Welfare,” Journal of Economic Perspectives, Vol. 33, No.
4 (2019), pp. 187–210; Kelan Lu, “e Spillover Eect of the Trade War
between Adversarial Dyads: Evidence from the Sino-US Investment Rela-
tionship,” Journal of Chinese Political Science, Vol. 25, No. 1 (2020), pp.
21–47; Badar Alam Iqbal, Nida Rahman, and Jonathan Elimimian, “e
Future of Global Trade in the Presence of the Sino-US Trade War,”
Economic and Political Studies, Vol. 7, No. 2 (2019), pp. 217–231.
17 Priyanka Vora, “US Colleges Look to Insure against Impact of Trade War,”
192 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
Financial Times, 10 September 2019, https://www.ft.com/content/7d0ead8c-d0d3-
11e9-99a4-b5ded7a7fe3f.
18 Ken Itakura, “Evaluating the Impact of the US–China Trade War,” Asian
Economic Policy Review, Vol. 15, No. 1 (2020), pp. 77–93.
19 Ibid.
20 Chris Woolston, “US-China Science Weathers Political Ill Wind,” Nature,
Vol. 575, No. 7783 (2019), pp. S26–S26.
21 Nana De Graa and Van Apeldoorn Bastiaan, “e Transnationalist US
Foreign-Policy Elite in Exile? A Comparative Network Analysis of the
Trump Administration,” Global Networks, Vol. 21, No. 2 (2021), pp.
238–264; Tang, Cao, Wang, and Zhou,“Decoupling in Science and Educa-
tion”; Qingnan Xie and Richard Freeman, “ Creating and Connecting US
and China Science: Chinese Diaspora and Returnee Researchers,” National
Bureau of Economic Research, No. w31306 ( 2023) ; Yu Xie, Xihong Lin, Ju
Li, and Junming Huang, “Caught in the crossfire: Fears of Chinese–
American scientists,” Proceedings of the National Academy of Sciences, Vol.
120, No. 27 (2023), e2216248120.
22 e China Initiative was terminated ocially on February 23, 2022.
23 Dan Bilefsky, “A Tech Executive Caught Between 2 World Powers,” e
New York Times, 7 March 2019, Section A, Page 6; Riham Bahi, “The
Geopolitics of COVID-19: US-China Rivalry and the Imminent Kindle-
berger Trap,” Review of Economics and Political Science, Vol. 6, No. 1 (2021),
pp. 76–94; Haiyong Sun, “US-China Tech War: Impacts and Prospects,”
China Quarterly of International Strategic Studies, Vol. 5, No. 2 (2019), pp.
197–212.
24 Emanuela Marrocu and Raaele Paci, “Dierent Tourists to Dierent Desti-
nations. Evidence from Spatial Interaction Models,” Tourism Management,
No. 39 (2013), pp. 71–83; Andreas Papatheodorou, “Why People Travel to
Dierent Places,” Annals of Tourism Research, Vol. 28, No. 1 (2001), pp.
164–179; Tomaso Pompili, Maurizio Pisati, and Eleonora Lorenzini, “Deter-
minants of International Tourist Choices in Italian Provinces: A Joint
Demand–Supply Approach with Spatial Eects,” Papers in Regional Science,
Vol. 98, No. 6 (2019), pp. 2251–2273.
25 Georey I. Crouch, “e Study of International Tourism Demand: A Review
of Findings,” Journal of Travel Research, Vol. 33, No. 1 (1994), pp. 12–23;
Haiyan Song and Stephen F. Witt, Tourism Demand Modelling and Fore-
casting: Modern Econometric Approaches ( Oxford: Taylor & Francis, 2000);
Pompili, Pisati and Lorenzini, “Determinants of International Tourist
Choices in Italian Provinces.”
26 Carla Massidda and Ivan Etzo, “The Determinants of Italian Domestic
Tourism: A Panel Data Analysis,” Tourism Management, Vol. 33, No. 3 (2012),
pp. 603–610; Marco Alderighi and Alberto A. Gaggero, “Flight Availability
e Impact of U.S.-China Tensions on People Mobility 193
and International Tourism Flows,” Annals of Tourism Research, No. 79 (2019),
102642.
27 Stephen Page, Ian Yeoman, Colin Munro, Joanne Connell, and Linda
Walker, “A Case Study of Best Practice—Visit Scotland’s Prepared Response
to an Inuenza Pandemic,” Tourism Management, Vol. 27, No. 3 (2006), pp.
361–393; Hsiao-I. Kuo, Chi-Chung Chen, Wei-Chun Tseng, Lan-Fen Ju, and
Bing-Wen Huang, “Assessing Impacts of SARS and Avian Flu on Interna-
tional Tourism Demand to Asia,” Tourism Management, Vol. 29, No. 5 (2008),
pp. 917–928; Catherine Cheung, Miki Takashima, Hyunjung Choi, Huijun
Yang, and Vincent Tung, “The impact of COVID-19 pandemic on the
psychological needs of tourists: Implications for the travel and tourism
in dustry,” Journal of Travel & Tourism Marketing, Vol. 38, No.2 (2021), pp.
155-166 .
28 Lars Hultkrantz, and Christina Olsson, “Chernobyl Eects on Domestic and
Inbound Tourism in Sweden—A Time Series Analysis,” Environmental and
Resource Economics, Vol. 9 (1997), pp. 239–258.
29 The dataset allows us to distinguish between transfer and destination
airports. In this research, all airports are destination airports of arrival.
30 Our air trac data, which is a proprietary database, only covers the period
till the end of 2019. However, the focus on this period, instead of more
recent years, can also be justied by the pandemic of COVID-19 since early
2020 which was the dominant factor that signicantly changed the patterns
of global travel at least in 2020 and 2021. With travel restrictions/outright
bans and health measures imposed on international travelers by many
countries around the world, COVID-related policies and health concerns
have distorted the normal patterns of international passenger ows. Even if
data on more recent years is available, it would be heavily contaminated by
the above factors, making it dicult to isolate the impact of geopolitical
events from the impact of the pandemic, thus unsuitable for the purpose of
this study.
31 United States Department of Transportation, “Air Passenger Travel Arrivals
in the United States from Selected Foreign Countries,” https://www.bts.gov/
content/air-passenger-travel-arrivals-united-states-selected-foreign-countries-
thousands-passengers.
32 In addition to the above three types of airports, 43 U.S. airports in the
sample are in the vicinity of both tourist attractions and universities that are
popular with Chinese students. To make our formal comparisons sharper
and cleaner, we must make a compromise on the sample size by dropping
these airports. Many of these airports are large airports such as John F.
Kennedy International Airport (in New York) and Los Angeles International
Airport (in Los Angeles) with at least one major university and one major
tourist destination nearby that are popular with Chinese students and/or
194 Zheng Wang, Li Tang, Cong Cao, and Zhuo Zhou
tourists. ese 43 U.S. airports in total account for around 48 percent of
international passenger arrivals and 67 percent of China-originated
passenger arrivals. e loss of these large airports, however, should not be a
major concern in this DDD setting as our estimation is based on airport-
time and country-airport variations. In addition, the robustness of the
distance criterion used is checked later where we change the distance criteria
from 100 miles to 50 or 150 miles and the results remain qualitatively stable.
Figure A1 in the Appendix section plots the locations of these university-
town and tourist city airports. Figure A2 demonstrates the monthly China-
originated passengers to the U.S. by dierent types of airports.
33 C. Textor,“Number of Visitors to the U.S. from China 2003–2024,” https://
www.statista.com/statistics/214813/number-of-visitors-to-the-us-from-
china/.
34 e country-year xed eects are not included in this DD specication as
they would absorb variations needed to identify βy. ey are included,
however, in the dierence-in-dierence-in-dierences strategy below, which
exploits more granular variations than the DD design.
35 As the US is the only destination country in this data, θiy eectively controls
for all factors that vary by country-country-year such as immigration policy
and other bilateral annual shocks.
36 e eect of interaction terms of Ty*Chinai, Ty*Treatedj, and Chinai*Treatedj
are absorbed by the use of xed eects θiy, ωjym, and δij, respectively, and
thus not listed in the specication model.
37 Patrick Gaulé and Mario Piacentini, “Chinese Graduate Students and U.S.
Scientic Productivity,” e Review of Economics and Statistics, Vol. 95, No.
2 (2013), pp. 698–701.
38 Ruixue Jia, Margaret E. Roberts, Ye Wang, and Eddie Yang, “e Impact of
US-China Tensions on US Science,” National Bureau of Economic Research,
No. w29941 (2022).
39 Richard Freeman and Wei Huang, “China’s ‘Great Leap Forward’ in Science
and Engineering,” in Global Mobility of Research Scientists: e Economics of
Who Goes Where and Why, ed. Aldo Geuna (London: Academic Press,
2015), pp. 155–175.
40 Rachel Nuwer, “Chinese students stay local as favour falls with study
abroad,” Nature, Vol. 620, No. 7973 (2023), pp. S11–S13.
41 Haiyan Song, Gang Li, Stephen F. Witt, and Baogang Fei, “Tourism Demand
Modelling and Forecasting: How Should Demand Be Measured?, ” Tourism
Economics, Vol. 16, No. 1 (2010), pp. 63–81; Tomaso Pompili, Maurizio
Pisati, and Eleonora Lorenzini, “Determinants of International Tourist
Choices in Italian Provinces: A Joint Demand–Supply Approach with Spatial
Eects,” Papers in Regional Science, Vol. 98, No. 6 (2019), pp. 2251–2273.
42 Riham Bahi, “e Geopolitics of COVID-19: US-China Rivalry and the
e Impact of U.S.-China Tensions on People Mobility 195
Imminent Kindleberger Trap,” Review of Economics and Political Science,
Vol. 6, No. 1 (2021), pp. 76–94; Guangyuan Hu, Rong Ni, and Li Tang, “Do
International Nonstop Flights Foster Inuential Research? Evidence from
Sino-US Scientic Collaboration,” Journal of Informetrics, Vol. 16, No. 4
(2022), 101348.
