The impact of the media on tourism development and income inequality

Abstract

This paper examines whether the relationship between tourism development and income inequality is sensitive to the media environment. Using panel data from 88 countries for the period 1996 to 2020, we find that countries with uncensored media environments experience lower income inequality as the tourism industry develops. We also find that a favourable media environment enhances tourism development. Further analysis shows that asymmetries in a hostile media environment; namely, media biasedness, media corruption, and harassment of journalists, inhibit tourism development, particularly in emerging countries. This paper calls for strong support for press freedom to develop the tourism industry as countries emerge from the adverse effects of the COVID-19 pandemic.

Keywords:

• Income inequality
• tourism development
• press freedom
• voice and accountability
• hostile media environment
• emerging countries

Introduction

A growing number of studies show the salient role of the tourism sector in economic growth and development, particularly in developing countries (see Tosun et al., 2003; Incera & Fernández, 2015). This is consistent with the view that tourism help countries attract foreign equity and foreign direct investment, foreign currency earnings and tax revenues Tourism contributes to employment opportunities in developing countries and accounts for around 10% of global GDP (World Travel and Tourism Council (WTTC), 2015).

Existing studies provide theoretical perspectives and empirical evidence on the impact of tourism development on income inequality. Among the existing literature, while one school of thought emphasizes that tourism development positively impacts income inequality in both the regional- and country level, the other takes the view that tourism development has negative effects on income inequality. Both these theoretical perspectives are supported by empirical evidence. For instance, Bartik (1991) postulates that tourism prompts economic activities and economic growth within a local area, which adversely impacts the local area’s income distribution. This is within the context that growth in economic activities in the local economy will lead to increased inflation and property prices.

Papatheodorou (2004) further highlights theoretically that tourism development will have a substantial negative impact on local employment and income distribution. Within this line of argument, other studies document that transnational companies (TNCs) dominate the local tourism industry and mainly engage in negative activities (see Schilcher, 2007; Stabler et al., 2010). These studies emphasize that TNCs exploit domestic workers and resources, whilst they repatriate profits to their home country, which further exacerbates income inequality and poverty in the local tourist zones.

Nonetheless, the second and apparently opposite studies argue that tourism through pro-poor measures reduces income inequality (Incera & Fernández, 2015). The government, through pro-poor tourism policies, ensures that tourism reduces income inequality and provides net benefits to the poor in the local tourist areas. This is within the context that tourism generates employment opportunities for the vulnerable in local tourist areas through the provision of goods and services. Further, the government can use tax revenues from tourism to redistribute income to enhance the welfare of the poor (Incera & Fernández, 2015).

Given the gap in the current tourism literature, we ask two research questions. First, we examine whether the relationship between tourism development and inequality is moderated by press freedom. Second, we investigate whether censored and uncensored media environments have varying impacts on tourism development. We are motivated to undertake this study due to the COVID-19 pandemic and the role the media will play in addressing the fears and concerns of tourists, which has implications for both the tourism industry’s competitiveness and its development.

In this study, we aim to partially address these theoretical and empirical tensions by linking the contrasting arguments and results to a strand of studies that investigate the role of the media environment of a country, and that may explain the link between tourism development and income distribution. We test our view by interacting the media environment of a country with tourism development to determine their combined effects on income inequality. Given the two opposing economic impacts of tourism development on income distribution, we extend this debate by investigating the effects of censored and uncensored media environments on tourism development.

We argue that important media heterogeneity through press freedom will exert pressure on the national government to redistribute tax revenue equitably. The uncensored media environment will also highlight exploitations in local tourist areas and influence the government to deliver better public services and create more open policy processes. Earlier studies document that the news media influences government policies and responsiveness to societal issues (Hollander, 2014). A recent study by Gottfried et al. (2017) finds that the media exert varying influences on the public’s attitude towards the national government; whilst Mickoleit (2014) documents that several policies of governments around the world are influenced by social media and online news environments. This is consistent with the view that the media could bring controversial social issues around the tourism sector into the public agenda. A public speech made by President Xi Jinping of China in which he indicated that social media has become “the main battlefield for public opinion struggle”.

This study contributes to the tourism literature and provides evidence of how press freedom could improve sustainable tourism and alleviate income inequality in local tourist areas. Given the salient role of the media in impelling government policies and the opposing views concerning the benefits of the domestic tourism sector, we make the following two contributions to the tourism literature.

First, as far as we are aware, this study is the first to present comprehensive evidence on whether the uncensored media environment, through press freedom, moderates the relationship between tourism development and income inequality. This adds to existing studies that have examined the benefits and negative effects of tourism development (Bartik, 1991; Papatheodorou, 2004; Schicher, 2007; Stabler et al., 2010; Incera & Fernández, 2015). The study also extends the work of Alam and Paramati (2016) who find that tourism development increases income inequality.

Second, we add to the tourism literature by examining whether restricted media environments have varying effects on cross-country variations in tourism development. A large number of prior studies in the broader tourism literature have examined the effects of information flow on tourism development (Buhalis, 1998; Crouch & Ritchie, 1999; Dwyer & Kim, 2003; Enright & Newton, 2004). Although there are many studies, research into the censored media environment on tourism development remains limited. The question that naturally arises is whether internet censorship, media self-censorship, print and broadcast media perspectives, media harassment, media biasedness, and media corruption impact tourism development.

We are motivated to undertake this study due to the adverse effects of the recent COVID-19 pandemic on the tourism industry. Information flow through the mass media will play a significant role in increasing the appeal and attractiveness of a country to tourists. This is in line with the view that some countries are accused of under-reporting COVID-19 infection and death rates. To our knowledge, previous studies have mainly examined how information flow improves business-related factors’ administrative efficiency, enhances social and economic conditions, lowers corruption levels to attain a better delivery of services to citizens, and improves transparency to reduce unethical behaviour and political instability in order to attract tourists (Ades & Di Tella, 1999; Bhatnagar, 2000; Treisman, 2000; Norris & Zinnbauer, 2002; Wei, 2000; Enright & Newton, 2004; DiRienzo et al., 2007).

Using panel data from 88 countries from 1996 to 2020, we find that the relationship between tourism development and income inequality is sensitive to press freedom. We also find that voice and accountability are used to proxy press freedom and interact with tourism development to reduce inequality. This is consistent with the view that the media will freely highlight controversial issues and exploitation in the local tourist zones, and bring them to the attention of the government. We further find that media biasedness, media self-censorship, and government censorship reduce the impact of tourism development on income inequality. Our examination of the implications of the media environment suggests that national governments should encourage press freedom. Further, developing countries experience lower tourism development or competitiveness as a result of higher media censorship through greater media biasedness, harassment of journalists, and internet censorship.

Literature review

Relationship between tourism development, income inequality, and the media

The existing literature on the benefits of tourism development offers a contrasting theoretical perspective and empirical evidence. For instance, Lee and Jang (2011) find that domestic tourism-dependent firms’ relative non-tourism dependent firms experience higher income inequality. An earlier study by Wen and Tisdell (1997) documents that international tourism exacerbates income inequality in China; whilst Tosun et al. (2003) find that tourism development makes a marginal contribution to economic development and reduces the living quality amongst rural and coastal regions and several income groups. Employing a Gini coefficient to proxy for income distribution, Marcouiller et al. (2004) find that water-based tourist areas have a greater prevalence of income inequality. Contributing to the debate on tourism development and income distribution nexus, Manyara and Jones (2007) find that tourism development does not reduce income inequality in Kenya. A more recent study by Truong et al. (2014) finds that tour operators are major benefactors of tourism in Vietnam.

The role the media can play to moderate the relationship between tourism development and income inequality is missing in the tourism literature. We fill this gap by investigating whether the relationship between tourism development and income inequality is sensitive to or could be explained by, the media environment. This is consistent with the view that the mass media can play a salient role in advocating and highlighting exploitations within the tourist industry.

There is a broad literature on the economic impact of the tourism industry (see Dwyer, 2015). Tourism has become the main driver for economic growth and foreign exchange and accounts for approximately 10% of the global GDP (World Travel and Tourism Council (WTTC), 2015). The COVID-19 pandemic has severely impacted the tourism industry due to restrictions imposed on traveling, and social distancing regulations. As part of the call for support from the government on tourism recovery, a further question is whether the media environment can contribute to the tourism industry’s competitiveness.

Over time, extensive literature suggests that the tourism industry is prone to disease outbreaks (Pham et al., 2021), and both terrorism risk and attacks. In light of the reported evidence that several national governments have under-reported COVID-19 infection and death rates, it is conceivable that the mass media could play a salient role in the tourism industry’s recovery post COVID-19 pandemic. This is in line with the view that ceteris paribus, tourists care about their safety and attribute great importance to information available at tourist destinations before deciding to visit a country (see Ortega & Rodriguez, 2007). Several national governments will want the media to provide a good report about the handling of the COVID-19 pandemic to the global audience. Some countries may be intolerant towards the media that are critical of their government’s handling of the COVID-19 pandemic and this may have serious ramifications for the recovery of the tourism industry. Tourists will then avoid visiting countries that are hostile to the media and journalists. This is consistent with the view that harassment of the media will not help provide adequate information about the host country. Uncensored media play a key role in generating awareness of the destination country as a place to visit. We, therefore, investigate whether hostile media environments (i.e. media harassment, media biasedness, media corruption) impact tourism industry development. Protection against disease and any harm in the destination is very important to tourists. This can be highlighted by the local media and global tourists will then depend on the mass media to provide the information. The media communicates tourist risk information concerning any uncertainties regarding whether tourists will be able to fulfill their traveling intentions without being exposed to disease or harm. Inaccurate reporting by biased and corrupt media affiliated with the national government could potentially have the unintended consequence of putting tourists at a greater risk of harm by catching the disease. Clift and Page (1996) posit that the health, civil unrest, and political instability of the destination country are genuine concerns for tourists.

A review of the tourism literature shows that terrorism and political instability exacerbate the fears of tourists (Sonmez & Graefe, 1998). Even though prior studies have examined the perennial challenges which pose a risk to the safety of tourists and the thorny issue of the tourism industry, the COVID-19 pandemic has exacerbated fear amongst tourists and they will therefore depend on the media to access information on the destination country in order to make their travel decisions.

Theoretical framework and hypothesis development

We postulate that the media environment has positive and negative actions: an uncensored environment (U) and a censored or hostile environment (C). We based our study on action C (e.g. media harassment, media biasedness, media corruption, journalists’ harassment) which can directly be observed by the general public. Let $\alpha ϵ$[$0,1].$ The opportunist government can select $U$ or $C.$ This is consistent with the view that an opportunistic government derives private rent when $R>0$ in the period, and does not earn anything if $U$ is selected. $R$ captures renting-seeking opportunistic government that does not use tourism revenue to develop local tourist areas.

If the government selects $U,$ the mass media observe signal $u$ with the probability of $\pi ϵ(1\backslash 2,1)$ and signal $c$ with a probability of $1-\pi .$ This suggests that the media can influence the government to use tax revenue to develop local tourist zones. This will subsequently address income inequality. However, when the government selects action $C,$ they observe signal $c$ with the probability of $\pi$ and observe signal $u$ with a probability of $1-\pi .$ Where $\pi$ denotes the ability of the mass media to process and organize information. A signal is a piece of information the mass media has access to through their interaction and observation. The information could not be considered hard or soft evidence until it is verified. The general public cannot observe the signal until it is processed by the mass media and interpreted by the general public. In interpreting the signal to the general public, a censored media environment can inherently misreport the news to the general public when the action of the government is ($C).$ This will make the international community discredit information they receive about the country and will undoubtedly affect their decision to visit those countries. Evidence suggests that ordinary people are ignorant about the general level of income inequality in society (see Gimpelson & Treisman, 2018) due to limited information (McCall, 2013). However, Coppini et al. (2018) document how the media can influence government policies to address income inequality in Colombia.

Following the above transmission mechanism through which the media can influence the national government policies, intuitively, the media can highlight income inequalities in local tourist areas. Even though existing studies show that tourism development exacerbates income inequality (Alam & Paramati, 2016), we argue that an uncensored media environment will expose abuses in local tourist areas and highlight income inequalities. Drawing on the evidence of the role the media can play in addressing income inequalities in local tourist zones, we test the following two hypotheses.

H1: The media interacts with tourism development to reduce income inequality.

H2: Hostile media environment reduces tourism development.

Data

We used unbalanced panel data from 88 countries from 1996 to 2020, due to the unavailability of data for several countries. In line with the definition provided by Morgan Stanley Capital International (MSCI), our sample consists of 16 developed countries and 72 emerging countries. Income inequality (IIE) is calculated by the Gini coefficient. A higher value suggests a higher IIE whilst a lower value indicates a lower IIE. We sourced data from the Standardized World Income Inequality Database (SWIID). We employ three variables to proxy tourism development: (1)(1) $$II{E}_{it}=\alpha + {\beta }_1.T{D}_{it}+{\beta }_2.V{A}_{it}+{\beta }_3.TD\times V{A}_{it}+{\beta }_4.\mathit{Ctl}{s}_{it}+{\beta }_5.TF{E}_t+{\beta }_6.CF{E}_i+{ϵ}_{it}$$(1) International tourism expenditure (ITE) captures the expenditure incurred by international visitors to destination countries. (2)(2) $$T{D}_{it}=\alpha + {\beta }_1.HM{E}_{it}+{\beta }_2.\mathit{Ctl}{s}_{it}+{\beta }_3.TF{E}_t+{\beta }_4.CF{E}_i+{ϵ}_{it}$$(2) International tourism numbers of departure (ITND) reflect expenditures of outbound international visitors in other countries for all services offered during international transportation by non-resident carriers. (3)(3) $$II{E}_{it}=a+{\beta }_1.II{E}_{it-1}+{\beta }_2.V{A}_{it}+{\beta }_3.\gamma Z_{it}+{\beta }_4.\mathit{Ctl}{s}_{it}+{\beta }_5.TF{E}_t+{\beta }_6.CF{E}_i+{ϵ}_{it}$$(3) International tourism receipt for passenger transport (ITRPT) captures the number of departures that people undertake from their home country of usual residence to any other country for reasons other than employment. This is mainly to ensure that our results are not sensitive to a particular proxy of tourism development and also to ensure the robustness of our results. We obtained data for the tourism development variables from The World Bank, World Tourism Organization, and Yearbook of Tourism Statistics.¹ Our key independent variables of interest are voice and accountability (VA) which we use to proxy for freedom of the press. Voice and accountability capture the independence of the media (see Bird et al., 2008) and the extent to which citizens’ voices are heard and are able to participate in the selection of government. Next, we use variables that capture hostile media environment: internet censorship (Mecenefi), media self-censorship (Meslfcen), harassment of the media and journalists (Meharjrn), print and broadcast media perspective (Merange), media corruption (Mecorrpt), and media biasedness (Mebias). We obtained data from Varieties of Democracy (V DEM) provided by the University of Gothenburg, V-Dem Institute. Consistent with existing literature, we control for the effects of economic growth on income inequality using GDP growth (GDPG). Rubin and Segal (2015), and Kuznets (1955) show that impacts income inequality. The economics literature documents that inflation (Infl) influences income inequality (see Cysne et al., 2005). Siami-Namini and Hudson (2019) postulate that inflation is a monetary phenomenon that affects the income inequality of a country. We control the judicial environment of a country on income inequality using rule of law (RL). This is consistent with Bennett and Nikolaev (2016) who show that rule of law impacts long-run net income inequality. Next, we control for the effects of control of corruption (CC). Gupta et al. (2002) find that corruption increases income inequality and poverty. Finally, we control for the effects of political stability (PS), and government effectiveness (GE) on income inequality. This is consistent with the view that stable and effective governments will be able to address poverty and income inequality.

Empirical results

Summary statistics

Table 1 provides summary statistics of all the variables employed in our analysis. Table 2 presents the mean statistics for the individual countries. There are substantial variations in income inequality across countries.

Table 1. Summary statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
IIE	1821	39.16	9.48	23.8	65.8
VA	2037	0.06	0.94	−1.89	1.80
ITE	1979	11.40	10.75	0.00	60.26
ITRPT	2029	7.83	1.75	100000	2.00
ITND	1406	1.52	2.72	6000	1.60
Mecenefi	2200	0.73	1.26	−3.46	2.10
Meslfcen	2200	1.03	1.13	−2.69	3.28
Meharjrn	2200	0.99	1.22	−1.82	4.08
Merange	2200	1.20	1.18	−2.62	2.84
Mecorrpt	2200	0.95	1.26	−2.13	3.26
Mebias	2200	1.14	1.15	−3.03	3.14
GDPG	2017	3.72	3.80	−30.15	26.42
Infl	2093	9.62	61.78	−27.05	2630.12
RL	2037	0.04	1.05	−2.32	2.10
CC	2037	0.07	1.09	−1.73	2.47
PS	2037	−0.11	0.93	−2.85	1.76
GE	2037	0.13	1.02	−1.88	2.44

This table reports the summary statistics for the variables employed in the regression estimations. The variables’ details are described in the Appendix.

Table 2. Mean statistics for individual countries, 1996–2020.

Country	IIE	VA	ITE	ITRPT	ITND	Mecenefi	Meslfcen	Meharjrn	Merange	Mecorrpt	Mebias	GDPG	Infl	RL	CC	PS	GE
Argentina	46.20	0.38	6687500000	1470000000	9900000	1.89	1.84	1.44	2.01	0.70	1.99	2.24	18.68	−0.52	−0.34	−0.07	−0.06
Armenia	32.25	−0.55	546833333	43583333	744650	1.45	0.25	0.45	0.49	−0.20	0.38	6.30	4.81	−0.37	−0.60	−0.21	−0.18
Australia	34.10	1.41	22933333333	4445833333	6375000	1.68	2.12	2.16	1.66	2.73	2.44	3.16	2.67	1.77	1.92	1.01	1.71
Austria	30.21	1.38	10560000000	1368000000	9930000	1.66	1.93	3.13	2.46	2.62	2.00	1.84	1.58	1.86	1.71	1.15	1.70
Bangladesh	32.78	−0.45	664583333	362500000	1508000	−0.55	−0.02	−0.36	1.60	0.93	1.12	5.92	6.21	−0.83	−1.06	−1.21	−0.73
Belarus	27.95	−1.45	695416666	84125000	7215000	0.10	−0.33	−0.40	−1.56	−1.70	−1.36	4.86	50.15	−1.02	−0.49	0.20	−0.78
Belgium	28.43	1.38	18611111111	1950000000	9212500	1.32	3.28	2.79	2.36	3.24	2.25	1.86	1.58	1.37	1.48	0.80	1.57
Bolivia	51.41	0.00	450041666	91666666	584375	1.44	1.24	1.38	1.42	0.96	1.10	4.09	5.29	−0.85	−0.63	−0.47	−0.47
Botswana	59.50	0.55	234250000	4207500	.NA	1.39	1.36	1.24	1.36	1.70	0.93	4.47	7.37	0.62	0.91	1.02	0.51
Brazil	55.48	0.44	13337500000	2155416667	5786957	1.78	1.59	1.21	1.73	1.32	1.75	2.26	7.84	−0.21	−0.11	−0.18	−0.12
Bulgaria	36.93	0.48	1418750000	287083333	4145833	1.79	0.80	0.88	1.61	0.11	1.47	2.50	45.91	−0.09	−0.16	0.28	0.12
Cameroon	43.98	−1.04	502500000	114875000	.NA	−0.24	0.38	0.70	1.15	−0.39	0.54	4.29	2.17	−1.11	−1.14	−0.72	−0.84
Canada	33.28	1.47	22250000000	4325000000	747000000	1.54	1.83	2.38	1.17	2.90	1.53	2.41	1.95	1.77	1.96	1.07	1.82
Chile	48.96	1.03	1747916667	425833333	2860000	2.02	1.82	2.04	1.54	2.78	1.96	3.89	4.32	1.27	1.38	0.52	1.17
China	40.34	−1.58	14387500000	1237500000	60625000	−2.01	−0.88	−0.97	−1.70	−1.17	−2.28	8.96	3.10	−0.45	−0.41	−0.43	0.11
Colombia	53.94	−0.15	3012500000	495416666	2370826	0.78	1.07	−0.26	1.19	1.18	1.35	3.25	7.85	−0.48	−0.30	−1.53	−0.12
Costa Rica	48.51	1.05	695416666	163958333	623875	1.76	2.43	1.77	1.48	2.09	2.39	4.01	8.44	0.55	0.65	0.66	0.32
Czech Republic	26.16	0.98	. 4255555556	127277777	7005882	1.83	2.40	1.70	2.43	2.09	2.24	2.62	2.86	0.96	0.42	0.96	0.91
Democratic Rep	42.15	−1.47	135357142	NA	220000	−1.58	0.60	−0.42	1.29	−0.42	0.87	3.45	178.96	−1.68	−1.42	−2.21	−1.64
Denmark	26.82	1.58	7630434783	7800000000	6658333	1.99	3.05	3.74	2.60	3.26	2.99	1.69	1.74	1.93	2.33	1.10	2.04
Dominican Rep	47.79	0.13	595416666	230416666	410166.7	1.92	1.13	0.93	1.24	−0.58	1.22	5.29	8.30	−0.58	−0.73	−0.02	−0.48
Ecuador	49.42	−0.17	808750000	247833333	831791.7	−0.57	0.99	0.77	1.57	0.88	1.04	3.00	3.76	−0.88	−0.71	−0.50	−0.64
Egypt	31.07	−1.10	2391666667	253833333	5316667	−1.36	−0.48	−0.59	−0.28	−0.40	−0.31	4.58	9.69	−0.27	−0.61	−0.91	−0.49
El Salvador	45.72	0.06	314090909	54545454	1250870	1.32	1.20	0.64	0.89	0.53	0.55	2.04	2.66	−0.66	−0.46	−0.02	−0.29
Ethiopia	32.00	−1.23	133583333	5766666	133000	−0.68	−0.77	−0.37	−0.86	0.56	0.37	8.17	9.85	−0.70	−0.60	−1.43	−0.67
Fiji	38.40	−0.29	108750000	15458333	116285	−0.55	−0.13	−0.30	0.49	1.53	0.46	2.37	3.88	−0.34	0.11	0.35	−0.33
Finland	27.49	1.55	4575000000	803913043	8520000	1.87	2.48	3.16	2.63	2.92	2.15	2.21	1.59	1.98	2.29	1.38	2.07
France	32.01	1.23	43909090909	8122727273	34071428	1.75	3.00	2.79	1.64	2.88	2.39	1.63	1.27	1.44	1.37	0.46	1.50
Gabon	40.10	−0.80	339473684	93105263	200333	0.96	0.29	0.75	1.57	0.19	1.08	2.09	4.48	−0.53	−0.83	0.20	−0.71
Germany	30.62	1.38	83250000000	8883333333	83636363	1.65	2.33	3.73	2.67	3.13	2.58	1.39	1.14	1.68	1.85	0.88	1.64
Ghana	42.20	0.35	716666666	355458333	.NA	1.57	1.38	1.18	1.46	0.70	1.82	5.81	22.80	0.01	−0.15	−0.02	−0.10
Guatemala	52.37	−0.30	660416666	NA	864708.3	1.09	0.97	0.60	0.93	0.66	0.88	3.56	4.73	−1.04	−0.71	−0.73	−0.62
Guinea	38.70	−1.06	70625000	49312500	.NA	−1.01	0.91	0.20	0.87	−0.33	1.09	4.45	11.73	−1.32	−1.02	−1.25	−1.04
Guinea-Bissau	43.15	−0.81	25713333	3566666	.NA	0.46	0.66	−0.06	1.05	0.22	1.25	2.47	9.42	−1.36	−1.27	−0.71	−1.30
Honduras	54.02	−0.34	357500000	86166666	1025556	1.25	1.07	−0.05	0.83	−0.22	1.05	3.79	6.94	−0.95	−0.85	−0.42	−0.64
Hungary	29.87	0.83	2533333333	374666666	16041666	1.78	1.36	2.12	1.99	1.00	0.87	2.64	6.49	0.75	0.43	0.84	0.72
India	35.17	0.42	13041176471	2398235294	11829166	−0.15	1.35	0.58	2.26	0.84	1.36	6.50	5.67	0.06	−0.38	−1.11	−0.03
Indonesia	35.28	−0.11	8355555556	2155555556	6185000	0.03	0.90	0.53	0.87	0.41	1.40	4.39	11.85	−0.59	−0.71	−1.05	−0.25
Iran, Islamic Rep	41.31	−1.36	5915909091	407136363	8300000	−1.91	0.00	−0.61	−0.66	−0.38	−0.65	2.57	19.74	−0.83	−0.64	−1.03	−0.48
Israel	40.31	0.68	5195833333	1128750000	4558333	1.03	0.93	1.53	1.88	1.80	1.55	3.55	3.01	0.99	0.96	−1.17	1.23
Italy	34.72	1.03	26000000000	4305882353	50304347	1.93	1.34	2.14	1.98	2.00	2.08	0.61	1.91	0.51	0.28	0.54	0.52
Jamaica	44.58	0.58	313913043	85434782	.NA	1.95	1.53	1.30	1.99	1.34	1.74	0.56	9.27	−0.33	−0.21	−0.03	0.20
Japan	33.27	1.02	36375000000	9404166667	16916666	1.67	1.11	1.66	2.22	2.55	1.90	0.87	−0.52	1.37	1.38	1.05	1.45
Jordan	34.72	−0.66	994166666	119583333	1621739	−0.28	−0.53	−0.11	0.39	0.58	0.96	4.19	4.03	0.31	0.16	−0.36	0.10
Kazakhstan	30.32	−1.11	1690833333	170791666	8128571	−0.97	−0.40	−0.10	−0.66	−0.59	−0.62	5.56	13.31	−0.77	−0.92	0.12	−0.43
Kenya	44.10	−0.33	175000000	26750000	.NA	1.45	1.04	0.46	1.76	0.74	1.55	4.35	9.09	−0.77	−0.98	−1.18	−0.51
Liberia	35.00	−0.49	43333333	22166666	.NA	1.11	1.20	1.02	1.04	−0.28	1.23	2.42	4.95	−1.20	−0.92	−1.08	−1.40
Malawi	48.98	−0.20	87916666	22958333	.NA	1.64	0.87	0.75	1.53	0.81	1.09	4.26	22.56	−0.26	−0.59	−0.07	−0.58
Malaysia	44.59	−0.37	7516666667	834583333	29000000	−0.65	−0.01	0.26	−1.83	−0.86	−0.59	4.89	3.39	0.47	0.22	0.22	1.01
Mali	37.27	−0.01	150130434	53869565	.NA	0.42	0.76	0.99	1.96	0.22	1.83	5.10	3.62	−0.46	−0.67	−0.60	−0.85
Mexico	49.48	0.12	9241666667	1652916667	105500000	0.85	1.22	0.52	1.07	0.20	1.48	2.63	7.67	−0.50	−0.46	−0.59	0.17
Moldova	32.95	−0.17	254583333	51041666	182036.4	0.93	0.61	0.71	0.58	0.25	1.08	3.17	12.55	−0.38	−0.72	−0.21	−0.58
Morocco	40.05	−0.62	1521666667	531458333	1929167	0.97	−0.20	0.07	1.03	0.16	0.46	4.23	1.06	−0.12	−0.24	−0.35	−0.13
Mozambique	50.05	−0.21	208047619	30619047	.NA	1.14	1.18	0.40	1.30	−0.12	0.63	7.28	8.67	−0.74	−0.59	−0.06	−0.59
Namibia	61.13	0.42	62545454	8272727	.NA	1.57	1.09	1.28	1.97	2.18	1.27	3.71	7.22	0.21	0.36	0.69	0.15
Nepal	38.30	−0.51	409166666	84916666	454000	0.09	1.07	0.54	1.29	0.17	1.62	4.48	7.44	−0.61	−0.69	−1.28	−0.79
Netherlands	28.54	1.55	17875000000	1582666667	16956521	1.92	1.70	1.81	2.20	2.64	1.95	2.02	1.79	1.81	2.05	1.08	1.87
Niger	36.88	−0.45	72826086	25521739	10000	−0.02	1.07	1.00	1.68	0.66	1.40	4.69	3.41	−0.60	−0.74	−0.74	−0.76
Nigeria	42.53	−0.68	5818260870	1550217391	.NA	1.00	1.84	0.52	1.92	0.33	1.90	5.15	11.93	−1.14	−1.15	−1.81	−1.03
Norway	27.32	1.62	12087500000	866250000	5388409	1.89	2.42	2.69	2.73	2.80	2.56	2.02	3.60	1.95	2.12	1.29	1.90
Panama	53.23	0.52	641666666	139750000	473333	1.46	1.62	0.98	0.84	0.87	1.24	5.83	2.53	−0.10	−0.34	0.12	0.14
Paraguay	51.00	−0.18	282916666	114375000	3485714	1.19	0.27	0.10	0.61	−0.30	0.89	3.00	8.35	−0.84	−1.04	−0.55	−0.88
Peru	45.89	0.07	5562500000	371125000	2682353	1.53	1.61	1.10	1.65	1.14	1.46	5.12	4.91	−0.40	−0.57	−1.22	−0.02
Philippines	47.88	0.05	1540833333	377500000	1815333	0.59	1.42	0.23	1.50	0.31	1.98	4.40	3.47	−0.58	−0.35	−0.75	−0.26
Poland	33.17	0.96	7120833333	421791666	46250000	1.39	2.29	2.42	2.43	1.78	1.84	4.10	4.04	0.63	0.54	0.69	0.60
Portugal	36.28	1.24	4179166667	739166666	2000000	2.00	2.33	2.60	2.02	2.27	2.01	1.43	2.28	1.14	1.05	1.02	1.11
Romania	36.31	0.43	2011250000	245583333	10758333	1.85	1.14	1.44	1.42	0.41	1.02	3.23	21.58	0.04	−0.24	0.21	−0.25
Rwanda	47.30	−1.28	155666666	38809523	.NA	−1.54	−0.88	−1.33	−0.62	−0.17	−1.36	8.12	6.38	−0.49	0.08	−0.63	−0.28
Senegal	40.23	0.09	206956521	95739130	.NA	1.58	2.35	1.25	2.51	0.61	2.28	4.54	1.86	−0.17	−0.18	−0.25	−0.29
Sierra Leone	36.63	−0.38	34565217	6343478	63173.91	1.71	0.94	0.92	1.22	−0.14	0.70	4.77	14.09	−0.98	−0.83	−0.52	−1.23
South Africa	62.12	0.66	5025000000	1856250000	.NA	1.51	1.32	1.10	2.04	1.68	2.06	2.61	6.83	0.11	0.24	−0.15	0.50
Spain	34.62	1.14	6533333333	1533333333	16384615	1.72	2.20	2.36	2.15	2.02	2.12	2.18	2.05	1.14	1.03	0.07	1.24
Sri Lanka	38.80	−0.30	1054583333	388375000	934083.3	−0.26	0.78	0.41	1.19	0.56	0.76	5.07	8.48	0.04	−0.27	−0.90	−0.17
Sudan	34.80	−1.74	455086956	622407358	.NA	−1.90	−0.77	−0.02	−0.34	−0.25	−0.35	4.57	21.03	−1.37	−1.30	−2.19	−1.33
Sweden	27.76	1.58	10300000000	1177333333	20333333	2.10	2.42	2.29	2.21	3.21	2.22	2.47	1.64	1.91	2.21	1.20	1.93
Switzerland	32.80	1.54	14087500000	2312500000	19846153	1.92	2.92	3.62	1.99	2.89	2.72	1.89	0.41	1.89	2.07	1.34	1.97
Tanzania	38.98	−0.31	713913043	43695652	148000	0.75	0.30	0.65	1.63	0.19	1.38	5.97	11.45	−0.40	−0.58	−0.38	−0.56
Thailand	39.40	−0.38	6979166667	1315833333	4570833	−0.42	0.11	0.40	0.49	0.85	0.98	3.31	2.63	0.04	−0.33	−0.70	0.29
Togo	43.77	−0.97	46565217	24130434	228000	−1.46	0.64	0.56	1.02	−0.64	0.94	4.02	2.87	−0.83	−0.88	−0.41	−1.25
Tunisia	36.78	−0.57	547500000	68083333	2209091	−1.60	0.08	0.25	0.49	0.17	0.02	3.60	4.52	−0.05	−0.12	−0.27	0.21
Turkey	40.77	−0.28	5020000000	621333333	6529167	−0.47	0.33	0.29	0.49	−0.25	0.68	4.70	27.75	−0.02	−0.11	−1.09	0.15
Uganda	42.59	−0.64	338125000	106421052	311050	−0.31	1.03	−0.01	1.77	1.12	1.15	6.37	8.32	−0.43	−0.93	−1.03	−0.52
Ukraine	26.92	−0.19	3700000000	302666666	22230769	0.85	0.05	0.24	1.22	−0.39	0.49	1.54	18.60	−0.80	−0.95	−0.68	−0.61
UK	34.85	1.33	36333333333	7233333333	66708333	1.56	2.14	2.65	2.43	2.38	2.10	2.10	1.95	1.71	1.84	0.47	1.66
US	40.72	1.17	1.13833E + 11	31625000000	110521739	1.28	2.33	2.56	1.76	2.76	1.51	2.47	1.89	1.58	1.46	0.47	1.58
Venezuela	47.75	−0.77	2509523810	880952381	1199682	0.16	0.47	0.18	0.70	1.03	0.67	2.42	31.52	−1.59	−1.17	−1.12	−1.11
Vietnam	36.13	−1.40	2713333333	1241351662	.NA	−0.95	−0.92	−0.78	−2.10	0.04	−2.52	6.57	7.70	−0.39	−0.56	0.24	−0.23
Zimbabwe	45.93	−1.31	304166666	81222222	1312810	−0.60	−0.05	−0.85	0.98	−0.37	−0.54	0.92	5.57	−1.51	−1.22	−0.94	−1.14

This table shows the summary statistics for all the variables across countries in the tests. The variable definitions are detailed in the Appendix. We used natural logarithm of ITE, ITRPT, and ITND in the regression.

The effects of press freedom on the relationship between tourism development and income inequality

In this section, we investigate whether voice and accountability moderate the relationship between tourism development and income inequality. We employ the Gini coefficient, which captures the extent to which national wealth is unevenly distributed, to proxy for income inequality. The Gini coefficient has a limitation as it is unable to differentiate categories of inequality (De Maio, 2007). However, the Gini coefficient has been widely used as a measure of income inequality in the economics and political science literature (see Piketty & Saez, 2003; Bartels, 2008). All significant coefficients are presented with asterisks and the t-statistics are presented in parentheses. We estimate the results using Equation (1)(1) $$II{E}_{it}=\alpha + {\beta }_1.T{D}_{it}+{\beta }_2.V{A}_{it}+{\beta }_3.TD\times V{A}_{it}+{\beta }_4.\mathit{Ctl}{s}_{it}+{\beta }_5.TF{E}_t+{\beta }_6.CF{E}_i+{ϵ}_{it}$$(1) . (1) $$II{E}_{it}=\alpha + {\beta }_1.T{D}_{it}+{\beta }_2.V{A}_{it}+{\beta }_3.TD\times V{A}_{it}+{\beta }_4.\mathit{Ctl}{s}_{it}+{\beta }_5.TF{E}_t+{\beta }_6.CF{E}_i+{ϵ}_{it}$$(1) where IIE is income inequality using the Gini coefficient. TD is a measure of tourism development (i.e. ITR, ITRPP, and ITND) entered successively. TD × VA is the interaction between tourism development, voice, and accountability. Ctls are control variables. TFE is time fixed effect. CFE is country fix effects.

Table 3 presents the results. As evident in models 1 to 3, the coefficients on the three proxies of tourism development are all positive and statistically significant at the 10% level; ITE 0.391 (t-statistics = 1.93) in model 1, ITRPT 0.305 (t-statistics = 1.75) in model 2, and ITND 0.209 (t-statistics = 3.06) in model 3. The results suggest that tourism development increases income inequality, and are consistent with Papatheodorou (2004), Schilcher (2007) who document that tourism development increases income inequality. The coefficients on VA are negative and statistically significant at the 1% level. The coefficients on VA are −0.247 (t-statistics=-6.85) in model 1, −0.296 (t-statistics=-8.96) in model 2, and −0.187 (t-statistics=-12.29) in model 3.

Table 3. Press freedom and tourism development.

	Model (1) IIE	Model (2) IIE	Model (3) IIE
ITE	0.391*
	(1.93)
ITRPT		0.305*
		(1.75)
ITND			0.209***
			(3.06)
VA	−0.247***	−0.296***	−0.187***
	(−6.85)	(−8.96)	(−12.29)
ITE*VA	−0.918***
	(−4.81)
ITRPT*VA		−1.076***
		(−7.08)
ITND*VA			−0.783**
			(−3.83)
GDPG	0.386***	0.374***	0.352***
	(14.18)	(14.50)	(11.12)
Infl	0.033*	0.022	0.089***
	(1.93)	(1.36)	(3.65)
RL	−0.120***	−0.119***	−0.955***
	(−10.27)	(−11.04)	(−8.01)
CC	−0.346***	−0.306***	−0.302***
	(−3.28)	(−3.21)	(−3.08)
PS	−0.380	0.336	−1.682***
	(−0.71)	(0.72)	(−3.40)
GE	−0.917	−0.761	−0.656
	(−0.76)	(−0.68)	(−0.54)
Constant	2.854***	2.986***	3.921***
	(7.34)	(7.91)	(12.79)
Number of observations	1779	1696	1287
Adjusted R-squared	0.49	0.50	0.54

This table reports the multivariate test of whether the relationship between tourism developments is sensitive to press freedom proxy by voice and accountability (VA). ITE is international tourism expenditure; ITR is international tourism receipt; ITRPT is international tourism receipt for passenger transport; ITND is international tourism numbers of departures; ITE*VA is the interaction between international tourism expenditure and voice and accountability; ITRPT*VA is the interaction between international tourism receipt for passenger transport and voice and accountability; ITND*VA is the interaction between international tourism numbers of departure and voice and accountability; GDPG is GDPG growth; Infl is inflation; RL is rule of law; CC is control of corruption; PS is political stability; GE government effectiveness. The t-statistics, reported in parentheses, are based on standard errors double clustered at the country and year levels. For tractable interpretation, all the coefficients are reported as partial elasticity and the statistical significance is reported against 10% (*), 5% (**) and 1% (***) significance levels.

Further analyses show that the interaction term between voice and accountability and all the tourism development proxies are negative and statistically significant at the conventional level. In models 1 to 3, the coefficients on the interactive terms are ITE*VA − 0.918 (t-statistics=-4.81) in model 1, ITRPT*VA − 1.076 (t-statistics=-7.08) in model 2, ITND*VA − 0.783 (t-statistics=-3.83) in model 3.

The effects of a censored media environment on tourism development

In this section, we investigate the effects of a hostile media environment (HME) on cross-country tourism development. We estimate the results using Equation (2)(2) $$T{D}_{it}=\alpha + {\beta }_1.HM{E}_{it}+{\beta }_2.\mathit{Ctl}{s}_{it}+{\beta }_3.TF{E}_t+{\beta }_4.CF{E}_i+{ϵ}_{it}$$(2) . (2) $$T{D}_{it}=\alpha + {\beta }_1.HM{E}_{it}+{\beta }_2.\mathit{Ctl}{s}_{it}+{\beta }_3.TF{E}_t+{\beta }_4.CF{E}_i+{ϵ}_{it}$$(2) where TD is tourism development. HME is a hostile media environment. Within the set of key explanatory variables of interest are the hostile media environment measures (i.e. Mecenefi, Meslfcen, and Meharjrn). Table 4 reports the results. In models 1 to 3, the coefficients are negative and statistically significant at the 1% level. The coefficients are Mecenefi − 0.371 (t-statistics=-12.17) in model 1, Meslfcen − 0.116 (t-statistics=-10.74) in model 2, and Meharjrn − 0.456 (t-statistics=-3.39) in model 3.

Table 4. Hostile media environment and tourism development.

	Model (1) ITND	Model (2) ITND	Model (3) ITND
Mecenefi	−0.371***
	(−12.17)
Meslfcen		−0.116***
		(−10.74)
Meharjrn			−0.456***
			(−3.39)
GDPG	−0.014	0.028***	0.007
	(−1.30)	(3.67)	(1.20)
Infl	−0.001*	−0.001	−0.001*
	(−1.64)	(−1.15)	(−1.89)
RL	1.671***	0.800***	0.911***
	(6.94)	(6.71)	(8.89)
CC	−0.808***	−0.036	−0.046
	(−3.85)	(−0.36)	(−0.52)
PS	−0.240**	−0.141***	−0.180***
	(−2.20)	(−2.66)	(−3.98)
GE	−0.138	−0.155	−0.112
	(−0.68)	(−1.56)	(−1.31)
Number of observations	1378	1204	1204
Adjusted R-squared	0.24	0.28	0.24

This table reports the multivariate test of the impact of a hostile media environment on tourism development. All variables employed in the regression are described in Appendix A. The t-statistics, reported in parentheses, are based on standard errors double clustered at the country and year levels. For tractable interpretation, all the coefficients are reported as partial elasticity and the statistical significance is reported against 10% (*), 5% (**) and 1% (***) significance levels.

Alternative measures of the media environment

In this section, we provide robustness to the impact of a hostile media environment on tourism development by employing alternative measures of the hostile media environment. We mainly use a print and broadcast media perspective (Merange), media corruption (Mecorrpt), and media biasedness (Mebias). We estimate the results using Equation (2)(2) $$T{D}_{it}=\alpha + {\beta }_1.HM{E}_{it}+{\beta }_2.\mathit{Ctl}{s}_{it}+{\beta }_3.TF{E}_t+{\beta }_4.CF{E}_i+{ϵ}_{it}$$(2) .

We present the results in Table 5. The coefficient on the alternative measures of a hostile media environment is negative and statistically significant at the 1% level. The estimated coefficients are Mecorrpt − 0.165 (t-statistics=-2.09) in model 1, Merange − 0.168 (t-statistics=-3.29) in model 2, Meharjrn − 0.179 (t-statistics=-2.30) in model 3, and Mebias − 0.142 (t-statistics=-2.66) in model 4. The results suggest the negative impact of a hostile media environment on tourism development is robust to alternative proxies.

Table 5. Alternative hostile media environment measures.

	Model (1) ITND	Model (2) ITND	Model (3) ITND	Model (4) ITND
Mecorrpt	−0.165**
	(−2.09)
Merange		−0.168***
		(−3.29)
Meharjrn			−0.179**
			(−2.30)
Mebias				−0.142***
				(−2.66)
GDPG	0.014***	0.015***	0.014***	0.015***
	(4.69)	(4.98)	(4.67)	(4.80)
Infl	−0.001	−0.002	−0.003	−0.000
	(−0.41)	(−0.58)	(−0.49)	(−0.35)
RL	0.001	0.006	0.029	0.018
	(0.00)	(0.09)	(0.41)	(0.25)
CC	−0.019	−0.005	0.004	0.0102
	(−0.31)	(−0.08)	(0.07)	(0.16)
PS	−0.075**	−0.077**	−0.060**	−0.066**
	(−2.47)	(−2.50)	(−1.98)	(−2.17)
GE	−0.083	−0.072	−0.064	−0.079
	(−1.37)	(−1.18)	(−1.07)	(−1.32)
Number of observations Adjusted R-squared	1253 0.15	1253 0.15	1253 0.17	1253 0.15

This table reports the multivariate test of the impact of a hostile media environment on tourism development using four alternative measures of a hostile media environment. All variables employed in the regression are described in Appendix A. The t-statistics, reported in parentheses, are based on standard errors double clustered at the country and year levels. For tractable interpretation, all the coefficients are reported as partial elasticity and the statistical significance is reported against 10% (*), 5% (**) and 1% (***) significance levels.

Dynamic generalized methods of moments

In this section, we address endogeneity as a result of reverse causality using the dynamic generalized method of moments (GMM). This is consistent with the view that countries with lower income inequality may have better press freedom. This provides robustness to our baseline regression as the dynamic GMM allows us to capture previous influences with a lagged value of the dependent variable (IIE). An extra benefit of employing the dynamic GMM model is to account for the persistence of income inequality. Following Wintoki et al. (2012), we use lagged ITRPT as an instrument; predetermined variables: LagITRPP, and GDPG; exogenous variables: Infl, CC, GE, PS, RE, RL, and country and year dummies. Instruments employed: endogenous variables are instrumented by lagged levels dated t-2 to t-3 (first-differences equation) and by lagged first first-differences (levels equation). The predetermined variables are instrumented by lagged levels dated t-1 to t-2 (first-differences equation) and by first-differences (levels equation). We, therefore, estimate the GMM results using equation (1)(1) $$II{E}_{it}=\alpha + {\beta }_1.T{D}_{it}+{\beta }_2.V{A}_{it}+{\beta }_3.TD\times V{A}_{it}+{\beta }_4.\mathit{Ctl}{s}_{it}+{\beta }_5.TF{E}_t+{\beta }_6.CF{E}_i+{ϵ}_{it}$$(1) . (3) $$II{E}_{it}=a+{\beta }_1.II{E}_{it-1}+{\beta }_2.V{A}_{it}+{\beta }_3.\gamma Z_{it}+{\beta }_4.\mathit{Ctl}{s}_{it}+{\beta }_5.TF{E}_t+{\beta }_6.CF{E}_i+{ϵ}_{it}$$(3) where, ${\mathit{IIE}}_{it}$ represents income inequality; ${\mathit{IIE}}_{it-1}$ represents lagged income inequality measure as an instrument; ${VA}_{it}$ represents voice and accountability; $Ct{ls}_{it}$ represents a vector of the control variables of country $i$ at time $t;$ TFE represents time effects; CFE represents country-fixed effects. Following Wintoki et al. (2012), we use the Hansen J test to estimate whether the GMM model employed is valid or not. The Hansen test results of overidentification are below the null, and these confirm that all the instruments used in the GMM model are valid. The results reported in Table 6 are not statistically different from the baseline regression in Table 3. This suggests that our main results are robust.

Table 6. Dynamic GMM.

	Model (1) IIE	Model (2) IIE	Model (3) IIE
ITE	0.087**
	(2.36)
VA	−0.173***	−0.484***	−0.303***
	(−2.90)	(−8.99)	(−6.68)
ITE × VA	−0.520*
	(−1.66)
ITRPT		1.141***
		(4.69)
ITRPT × VA		−1.955***
		(−7.80)
ITND			1.128***
			(4.94)
ITND × VA			−1.529***
			(−5.39)
GDPG	−0.386***	−0.359***	−0.326***
	(−13.63)	(−13.29)	(−9.47)
Infl	0.022	0.012	0.078***
	(1.28)	(0.72)	(3.15)
RL	−0.140***	−0.147***	−0.134***
	(−11.16)	(−12.67)	(−10.15)
CC	−0.856***	−0.864***	−0.914***
	(−7.21)	(−7.68)	(−7.36)
PS	−0.633	−0.070	−1.959***
	(−1.08)	(−0.13)	(−3.55)
GE	−0.209	−0.392***	−0.211
	(−1.61)	(−3.08)	(−1.51)
Constant	3.618***	6.266***	5.481***
	(7.92)	(11.98)	(15.51)
Number of observations	1600	1633	1269
AR1 (P-value) AR2 (P-value) Hansen J Statistics	0.002 0.207 0513	0.001 0.186 0.455	0.005 0.512 0.718

This table reports the multivariate test of whether the relationship between tourism developments is sensitive to press freedom proxy by voice and accountability (VA) using dynamic GMM. All variables employed in the regression are described in Appendix A. The t-statistics, reported in parentheses, are based on standard errors double clustered at the country and year levels. For tractable interpretation, all the coefficients are reported as partial elasticity and the statistical significance is reported against 10% (*), 5% (**) and 1% (***) significance levels.

Emerging countries

In this section, we partition our sample countries into emerging countries. This is to determine whether the media environment has a varying impact on tourism development across developed and emerging countries. Table 7 reports the results of emerging countries. The evidence in Table 7 reinforces our main result findings in Table 4. The coefficient on the three hostile environment measures, i.e. media corruption (Mecorrpt), harassment of journalists (Meharjrn), and media biasedness (Mebias) are negative and statistically significant. The coefficients are Mecorrpt − 0.815 (t-statistics=-1.91) in model 1, Meharjrn − 0.212 (t-statistics=-4.92) in model 2, and Mebias − 0.614 (t-statistics=-1.85) in model 3. Overall, the separate regression results confirm that hostile media environments, such as the corruption of the media, harassment of journalists, and media biasedness, inhibit tourism development in emerging countries.

Table 7. Emerging countries.

	ITE	ITE	ITE
Mecorrpt	−0.815*
	(−1.91)
Meharjrn		−0.212***
		(−4.92)
Mebias			−0.614*
			(−1.85)
GDPG	0.337	0.308	0.363
	(1.33)	(1.23)	(1.43)
Infl	−0.021*	−0.021*	−0.020*
	(−1.96)	(−1.96)	(−1.91)
RL	1.169	1.652	1.300
	(0.91)	(1.30)	(1.00)
CC	0.733***	0.734***	0.683***
	(5.77)	(6.02)	(5.57)
PS	0.030	0.532	0.024
	(0.06)	(0.98)	(0.05)
GE	−1.533	−1.680	−1.618
	(−1.32)	(−1.46)	(−1.39)
Constant	1.606***	1.749***	1.604***
	(22.17)	(23.19)	(22.12)
Number of observations	1558	1558	1558
Adjusted R-squared	0.10	0.12	0.10

This table reports the multivariate test of the impact of a hostile media environment on tourism development in emerging countries. All variables employed in the regression are described in Appendix A. The t-statistics, reported in parentheses, are based on standard errors double clustered at the country and year levels. For tractable interpretation, all the coefficients are reported as partial elasticity and the statistical significance is reported against 10% (*), 5% (**) and 1% (***) significance levels.

Discussion

The results suggest that countries with press freedom reduce the negative effects of tourism development on income inequality. This is consistent with the view that the mass media will put pressure on the local and national authorities to use tourism revenues for the benefit of the local people living in the tourist areas/zones. Further, the media will freely highlight any form of exploitation in the tourist zones.

Tourism development can lead to income inequality which is consistent with Manyara and Jones (2007) who provide evidence that tourism does not enhance income distribution in Kenya’s tourist areas. However, evidence from the results shows that the adverse effects of tourism development can be mitigated when there is greater freedom of the press. The results have implications for national governments. For instance, developing countries seeking policies and reforms to address income inequalities following the COVID-19 pandemic can enhance press freedom.

The analysis shows that a policy on media freedom will be needed to reduce the adverse effects of tourism development on income inequality. This will potentially improve tourist guides, transport, hotel, and restaurant industry as countries attract more international tourist receipts.

The findings of the study indicate that a hostile media environment reduces development. For instance, internet censorship, media self-censorship, and the harassment of journalists have negative ramifications for tourism development. This is very profound and has implications for the tourism industry in developing countries as they emerge from the COVID-19 pandemic. This is consistent with the view that tourists depend on the domestic media to assess tourist destinations and safety for tourists. More importantly, if tourists cannot rely on the media to independently provide information on how the country is addressing the COVID-19 pandemic, it will adversely influence tourists’ decisions to visit those countries.

The results are in line with Demir and Gozgor (2019) who find that press freedom enhances inbound tourism. The results imply that, if countries seek to develop their tourist sector, the media must be fair and unbiased. Hostile media impedes tourism development and increases income inequality. This has implications as countries recover from the adverse effects of the COVID-19 pandemic on the tourism sector. As countries seek to eradicate income inequality, evidence suggests that the COVID-19 pandemic has globally moved more than 100 million people into extreme poverty (World Bank, 2020). On the policy side, reduction of the adverse effects of the COVID-19 pandemic on tourism development, poverty, and income inequality.

The findings of the study further show that a hostile media environment is prevalent in emerging countries. This will inhibit governments’ efforts to recover the tourism industry in the post-COVID-19 pandemic. Policies will be required to reduce media biasedness, corruption of media, and harassment of journalists.

The concept of income inequality has received much attention in the literature. Whilst this study focuses on how tourism development and the media impact income inequality, there are some significant insights from this study that advance the concept of income inequality. The study improves our understanding of how the media influence the relationship between tourism development and income inequality. Future studies modeling income inequality need to incorporate the role of press freedom and a hostile media environment. The results of this study have implications for practice. A hostile media environment exacerbates income inequality. Governments should avoid media censorship and rather enhance a free media environment, as it will help build resilience in domestic tourism and lower income inequality.

Conclusion

The existing tourism literature provides contrasting theoretical arguments and empirical evidence on the effects of tourism development on income inequality. While one school of thought argues that tourism development reduces income inequality, the other offers a contrasting view that tourism development increases income inequality. Both of these theoretical views are perspectives supported by empirical evidence. Using panel data from 88 countries from 1996 to 2020, we test whether the relationship between tourism development and income inequality is sensitive to the censored and uncensored media environment of the country. We find that press freedom interacts with tourism development to reduce income inequality. Further analysis shows that an uncensored media environment enhances tourism development. We also find that a hostile media environment; including, media corruption and media biasedness reduces tourism development.

The findings, supported by extensive robustness checks, imply that in countries where there is press freedom, journalists can highlight the abuses in tourist zones for the government to act upon or for the local authorities to use taxes generated from tourism to develop those areas. The policy implications of the study are that, as countries seek to develop their tourism industry following the adverse effects of the COVID-19 pandemic, emerging countries need to improve press freedom to attract international tourists. This is consistent with the view that a hostile media environment contributes to information asymmetries, which then deter tourists from visiting countries amid the COVID-19 pandemic. Countries seeking to attract tourists and address income inequality should promote policies and reforms to ensure press freedom and reduction of media biasedness, harassment, and censorship. Independent media commissions should be established to regulate the media, ensuring they are devoid of political interference. The study opens avenues for future research to investigate how press freedom could enhance the tourism industry’s recovery post the COVID-19 pandemic.

Notes

1 For the definition and how tourism development proxies are measured, please see The World Bank, World Tourism Organization, and Yearbook of Tourism Statistics.

Appendix

Table A1. Description of variables.

Variable	Abbreviation	Source
Income inequality	IIE	Standardized World Income Database (SWIID) calculated using Gini coefficient.
Voice and accountability	VA	World Governance Indicators.
International tourism receipt	ITR	The World Bank, World Tourism Organization, Yearbook of Tourism Statistics.
International tourism receipt for passenger transport	ITRPT	The World Bank, World Tourism Organization, Yearbook of Tourism Statistics.
International tourism numbers of departures	ITND	The World Bank, World Tourism Organization, Yearbook of Tourism Statistics.
Internet censorship	Mecenefi	Varieties of Democracy (V DEM) provided by the University of Gothenburg, V-Dem Institute.
Media self-censorship	Meslfcen	Varieties of Democracy (V DEM) provided by the University of Gothenburg, V-Dem Institute.
Harassment of media and journalists	Meharjrn	Varieties of Democracy (V DEM) provided by the University of Gothenburg, V-Dem Institute.
Print and broadcast media perspective	Merange	Varieties of Democracy (V DEM) provided by the University of Gothenburg, V-Dem Institute.
Media corruption	Mecorrpt	Varieties of Democracy (V DEM) provided by the University of Gothenburg, V-Dem Institute.
Media biasedness	Mebias	Varieties of Democracy (V DEM) provided by the University of Gothenburg, V-Dem Institute.
GDP growth	GDPG	World Development Indicators (WDI).
Inflation	Infl	World Development Indicators (WDI).
Rule of law	RL	World Governance Indicators.
Control of corruption	CC	World Governance Indicators.
Political stability	PS	World Governance Indicators.
Government stability	GE	World Governance Indicators.