Final Paper-Development Economics

The Effect of Tourism on Income Inequality

Rafael Wu

Advisor: Prof. Piyush Chandra

Abstract

Several types of research have shown that the tourism sector is among the fast growing

sectors in the economies of several countries across the world. A good number of

international agencies have considered this factor considering the outward-oriented

growth strategy and promotes the sector as a means for economic development in

countries around the world. Statistics have proven that tourism has brought a positive

impact to the growth and development of several countries. On the other hand, little

research has been conducted to analyses the distributional consequences of the

development and growth of countries as a result of tourism. In our discussion, we

concentrate on the impact of tourism on the income inequality by applying panel data and

cross-country regression. According to Page & Connell (2010), the samples of countries

that were considered in the analysis, the data found have shown that tourism has been a

contributing factor towards the decrease in the gross income inequality.

I. Introduction

1. Tourism

Kulkarni (2010) defines tourism as an activity undertaken by a person that involves

travelling to a place outside their normal environment and staying over there for less than

a year. He further stresses that the motive behind the travel is leisure. The increase in the

per-capita income and reduce working hour has been a major factor contributing to the

increase in tourism accrues the world. It evident that all social classes currently

participate equally in the tourism sector thus the increase in both the domestic and

international tourism (Walker & Harding 2007). Other factors that have positively

contributed to the growth of the tourism sector include the low transport costs and fast

and affecting ways of moving from one region to another (Page & Connell, 2010).

The contribution of the tourism sector to the economies of several countries has been

positive, and the outcome has been seen in the increase levels of development. According

to Tourism and Poverty (2010), expansion of tourism sector in the developing countries

will be a step forward toward the outward-oriented development strategy. In our

document, we analyze the impact of tourism on the economy’s income inequality of

countries and also compare the impact that the domestic and international tourism have in

the income inequality.

2. Income Inequality

Tang, Selvanathan & Selvanathan (2008) explain that income inequality has a direct

relationship with poverty. The definition of poverty can be regarded as the pronounce

denial in the provision of a sustainable life for an individual. Other factors that can be

used as a measure of poverty include poor health, low levels of education, high exposure

to risk conditions and powerlessness. However, Cole & Morgan (2010) ranks income

inequality as the better measure of the economic state of a country than poverty. He

explains that inequality captures the entire population and gives a distinct and broader

measure of the economic state, unlike poverty which only entails a smaller section of the

economy.

In our we study, however, we concentrate of the income inequality which is in

contrary to the actual study of measures inequality since several other factors have the

almost same effect to the extent of inequality of income. Moreover, applying the concept

of cross-country data, it is difficult to incorporate other elements apart from income in the

study of inequality since they are difficult to quantify and almost impossible to collect

sufficient data in them (Cole & Morgan, 2010). On the other hand, tourism has little or no

direct relationship to the elements and any relationship that may exist might be due to the

variation in income distribution.

3. Measures of Income Inequality

Several measures that have been incorporated into many types of research to measure

the extent of inequality are based mathematical concepts. Initially, the extent of income

inequality was carried out by determining the highest and the lowest income in the

market in a given population sample. The limitation that this approached had was the fact

that it gives only two observations leaving out important factors such as the population of

a given regions. Therefore, there was the need to apply complex measuring approaches so

as to perform an efficient analysis.

According to Cole & Morgan (2010) the best measure of income inequality is

composed of the following characteristics; mean independence such that any variation in

the mean income due to change in some incomes should not have any effect on the

inequality measure. Secondly, must have a population size independences and symmetry

properties. Cole & Morgan (2010) add that measure should build on Pigou-Dalton

transfer sensitivity which implies that if the income of the rich is passed to the poor, there

will be a proportional decrease in the inequality measure. The most commonly used

measure of economic inequality in the cross-country datasets includes the Thiel’s T

statistics and Gini coefficient.

To determine the Gini coefficient, a mathematical formulation is applied. About

the Lorenz curve showing the relationship between aspects of income inequality, the Gini

coefficient is taken as the degree of deviation from perfect equality after superposition is

done on the Lorenz curve by a perfect equality line. With the recent development, Thiel’s

T statistic method has been used in place of the Gini coefficient method due to it great

flexibility when measuring income inequality. Also, OECD Factbook 2008 (2008) states

that the method incorporated all the good characteristics of a measurement element stated

above and even much more. However, in our analysis of the impact of tourism on the

income inequality, there is little decomposition of the inequality variable making the use

of the Gin coefficient more efficient. Also, several countries have this coefficient, and the

time-period makes more reliable for these study.

II. Literature Review

There are only a handful of studies regarding the impact of tourism on economic

growth, and even fewer papers on tourism and a specific economic issue: income

inequality. Furthermore, there have not been any cross-country studies that investigate

the impact of tourism on income inequality. There are only a few publications focusing

on the domestic impact of tourism industry on income inequality.

Blake et al. (2009), developing a Computable General Equilibrium (CGE) model of

tourism, including earnings by different types of labor in the tourism industry, examines

the issue of how tourism affects poverty in the context of its effects on an economy as a

whole and on particular sectors within it. With a dataset that is unique in the context of

developing countries, the CGE model shows tourism does benefit lower-class people in

brazil in lowering income inequality. With their model on this study, they expand the

research to some developed countries like Australia and Spain, tourism still generates an

impact on reducing income inequality. Actually, industry lowering income inequality

reduces government’s role in tax. They redistribute the resources from upper class to

lower class.

Lee and Kang (1998), using the data on wages of South Korea from 1985 to 1995,

measure the degree of earning inequality in tourism employees in South Korea. With Gini

coefficient and Lorenz Curve, they present the different income distribution across all the

industries. In this study, tourism performs more equal earning distribution than most of

traditional studies. However, the authors focus on the income inequality inside every

industry, and they do not figure out the reasons of the results. Their study focuses on the

disproportional distribution to lower-class rather than implying the identity. To modify

the study, it would be better off if they run the time-series with every industry’s scale and

income inequality. Some industry’s earning distribution might be more equal with the

industry’s expanding.

III. Data

1 Experimental Variables

As the topic showed in this paper, the experimental variables we set should be income

inequality variable and tourism variable.

As the dependent variable in the whole model, we choose Gini coefficient

representing the income inequality level. There are quite a few data sources of Gini

coefficient among various research institutes, and the most widely cited dataset is the

Deininger and Squire dataset (1996) compiled by Klaus Deininger and Lyn Squire for the

World bank. However, the D&S dataset is found to be inadequate for a pure cross-

country panel analysis. Beyond that, we also have World Income Inequality Database

(WIID2), Estimated Household Income Inequality Data Set (EHII) and so on. Among all

the databases, Standardized World Income Inequality works best in this studies since it

contains 4340 Gini coefficients for 153 countries in the sample.

SWIID has two methodologies to calculate Gini coefficients, gross income inequality

and net income inequality. As the term implies, gross/net income inequality is calculated

over gross income and net income, respectively. In the following model, we focus on the

gross income across countries, so we choose gross Gini coefficient as the dependent

variable.

As for tourism variable, what we want is the role of tourism industry plays among all

the economic activities. In other words, we should calculate how much tourism

contributes to GDP. World Travel and Tourism Council (WTTC) presents tourism

industry’s direct contribution as a percentage of the GDP, then we add the dataset into the

model.

2. Control Variables

To select control variables affecting income inequality, firstly we should classify

potential factors into different groups, such as education, economics and politics.

For education sector, we use Barro-Lee (2011) dataset, measuring the effect of no

schooling percentage, primary school percentage, secondary school percentage and

average years of schooling on income inequality.

When it comes to equality issue, politics is always crucial. In this study, “policy2”

variable indicates the political form of governance with rating from Center for Systemic

Peace (2011). CSP grades the political form of governance from -10 to 10 regarding the

institution.

As for economic sector, we pick real income per capita variable from Penn World

Table (2011). According to Kuzents Hypothesis (1955), income inequality has an inverse

U-shaped relationship between income inequality and economy. Based on the income

classifications by World Bank (2016), upper middle-income countries are expected to

have higher income inequality, and low-income and high-income countries are predicted

to be more “equal” economically. Therefore, realincome square should be included in the

regression. Unfortunately, the realincome is not discovered in Penn World Table (2011),

and we adopt real GDP PPP per capita instead. Another economic variable we use is the

openness of economics for a country.

In addition to education, politics and economic issues, we also need add some

sociologic issues including labor and urbanization, which seem to be important in all the

previous studies on income inequality.

3. Summary

Table 3.1 Variable sources

Variable Source ………………………………………………

Grossgini Standardized World Income Inequality Database (2011)

TourismGDP2 World Development Indicators (2011)

Laborrate World Development Indicators (2011)

agedependency World Development Indicators (2011)

femalelabor World Development Indicators (2011)

urbanpop World Development Indicators (2011)

urbanprimacy World Development Indicators (2011)

noschooling Barro-Lee Dataset (2011)

primarysch Barro-Lee Dataset (2011)

secondarysch Barro-Lee Dataset (2011)

yearschool Barro-Lee Dataset (2011)

RealGDP PPP per Capita Penn World Table (2011)

realincome-squared Penn World Table (2011)

openk Penn World Table (2011)

polity2 Center for Systematic Peace (2011)

Table 3.2.x Summary of variables

3.3 Time-series analysis of experimental variables

Figure 3.3.1 Gini Coefficient-China

From Figure 3.3.1, the Gini coefficient of China perfectly meets Kuzents Hypothesis.

Before the economic reform happened in 1978, this country suffered from Anti-Rightest

Campaign, Great Leap Forward and Culture Revolution, and the Gini coefficient reduced

to a low level since the economic development of China was at the lowest level

throughout the world. After the reform, the Gini coefficient raised rapidly along with the

rapid development of this country as a result of marketing economy. The Gini coefficient

of China will keep rising as long as China is still a middle-income developing country.

Figure 3.3.2 Tourism’s Contribution to GDP (%) – World (From Koenma Database)

From 3.3.2, it seems that tourism’s contribution to GDP was rising rapidly before

1998. It steps down a little bit after 1998, perhaps it is due to the financial crises in 1998

and 2008. In general, tourism’s contribution to GDP will maintain at a certain level as

time goes by.

IV. Replication

Based on all the variables mentioned last part, we try to run the regression as the

original publication. A fixed-effect estimation is presented as follows:

Table 4.1 Fixed-effect Replication

Table 4.2 Fixed-effects estimation results:

Variable Grossgini

TourismGDP2 -.2664629

(.1268513)

Laborate -.2986885

(.2056305)

Polity2 .0260885

(.0938379)

Agedependency 0.264343

(0.736555)

Rgdpl2 .0004745

(.00035)

Rgdp2sq -3.82e-09

(-4.82e09)

Openk .0106121

(.0190147)

Noschooling .2679628

(.4316772)

Yearschool -1.645246

(2.350404)

Femalelabor .5769924

(.2812813)

Urbanpop .1045371

(.1748709)

Urbanprimacy .1643134

(.1660433)

Primarysch .1906528

(.2769215)

Secondarysch .1846698

(.181899)

Constant 19.46196

(40.87753)

R-squared 0.1661

Despite some coefficients seem to be a little bit far from the original paper

(Coefficient table presents in appendix B), most of the variables’ coefficients are closed

to the original results. Especially, the results for the experiment variable are well fitted.

The coefficient of tourismGDP2 is -.266, which implies the Gini coefficient will

reduce .266 with one percentile creeping-up of tourism’s contribution to GDP.

Furthermore, the p-value of its coefficient is less than 0.05, which indicates statistical

significance of tourism variable.

The measurement of “polity2” seems to be against expectation. In general, countries

with better political form are expected to have better social equality. The coefficient of

“polity2” is positive, which implies better politic-organized country has poorer income

inequality.

Similar with the results of original paper, the “rgdpl2” variable, as well as its square,

does not show significant effects in the regression. With the polynomial function of

income level, it would be hard to generate an inverse U-shape in the regression. We will

try to find a new variable instead of it next part.

For the education variables, all of them are statistically insignificant as the original

paper. These four variables seem to have multicolliearity.

The value of rho for this fixed-effect estimation is 0.894, which means that greater

than 89% of the variation in income inequality is due to the difference across countries in

the sample. Table 4.2 above summarizes the results without presenting the equation

mathematically.

V. Twist

Since the regression we replicated last part is concerned with Fixed-effects

estimation, we will try to figure out the feasibility using Random-effects estimation:

Table 5.1 Random-effect Estimation

Since the p-value is less than 0.05 in Hausman Test, we reject the null hypothesis.

Fixed-effect estimations should be the better methodology as the author used in the

original publication.

The result of Hausman Test implies the correlation between variables and

interception term and correlation between variables themselves. Since we use more than

one variable in education, economic and political sectors in the model, there is a potential

possibility of multicollinearity.:

Table 5.2 Collin Test (multicollinearity)

The Collin Test shows there is an extremely serious situation of multicollinearity

among the education variables. Obviously, year of schooling, and no schooling are

almost perfectly negative correlated based on our common sense. Moreover, primary

school and secondary school participation rate also seem to be closely related. We just

keep one of the variables, years of schooling, to indicate the education sector.

Another multicollinearity is under our expectation, rgdpl2 and rgdpl2sq are

absolutely correlated. However, when we replicate the data, since we cannot find “real

income” as the author did in the publication, this economic variable does not play a

significant role in the regression. The polynomial expression of real GDP cannot present

the Kuzents’s inverse U-shape perfectly. We consider adding a new economic variable,

inflation, to the regression from the World Bank database.

According to Kuzents Hypothesis, rich countries and poor countries are expected to

have low income inequality. Rich countries and extremely countries have similar features

with low inflation. Then, we add inflation as a new variable instead of rgdpl2, and run the

fixed-effect regression again:

5.3 New Fixed-effect Estimation

As expected, inflation is positively related to Gini coefficient with statistical

significance. Beyond that, the dropping of education variables and the adding of inflation

increases the variables which are statistically significant in the regression. TourismGDP2,

inflation, years of schooling and female labor are significant to gross Gini coefficient.

Since we have quite a few variables in this regression, the change of variables does not

really change the coefficients numerically.

Under ceteris puribus assumption, we could interpret the coefficients of all variables

as below:

1. One percentile increase in tourism’s contribution to GDP leads to .264 percentile

reduce in Gini coefficient;

2. One percentile increase in labor rate leads to .346 percentile reduce in Gini

coefficient;

3. One-point increase political form of governance rating leads to .036 percentile

increase in Gini coefficient, which is against the expectation;

4. One percentile increase in age dependency leads to .034 percentile increase in Gini

coefficient;

5. One percentile increase in inflation leads to .015 percentile increase in Gini

coefficient;

6. One-point increase in economic openness index leads to .024 percentile increase in

Gini coefficient;

7. One more year increase of schooling leads to 1.839 percentile reduce in Gini

coefficient;

8. One percentile increase in female labor rate leads to .683 percentile increase in

Gini coefficient;

9. One percentile increase in urban population percentage leads to .083 percentile

increase in Gini coefficient;

10. One percentile increase in urban primacy leads to .239 percentile increase in Gini

coefficient.

In the process of modification, the coefficient on the tourismGDP2 variable is

negative in all cases and is statistically significant in all the cases. Hence, the fixed-effect

estimation conclusively shows the direct contribution of the tourism sector to income

inequality.

VI. Conclusion

The results in this study indicate the negative relationship between the contribution

of the tourism sector to GDP and income inequality. To conclude, tourism not only

contributes to the economic development, but makes sense to the equality of human

beings.

Based on microeconomic theory, tourism increases the marginal revenue to some

sub-industries, like restaurant, hotel, transportation and service. Along with the marginal

revenue and marginal profit, tourism benefits a large number of small business owners.

From macroeconomic side, tourism increases the job opportunities in the related sub-

industries, where lower-class people can find a job and earn money. It is how tourism

reduces the income inequality behind the studies with cross-country and panel data.

According to the “twist” model, this study also compares the effect on income

inequality between tourism and some traditional sectors like economics and politics.

Comparatively, tourism has a large impact on lowering income inequality among all the

sectors of the regression. For instance, inflation even has a positive relationship with

income inequality. When inflation occurs, rich people generally have better knowledge of

investment, as they always perform better skills to manage their wealth. In other words,

as traditional industry, finance seems to enlarge the income inequality in spite of their

contribution to economic growth.

As time goes by, more and more social research focuses on people’s life quality.

Researchers would like to evaluate something like “Happy Planet Index” with a

quantitative analysis methodology. Actually, equality is believed to affect people’s

happiness in daily lives. Similar to this empirical study, new industry could be involved

in such an econometric model with income inequality. Hopefully, we could evaluate its

impact to the equality of human beings in addition to its contribution to the economic

development.

Biobiography

Blake, A., Arbache, J. S., Sinclair, M. T., & Teles, V. (2008). Tourism and poverty

relief. Annals of Tourism Research, 35(1), 107-126.

Brohman, J. (1996). New directions in tourism for third world development.Annals of

tourism research, 23(1), 48-70.

Cameron, A. C., & Trivedi, P. K. (2010). Microeconometrics using stata (Vol. 2). College

Station, TX: Stata Press.

Cole, S., & Morgan, N. (Eds.). (2010). Tourism and inequality: Problems and prospects.

CABI.

Edwards, S. (1997). Trade policy, growth, and income distribution. The American

Economic Review, 87(2), 205-210.

Lee, Choong-Ki, and Seyoung Kang. "Measuring earnings inequality and median

earnings in the tourism industry." Tourism Management 19, no. 4 (1998): 341-348.

Lundberg, D. E., Krishnamoorthy, M., & Stavenga, M. H. (1995). Tourism economics.

John Wiley and sons.

Page, S., & Connell, J. (2006). Tourism: A modern synthesis. Cengage Learning EMEA.

Scheyvens, R. (2012). Tourism and poverty. Routledge.

Tang, S., Selvanathan, E. A., & Selvanathan, S. (2008). Foreign direct investment,

domestic investment and economic growth in China: a time series analysis. The World

Economy, 31(10), 1292-1309.

Appendix A

List of Countries:

Appendix B

Replicated Table from Original Publication:

Appendix C

Stata Code:

sort countryname

rename countryname country

drop country


gen id=_n

reshape long y, i(id) j(year)

rename y tourismGDP2

rename country countryname

save "\\Client\H$\Downloads\yuxiaoze.dta"

replace countryname="Slovak Republic" if countryname=="Slovakia"

merge 1:1 countryname year using "\\Client\H$\Downloads\yuzeyu.dta"

drop _merge

merge 1:1 countryname year using "\\Client\H$\Downloads\xiaoze.dta"

drop _merge

merge 1:1 countryname year using "\\Client\H$\Downloads\jianrujing.dta"

drop _merge

merge 1:1 countryname year using "\\Client\H$\Downloads\howlongwilliloveyou.dta"

rename y femalelabor

drop _merge

merge 1:1 countryname year using "\\Client\H$\Downloads\cristiana.dta"

drop _merge

merge 1:1 countryname year using "\\Client\H$\Downloads\ieu.dta"


merge 1:1 country year using "\\Client\H$\Downloads\ieu.dta"

replace country="Slovakia" if country=="Slovak Republic"

save "\\Client\H$\Downloads\yuxiaoze.dta", replace

drop _merge

merge 1:1 country year using "\\Client\H$\Downloads\BL(2010)_MF1599_v1.2 (1).dta"

drop if id==.

save "\\Client\H$\Downloads\teamo.dta"

drop _merge

merge 1:1 country year using "\\Client\H$\Downloads\woaini.dta"

drop if id==.

save "\\Client\H$\Downloads\teamo.dta", replace

xtset country year

merge 1:1 country year using "\\Client\H$\Downloads\yuzeyushishabi.dta"

drop _merge

merge 1:1 country year using "\\Client\H$\Downloads\yuzeyushishabi.dta"

drop if id==.


xtset gini_gross year

sort country time

sort country year

isid country year

rename femalelabor agedependency

drop _merge

merge 1:1 country year using "\\Client\H$\Downloads\yuzeyuwoaini.dta"

rename y femalelabor

rename femalelabor y

drop _merge

merge 1:1 country year using "\\Client\H$\Downloads\yuzeyuwoaini.dta"

isid country year

xtset id year

xtreg gini_gross tourismGDP2 laborrate polity2 agedependency openc lu yr_sch

femalelabor urbanpop urbanprimacy lp ls, fe robust

use "\\Client\H$\Downloads\teamo.dta", clear

drop _merge

sum tourismGDP2

sum laborrate

sum agedependency

sum femalelabor

sum urbanpop

sum urbanprimacy

rename lu noschooling

rename lp primarysch

rename ls secondarysch

rename yr_sch yearschool

sum noschooling

sum primarysch

sum secondarysch

sum yearschool

rename openc openk

sum openk

sum policy2

sum polity2

rename gini_gross Grossgini

sum Grossgini

xtset id year

xtline Grossgini

twoway (tsline Grossgini)

gen rgdpttsq= rgdptt^2

xtreg Grossgini tourismGDP2 laborrate polity2 agedependency rgdptt rgdpttsq openk

noschooling yearschool femalelabor urbanpop urbanprimacy primarysch secondarysch, fe

robust

xtreg Grossgini tourismGDP2 laborrate polity2 agedependency rgdptt rgdpttsq openk

femalelabor urbanpop urbanprimacy , fe robust

gen rgdpl2sq= rgdpl2^2

xtreg Grossgini tourismGDP2 laborrate polity2 agedependency rgdpl2 rgdpl2sq openk


robust

estimates store fixed


noschooling yearschool femalelabor urbanpop urbanprimacy primarysch secondarysch, re

robust



robust



estimates store fixed


noschooling yearschool femalelabor urbanpop urbanprimacy primarysch secondarysch, re

estimates store random

hausman fixed random



robust

help collin

findit collin

collin id tourismGDP2 laborrate polity2 agedependency rgdpl2 rgdpl2sq openk yearschool

femalelabor urbanpop urbanprimacy primarysch


merge 1:1 country year using "\\Client\H$\Desktop\ainixiaoze.dta"

xtreg Grossgini tourismGDP2 laborrate polity2 agedependency inflation openk yearschool

femalelabor urbanpop urbanprimacy primarysch, fe robust

xtreg Grossgini tourismGDP2 laborrate polity2 agedependency inflation openk yearschool

femalelabor urbanpop urbanprimacy, fe robust

Final Paper-Development Economics

Documents