The Influence of Uber on the Tourism Industry in Sub-Saharan Africa

Abstract

An unreliable and inefficient public transportation system can be a barrier to the successful development of a destination’s tourism industry. Uber, a convenient ride-hailing service, can complement underdeveloped public transport and play a significant role in stimulating the tourism economy by increasing tourists’ mobility and accessibility to attractions and service facilities. Using the data of 48 sub-Saharan African countries, this study conducted propensity score matching and difference-in-differences analysis to empirically examine the influence of Uber on a country’s tourism industry. The results showed that Uber contributed $20 million annually in total tourist spending—$24 per tourist spending—on average, to a country’s economy between 2013 and 2016. However, it did not have a significant influence on the number of international arrivals. The findings of this study provided insights into the benefits of Uber service in promoting per tourist spending by providing a reliable and efficient means of travel.

Keywords

Uber sharing economy sub-Saharan Africa difference-in-differences propensity score matching tourist expenditure

Transportation is an indispensable element in destination development and management (Prideaux 2000). An efficient and convenient transportation system within a destination can improve tourists’ satisfaction and increase the number of attractions visited, leading to higher tourist expenditure, better destination image, and eventually greater tourist volume (Gutiérrez and Miravet 2016; Le-Klähn and Hall 2015; Prideaux 2000; Xiao, Jia, and Jiang 2012; Albalate and Bel 2010; Israeli and Mansfeld 2003). It also allows for effective management of visitors and reduces traffic congestion and crowding (Gutiérrez and Miravet 2016). An unreliable and inefficient transportation system, on the other hand, can work as a barrier to the tourist mobility and accessibility (Israeli and Mansfeld 2003; Prideaux 2000; Lew and McKercher 2006), which deters the development of the destination’s tourism industry.

Many countries in sub-Saharan Africa (SSA) suffer from an underdeveloped public transportation system (Economic Commission for Africa 2009; Trans-Africa Consortium 2008). Though the specific situation differs from country to country, transportation systems are considered to be poorly organized across the SSA, mainly because of the lack of transportation infrastructure and poor maintenance of the existing infrastructure (Trans-Africa Consortium 2008). Intra-city transportation in many of the SSA regions is often unstable and inactive, which increases waiting time and decreases passenger mobility (Trans-Africa Consortium 2008). Safety and security are also critical barriers to utilizing public transportation in SSA countries (Economic Commission for Africa 2009). Insecure and unreliable public transportation systems can deter the development of tourism industry as it can hinder tourists’ accessibility to attractions and service facilities and negatively impact their overall experience (Israeli and Mansfeld 2003; Prideaux 2000; Lew and McKercher 2006).

The best solution would be to build a quality public transportation system. However, infrastructure development requires a considerable amount of resources and time. Uber, which is a convenient ridesourcing mobile platform that helps people travel by connecting them to private drivers (Tham 2016a), can provide an alternative solution to improve accessibility. Ridesourcing refers to “transportation services that connect community drivers—people who drive private cars instead of commercial vehicles—with passengers via mobile devices and applications” (Jin et al. 2018, p. 96). Ridesourcing does not require substantial capital investment as it uses already existing vehicles. Individuals can register themselves as drivers after getting screened for their driving record and criminal history. Passengers can request a ride between locations using the app on their mobile devices (https://www.uber.com).

Since its introduction to SSA in 2013, Uber has experienced continued growth and is believed to provide a safe, transparent, and efficient transportation mode for international tourists visiting the area (Henama and Sifolo 2017; Wyk 2016). As of 2018, there are more than 1.3 million users and 36,000 drivers of Uber in SSA, and Uber continues its effort to expand the service across the continent (Dahir 2017; Akwagyiram 2019). For the countries in SSA where the public transportation system is relatively underdeveloped, an alternative intra-destination transportation system, such as Uber, can play a significant role in increasing tourist mobility and vitalizing the tourism industry. However, research on the economic importance of intra-destination transportation, mainly nonpublic services like Uber, is scarce (Albalate and Bel 2010; Truong and Shimizu 2017).

In this article, we examine the role of Uber in the tourism economy for the destinations with an underdeveloped public transportation system using country-level data in SSA. The current study takes a macro approach by investigating the relationship between Uber and the tourist arrivals and the tourism expenditure of 48 countries in SSA. We do not investigate the changes in behaviors or perceptions of individual tourists before and after Uber. We expect to find that Uber positively influenced the tourism industry, assuming that its availability increases tourist accessibility to attractions and service facilities. Better accessibility can lead to increased expenditures and their satisfaction with the destination, resulting in a positive destination image and repeated visits (Chi and Qu 2008; Virkar and Mallya 2018).

In SSA, international tourism, in terms of tourist arrivals and receipts, has steadily increased in the past two decades. The tourist volume had more than tripled from 13 million in 1995 to 42 million in 2017. During the same period, the amount of spending by international visitors increased from 7.3 billion to 34.4 billion (World Bank n.d.-a, n.d.-b). In an attempt to examine the impact of Uber on the tourism industry, the current study compares the international tourist expenditures and arrivals of the countries that adopted Uber against those of the countries that have not. The study uses expenditures and arrivals data from 1995 to 2016 to examine the tourism industry before and after the introduction of Uber in 2013. Observing the tourism economy pattern across 20 years will allow us to eliminate the bias from the different sizes and random shocks of the economy.

Since the introduction of Uber was an intervention under a natural setting, this article employed the difference-in-differences (DID) method to analyze the causal effect. The DID is often used to estimate the effect of policy interventions by comparing the changes in outcomes of a treatment group (a population that experienced the intervention) to those of a control group (a population that did not experience the intervention) before and after the policy implementation (Bertrand, Duflo, and Mullainathan 2004). We also employed the propensity score matching (PSM) method to match countries with similar characteristics, such as tourism service infrastructure, and remove outliers to minimize the differences in the outcomes due to the factors other than the treatment, that is, Uber.

The current study contributes to the existing literature by empirically investigating the role of Uber and quantifying its impact on the tourism industry, especially its influence on the tourist expenditures and arrivals, for the first time. Understanding the influence of Uber on the tourism economy can provide crucial insight for the countries with limited public transport as it can ease the discomfort of the tourists with existing infrastructure, and policy guidance on encouraging Uber adoption for the sake of the tourism industry. The following section explores the previous literature on the role of transportation in the international tourist arrivals and expenditures and the benefits of Uber.

Literature Review

Transportation and the Tourism Economy

The development of a public transport network between and within tourist destinations is an essential factor in the creation and development of new destinations and the growth of existing ones, as well as contributing to the economy (Hall 1999; Prideaux 1996, 2000; Marlina and Natalia 2017a; Khadaroo and Seetanah 2007). Transportation infrastructure affects tourist movement at various levels, from the highest level of country-to-country to the lowest level of within a tourist attraction (Hall 1999; Prideaux 2000). The current study focuses on the within-destination transport that provides accessibility inside a destination, such as between attractions and lodging.

Though scholars have extensively researched the importance of the transportation systems that link countries and wider regions (e.g., Khadaroo and Seetanah 2008; Lundgren 1982; Rey, Myro, and Galera 2011), few have focused on the systems inside a destination (Albalate and Bel 2010). Hall (1999) considers within-destination mobility and accessibility as one of the four roles of tourist transportation. Researchers have repeatedly found that intra-destination transport affects tourist satisfaction and destination image (Avgoustis and Achana 2002; Echtner and Jamal 2002; Thompson and Schofield 2007; Virkar and Mallya 2018; Sarma 2003).

In their review of literature, Echtner and Ritchie (1993) identified local infrastructure and transportation as one of the top features that researchers used to measure destination image. More recent studies have also considered local transportation as an attribute to measure destination image and satisfaction (Avgoustis and Achana 2002; Thompson and Schofield 2007; Virkar and Mallya 2018; Sarma 2003). Thompson and Schofield (2007) and Virkar and Mallya (2018) provide an extensive review of transportation’s role in tourist satisfaction and destination image.

Positive destination image can lead to the increased tourist satisfaction and loyalty (Chon 1990; Chi and Qu 2008; Hernández-Lobato et al. 2006; Zhang et al. 2014), which can lead to repeat visits (Campo-Martínez, Garau-Vadell, and Martínez-Ruiz 2010; Zhang et al. 2014) as well as new visitors via word of mouth (Garín-Muñoz 2006; Song, Wong, and Chon 2003). The increased number of visitors can potentially grow the total tourism receipt, assuming that the spending pattern persists. Hence, the current study hypothesizes that quality local transportation can result in a greater number of tourists and tourism spending by building a positive destination image.

Another path through which an efficient local transportation system can increase tourist arrivals and receipts is by improving tourist mobility and accessibility. Better mobility and accessibility can boost tourist expenditure as tourists can visit more attractions and facilities; the number of attractions visited is positively correlated with expenditures (Spotts and Mahoney 1991; Leones, Colby, and Crandall 1998; Marlina and Natalia 2017b).

The mode of travel has often been associated with tourist expenditures in previous studies (e.g., Downward and Lumsdon 2004; Lee, Var, and Blaine 1996; Marcussen 2011; Mok and Iverson 2000; Wang et al. 2006). However, most of them examine the expenditure differences between modes of travel to a destination, such as air and road (Lee, Var, and Blaine 1996; Marcussen 2011; Wang et al. 2006). One exception is Downward and Lumsdon (2004) that compared the spending of the tourists who drove cars to those who used public transportation inside the North York Moors National Park, UK, and found that the former spent more than the latter.

Uber and the Tourism Economy

Uber is a private mobile platform that can affect the movement of tourists within a destination. In the mobile application, tourists type in their desired location and they are matched with a driver within seconds. They are able to travel wherever and whenever they want at their fingertips. The application provides information about the vehicle, itinerary, travel cost, and driver such as their picture and reviews from other customers. Uber enables tourists to visit the attractions and facilities that are inaccessible or cumbersome to reach using public transportation.

Moreover, Uber can compensate for a weak transportation system by providing safe, transparent, and efficient transportation (Henama and Sifolo 2017; Wyk 2016). Mohamad et al. (2016) described four benefits of using Uber: safety, price, convenience, and accessibility. All of which are some of the critical factors that tourists take into consideration when choosing a transportation mode (Westlake and Robbins 2005). Previous studies have confirmed that passengers prefer Uber to traditional taxi services and public transit due to reduced time and cost, along with greater convenience, accessibility, and flexibility (Ngo 2015; Rayle et al. 2016; Tham 2016b; Mohamad et al. 2016). Rayle et al. (2016) surveyed residents in San Francisco and identified ease of payment, short wait time, time to reach the destination, ease of calling a ride, reliability, comfort/safety, and cost as some of the reasons they utilize ride-shares such as Uber. Ngo (2015) found that Uber improved overall transportation service and increased travel options; the number of taxi complaints in Chicago and New York decreased as a result of increased service quality as a response to the competition from Uber.

Though research on the relationship between the transportation and the destination image and that between the destination image and the tourism economy has accumulated extensively, the studies that directly link the local transportation and the tourism economy is still limited (Virkar and Mallya 2018; Thompson and Schofield 2007). The literature is even more scare for nonpublic intra-destination transport infrastructure, even though private transportation systems can compensate for underdeveloped infrastructure and can lead to growth in a country’s tourism industry (Page 2004).

Figure 1 summarizes the relationship between Uber and the tourism economy based on the past studies. Studies have shown that at the micro-level, Uber can increase tourist mobility and improve destination image, which can lead to (re)visits and a greater number of attractions visited by the tourists. Such micro-level changes can result in macro-level outcomes, including higher tourist expenditure and tourist arrivals. In the current study, we do not examine the micro-level change due to Uber, such as tourist travel patterns and changes in perception of the destinations. Instead, we presume the mechanism and investigate the macro outcomes of tourist volume and spending (Figure 1). We empirically investigate the relationship between Uber and the tangible outcomes of the tourism industry using a quasi-experimental design with the data from countries in SSA. The following section elaborates on the data and methods employed.

Figure 1.

A conceptual framework for the role of Uber in the tourism industry.

Research Method

This study examined the effect of Uber on a country’s tourism economy using tourist arrivals and expenditure data of 48 countries in SSA between 1995 and 2016. PSM and DID methods were employed to measure the Uber’s influence as a form of intervention within groups of similar characteristics. PSM was conducted to match countries with similar characteristics and mitigate a selection bias. For example, it is logical to assume that Uber is prone to select more developed markets with a strong demand for its service.

Then, a DID approach was taken to measure the difference in outcomes between the treatment and control groups that resulted from an intervention, that is, Uber. DID can remove biases in the post-intervention comparisons between the two groups that could arise from their initial gap by including the counterfactual outcome. Moreover, it separates the differences in groups that could be the result of the passing of time to correctly capture the influence of Uber.

There are two reasons for the choice of SSA countries as a study area. First, the region of study should be eligible for a quasi-natural experiment. That is, Uber should have entered some countries in the region when not the others so that a comparison can be made between the two groups. SSA provides an ideal setting for such a quasi-natural experiment. As of 2018, only 6 of 48 countries in SSA have Uber operating. Therefore, we can compare the growth of the tourism industry in the six countries to that of the rest to investigate the impact of Uber’s introduction.

Second, it is easier to detect the impact of Uber on the countries in SSA compared to the highly developed countries where public transportation systems are already well-developed. The increase in marginal utility from Uber may be minimal. As SSA overall has limited public transit (Economic Commission for Africa 2009; Trans-Africa Consortium 2008), the advantage of Uber should be more prominent. Hence, SSA provides an ideal condition to observe the influence of Uber over time as Uber can complement an underdeveloped public transport system and as we can compare the countries that have had Uber against those that have not.

Data and Variables

Dependent variables

The dependent variables in the current study were the tourist arrivals and expenditure (Table 1). First, the volume of tourist arrival was measured via annual numbers of international inbound tourist arrivals obtained from the World Bank, which was sourced from the World Tourism Organization’s Yearbook of Tourism Statistics and Compendium of Tourism Statistics (World Bank 2018a). The World Bank defines international inbound tourists as those “who travel to a country other than that in which they have their usual residence, but outside their usual environment, for a period not exceeding 12 months and whose main purpose in visiting is other than an activity remunerated from within the country visited (para.1).”

Table1.

Descriptive Statistics.

Category	Variable	Mean	SD	Min	Max
Dependent	International tourist expenditure (million USD)	400	1,228	31	11,109
	Expenditure per tourist (USD)	745	645	40	4,130
	Number of tourist arrival (1,000)	597	1,208	2.9	9,592
Treatment	Uber service availability	0.13	0.33	0	1
Independent	Political stability	34.40	23.10	0.50	93.80
	ln (FDI) (USD)	18.46	2.32	4.60	23.02
	Internet usage (%)^a	5.30	8.90	0.0	54.30
	Urbanization (%)^a	36.2	15.30	7.20	87.20
	Ground and port infrastructure^b	2.90	0.70	0.80	4.70
	Tourist service infrastructure^b	2.10	0.90	1	5.80
	Price competitiveness^b	4.60	0.60	3.50	5.80

Source: Data from World Bank, Uber, and World Economic Forum Travel and Tourism Competitiveness Index.

Note: FDI = foreign direct investment; USD = US dollar.

Percentage of population.

Score ranges from 1 to 7.

Tourist expenditure used a collection of records from the World Tourism Organization (WTO), International Money Fund (IMF), and World Bank import estimates (World Bank 2018b). In addition to the goods and services expenditures made by inbound tourists, the data also included payments to national carriers for international transport and receipts from same-day visitors and passenger services performed within an economy by foreign carriers (World Bank 2018b). We examined the expenditure per tourist and the total expenditure to understand whether the increase in the spending was due to a higher number of visitors or greater spending by each visitor. Both tourist arrivals and expenditure data were collected for the years from 1995 to 2016.

Independent variables

The availability of Uber, our primary independent variable, was represented in a binary format—1 if Uber was available in a country at a given year and 0 if not. From Uber’s official website (https://www.uber.com/cities), we collected the dates of Uber’s introduction in the SSA region. Of 48 countries in the SSA, 6 have Uber operating in them since 2013: Ghana (Accra, Kumasi),¹ Kenya (Nairobi, Mombasa), Nigeria (Abuja, Lagos), South Africa (Cape Town, Durban, Johannesburg, Pretoria, Port Elizabeth), Tanzania (Dar Es Salaam), and Uganda (Kampala). Figure 2 graphically marks the locations on a map.

Figure 2.

Uber’s operating countries in sub-Saharan Africa marked in gray as of December 2018.

The intervening variables that can influence tourist arrivals and expenditures were added to the analyses to account for any differences due to the country-specific characteristics other than Uber availability. We controlled for five global economic and societal development variables that are critical determinants of the tourism industry: foreign direct investment (FDI), Internet usage, urbanization rate, political stability and absence of violence and terrorism, and tourism competitiveness (Naudé and Saayman 2005; Ivanov et al. 2017).

Previous research has found that price competitiveness, political stability, tourism infrastructure such as the number of hotel rooms, Internet usage, and urbanization rate affect tourist arrivals and spending in African countries (Naudé and Saayman 2005; Seetanah, Durbarry, and Ragodoo 2010; Adeola, Boso, and Evans 2018; Viljoen, Saayman, and Saayman 2019; Saayman and Saayman 2008). Naudé and Saayman (2005) conducted a panel analysis on 1996–2000 tourism arrivals data of 43 African countries. They showed that political stability, Internet usage, and urbanization rate as significant factors that influenced the total tourist arrivals from the Americas and Europe. Adeola, Boso, and Evans (2018) and Viljoen, Saayman, and Saayman (2019) used more recent data sets and found FDI and price competitiveness as critical factors in addition to those Naudé and Saayman (2005) identified.

FDI data came from the World Bank database and was the sum of equity capital, reinvestment of earnings, other long-term capital, and short-term capital, as shown in the balance of payments. Internet usage was measured as the percentage of the population who are Internet users and is acquired from the International Telecommunication Union’s World Tele-communication/ICT Indicators Database. The urbanization rate indicated the percentage of the total population living in urban areas, representing the level of industrialization and societal development (World Bank n.d.-c). The source of data was the United Countries Population Division’s 2018 revision of World Urbanization Prospects. The political stability and absence of violence and terrorism was one of the Worldwide Governance Indicators from the World Bank and assessed the “perceptions of the likelihood that the government will be destabilized or overthrown by unconstitutional or violent means” (World Bank n.d.-d, para. 1). All data sets were collected for the years from 1995 to 2016.

In addition, we took the tourism industry competitiveness into account as the increase in tourist volume and expenditure can depend on the level of tourism industry development (Naudé and Saayman 2005; Adeola, Boso, and Evans 2018; Seetanah, Durbarry, and Ragodoo 2010). As a proxy for tourism competitiveness, we used a subset of the World Economic Forum’s (WEF’s) Travel & Tourism Competitiveness Index (TTCI). The TTCI provides 14 variables that evaluate the competitiveness of a country’s tourism industry, from its business environment to natural resources. Of the 14, we used 3 variables that the previous studies have found most relevant to Uber: the ground and port infrastructure index, the tourism service infrastructure index, and the price competitiveness index (Naudé and Saayman 2005; Adeola, Boso, and Evans 2018; Seetanah, Durbarry, and Ragodoo 2010).

The ground and port infrastructure index measured the quality of the transportation infrastructure, such as paved road density and roads’ quality. The tourism service infrastructure index included the quality of tourism infrastructure, the number of hotel rooms, the presence of major car rental companies, and the number of automated teller machines (ATMs) per adult. The price competitiveness index, which measured how costly it is to travel or invest in a country, incorporated ticket taxes and airport charges, a hotel price index, purchasing power parity, and fuel price levels. The three variables were available every two years from 2007 and 2015, and their scale ranged from one to seven. Because of the significant presence of missing data, we calculated the average indexes for the years 2007, 2009, and 2011 to create a tourism competitiveness index before Uber and the average indexes for the years 2013 and 2015 to generate an index after Uber. Table 1 provides the descriptive statistics of all the variables.

Analysis

Propensity score matching

The countries’ attributes can cause a selection bias since they can influence the countries’ tourism industry and the probability of being selected by Uber. PSM mitigates this selection bias by matching the countries with alike characteristics for comparison as closely as possible. Figure 3 graphically explains how the PSM works.

Figure 3.

Propensity score matching (PSM) method. White circles represent countries that have Uber operating (treatment group) and black circles that do not have Uber operating (control group). The size of a circle indicates the Propensity Score calculated based on its political stability, Internet usage, urbanization rate, ground and port infrastructure, tourist service infrastructure, and price competitiveness.

The white circles of the figure represent the treatment group, that is, the countries with Uber, while the black circles represent the countries in the control group. The size of a circle represents a country’s Propensity Score (PS) based on the intervening variables: FDI, Internet usage, political stability, urbanization rate, ground and port infrastructure, tourist service infrastructure, and price competitiveness. Then, countries with the closest PS are matched for later analysis. Since it is challenging to find observations that exactly match, we adopted a kernel matching method that calculates a logistic model of covariates between treatment and control groups and matches the sample of observations with the nearest PS (Smith and Todd 2005).

The bottom box of Figure 3 represents the result after PSM. Since the countries that do not have corresponding PS were excluded to reduce bias, the total sample size of both the treatment and control groups was reduced. The number of control groups decreased from 42 to 20, and South Africa was dropped from the first six countries in the treatment group.

Table 2 compares the characteristics between the control and treatment groups before and after PSM. We can observe that the intervening variable values became more similar between the control and treatment groups after the matching process. Political stability showed the most considerable change. The gap in average value decreased from 9.09 to 2.36. After matching, all of the variables showed less than one-unit difference. The discrepancy in ground and port infrastructure between the groups increased slightly from 0.02 to 0.09, but on average, the three tourism competitiveness indexes became more homogenous after PSM.

Table 2.

Comparison of Variable Averages before and after Propensity Score Matching.

	Before Matching		After Matching
	Treatment	Control	Treatment	Control
ln(Foreign direct investment) ($)	20.84	18.41	20.42	19.67
Internet usage^a	7.18	4.92	4.72	4.28
Political stability^b	25.90	34.99	27.11	24.75
Urbanization^a	36.46	34.44	35.73	34.62
Price competitiveness^c	4.75	4.57	4.78	4.81
Ground and port infrastructure^c	2.93	2.95	2.91	3.00
Tourist service infrastructure^c	2.32	2.12	2.14	2.05

Note: The initial data set before matching included 6 countries in the treatment group (Ghana, Kenya, Nigeria, Tanzania, Uganda, and South Africa) and 42 Sub-Saharan African countries (without Uber services as of 2019) in the control group. After the propensity score matching, South Africa was excluded from the initial treatment group and the size of the control group decreased to 20 countries.

Percentage of population.

Scale 1 to 100.

Scale 1 to 7.

Difference-in-differences estimation

With the matched data, DID was conducted to measure the influence of Uber on tourist arrival and receipt. An experimental design is an ideal method for evaluating the introduction of a new system or policy, that is, a treatment. In an identically controlled environment, the effectiveness of a treatment can be determined by comparing the group that received the treatment (treatment group), and that did not (control group). However, it is nearly impossible to experiment with a policy in real-world settings because of ethical issues and limited resources (Asgari and Baptista Nunes 2011).

The DID estimation is a quasi-experimental method that utilizes repeated cross section and time series to examine the impact of an intervention such as policy implementation in natural settings (Bryman 2016). DID is one of the most widely used methods for estimating the effectiveness of policies since it complements the shortcomings of both the cross-section and time-series analyses. The cross-section analysis measures policy effects by comparing the treatment and control groups in the same period. Thus, the difference caused by an unobserved time difference can be controlled. However, the method does not capture the differences that occur from the different characteristics between the treatment and control groups.

On the other hand, the time-series method estimates a policy’s effects by comparing the before and after characteristics of a treatment group. It is advantageous for controlling the unobserved characteristics of a treatment group. Nevertheless, the estimation results of time series analysis cannot separate the unobserved impact from the lag of time from policy effectiveness as there is no control group for comparison.

The DID controls for both the unobserved time and group characteristics differences and observed or complementary information (Angrist and Pischke 2009). The method includes a time-invariant assumption. That is, the unobserved differences between treatment and control groups are the same over time in the absence of treatment. With this assumption, DID estimates the causal effect of specific intervention by comparing the changes in outcomes over time between the treatment and control groups (Lechner 2011).

Figure 4 graphically describes the primary mechanism of DID. The line T1–T2 represents the treatment group’s outcome, whereas the line C1–C2 represents the outcome of the control group. The outcomes of both groups are measured before either group has received the treatment, represented by T1 and C1. Their difference is marked as D1. The treatment group then receives the treatment, marked as “Uber” in Figure 4, and the outcomes of both groups are measured again, presented as T2 and C2.

Figure 4.

Difference-in-differences (DID) method. T1 and C1 respectively represent the initial value of tourism receipt/number of international tourist arrivals of the country that has Uber operating (treatment group), and that does not (control group) before the introduction of Uber. T2 and C2 represent the tourism receipt/number of international tourist arrivals of the treatment and the control group after the introduction of Uber. Q is the counterfactual outcome of the treatment group, reflecting the change of the control group as well as its trend.

The DID then calculates the counterfactual difference in the outcome variable between the two groups—the difference would still exist if neither group experienced the treatment. D2 represents the counterfactual difference, the difference between points C2 and Q. Q is the outcome of the treatment group, reflecting the change of the control group with the assumption of no Uber. The DID effect is the impact of Uber after eliminating the difference from time and country characteristics. It is calculated as the difference between the observed outcome and the counterfactual outcome, T2 – Q ( $β_{3}$ ). Table 3 provides mathematical expressions for each denotation.

Table 3.

Difference-in-Differences Identification.

	Preintervention(t < 2013, T=0)	Postintervention(t ≥ 2013, T=1)	Time Difference
Treatment group^a (Uber, D=1)	T1 β₀ + β₂ + γ (E₁₀)	T2 β₀ + β₁ + β₂ + β₃ + γ (E₁₁)	T2–T1 β₁ + β₂ + γ (E₁₁ − E₁₀)
Control group (No Uber, D=0)	C1 β₀ + γ (E₀₀)	C2 β₀ + β₁ + γ (E₀₁)	C2–C1 β₁ (E₀₁ − E₀₀)
Group difference	D1 (T1–C1) β₂ (E₁₀ − E₀₀)	D2 (T2–C2) β₂ + β₃ (E₁₁ − E₀₁)	D2–D1 β₃ (E₁₁ − E₀₁) − (E₁₀ − E₀₀)

Ghana, Kenya, Nigeria, Tanzania, and Uganda.

Since Uber was introduced to SSA in 2013, the article used the data from 1995 to 2012 as the before-treatment period and those from 2013 to 2016 as the after-treatment period. The treatment group included five countries: Ghana, Kenya, Nigeria, South Africa (removed after Propensity Score Matching), Tanzania, and Uganda. We denote tourist arrivals and expenditures in the treatment group after Uber as T2. Therefore, (T2 − T1) represents the change in tourist expenditure or volume in the treatment group before and after Uber’s introduction. The control group includes 42 countries (20 after PSM), and their tourist expenditure or volume before and after 2013 are represented as C1 and C2, respectively. D1 and D2 account for the differences between the tourist expenditure or volume of the treatment and control groups pre- and post-Uber introduction, and D2 − D1 represents the treatment effect (Table 2).

Linear estimation of DID

The equation below is the linear model used for the DID analysis:

t o u r_{i t} = β_{0} + β_{1} T_{t} + β_{2} D_{i} + β_{3} (T_{t} \times D_{i}) + X_{i} γ + e_{i}

(1)

The dependent variable tour_it represents the total or per capita expenditure by tourist(s) or tourist arrivals for country i at year t. We define the treatment as the availability of Uber to the country and assume that there exist two periods, before treatment, T_t = 0 (t < 2013), and after treatment, T_t = 1 (t ≥ 2013). For the qualification variable (D_i), we assign the value 1 if a country belongs to the treatment group (D_i = 1) and 0 if not (D_i = 0). X_iγ is the covariates of the country.

According to time and group variables, we identify the conditional value of equation (1) as the following:

\begin{array}{l} 1) E_{11} = E (t o u r | D = 1, T = 1) = β_{0} + β_{1} + β_{2} + β_{3} + γ \\ 2) E_{01} = E (t o u r | D = 0, T = 1) = β_{0} + β_{1} + γ \end{array}

(2)

\begin{array}{l} 3) E_{10} = E (t o u r | D = 1, T = 0) = β_{0} + β_{2} + γ \\ 4) E_{00} = E (t o u r | D = 0, T = 0) = β_{0} + γ \end{array}

Table 3 describes equation (2) for each condition. The pure effect of the treatment is defined as β₃ since it is the difference between the change in the treatment and the control groups: (E₁₁ − E₁₀) − (E₀₁ − E₀₀) = (β₂ + β₃) − β₂ = β₃. Thus, the coefficient of the interaction term between the time variable and the qualification variable (T_i × D_i) can be interpreted as the pure effect of treatment.

As mentioned in the Data section, the Travel & Tourism Competitiveness Index (TTCI) data were only provided for limited years compared to Internet usage, political stability, and urbanization. Hence, when we controlled for TTCI variables, many observations were lost. To mitigate the loss of observations, we ran DID three times: first without any covariates; second with FDI, political stability, Internet usage, and urbanization; and third with all covariates.

Results

Table 4 presents the analysis results. First, we examine the results for the tourism receipts in columns (0) through (4). The first column (0) shows the simple DID estimation result without controlling any covariates. We observed a significant difference between the control and treatment groups pre- and post-Uber service availability, which encouraged further investigation. From the columns (1) and (3), we can observe that the impact of Uber on total and per capita tourist expenditure is approximately 136 million USD and 123 USD, respectively, since 2013. On a yearly average, the countries that had Uber operating gained 34 million USD in total tourist expenditure and 31 USD in per capita tourist expenditure. When averaged, the tourists who visited the countries that had Uber operating spent 31 USD more per year, which contributed to 34 million USD gain for the tourism industry as a whole.

Table 4.

Difference-in-Differences Estimation Results after Propensity Score Matching.

	Total Tourist Expenditure			Per Tourist Expenditure		Tourist Arrival
Variable	(0)	(1)	(2)	(3)	(4)	(5)	(6)
Time	84.74	−11.48	−8.02	−31.48	−18.49	−214.44	−179.37
Time	(1.31)	(1.49)	(1.38)	(1.17)	(1.57)	(1.42)	(1.53)
Treatment	138.17***	88.18***	108.41***	28.74***	28.17***	1410.37***	910.46***
Treatment	(5.92)	(4.18)	(3.99)	(3.86)	(4.13)	(4.51)	(4.69)
Treatment × Time	130.30***	136.47***	79.39***	123.37***	96.18*	488.14	332.42
Treatment × Time	(2.59)	(2.98)	(2.67)	(2.24)	(2.01)	(1.74)	(1.82)
Political stability		−0.023 ***	−0.038 ***	−0.022***	−0.041***	−0.019***	−0.041***
Political stability		(2.87)	(2.59)	(2.78)	(2.62)	(2.75)	(2.65)
Foreign direct investment		0.078***	0.062***	0.046***	0.043***	0.033**	0.016**
Foreign direct investment		(2.56)	(2.63)	(2.41)	(2.37)	(2.38)	(2.39)
Internet usage		0.031***	0.028***	0.009***	0.012**	0.031***	0.013***
Internet usage		(4.46)	(3.19)	(3.94)	(2.86)	(3.84)	(2.83)
Urbanization		0.005	0.001	0.002	0.002	0.001	0.001
Urbanization		(0.42)	(0.31)	(0.47)	(0.27)	(0.44)	(0.28)
Ground and port infrastructure			0.031		0.040		0.028
Ground and port infrastructure			(1.48)		(1.61)		(1.58)
Price competitiveness			0.092***		0.076***		0.094**
Price competitiveness			(2.62)		(2.73)		(2.28)
Tourist service infrastructure			0.414**		0.372**		0.418**
Tourist service infrastructure			(2.45)		(2.35)		(2.37)
N	793	473	318	437	301	452	302

Note. Time value is 1 since the year 2013 and 1 before. Treatment value is 1 if a country had Uber service and 0 if not. Treatment × Time value is 1 if a country had Uber service since the year 2013 and 0 otherwise. Unit of values for Columns (0), (1), and (2) are million USD and that for columns (3) and (4) are USD. t values are in parentheses.

p < 0.10, **p < 0.05, ***p < 0.01.

The coefficient value of estimation was lower but consistent when we included the TTCI variables. Columns (2) and (4) are the estimation result when we control all explanatory variables. The DID effect was positive and significant. For the countries with Uber, the total and per capita tourist expenditure is higher than those without Uber by 79 million USD and 96 USD. That is, since 2013, tourists to the countries with Uber spent 24 USD more on yearly average than those to the countries without Uber, resulting in 20 million USD average tourism receipts annually. The findings affirm our hypothesis that the existence of Uber promotes tourist spending with an assumption that it increases mobility and accessibility.

The analysis of Uber’s effect on the number of tourist arrivals told a different story from our initial hypothesis. Columns (5) and (6) of Table 4 indicate that the coefficient between treatment and time is insignificant. The insignificant coefficient indicates that Uber did not significantly contribute to attracting more tourists to the countries with Uber. Intervening variables except for urbanization and ground and port infrastructure had significant influence on tourist arrival and expenditure.

In order to make our estimation result more reliable, a parallel assumption should hold over time. Since the violation of the parallel assumption may lead to biased estimation results of the causal effect, the difference between the treatment and control group should be constant over time in the absence of treatment. Although there is no statistical test to confirm the assumption, we have enough observations over many periods to present a visual inspection. Figure 5 shows the trends in expenditure and expenditure per tourist. Until 2013, there was a similar trend between the control and treatment groups. However, after 2013, Uber was introduced for the treatment group, and the trend between expenditure and expenditure per tourist went in a different direction for the treatment and the control groups.

Figure 5.

Parallel assumption test. The above figure is the parallel assumption test for total expenditure between the treatment and control group. The below figure is the parallel assumption test for expenditure per tourist. The solid line represents the average values of the treatment group, whereas the dashed line represents those of the control group.

We also carried out a placebo test to confirm that results were not driven purely by the research design. To conduct a placebo test, we eliminated all observations for the treated group after receiving treatment for the first time. Then, we created a simulated treatment group, that is, a placebo group, randomly assigning members of the control group to the treatment group. Then, we ran the DID estimation on the placebo group against the control group’s remaining members. If the research design did not drive our treatment, the estimation results should be insignificant. Table 5 is the result of DID estimation results with placebo treatment groups.

Table 5.

Placebo Test.

	Total Tourist Expenditure		Per Tourist Expenditure
Variable	(1)	(2)	(3)	(4)
Time	−4.51	−3.82	−8.30	−6.47
Time	(1.37)	(1.45)	(1.25)	(1.38)
Treatment	−10.42	−9.13	−6.18	−5.87
Treatment	(1.24)	(1.39)	(1.72)	(1.64)
Treatment × Time	−4.17	−3.82	−8.93	−6.81
Treatment × Time	(0.56)	(0.74)	(1.04)	(1.12)
Political stability	−0.012***	−0.024***	−0.025***	−0.029***
Political stability	(2.55)	(2.63)	(2.87)	(2.74)
Foreign direct investment	0.052*	0.048*	0.027*	0.038*
Foreign direct investment	(2.14)	(2.28)	(2.06)	(2.11)
Internet usage	0.030***	0.026***	0.008***	0.009**
Internet usage	(3.21)	(3.49)	(3.72)	(3.12)
Urbanization	0.003	0.002	0.001	0.001
Urbanization	(0.29)	(0.28)	(0.34)	(0.38)
Ground and port infrastructure		0.020		0.032
Ground and port infrastructure		(1.39)		(1.52)
Price competitiveness		0.083***		0.069***
Price competitiveness		(2.75)		(2.87)
Tourist service infrastructure		0.414*		0.372*
Tourist service infrastructure		(2.25)		(2.18)
N	428	265	416	257

Note: Time value is 1 since the year 2013 and 1 before. Treatment value is 1 if a country had Uber service and 0 if not. Treatment × Time value is 1 if a country had Uber service since the year 2013 and 0 otherwise. Unit of values for columns (0), (1), and (2) are million USD and that for columns (3) and (4) are USD. t values are in parentheses.

p < 0.10, **p < 0.05, ***p < 0.01.

Columns (1) and (2) present the DID estimation results of total expenditure and columns (3) and (4) list the DID estimation results of total expenditure per tourists. The coefficient value is negative and not statistically significant. It provides the robustness to our empirical results by showing that the treatment effect is only seen in the countries that received the treatment. It also implies that our previous DID estimation results come not from the research design but from the treatment effect of introducing Uber.

Conclusion and Discussion

Using PSM and DID method, this research empirically measured the influence of Uber availability on the tourism economy of SSA countries. According to the study results, Uber service brought in 79 million USD tourist spending to sub-Saharan African countries since 2013. The extra income is due to the increase of 96 USD per tourist spending and not from the increased tourist arrivals. The results confirmed our hypothesis that Uber service could improve a country’s tourism economy by providing cheaper and more convenient services and breaking down transportation barriers.

The results demonstrated that the establishment of Uber had a positive impact on tourist expenditures because first, Uber helps tourists overcome insecure public transportation systems in SSA countries. Uber can provide a reliable and efficient alternative means to travel intra-destination, as many countries in SSA have inefficient and inadequate quality public transportation (Economic Commission for Africa 2009; Trans-Africa Consortium 2008).

At the same time, Uber provides a familiar, accessible, and transparent travel mode for international tourists (Ngo 2015; Rayle et al. 2016; Tham 2016b; Wan et al. 2016). Tourists can request a ride directly from their location and inquire about pricing, even if they cannot speak the native language. They can also pay for the service using Uber’s mobile app, which increases convenience. When tourists are more comfortable with a transportation system, they are more likely to travel to more places where they are likely to spend money.

However, contrary to our initial thoughts, Uber did not attract more international tourists through an improved destination image. The limited impact might be due to the short time since Uber’s introduction. Improved destination image may need more time to propagate through customer word-of-mouth; thus, the impact has not materialized yet. It will be interesting to conduct similar research a few years into the future to validate this assumption.

The findings from this study call for SSA countries to be open to the introduction of new technologies to mitigate their limited local transportation system. The researchers are not aware of the criteria for Uber’s decision to enter a particular country. However, the substantial growth in tourist spending makes it worthwhile to invest in those conditions, making its introduction viable. Even though Uber is facing several lawsuits in South Africa and around the world because of its disruption to the traditional transportation industry (Marcano 2018), the results of the current study confirm its value to a country’s tourism economy, at least in developing countries where transportation infrastructure is usually inferior.

Uber should be considered a partner of a country’s tourism and hospitality industry, not a foe. The conclusions are also valuable to the company Uber for justifying its value not only by fulfilling local transportation needs but also by aiding the tourism and hospitality industry in bringing in more dollars from international visitors and creating more jobs and boosting a country’s economy.

Limitations and Future Research

The limitation of the study lies in its limited scope in sub-Saharan African countries and a macro approach. First, it is difficult to confirm that Uber would have the same effect on tourist spending in other developing or developed countries with similar or relatively superior transportation infrastructure. The analysis was constricted to the countries in the sub-Saharan Africa region, and hence the results cannot be generalized to the other nations with different socioeconomic characteristics. Further research on the relationship between Uber and the tourism industry in different countries and regions will provide a more comprehensive understanding of Uber’s role in tourism.

Another limitation is the study’s macro-level analysis of Uber’s impact on the tourism economy. Individual-level data analysis was not carried out, such as comparing the spending or re-visiting pattern of tourists who used Uber to those who did not. The purpose of the current study was to explore the impact of Uber on the number of tourist arrivals and the number of tourist receipts at the country-level. The macro approach allowed the study to compare the countries that adopted Uber against those that have not over time. Future research that examines micro-level data such as tourists’ travel patterns using Uber, their loyalty toward the destination, and their spending pattern compared to the other tourists would validate the conceptual model in Figure 1 and provide meaningful insights from the tourism industry.

There could be a list of exciting and meaningful future research directions. First, since 2015, multiple startup companies have launched ride-hailing services in sub-Saharan Africa, such as Bolt and Bodapp. As our analysis examined the impact of Uber between 1995 and 2016, we did not consider the impact of emerging competitors. Examining the effect of other ride-hailing services on the tourism industry and Uber would provide an insight into the role of ride-hailing services.

Other future research opportunities include measuring the number of jobs created due to Uber’s introduction while considering the number of jobs lost in the traditional transportation industry; quantifying the impact of other sectors of the sharing economy, such as Airbnb or bike-sharing services, on the tourism industry could also be fruitful; confirming the process through which Uber impacts the spending of the tourists and the destination image.

Footnotes

Author Note

So Young Park (Soyoung Park) is now an Assistant Professor of Hospitality and Tourism Management Program, in the Department of Marketing at Florida Atlantic University.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

So Young Park

Notes

Author Biographies

So Young Park (Soyoung Park), is an Assistant Professor of Hospitality and Tourism Management Program in the Department of Marketing at Floria Atlantic University. Her research utilizes advanced data analytics and unstructured data to examine social issues in tourism and hospitality involving children and minorities.

JunYun Kim is a doctoral candidate in the Department of Economics at Korea University. His research interests include applied microeconomics, especially focusing on international trade and development.

Bing Pan, is an Associate Professor in the Department of Recreation, Park, and Tourism Management at Pennsylvania State University. He is also a faculty affiliate of the Institute for Computational Data Sciences and the Graduate Program in Social Data Analytics. His research interests focus on data analytics, tourism big data, destination marketing, information technology, and travel benefits.

References

Adeola

Ogechi

Boso

Nathaniel

Evans

Olaniyi

. 2018. “Drivers of International Tourism Demand in Africa.” Business Economics 53 (1): 25–36.

Akwagyiram

Alexis

. 2019. “Uber Targets Expansion in Fast-Growing West African Markets.” Reuters. https://www.reuters.com/article/us-uber-africa/uber-targets-expansion-in-fast-growing-west-african-markets-idUSKCN1TS1TM.

Albalate

Daniel

Bel

Germà

. 2010. “Tourism and Urban Public Transport: Holding Demand Pressure under Supply Constraints.” Tourism Management 31 (3): 425–33.

Angrist

Joshua D.

Pischke

Jörn-Steffen

. 2009. Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton, NJ: Princeton University Press.

Asgari

Baptista Nunes

J. M. B.

2011. “Experimental and Quasi-Experimental Research in Information Systems.” Paper presented at: IADIS International Workshop Information Systems Research Trends: Approaches and Methodologies (ISRTAM 2011); Italy.

Avgoustis

S. H.

Achana

2002. “Designing a Sustainable City Tourism Development Model Using an Importance Performance (IP) Analysis.” Paper presented at: City Tourism 2002: Proceedings of European Cities Tourism’s International Conference in Vienna, Austria, 2002, 139–49.

Bertrand

Marianne

Duflo

Esther

Mullainathan

Sendhil

. 2004. “How Much Should We Trust Differences-in-Differences Estimates?” Quarterly Journal of Economics 119 (1): 249–75.

Bryman

Alan

. 2016. Social Research Methods, 5th ed. Oxford: Oxford University Press.

Campo-Martínez

Sara

Garau-Vadell

Joan B.

Martínez-Ruiz

María Pilar

. 2010. “Factors Influencing Repeat Visits to a Destination: The Influence of Group Composition.” Tourism Management 31 (6): 862–70.

10.

Chi

Christina Geng Qing

Hailin

. 2008. “Examining the Structural Relationships of Destination Image, Tourist Satisfaction and Destination Loyalty: An Integrated Approach.” Tourism Management 29 (4): 624–36.

11.

Chon

Kye-Sung

. 1990. “The Role of Destination Image in Tourism: A Review and Discussion.” Tourist Review 45 (2): 2–9.

12.

Dahir

Abdi Latif

. 2017. “Uber Is Marking Four Years in Africa.” Quartz Africa. https://qz.com/africa/1090738/uber-is-marking-four-years-in-africa/.

13.

Downward

Paul

Lumsdon

Les

. 2004. “Tourism Transport and Visitor Spending: A Study in the North York Moors, National Park, UK.” Journal of Travel Research 42 (4): 415–20.

14.

Echtner

Charlotte M.

Brent Ritchie

J. R.

1993. “The Measurement of Destination Image: An Empirical Assessment.” Journal of Travel Research 31 (4): 3–13.

15.

Echtner

Charlotte M.

Jamal

Tazim B.

2002. “The Disciplinary Dilemma of Tourism Studies.” Annals of Tourism Research 24 (4): 868–83.

16.

Economic Commission for Africa. 2009. “Africa Review Report on Transport.” Addis Ababa, Ethiopia.

17.

Garín-Muñoz

Teresa

. 2006. “Inbound International Tourism to Canary Islands: A Dynamic Panel Data Model.” Tourism Management 27 (2): 281–91.

18.

Gutiérrez

Aaron

Miravet

Daniel

. 2016. “The Determinants of Tourist Use of Public Transport at the Destination.” Sustainability (Switzerland) 8 (9).

19.

Hall

Derek R.

1999. “Conceptualising Tourism Transport: Inequality and Externality Issues.” Journal of Transport Geography 7 (3): 181–88.

20.

Henama

Unathi Sonwabile

Sifolo

Siyanda

. 2017. “Uber: The South Africa Experience.” African Journal of Hospitality, Tourism and Leisure 6 (2).

21.

Hernández-Lobato

Lucio

Solis-Radilla

Maria Magdalena

Moliner-Tena

Miguel Angel

Sánchez-García

Javier

. 2006. “Tourism Destination Image, Satisfaction and Loyalty: A Study in Ixtapa-Zihuatanejo, Mexico.” Tourism Geographies 8 (4): 343–58.

22.

Israeli

Yechezkel

Mansfeld

Yoel

. 2003. “Accès Par Les Transports Aux Attractions Touristiques de La Vieille Ville de Jérusalem.” Tourism Geographies 5 (4): 461–81.

23.

Ivanov

Stanislav

Gavrilina

Margarita

Webster

Craig

Ralko

Vladyslav

. 2017. “Impacts of Political Instability on the Tourism Industry in Ukraine.” Journal of Policy Research in Tourism, Leisure and Events 9 (1): 100–27.

24.

Jin

Scarlett T.

Kong

Hui

Rachel

Sui

Daniel Z.

2018. “Ridesourcing, the Sharing Economy, and the Future of Cities.” Cities 76 (October 2017): 96–104.

25.

Khadaroo

Seetanah

2007. “Research Note: Does Transport Infrastructure Matter in Overall Tourism Development? Evidence from a Sample of Island Economies.” Tourism Economics 13 (4): 675–84.

26.

Khadaroo

Seetanah

2008. “The Role of Transport Infrastructure in International Tourism Development: A Gravity Model Approach.” Tourism Management 29 (5): 831–40.

27.

Le-Klähn

Diem Trinh

Hall

C. Michael

. 2015. “Tourist Use of Public Transport at Destinations—A Review.” Current Issues in Tourism. London: Routledge.

28.

Lechner

Michael

. 2011. “The Estimation of Causal Effects by Difference-in-Difference Methods.” Foundations and Trends® in Econometrics 4 (3): 165–224.

29.

Lee

Choong Ki

Var

Turgut

Blaine

Thomas W.

1996. “Determinants of Inbound Tourist Expenditures.” Annals of Tourism Research 23 (3): 527–42.

30.

Leones

Julie

Colby

Bonnie

Crandall

Kristine

. 1998. “Tracking Expenditures of the Elusive Nature Tourists of Southeastern Arizona.” Journal of Travel Research 36 (3): 56–64.

31.

Lew

Alan

McKercher

Bob

. 2006. “Modeling Tourist Movements: A Local Destination Analysis.” Annals of Tourism Research 33 (2): 403–23.

32.

Lundgren

Jan O. J.

1982. “The Tourist Frontier of Nouveau Quebec: Functions and Regional Linkages.” Tourist Review 37 (2): 10–16.

33.

Marcano

Isaiah J.

2018. “E-hailing and Employement Rights: The Case for an Employment Relationship Between Uber and its Drivers in South Africa.” Cornell International Law Journal 51: 273–295.

34.

Marcussen

Carl H.

2011. “Determinants of Tourist Spending in Cross-sectional Studies and at Danish Destinations.” Tourism Economics 17 (4): 833–55.

35.

Marlina

Endy

Natalia

Dita A. R.

2017a. “Land Transportation and Tourism Development.” International Journal of Economic Perspectives 11 (2): 542–54.

36.

Marlina

Endy

Natalia

Dita A. R.

2017b. “Land Transportation and Tourism Development.” International Journal of Economic Perspectives 11 (2): 542–54.

37.

Mohamad

Wan Nor Ainun Al Baiyah Wan

Fuad

Anis Fatihah Mahamad

Shahib

Nor Syafiqah

Azmi

Azila

Bahari

Saiful

Kamal

Mohd

Abdullah

Dahlan

. 2016a. “A Framework of Customer’s Intention to Use Uber Service in Tourism Destination.” International Academic Research Journal of Business and Technology 2 (2): 102–6.

38.

Mok

Connie

Iverson

Thomas J.

2000. “Expenditure-Based Segmentation: Taiwanese Tourists to Guam.” Tourism Management 21 (3): 299–305.

39.

Naudé

Willem A.

Saayman

Andrea

. 2005. “Determinants of Tourist Arrivals in Africa: A Panel Data Regression Analysis.” Tourism Economics 11 (3): 365–91.

40.

Ngo

Victor

. 2015. Transportation Network Companies and the Ridesourcing Industry: A Review of Impacts and Emerging Regulatory Frameworks for Uber. Vancouver, BC.

41.

Page

Stephen

. 2004. “Transport and Tourism.” In A Companion to Tourism, edited by Lew

Alan A.

Hall

Michael C.

Williams

Allan M.

, 146–58. Malden, MA: Blackwell.

42.

Prideaux

Bruce

. 1996. “Recent Developments in the Taiwanese Tourist Industry—Implications for Australia.” International Journal of Contemporary Hospitality Management 8 (1): 10–15.

43.

Prideaux

Bruce

. 2000. “The Role of the Transport System in Destination Development.” Tourism Management 21 (1): 53–63.

44.

Rayle

Lisa

Dai

Danielle

Chan

Nelson

Cervero

Robert

Shaheen

Susan

. 2016. “Just a Better Taxi? A Survey-Based Comparison of Taxis, Transit, and Ridesourcing Services in San Francisco.” Transport Policy 45 (January): 168–78.

45.

Rey

Belén

Myro

Rafael L.

Galera

Asun

. 2011. “Effect of Low-Cost Airlines on Tourism in Spain. A Dynamic Panel Data Model.” Journal of Air Transport Management 17 (3): 163–67.

46.

Saayman

Andrea

Saayman

Melville

. 2008. “Determinants of Inbound Tourism to South Africa.” Tourism Economics 14 (1): 81–96.

47.

Sarma

Mrinmoy K.

2003. “Towards Positioning a Tourist Destination: A Study of North East India.” ASEAN Journal on Hospitality and Tourism 2.

48.

Seetanah

Boopen

Durbarry

Ramesh

Ragodoo

J. F. Nicolas

. 2010. “Using the Panel Cointegration Approach to Analyse the Determinants of Tourism Demand in South Africa.” Tourism Economics 16 (3): 715–29.

49.

Simth

Jeffrey A.

Todd

Petra E.

2005. “Does Matching Overcome LaLonde’s Critique of Nonexperimental Estimators?” Journal of Econometrics 125 (1-2): 305–353.

50.

Song

Haiyan

Wong

Kevin K. F.

Chon

Kaye K. S.

2003. “Modelling and Forecasting the Demand for Hong Kong Tourism.” International Journal of Hospitality Management 22 (4): 435–51.

51.

Spotts

Daniel M.

Mahoney

Edward M.

1991. “Segmenting Visitors to a Destination Region Based on the Volume of Their Expenditures.” Journal of Travel Research 29 (4): 24–31.

52.

Tham

Aaron

. 2016a. “Overcoming Resistance with Resilience When Researching Controversial Tourism Topics—The Case of Uber.” Paper presented at: Proceedings of the 2016 Travel and Tourism Research Association Asia-Pacific Chapter Conference; December 2-4, Dubai, United Arab Emirates.

53.

Tham

Aaron

. 2016b. “When Harry Met Sally: Different Approaches towards Uber and AirBnB—An Australian and Singapore Perspective.” Information Technology and Tourism 16 (4): 393–412.

54.

Thompson

Karen

Schofield

Peter

. 2007. “An Investigation of the Relationship between Public Transport Performance and Destination Satisfaction.” Journal of Transport Geography 15 (2): 136–44.

55.

Trans-Africa Consortium. 2008. Overview of Public Transport in Sub-Saharan Africa. Brussels, Belgium.

56.

Truong

Nguyen Van

Shimizu

Tetsuo

. 2017. “The Effect of Transportation on Tourism Promotion: Literature Review on Application of the Computable General Equilibrium (CGE) Model.” Transportation Research Procedia, 25:3096–3115.

57.

Viljoen

Armand

Saayman

Andrea

Saayman

Melville

. 2019. “Determinants Influencing Inbound Arrivals to Africa.” Tourism Economics 25 (6): 856–83.

58.

Virkar

Anjali R.

Mallya

Prita D.

2018. “A Review of Dimensions of Tourism Transport Affecting Tourist Satisfaction.” Indian Journal of Commerce & Management Studies 9 (1): 72–80.

59.

Wan

W. N. A. A. B.

Mohamad

A. F. M. F.

Shahib

Nor Syafiqah

Azmi

Azila

Mohd Kamal

Saiful Bahari

Abdullah

D. A.

2016. “Framework of Customer’s Intention to Use Uber Service in Tourism Destination.” International Academic Research Journal of Business and Technology 2 (2): 102–106.

60.

Wang

Youcheng

Rompf

Paul

Severt

Denver

Peerapatdit

Nichakarn

. 2006. “Examining and Identifying the Determinants of Travel Expenditure Patterns.” International Journal of Tourism Research 8 (5): 333–46.

61.

Westlake

Robbins

2005. “Transportation.” In Tourism: Principles and Practice, 3rd ed., edited by Cooper

Fletcher

Gilbert

Fyall

Wanhill

Essex, UK: Pearson Education.

62.

World Bank. n.d.-a. “International Tourism, Number of Arrivals.” Accessed November 6, 2019a. https://data.worldbank.org/indicator/st.int.arvl.

63.

World Bank. n.d.-b “International Tourism, Receipts for Passenger Transport Items (Current US$).” Accessed November 6, 2019b. https://data.worldbank.org/indicator/ST.INT.TRNR.CD.

64.

World Bank. n.d.-c “Urban Population (% of Total Population).” Accessed November 6, 2019b. https://data.worldbank.org/indicator/SP.URB.TOTL.IN.ZS.

65.

World Bank. n.d.-d “Political Stability and Absence of Violence/Terrorism.” Accessed November 6, 2019b. http://info.worldbank.org/governance/wgi/Home/downLoadFile?fileName=pv.pdf.

66.

Wyk

Cari Van

. 2016. “Uber Benefits for Tourists and Tourism.” Bizcommunity. https://www.bizcommunity.com/Article/196/747/141808.html.

67.

Xiao

Suili

Jia

Liming

Jiang

Linlin

. 2012. “Forest Recreation Opportunity Spectrum in the Suburban Mountainous Region of Beijing.” Journal of Urban Planning and Development 138 (4): 335–41.

68.

Zhang

Hongmei

Xiaoxiao

Cai

Liping A.

Lin

. 2014. “Destination Image and Tourist Loyalty: A Meta-analysis.” Tourism Management 40 (February): 213–23.