Gender Wage Gap in Hospitality

Abstract

This article analyzes the gender wage gap in the hospitality sector. First, it explores whether the gender wage gap is partly explained by the economic sector. Second, it measures how this gap changes across the wage distribution using quantile regression. Third, it decomposes the gender wage gap in the hospitality sector to distinguish which part can be explained by observed attributes and which part is explained by other factors (unobserved characteristics or gender discrimination). Methodologically, this article introduces the use of quantile regression to the analysis of the gender wage gap and its decomposition in the hospitality sector. The main findings are as follows. First, on average in the hospitality sector, wages (without taking into account worker skills) are below the overall average wages. However, if a deeper look is taken, this research reveals that unskilled workers are better paid in hospitality than in most of the other sectors. The opposite is true for skilled workers, since mid- and high-wage workers in the hospitality sector receive wages below their counterparts in other sectors. Second, the gender wage gap is particularly low in the hospitality sector and the gap changes across the wage distribution. Third, a large part of the gender wage gap in hospitality is not explained by worker or company characteristics. The segregation of women into worse-paid jobs and gender discrimination (or unobserved characteristics) seem to be the main sources of the gender wage gap.

Keywords

gender wage gap unconditional and conditional quantile regression decomposition hospitality

Introduction

“No matter which western country we focus on or how earnings are measured, whether it be hourly, monthly or annually, it is a well-established fact that men tend to receive higher earnings than women on average.”

—Thrane (2008, p. 515)

Even though European countries and the United States have labor legislation to combat gender differences, the gender wage gap is still a reality. Several theories attempt to account for the existence of a difference in wages between men and women from an economic point of view. Fleming (2015) and Sparrowe and Iverson (1999) reviewed the main theories. The human capital theory attributes the difference to differing skills, based on the workers’ experience and education (Becker, 1975, 1985; Mincer, 1958). The new home economics approach (Becker, 1981) applies the comparative advantage theory to the market and household sectors, arguing that a member with a comparative advantage in one of the sectors will specialize in it. These first two theories base the difference in wages on different worker characteristics.

However, empirical studies also reveal a gender wage gap that cannot be explained by different worker characteristics and might be a consequence of gender discrimination. The structural theory of occupational crowding (Groshen, 1991; Sorensen, 1989) is based on occupational segregation. Due to employer discrimination, women tend to be placed in occupations with lower wages and so they are crowded out of better paid jobs. This theory associates productivity with the employment position and not just with worker characteristics. Similarly, the devaluation of female work theory (England, 1992) also upholds the idea that women are crowded out to low-wage jobs, but in this case, due to the devaluation of female employment. According to the social closure theory (Tomaskovic-Devey, 1993), women are crowded out to sectors or positions with lower wages because male employees use their power to preserve their privileges and to keep the best jobs.

Figueroa-Domecq et al. (2015), Morgan and Pritchard (2019), and Rivera (2018) evidenced the scare literature on tourism gender research. This article pretends to contribute to the literature analysing the wage disparities between men and women in the hospitality sector. This study focuses on Spain due to the importance of tourism in this country. Spain ranks second in the number of international tourist arrivals (World Tourism Organization, 2018). According to the Active Population Survey 2017, drawn up by the Spanish National Institute of Statistics (the INE according to its Spanish acronym), 13.34% of the labor force work in the hospitality sector. Although some features of the labor market are due to country-specific institutional factors, we believe that the analysis is relevant to other countries too.

Workers in the tourism and hospitality industries are characterized by their low level of qualifications, receiving below-average wages. In Spain, excluding staff working for the public administration, 26.8% of all workers hold a university degree, while in the hospitality sector only 5.3% of the workforce are graduates. The average wage in Spain is 11.4 euros an hour, while in the hospitality sector the average wage is substantially lower: 8.5 euros an hour.¹ In addition, it is a sector with a greater share of women (56.6% in the hospitality sector) than the overall average (48%). All these facts make the hospitality sector a very attractive choice for this study.

The aim of this article is to shed light on three relevant issues that has not been addressed with latest methodological advances. First, it analyzes whether the hospitality sector pays lower wages than other sectors or not across the wage distribution. Some Spanish studies found that there is a wage penalty in the hospitality sector, both on average (Lillo-Bañuls & Casado-Díaz, 2015) and across the distribution (Casado-Díaz & Simón, 2016). This result also applies to other economies, like Portugal (Santos & Varejao, 2007) or South Korea (Lee & Kang, 1998). Interestingly, Webster (2014) also found lower wages in the hospitality sector, although there was a wage premium in the case of the worst paid jobs. The article explores whether the hospitality sector pays lower or higher wages than the other sectors (conditional on the observed characteristics of the worker, company, and position). It also investigates whether the wage premium/penalty changes across the wage distribution. Regarding this point, this article contributes to the existing literature in two ways. First, as in Casado-Díaz and Simón (2016), it estimates the hospitality premium/penalty across the wage distribution, but in contrast to aforementioned authors, this work compares hospitality between each other sector. The rest of the article compare hospitality with the rest of the economy (all together, without analysing each of the sectors) and/or they only examine what happens on the average, not across the wage distribution.

Second, this research explores whether the gender wage gap is higher in the hospitality sector than in others across the wage distribution. There is little literature on the subject and a lack of a consensus. Muñoz-Bullón (2009) found that there is a bigger mean gender wage gap in Spain, while Santos and Varejao (2007) came to just the opposite conclusion in Portugal. To our knowledge, previous studies (Campos-Soria et al., 2009, 2015; Fleming, 2015; Kortt et al., 2018) measured gender wage gap in hospitality at the mean, but this is the first attempt to also analyze it across the wage distribution in hospitality, comparing its relative position with the other sectors of the economy and using conditional quantile regression (CQR) and unconditional quantile regression (UQR).

Finally, this piece of work uses the recent advances in decomposition techniques (Firpo et al., 2018; Fortin et al., 2011) to determine the causes of the gender wage gap in hospitality. This technique, that will be referred as FFL (Firpo, Fortin, and Lemieux and decomposition) hereinafter, distinguishes the part corresponding to differences in worker, job, and company characteristics (the proportion of the gender wage gap attributable to differences in productivity) and the part corresponding to differences in the coefficients. There is no consensus on the causes of the gender wage gap in previous literature using other decomposition techniques. Some studies mainly attribute the wage gap to differences in observed characteristics (as in Muñoz-Bullón, 2009; or Santos & Varejao, 2007), while others state that the wage gap is primarily due to gender discrimination or unobserved characteristics (such as Campos-Soria et al., 2009, 2015; or García-Pozo et al., 2012). Previous studies only decompose the gender wage gap at the mean. In contrast, this work explores the entire wage distribution. That way, it can be observed different determinants of the gender wage gap for low-wage workers and high-wage workers.

Methodologically, the article uses quantile regressions to measure gender wage gap and to decompose it. Recently, Assaf and Tsionas (2018) admitted that this is a technique with few applications in tourism and aims “to encourage more use of Quantile Regressions (QRs) in hospitality and tourism research” (p. 140). These authors focus on Bayesian estimation of QRs and study social responsibility and firm value. QRs permit to explore whether the wage gap changes and provides a more accurate image of the results obtained when the interest is across the wage distribution.

In the present study, both a CQR and UQR are performed. Hence, the differences between these two methods are highlighted, something that is new to tourism literature. UQR were introduced by Firpo et al. (2009) and can be used to evaluate the marginal effect across the wage distribution, similar to what it is done by using an ordinary least squares (OLS) regression at the mean. However, using a CQR, just allowed to analyze within-group disparities. The following example emphasizes the differences. Suppose we focus on the 90th percentile and analyze the gender wage gap, the CQR examines what happens to those male and female workers who receive high wages conditional on the other covariates. That is, the 90th percentile could include workers with low wages who perform better than their counterparts (other workers with the same observed characteristics). In contrast, if a UQR is performed (also focusing on the 90th percentile), it will explore wage disparities between males and females who earn high wages independently from the other covariates. As a result, even though both methods are valid, great care must be taken in interpreting the results. Then, to explore the effect of gender on wages across the distribution, UQR should be used as a tool.

Second, the use of FFL decomposition, which is based on the UQR, to know which part of the gender wage gap is attributable to differences in observable characteristic between men and women and which one to gender discrimination. Moreover, this technique allowed us to analyze these effects on workers with low salaries and high salaries (across the wage distribution).

As Fleming (2015) and Sparrowe and Iverson (1999), this research also relates the explanatory variables introduced in the Mincer-type wage equation with the aforementioned economic theories that justify the existence of a gender wage gap connecting theory with data. But this research extends their analysis in several ways. First, more explanatory variables are included that can explain the gender wage gap. Second, it explores the effects across the wage distribution. Finally, the gender wage gap is decomposed to deeper in its causes and present further evidence to the connection of each theory.

The results show that, on average, the mean wages earned in the hospitality sector are below the overall average when worker characteristics are not taken into account. However, if observed characteristics are controlled for, the less productive workers (those in the lower part of the wage distribution) are better paid in hospitality than in the other sectors (ranking in 1st position out of the 18 sectors at the 20th percentile). The opposite is true for skilled workers, since mid- and high-wage workers in the hospitality sector are paid wages below their counterparts (ranking 15th out of 18 at the 80th percentile). Second, the gender wage gap is particularly low in the hospitality sector and the gap changes across the wage distribution. Third, a large part of the gender wage gap in hospitality is not explained by worker or company characteristics. Hence, the human capital theory and new home economic do not explain the gender wage gap in hospitality. However, the segregation of women into worse-paid jobs and gender discrimination (or unobserved characteristics) seem to be the main sources of the gender wage gap. Therefore, the occupational crowding, the devaluation of female work, and the social closure theory seems to explain these wage differences in the hospitality sector.

Theoretical Background

Following Fleming (2015) and Sparrowe and Iverson (1999), economic theories, which strive to explain the gender wage gap, can be summarized into the following four main theories.²

First, some authors focus on the worker characteristics that account for different productivities. The human capital theory (Becker, 1975; Mincer, 1958) suggests that the higher the human capital, the higher a worker’s productivity and, consequently, the more wages he or she receives. That is, according to the human capital theory, the gender wage gap can be explained by men and women’s differing level of endowments. Hence it follows that if women are less educated or have less experience, they will be paid lower wages than men. The new home economics approach, pioneered by Mincer and Becker, consists of economic theories on household decisions. Becker (1981) applies the comparative advantage theory to households and differentiates between household human capital and market human capital. Thus, according to Theorem 2.1,

“If all members of an efficient household have different comparative advantages, no more than one member would allocate time to both the market and household sectors. Everyone with a greater comparative advantage in the market than this member’s would specialize completely in the market, and everyone with a greater comparative advantage in the household would specialize completely there.”

Based on this theorem, Becker postulates theorem 2.2, according to which the household member with a comparative advantage in the market sector would invest in market capital, while the member with a comparative advantage in the household sector would invest in household capital. Based on this theory, gender wage gaps could be partly explained by the fact that “women invest less than men in market human capital, while the productivity of household time is presumably greater for women, partly because they invest more than men in household capital” (Becker, 1981, p. 42). As Sparrowe and Iverson (1999) suggest, part-time employment should be an indicator of this theory. Women choose part-time jobs to juggle their household tasks with earning a living.

Following these theories, others appeared that associate productivity with the employment position (and not just with worker characteristics). Occupational crowding (Groshen, 1991; Sorensen, 1989), also referred to as the status composition theory, is based on the crowding hypothesis (Bergmann, 1974). Crowding is a phenomenon that occurs when women are placed in lower paid typically female jobs and men are placed in higher paid typically male jobs. This theory suggests that the more women there are in a job, the lower the wages. Therefore, typically female jobs will have lower wages than typically male ones. Similarly, according to the devaluation of female work theory (England, 1992; Sorensen, 1994; Steinberg, 2001), jobs with a higher ratio of women will be less well paid simply because employers place less value on “female occupations.”

Finally, the social closure theory (Tomaskovic-Devey, 1993) states that male employees use their power to preserve their privileges and to keep the best jobs, crowding out women to sectors or jobs with lower wages. With this theory, it is the employees who push women out of better paid jobs.

Consequently, the gender wage gap can be caused by different productivities (associated with either workers or jobs) or by pure gender discrimination. One of the goals of this article is to decompose the wage gap to see which theories fit in better with the data. The hospitality sector is an interesting choice because, typically, it has a higher ratio of women than other sectors, as shown in Figure 1.

Figure 1

Relationship Between Gender Wage Gap and Share of Female Workers in Each Sector

Hypothesis and Previous Empirical Results

This article’s contribution resides in its analysis of the gender wage gap among sectors, focusing specially on the hospitality industry and exploring the reasons that explain the gap (worker endowments, the company characteristics, job, discrimination, etc.). First, given the observed characteristics, the study compares the estimated wages and gender wage gap in the hospitality industry with other sectors. Second, it focuses on the gender wage gap in the hospitality sector in order to disentangle which part of the wage gap is due to observed characteristics and which part is accounted for by gender discrimination (or unobserved characteristics).

More specifically, the main objectives can be summarized by the following three hypotheses.

Hypothesis 1 (H1₀): Wages in the hospitality sector are not statistically different from wages in other sectors, once the observed characteristics of the worker, job, and the company are taken into account.

Hypothesis 1 (H1_a): Wages in the hospitality sector are different from wages in other sectors.

Wages in the hospitality sector are lower than average, as it was pointed out. This is what is called the “raw” wage gap, without considering observed characteristics. Perhaps this difference can be explained by the fact that workers in the hospitality sector are less productive due to their characteristics or other factors. Hence, this research aims to check the sector’s wages across the entire distribution and to confirm whether there still is a gap between hospitality workers and those of other sectors once observed characteristics are controlled for. Previous studies tend to find a lower mean wage for jobs in the hospitality sector. Lillo-Bañuls and Casado-Díaz (2015, p. 6) pointed out that “The wages earned by the workers in the Spanish tourism sector are, on average, lower than those of the economy as a whole.” Several authors have analyzed this wage discount in hospitality. In Spain, Casado-Díaz and Simón (2016) demonstrated that wages in the hospitality sector are lower than in the rest of the economy and these differences increase as one moves across the wage distribution. Campos-Soria et al. (2011) and García-Pozo et al. (2012) also found mean wage penalties for workers in this sector. In Portugal, Santos and Varejao (2007) showed that mean wages in the hospitality sector are lower than in other nontourism sectors. Silva and Guimarães (2017) evidenced a penalty of 9.3% for workers in the tourism sector compared with nontourism sector in Brazil. In South Korea, Lee and Kang (1998) stated that sectors like the hospitality industry, with a huge proportion of female workers, tend to be characterized by lower mean wages. Although in the United States, Webster (2014) concluded that, in general, wages in the hospitality sector are lower; however, in the worst paid hospitality jobs, there is a wage premium, while in the best paid ones there is a wage discount. Additionally, Brandt (2018) achieved similar results in Sweden, where he found lower mean wages for workers in the tourism sector compared with the rest of the economy using panel data.

Testing Hypothesis 1 contributes to the empirical literature by measuring the relative wage position that hospitality has among each one of the sectors of the economy and whether its position changes across the wage distribution.

Hypothesis 2 (H2₀): The gender wage gap in the hospitality sector is statistically equal to that of other sectors.

Hypothesis 2 (H2_a): The gender wage gap is not equal in the hospitality sector than in other sectors.

It is interesting to check whether economies that specialize strongly in the hospitality sector help reduce gender wage inequalities. Few studies can be found that analyze this issue and they obtain contradictory results. In Spain, Muñoz-Bullón (2009) using Tobit and Oaxaca–Blinder decomposition found that the mean gender wage gap in the hospitality sector stands at around 6.7%, while in the other sectors it is slightly lower (4.8%). However, in Portugal, Santos and Varejao (2007) showed that the mean gender wage gap is lower in the hospitality sector than in the other ones using Probit and Oaxaca–Blinder decomposition. In South Korea, Lee and Kang (1998) using the Gini index and the Lorenz curve also found there to be a lower median gender wage gap (the second lowest wage gap between men and women).

Hypothesis 2 contrast the relative gender wage gap position in hospitality among each one of the sector and adds knowledge on the issue at different parts of the wage distribution (low, mid, and high wages) as it is the first study that analyzes this issue with CQR and UQR.

Hypothesis 3 (H3₀): The gender wage gap in the hospitality sector is a consequence of different productivities. Then, employers pay lower salaries for women because their qualities or characteristics make them less productivities.

Hypothesis 3 (H3_a): The gender wage gap is caused by gender discrimination (or unobserved characteristics).

In Spain, Campos-Soria et al. (2009, 2015) and García-Pozo et al. (2012) stated that wage differences between male and female workers in the Andalusian hospitality industry and in different Spanish regions are mainly due to gender discrimination. However, Muñoz-Bullón (2009) found that, in Spain, these differences are mostly explained by differences in observable characteristics. Santos and Varejao (2007) achieved similar results for Portugal. Mention must also be made of the Brazilian-based study by Guimarães and Silva (2016) whose authors showed that men and women’s differing characteristics reduce the gender wage gap, although wage discrimination increases it. All these authors use the Oaxaca–Blinder decomposition that just explained the differences at the mean. In the case of the United States, Fleming (2015) and Sparrowe and Iverson (1999) showed that a huge part of the gender wage gap is not explained by the variables that economic theories attribute to the possible causes of these wage differences and that it is thus due to wage discrimination in line with the social closure theory. These authors just use an OLS regression and do not perform any decomposition.

This is the first article that investigate the causes of gender wage gap in hospitality (Hypothesis 3) through FFL techniques, which allow to explore whether the effect is different for low and high wages.

Data

This section outlines the data used in the study and the descriptive statistics. The data were taken from the Spanish WSS, using the last available wave: 2014. The WSS is a survey on the structure and distribution of wages, conducted every 4 years by the Spanish National Institute of Statistics (the INE according to its Spanish acronym). It is based on a harmonized methodology and contents for all European Union countries.³

The WSS does not include information on self-employed workers or those not registered with the Social Security (i.e., most civil servants are excluded). The WSS gathers information on the wages earned during the month of October⁴ of the reference year. The wage information is individually collected in a questionnaire, together with a large number of variables relating to the worker, their job and the company where they work. All the estimations and the descriptive analysis were conducted taking into account the factor variable that facilitates the shift from sample to population results.

The sample for all the sectors is made up of 208,853 employees, representing a population of 10,925,777. 7,058 members of the sample work in hospitality, which represents 889,400 of the population. For the empirical analysis, some observations are not included in the sample. First, those workers with hourly wage rates below the legal minimum wage⁵ (490 out of 208,853) are dropped. Second, for the purposes of the study, anyone with an hourly wage of over 400 euros⁶ (8 observations) or without a wage rate (44 observations) is considered. Third, 41 workers in military occupations are excluded.⁷

The descriptive statistics for the variables in the study are provided in terms of means and standard deviations for the continuous variables and frequencies and percentages for the categorical variables, shown in Table A1 in the appendix.⁸ Columns 1 to 3 (all workers, women and men, respectively) show the descriptive statistics for all the sectors and columns 4 to 6 for the hospitality sector. The main conclusions are the following. First, the “raw” mean hourly wage for women is lower and statistically different⁹ from that of men (10.51 vs. 12.22), and the same is true in the hospitality sector (8.2 vs. 8.96). However, the “raw” gender wage gap is smaller in this last sector. Second, the ratio of women working in hospitality is higher (56.68%), while in the economy as a whole the opposite is true (47.96%). Third, low relative levels of education prevail in the hospitality sector (92.62% of the workers do not have a higher education). However, no educational differences can be noted between women and men. Fourth, females are less likely to hold a position of responsibility. This is in line with glass ceiling theories. In hospitality, the ratio of women in a position of responsibility is below the average figure for the economy as a whole. Fifth, the average tenure is shorter in the hospitality sector (5.54 years vs. 9.63 years for all the sectors) and even shorter for women in hospitality. These figures suggest that worker rotation is greater than in the other sectors, as pointed out by García et al. (2012) and Marchante et al. (2005). Finally, women in hospitality mostly work in part-time jobs (58.7%). The last two conclusions indicate higher instability for women, especially in the hospitality industry.

Methodology

One of the goals is to analyze the gender wage gap across the wage distribution. OLS offers an insight into what happens at the mean, but not in the tails. It is also important to analyze the effect at different quantiles to consider the fact that the wage gap can be affected by the legal minimum wage or by the glass ceiling.¹⁰

Traditionally, the effects across the distribution have been analyzed using CQR (Koenker & Bassett, 1978). However, there is a newer technique which allows for the computation of the UQR (Firpo et al., 2009). Both techniques are valid, but the coefficients and their interpretations are different. One of the objectives of this article is to compare the results of both. If we are interested in the marginal effects, the advantage of the UQR is that the interpretation of the coefficients is similar to OLS, while CQR has a less intuitive interpretation. If we are interested in the overall effect of gender on wage dispersion, the UQR should be used to obtain the effects of gender at different quantiles of the unconditional wage distribution.

In the following section, the equations and variables used to test our hypothesis are explained and, briefly, the CQR and UQR methodologies are summarized. Finally, the decomposition methodology, which is based on the Oaxaca decomposition procedure, is reviewed.

Model Specification and Variables

To measure wages and the gender wage gap in different activity sectors, the study starts by estimating a Mincer-type wage equation, where wages are regressed on a list of variables that are potential determinants of the workers’ wages.

w_{i} = β_{0} + s e c t o r_{i} β_{1} + (f e m a l e_{i} * s e c t o r_{i}) β_{2} + X_{i}^{c} β_{3} + ε_{i}

(1)

where w_i is the natural logarithm of the gross hourly wage and i is the worker identifier; $β_{0}$ is a constant term; $s e c t o r_{i}$ is a variable for the different sectors; $(f e m a l e_{i} * s e c t o r_{i})$ is the interaction of a female dummy with the different sectors which captures the wage gap in each sector; $X_{i}^{c}$ are different variables for worker, job, and company characteristics which are also potential determinants of the workers’ wages; $β_{1}$ , $β_{2}$ , and $β_{3}$ are vectors of coefficients; and, finally, ε_i is the error term.

The dependent variable is the logarithm of the wage rate. Following the survey’s own recommendations, first of all, extra and overtime payments is added to the base wage for the month of October (the month taken as a reference in the WSS). Then, this amount is divided by the total (regular and extra) work hours that month.

The explanatory variables $s e c t o r_{i}$ and $(f e m a l e_{i} * s e c t o r_{i})$ allow to compare the mean wage and the gender wage gap in each sector, and thus to test Hypotheses 1 and 2. To control for other variables that influence wages, other covariates, $X_{i}^{c}$ , are included in the regression:

The effects of the human capital theory should be captured by the variables “educational attainment” and “workforce experience.” Although the WSS does not include the workers’ experience, age, and labor seniority are included as proxies. Age is a discrete variable with five possible age brackets (younger than 29 years, 30 to 39 years, 40 to 49 years, 50 to 59 years, and older than 59 years). The level of education is divided into seven categories (less than primary, primary, secondary I, secondary II, advanced vocational training, university diploma [3 years], and bachelor’s degree [>3 years or higher]). Seniority is the number of months of the worker in the company. This is converted into years dividing by 12. A quadratic term is also included in the regressions to control for diminishing returns.

Another worker characteristic that could determine wages is the worker’s nationality. Previous literature has found a negative impact for foreigners across the distribution but not at the very top (Canal-Domínguez & Rodríguez-Gutiérrez, 2008; Casado-Díaz & Simón, 2016). Hence, a dummy variable for nationality is included that takes a value of 1 if the worker is Spanish and 0 otherwise.

The new home economies approach claims that women tend to specialize in household production. Thus, a higher ratio of women in part-time jobs is expected and this could affect salaries. The variable “full-time” is introduced to capture this effect. It takes a value of 1 if the employee is working full-time and 0 if he or she works part-time.

The structural theory of occupational crowding and the devaluation of female work argues that women are placed in lower paid jobs. Consequently, a new variable called “segregation” is created, which is constructed as the ratio of women (to the total workers) in a certain occupation_j and sector_k in relation to the total ratio of women in the whole sample. Those combinations of sectors and occupations with a share of women higher (lower) than the average in the whole sample takes values greater (lower) than one.

Other employment characteristics incorporated in the wage regression are as follows: the occupation, type of contract, and whether the worker has a position of responsibility. In the WSS, occupations are grouped into 16 categories in accordance with the National Classification of Occupations (CNO11)¹¹. The type of contract is included as a dummy variable, taking a value of 1 for open-ended contracts and 0 for temporary contracts. Likewise, the position of responsibility variable takes a value of 1 if the worker carries out supervisory duties and 0, otherwise.

The estimation also incorporates variables for those company characteristics that we think might affect wages: the sector, company size, type of collective bargaining agreement, type of market, company ownership, and region. In the WSS, there are 27 economic activities in accordance with CNAE-2009 codes. Some of these categories are very similar to others or else they contain few observations. Hence, we decided to aggregate them into 18 categories for a better insight into the results.¹² The companies are classified into three intervals depending on their size: 1-49, 50-199, and >200 workers. Collective bargaining plays an important role in explaining unemployment and wages. The type of collective bargaining agreement (state, sector, company, or work centre) is incorporated. Gayà and Groizard (2015) discovered a wage premium for exporter companies, and to control for that, it is included a variable called “type of market,” which distinguishes where the company operates (at a local, national, or international level). A dummy variable for company ownership is included that takes a value of 1 for private ownership and 0 for public ownership. The region where the company operates is based on the NUTS 1 classification,¹³ which divides Spain into seven regions.

As pointed out by Fleming (2015): “ . . . finding that an unexplained wage gap persists after accounting for human capital differences, new home economics, and structural discrimination suggests that social closure practices may be operating.”

Conditional Quantile Regression

A CQR (see Koenker & Bassett, 1978, for a more detailed description) performs different regressions at different points of the distribution of the logarithm of wages. The study borrows the notations used by Koenker and Bassett (1978), adapted to the aforementioned Mincer-type wage Equation (1):

w_{i} = δ_{0 θ} + s e c t o r_{i} δ_{1 θ} + (f e m a l e_{i} * s e c t o r_{i}) δ_{2 θ} + X_{i}^{c} δ_{3 θ} + μ_{θ i}

(2)

where θ is the quantile; $δ_{0 θ}$ is the constant term; $δ_{1 θ}$ , $δ_{2 θ}$ , and $δ_{3 θ}$ are vectors of coefficients and $μ_{θ i}$ is the error term.

For notational simplicity, all the explanatory variables are grouped into matrix Χ.

\begin{array}{l} X = {s e c t o r_{i}, f e m a l e_{i} * s e c t o r_{i}, X_{i}^{c}} \\ t h e n Q_{θ} (w_{i} | X = x) = δ_{0 θ} + s e c t o r_{i} δ_{1 θ} + (f e m a l e_{i} * s e c t o r_{i}) δ_{2 θ} + X_{i}^{c} δ_{3 θ} \end{array}

$Q_{θ} (w_{i} | X = x)$ represents the conditional quantile θ of w_i conditional on explanatory variables Χ; for example, when θ = 0.2, $Q_{θ} (w_{i} | X = x)$ describes the second decile of the wage distribution, given Χ.

The coefficients of the quantiles, $δ_{θ}$ , are interpreted as the effect on a conditional quantile of a small increase of explanatory variable z of conditional quantile w_i in relation to one of the explanatory variables, $δ_{z} = \frac{d Q_{θ} (w_{i} | X)}{d X_{i z}}$ . This derivative indicates the marginal change in conditional quantile θ brought about by a marginal change in Ζ. If Χ is made up of Κ different variables, then this derivative is simply the coefficient of variable z. The coefficient on a covariate represents the marginal effect (for continuous covariates) or partial effect (for binary covariates) of the wage distribution on the conditional quantile, conditional on the values of the other covariates.

Unconditional Quantile Regression

The UQR was introduced by Firpo et al. (2009). It consists of running a regression of a transformation of the wage on the explanatory variables, $X_{i}$ . More specifically, in the Mincer-type wage Equation (2), the wage is replaced with a transformation of it: the recentered influence function (RIF). A RIF-regression is similar to a standard regression, except that the logarithm of the wage rate is replaced with the RIF of the statistic of interest; in this research, the quantiles.

Borrowing the notations used by Firpo et al. (2009) and adapting them to the Mincer-type wage equation, the θth quantile of the (marginal) unconditional distribution of log wages, F_w, is defined as the function, $Q (F_{w}, θ) = \inf {w | F (w) \geq θ}$ , or as $Q_{θ} f o r s h o r t$ and its influence function (IF) is

I F (w; Q_{θ}, F_{w}) = \frac{θ - I {w \leq Q_{θ}}}{f_{w} (Q_{θ})}

(3)

where $Q_{θ}$ refers to the θth quantile of the unconditional distribution of w; $f_{w} (Q_{θ})$ is the probability density function of w evaluated at $Q_{θ}$ and $I {w \leq Q_{θ}}$ is an indicator function to denote whether w is less than $Q_{θ}$ or not and it is the only part that varies across observations. The influence function is simply a dichotomous variable that takes on the value $- (1 - θ) / f_{w} (Q_{θ})$ when w is below the quantile $Q_{θ}$ and $θ / f_{w} (Q_{θ})$ when w is above the quantile $Q_{θ}$ .

By definition the RIF of the $θ$ th quantile is,

\begin{array}{l} R I F (w; Q_{θ}, F_{w}) = Q_{θ} + I F (w; Q_{θ}, F_{w}) = \\ Q_{θ} + \frac{θ - l {w \leq Q_{θ}}}{f_{w} (Q_{θ})} = \frac{l {w > Q_{θ}}}{f_{w} (Q_{θ})} + Q_{θ} - \frac{1 - θ}{f_{w} (Q_{θ})} = c_{1, θ} * l {w > Q_{θ}} + c_{2, θ} \end{array}

(4)

where $c_{1, θ} = \frac{1}{f w (Q_{θ})}$ , $c_{2, θ} = Q_{θ} - c_{1, θ} * (1 - θ)$ . Thus, the conditional expectation of the $R I F (w; Q_{θ}, F_{w})$ is

E {RIF (w; Q_{θ}; F_{w}) | X = x} = c_{1, θ} * P r [w > Q_{θ} | X = x] + c_{2, θ}

(5)

The estimation of Equation 5 can be seen as the estimation of a conditional probability model, indicating the probability of being below or above a given quantile, $Q_{θ}$ , rescaled by a factor, $c_{1, θ}$ , to reflect the relative importance of the quantile to the distribution, and recentered by a constant, $c_{2, θ}$ .

Then, the conditional expectation of the $R I F (w; Q_{θ}, F_{w})$ can be modelled as a linear function of the explanatory variables in Equation 2 (i.e., the independent variable, w_i, in Equation 2 is replaced by the estimated values of the RIF in Equation 5):

\begin{array}{l} E {RIF (w; Q_{θ}; Fw) | X = x} = γ_{0 θ} + s e c t o r_{i} γ_{1 θ} + (f e m a l e_{i} * s e c t o r_{i}) γ_{2 θ} \\ + X_{i}^{c} γ_{3 θ} + ϵ_{θ i} \end{array}

(6)

This regression 6 is what Firpo et al. (2009) call the UQR and allows to compute the partial effects of covariates, $X_{i},$ on any unconditional quantile of the wage distribution.

The implementation of the UQR is similar to OLS, but UQR uses the estimated RIF for each observation w_i in the database as the independent variable, and regresses it against all the dependent variables. Summarizing, for a specific quantile, θ, first it is estimated the RIF of the θth quantile of the wage following Equations 3, 4, and 5. Then an OLS regressions of the RIF on the explanatory variables is carried out (following Equation 6). The unconditional quantile partial effect that measures the effect of an explanatory variable on the wage at the specific quantile in the linear model is equal to ${\hat{γ}}_{j}$ . These estimated coefficients are interpreted as the effect of a small increase in the explanatory variable (given the rest of $X_{i}$ ) on the unconditional¹⁴ wage distribution for a given quantile $, Q_{θ}$ .

These estimations allow to test Hypotheses 1 and 2 at the mean (OLS) and across the distribution (CQR and UQR).

The next step consists of restricting the sample to those employees that work in the hospitality industry. The study repeats the previous estimations to check whether wages are driven by the same factors. Since there is only one sector, the ratio of women in each sector (segregation variable), the sector dummies or the cross product of gender and sector cannot be incorporated. However, an indicator function that takes a value equal to 1 if the worker is a female is included. This dummy captures the extent to which wage differences between men and women remain unexplained after controlling for other observed worker, job, and company characteristics with the same returns for both genders. This research estimates the new versions of Equations 1 and 2 by OLS and by CQR and UQR, respectively, and then, the gender wage gap decomposition is performed as explained below.

Decomposition

The gender wage gap is decomposed in two parts: the part of the gender wage gap corresponding to differences in observed worker, job, and company characteristics (the composition effect) and the part corresponding to differences in the coefficients (the wage structure effect). There are different techniques for decomposing the wage gap (e.g., Blinder, 1973; Chernozhukov et al., 2013; Machado & Mata, 2005; Oaxaca, 1973), but this research use the FFL (see Firpo et al., 2018; Fortin et al., 2011). It offers some appealing features. First, it is founded on the UQR, and so the decomposition is performed across the unconditional wage distribution. Second, it is path independent and so the results of the decomposition do not depend on the order in which it has been carried out. Finally, the composition and wage structure effect is shown for every explanatory variable, which allows us to distinguish between the contribution of the worker, job, and company to the gender wage gap.

The FFL methodology entails an in-depth two-step decomposition of the quantiles in the unconditional distribution of the wage gap. First, it estimates the UQR, as explained above. Second, it performs a standard Oaxaca–Blinder decomposition for each quantile, based on the UQR estimations. The decomposition is as follows:

Δ Q_{θ} = ({\bar{X}}^{m} - {\bar{X}}^{f}) {\hat{γ}}_{Q_{θ}}^{m} + {\bar{X}}^{f} ({\hat{γ}}_{Q_{θ}}^{m} - {\hat{γ}}_{Q_{θ}}^{f})

(7)

$Δ Q_{θ}$ is the difference between the wage distributions of males and females, respectively, at quantileθ; ${\bar{X}}^{m} a n d {\bar{X}}^{f}$ are the means of the observed worker, job and company characteristics for males (m) and females (f); and ${\hat{γ}}_{Q_{θ}}^{m}$ , ${\hat{γ}}_{Q_{θ}}^{f}$ are the estimated coefficients of the UQR; that is, the RIF regression of the group of explanatory variables run on quantile θ.

The first part of Equation 7, $({\bar{X}}^{m} - {\bar{X}}^{f}) {\hat{γ}}_{Q_{θ}}^{m}$ , is called the composition effect and it shows the differences in wages due to different observed worker, job and company characteristics. The second part, ${\bar{X}}^{f} ({\hat{γ}}_{Q_{θ}}^{m} - {\hat{γ}}_{Q_{θ}}^{f}),$ is called the wage structure effect and it measures the differences between men and women in the returns on worker, job, and company characteristics; that is, the differences not explained by male and female differences in the observed worker, job, and company characteristics.

Results

This section discusses the results and test our hypotheses. First, it presents the results of the mean wage and gender wage gap for all sectors across the wage distribution. Then, it focuses on the hospitality sector and explore the factors that account for different wages. Finally, it decomposes the gender wage gap in the hospitality sector to find out whether it is caused by male and female differences in endowments or not.

All the Sectors

Figure 1 illustrates the relationship between the ratio of women working in a particular sector and the gender wage gap. The aim is twofold: first, to highlight the fact that the wage gap changes substantially across sectors and, second, to show that there is a connection between the wage gap and the ratio of females working in that particular sector. From Figure 1, it can be concluded that in sectors where most of the workers are men—what we will call “male sectors” (such as construction workers, the extractive industry, water, and energy)¹⁵—and women’s wages are 10% to 20% lower than those of males. Nevertheless, sectors with higher ratio of women tend to have a lower wage gap. The hospitality sector (with a ratio of 56.68%) or education sector (66.58%) are the ones with a lower wage gap (between 2% and 3%). Consequently, it is particularly relevant to analyze the case of the hospitality sector and to compare it with other sectors.

Our first hypothesis aims to compare wages in the hospitality and other sectors. If “raw” wages are taken, as shown in the appendix in Table A1¹⁶, the average wage for the whole economy is 11.4 euros per hour, while in the hospitality industry the average is significantly lower: 8.53 euros per hour. However, this difference could be attributed to the fact that workers in hospitality are less skilled (with less experience, lower educational studies, etc.). Consequently, Equation 1 is estimated by OLS. Figure 2 shows that once the observed characteristics are taken into account (see the previous section for a description of the variables), workers in the hospitality sector are relatively well paid on average. The top 5 sectors are energy, finance, the extractive industry, water, and hospitality, respectively. What is striking is the fact that three of them are what we describe as “male sectors,” with a very low ratio of female workers, and hospitality is a “female sector.” On the other hand, education—a sector with a high ratio of female workers, like the hospitality sector—ranks 17th out of 18.

Figure 2

Estimates of the Mean, Female and Male Log-Wages for All Sectors by OLS and CQR

On top of that, this work is interested in ascertaining whether this ranking system holds or whether it changes across the wage distribution. This is done using CQR and UQR for percentiles 20, 50, and 80, shown in Figures 2 and 3, respectively.¹⁷

Figure 3

Estimates of the Mean, Female and Male Log-Wages for All Sectors by OLS and UQR

Taking the CQR estimations, Figure 2 shows that hospitality workers tend to rank in the same positions in the distribution (between 4th and 6th place). The CQR results are difficult to interpret because the effect conditional is measured on the remaining covariates. For example, our estimates, using CQR, indicate that the effect of being a woman at the 40th percentile is greater than it is at the 60th percentile. This simply means that the within-group dispersion—where the “group” consists of workers who share the same covariate values (other than gender)—is greater at the 40th percentile, although this does not necessarily imply that being a woman would reduce the overall wage dispersion. The overall wage dispersion is measured by the UQR. In our case, moving from percentile 40 to percentile 60 reduces the estimated UQR dispersion. UQR includes both the within-group and between-group effect (adapted from Firpo et al. 2009, p. 963). Consequently, UQR shows a different picture. The results of the CQR and UQR are presented in figures and tables to highlight the fact that they can lead to different results, although our comments will focus on the UQR, because we understand that they offer a more interesting interpretation.

The results shown in Figure 3 for the UQR lead us to draw the following conclusions. On the one hand, workers with low wages, those in the 2nd decile, are better paid in hospitality than in the rest of the sectors (ranking in 1st position). On the other hand, in the case of workers earning mid and high wages, they are paid less in the hospitality sector, taking into account the fact that the observed characteristics in the hospitality sector are different from those of other sectors. At the median and 80th percentile, hospitality workers’ wages worsen in relative terms (with them ranking 13th and 15th out of 18). In other words, the hospitality sector is appealing for unskilled workers who earn low wages, but it is not attractive for skilled or highly skilled workers. In all the estimations, the results of the wage differences among sectors turned out to be statistically significant.¹⁸ Consequently, wages in the hospitality sector are different from those earned in other sectors. Hypothesis 1₀ can therefore be rejected.

Previous studies (Campos-Soria et al., 2011; Casado-Díaz & Simón, 2016; García-Pozo et al., 2012) systematically found lower salaries in hospitality when compared with the remaining sectors as a whole. Interestingly, when it is compared with the other sectors on a separate basis, on average, hospitality pays relatively high wages. Only at the upper part of the distribution, there is a wage penalty, as it is found at the mean in previous studies.

Figures 2 and 3 also show the gender wage gap. The main findings from the UQR are the following. First, women earn lower salaries than men in all the sectors and across the distribution (the 20th, 50th, and 80th percentiles). Second, although the regression controls for characteristics, hospitality is still the sector with the narrowest gap, on average, after education (woman earn 2.68% less than men). Third, the wage gap tends to diminish in the upper part of the distribution; that is, wealthier workers tend to display a smaller gender wage gap in hospitality. The gap moves from 3.38% at percentile 20 to a gap of 0.34% (the lowest gap across the sector) at the 80th percentile. In other sectors, like finance, the opposite is true: the higher the wage, the bigger the wage gap. Fourth, the coefficients for interaction between the hospitality sector and women always turned out to be statistically significant (except for the energy sector at the 20th percentile¹⁹). The interpretation of these results leads us to conclude that the gender wage gap is different across sectors, and so Hypothesis 2₀ is rejected.

Our results for the gender wage gap on average are in line with those obtained by Santos and Varejao (2007) in Portugal and Lee and Kang (1998) in South Korea, but they contradict those of Muñoz-Bullón (2009) in Spain.

The remaining coefficients display the expected signs in accordance with economic theory and previous studies, as shown in the supplemental appendix Table A2 (available online). Wages rise with worker experience (estimated by age and seniority), the level of education, and whether workers have an open-ended contract, work for a public company, have a position of responsibility or are Spaniards.²⁰ On the contrary, working in a sector with a high ratio of females has a negative impact on wages.

Hospitality

Our next step is to restrict the sample to workers in the hospitality sector for two reasons: first, to check the determinants of wages in hospitality; and second, to decompose the gap and find out which part is due to different observed characteristics and which part is explained by other factors.

Figure 4 displays the estimates of the gender wage gap for hospitality. The following results are underlined. First, women tend to earn lower salaries than men of the same characteristics across the whole distribution. Second, as we move to higher wages, the gender wage gap increases (5.5% and 14% at the 80th and 90th percentiles, respectively). This seems to provide evidence of a glass ceiling for women. Third, there are clear differences between the CQR and UQR. Both the CQR and UQR estimates show that women tend to receive lower wages. The gap is smaller in the middle part of the distribution. In the tails of the wage distribution, the gap is wider, especially with the UQR. This could seem to contradict our previous results when all the sectors are included to estimate the wage gap. Two points should be taken into account. First, the effects of the other covariates change significantly when we move from the total sample to the hospitality sector, as described in the next paragraph, and this also affects the estimated gender wage gap. Second, the wage distribution for all the sectors is completely different from the distribution for just hospitality workers. That is, the 80th percentile of the total sample does not correspond to the 80th percentile for a hospitality worker. Given that hospitality workers are less skilled and earn lower salaries on average, a worker in the 80th percentile for the hospitality sector will probably rank lower when all the sectors are included.

Figure 4

Gender Wage Gap Across the Wage Distribution: Conditional Versus Unconditional Quantile Regressions

Some of the other covariates are shown in Table A3 in the appendix.²¹ Surprisingly, age and education do not have a clear effect on the hospitality sector. This seems to go against the human capital theory. Employers in hospitality appear to take into account other factors not captured by the observed characteristics included when they set wages. The languages spoken by the employee could be one of these variables that is taken into account by employers but it is not observed in the sample. It is also striking that full-time workers tend to receive a lower wage.²² According to the new home economy theory, women tend to specialize in home production and they are more likely to take part-time jobs, as shown in Table A1 in the supplemental appendix (available online). This theory predicts lower wages if the ratio of females is higher, but this is not true of our data.²³ As expected earlier, workers tend to earn more if they are hired with an open-ended contract, they are Spaniards, it is a public company and the worker carries out supervisory duties. Roughly speaking, these findings are also true across the wage distribution.

Decomposition

Finally, the gender wage gap in hospitality is decomposed into a composition effect (the part of the gap explained by differences in the observable variables) and a wage structure effect (the differences in the coefficients). The upper part of Figure 5 shows that the “raw” gender wage gap displays an inverted U-shaped curve. It starts from 5.9% at the 10th percentile and then drops down to the median (with a gap of 2.8%), where starts to rise (especially at deciles 8 and 9, where the gap is 8.9% and 18%, respectively). The wage gap is relatively small for low and mid wages (compared with other sectors, as it is seen in Figure 3). Small differences in the lower part of the distribution can be explained by the existence of a minimum wage or collective bargaining agreement, which do not allow for large divergences in low wages.

Figure 5

Decomposition of the Gender Wage Gap in Hospitality

Economists aspire to disentangle which part of the gap is due to discrimination and which part is accounted for by worker endowments. This could be tricky. On the one hand, employers might value certain other qualities that are not included in the database (such as languages, enthusiasm, etc.) and regard them as justifying a reward. This will upwardly bias what it is measured as gender discrimination, since it cannot be separated discrimination from unobserved characteristics. On the other hand, as upheld in the occupation crowding, devaluation of female work and social closure theories, woman might be relegated to certain occupations or activities. Consequently, despite the fact that the results explain wage dispersion as a consequence of observed job and company-related characteristics, there is a part that can be considered to be attributable to gender discrimination.

The upper panel of Figure 5 shows that the main part of the wage gap does not stem from the composition effect, but from the wage structure effect. This result (given the above consideration) indicates that, even if the gap is small, it is not explained by heterogeneity in the observed characteristics. Indeed, it is decomposed the part explained by the observed characteristics (the composition effect) in the lower panel of Figure 5. A large part of the explained gap is due to employment characteristics (seniority, the occupation, having an open-ended contract, or holding a position of responsibility). The company (the size, region, whether it operates at a national or international level, whether it is privately or publicly owned, or how wages are negotiated) and worker characteristics (age, education, and nationality) have little impact on the wage gap and, in most quantiles, they contribute to a slight reduction in the differences between male and female wages. Additionally, it can be observed larger differences in the upper part of the distribution, boosting the gap to 18% at the 90th percentile. This supports the existence of a glass ceiling for women who merit recognition of their skills. As it have been shown, this is not a consequence of worker or company characteristics.

To sum up, the results seem to be in accordance with the social closure theory as they point to the existence of gender discrimination and lower wages for women not accounted for by qualities or characteristics that make them less productive. Consequently, we also reject Hypothesis 3₀.

Final Remarks

The gender wage gap is a concern in many countries and across all sectors of activity. This article focuses on the hospitality industry. A large part of the staff working in the hospitality sector are known to be low-skilled workers. This has an impact on their wages which could also affect gender wage dispersion.

It is used a Spanish database of wages reported by companies, with detailed information on wages, jobs, the company, and worker characteristics. The obtained evidence allowed us to test the hypothesis. The results show that average “raw” wages in hospitality are below the overall average. The aim of the first hypothesis was twofold. On the one hand, we wanted to check whether this is a consequence of different levels of worker skills. On the other hand, it is analyzed wage dispersion across sectors to see whether the difference held across the wage distribution. The main findings suggest that workers earning lower wages are relatively well paid in the hospitality sector, once it have been controlled for the observed characteristics. However, in the case of mid and high wages, the opposite is true. The second hypothesis analyzed the gender wage gap across the wage distribution. It is concluded that the wage gap in hospitality is particularly small when compared with the other sectors. The goal of the third hypothesis was to find out the causes of the gender wage gap. One of the main worries was how much of the wage gap can be attributed to gender discrimination. The results show that even though the gap is smaller than that of other sectors, it is not caused by differences in worker or company characteristics. The human capital theory and new home economic do not explain the gender wage gap in hospitality. It seems to be mainly caused by different job characteristics and gender discrimination, with a particularly relevant impact in the case of higher wages. Therefore, the occupational crowding, the devaluation of female work, and the social closure theory seems to explain these wage differences in the hospitality sector.

Methodologically, the novelty of this article is twofold. First, compare CQR and UQR techniques as a tool in analysing the wage gap at several points of the wage distribution. It is concluded that, in certain cases, these estimation methods lead to differing results. Indeed, UQR has a more useful interpretation in terms of the marginal effects of the variables. We strongly recommend this last method if it is suspected that the effect could be different across the distribution. Moreover, the results confirm that when only OLS estimates are presented, this deprives us of a more accurate view of the effect of gender on wages. Second, the use of FFL decomposition, which is based on the UQR, allows to know which part of the gender wage gap is attributable to differences in observable characteristic between men and women and which one to gender discrimination. Moreover, this technique analyzes these effects on workers with low and high salaries (across de wage distribution).

Analyses of the gender wage gap are fundamental in order to implement policies designed to promote equal opportunities for men and women and to reduce gender discrimination. In particular, policies should promote women to work in sectors with average higher wages (which traditionally are done by men). Moreover, since most of the wage gap is due to job characteristics, authorities should promote the participation of women in occupations where they are scarce nowadays.

Finally, some limitations should be mentioned. First, it is true that institutions play a very important role in understanding the behavior of the labor market, with repercussions on the wage structure. Nonetheless, we think that most of the conclusions can be extended to other countries with an important hospitality industry and certainly, the methodology can be extended to them. The extension of the analysis to other countries is beyond the scope of this article and it must be left for further research. Second, some important variables not included in the database could help explain the differences between men and women. For example, we do not know the household composition. Mothers might tend to sacrifice a higher wage in exchange for other advantages like a better timetable, more flexibility, and so on. Third, the database does not include unemployed workers, and there could be a self-selection bias if unemployment affects males and females differently.

Supplemental Material

Supplemental_material – Supplemental material for Gender Wage Gap in Hospitality

Supplemental material, Supplemental_material for Gender Wage Gap in Hospitality by Xisco Oliver and Maria Sard in Journal of Hospitality & Tourism Research

Footnotes

Authors’ Note:

The authors acknowledges the useful comments of William Nilsson.

ORCID iDs

Xisco Oliver

Maria Sard

Supplemental Material

Supplemental material for this article is available online.

Notes

Xisco Oliver (e-mail: xisco.oliver@uib.es) is an associate professor at the Department of Applied Economics at the University of the Balearic Islands. Maria Sard (e-mail: maria.sard@uib.es) is an associate professor at the Department of Applied Economics at the University of the Balearic Islands.

References

Assaf

A. G.

Tsionas

(2018). Changing the basics: Towards more use of quantile regressions in hospitality and tourism research. International Journal of Hospitality Management, 72, 140-144. https://doi.org/10.1016/j.ijhm.2018.01.009

Bakas

F. E.

Costa

Breda

Durao

(2018). A critical approach to the gender wage gap in tourism labor. Tourism, Culture & Communication, 18(1), 35-49. https://doi.org/10.3727/109830418X15180180585167

Becker

G. S.

(1975). Human capital. University of Chicago Press.

Becker

G. S.

(1981). A treatise on the family. Harvard University Press.

Becker

G. S.

(1985). Human capital, effort, and the sexual division of labor. Journal of Economic Literature, 3(1), 33-58. https://doi.org/10.1086/298075

Bergmann

B. R.

(1974). Occupational segregation, wages and profits when employers discriminate by race and sex. Eastern Economic Journal, 1(2), 103-10.

Blinder

A. S.

(1973). Wage discrimination: Reduced forms and structural estimates. Journal of Human Resources, 8(4), 436-455. https://doi.org/10.2307/144855

Brandt

(2018). Wage determinants in the Swedish tourism sector. Scandinavian Journal of Hospitality and Tourism, 18(1), 18-38. https://doi.org/10.1080/15022250.2016.1206832

Campos-Soria

J. A.

García-Pozo

Sánchez-Ollero

J. L.

(2015). Gender wage inequality and labour mobility in the hospitality sector. International Journal of Hospitality Management, 49, 73-82. https://doi.org/10.1016/j.ijhm.2015.05.009

10.

Campos-Soria

J. A.

García-Pozo

Sánchez-Ollero

J. L.

Benavides-Chicón

C. G.

(2011). A comparative analysis on human capital and wage structure in the Spanish hospitality sector. Journal of Service Science and Management, 4(4), 458-468. https://doi.org/10.4236/jssm.2011.44052

11.

Campos-Soria

J. A.

Ortega-Aguaza

Ropero-García

M. A.

(2009). Gender segregation and wage differences in the hospitality industry. Tourism Economics, 15(4), 847-866. https://doi.org/10.5367/000000009789955152

12.

Canal-Domínguez

J. F.

Rodríguez-Gutiérrez

(2008). Analysis of wage differences between native and immigrant workers in Spain. Spanish Economic Review, 10(2), 109-134. https://doi.org/10.1007/s10108-007-9033-3

13.

Casado-Díaz

J. M.

Simón

(2016). Wage differences in the hospitality sector. Tourism Management, 52, 96-109. https://doi.org/10.1016/j.tourman.2015.06.015

14.

Chernozhukov

Fernández-Val

Melly

(2013). Inference on counterfactual distributions. Econometrica, 81(6), 2205-2268.

15.

England

(1992). Comparable worth: Theories and evidence. Routledge.

16.

Figueroa-Domecq

Pritchard

Segovia-Pérez

Morgan

Villacé-Molinero

(2015). Tourism gender research: A critical accounting. Annals of Tourism Research, 52(1), 87-103. https://doi.org/10.1016/j.annals.2015.02.001

17.

Firpo

Fortin

N. M.

Lemieux

(2009). Unconditional quantile regressions. Econometrica, 77(3), 953-973.

18.

Firpo

Fortin

N. M.

Lemieux

(2018). Decomposing wage distributions using recentered influence functions regressions. Econometrica, 6(2), Article 28. https://doi.org/10.3390/econometrics6020028

19.

Fleming

S. S.

(2015). Déjà vu? An updated analysis of the gender wage gap in the U.S. hospitality sector. Cornell Hospitality Quarterly, 56(2), 180-190. http://scholarship.sha.cornell.edu/articles/595/

20.

Fortin

Lemieux

Firpo

(2011). Decomposition methods in economics. In Card

Ashenfelter

(Eds.), Handbook of labor economics (Vol. 4, pp. 1-102). Elsevier.

21.

García-Pozo

Campos-Soria

J. A.

Sánchez-Ollero

J. L.

Marchante-Lara

(2012). The regional wage gap in the Spanish hospitality sector based on a gender perspective. International Journal of Hospitality Management, 31(1), 266-275. https://doi.org/10.1016/j.ijhm.2011.06.007

22.

Gayà

Groizard

J. L.

(2015). Exporters versus domestic wage adjustment during the Great Recession in Spain. Economics Bulletin, 35(3), 1598-1606.

23.

Groshen

E. L.

(1991). The structure of the female/male wage differential: Is it who you are, what you do, or where you work? Journal of Human Resources, 26(3), 457-473. https://doi.org/10.2307/146021

24.

Guimarães

C. R. F.

Silva

J. R.

(2016). Pay gap by gender in the tourism industry of Brazil. Tourism Management, 52, 440-450. https://doi.org/10.1016/j.tourman.2015.07.003

25.

Koenker

Bassett

Jr. (1978). Regression quantiles. Econometrica, 46(1), 33-50. https://doi.org/10.2307/1913643

26.

Kortt

M. A.

Sinnewe

Pervan

S. J.

(2018). The gender wage gap in the tourism industry: Evidence from Australia. Tourism Analysis, 23(1), 137-149. https://doi.org/10.3727/108354217X15143857878697

27.

Lee

Kang

(1998). Measuring earnings inequality and median earnings in the tourism industry. Tourism Management, 19(4), 341-348. https://doi.org/10.3727/108354217X15143857878697

28.

Lillo-Bañuls

Casado-Díaz

J. M.

(2015). Exploring the relationship between educational mismatch, earnings and job satisfaction in the tourism industry. Current Issues in Tourism, 18(4), 361-375. https://rua.ua.es/dspace/bitstream/10045/58284/5/2015_Lillo_Casado_CurrentIssTourism_rev.pdf

29.

Machado

Mata

(2005). Counterfactual decomposition of changes in wage distributions using quantile regression. Journal of Applied Econometrics, 20(4), 445-465. https://doi.org/10.1002/jae.788

30.

Marchante

A. J.

Ortega

Pagán

(2005). Educational mismatch and wages in the hospitality sector. Tourism Economics, 11(1), 103-117. https://doi.org/10.5367/0000000053297149

31.

Mincer

(1958). Investment in human capital and personal income distribution. Journal of Political Economy, 66(4), 281-302. https://doi.org/10.1086/258055

32.

Morgan

Pritchard

(2019). Gender matters in hospitality. International Journal of Hospitality Management, 76(Pt. B), 38-44. https://doi.org/10.1016/j.ijhm.2018.06.008

33.

Muñoz-Bullón

(2009). The gap between male and female pay in the Spanish tourism industry. Tourism Management, 30(5), 638-649. https://doi.org/10.1016/j.tourman.2008.11.007

34.

Oaxaca

(1973). Male-female wage differentials in urban labour markets. International Economic Review, 14(3), 693-709. https://doi.org/10.2307/2525981

35.

Rivera

(2018). Turismo, brecha salarial y desigualdades laborales de género en espacios rurals de Andalucía (España) [Tourism, wage gap and gender labor inequalities in rural areas of Andalusia (Spain)]. Cuadernos Geográficos, 57(3), 291-312. https://doi.org/10.30827/cuadgeo.v57i3.6459

36.

Rodríguez

R. E.

Limas

(2017). Propuestas teóricas y evidencia empírica sobre desigualdad salarial por género [Theoretical proposals and empirical evidence on wage inequality by gender]. Revista de Economía Laboral, 14(2), 104-136. https://doi.org/10.21114/rel.2017.02.05

37.

Santos

L. D.

Varejao

(2007). Employment, pay and discrimination in the tourism industry. Tourism Economics, 13(2), 225-240. https://doi.org/10.5367/000000007780823186

38.

Silva

J. R.

Guimarães

C. R. F. F.

(2017). Wage differentials in Brazil: Tourism vs. other service sectors. Cogent Economics & Finance, 5, Article 1319606. https://doi.org/10.1080/23322039.2017.1319606

39.

Sorensen

(1989). Measuring pay disparity between typically female occupations and other jobs: A bivariate selectivity approach. Industrial and Labor Relations Review, 42(4), 624-639. https://doi.org/10.1177/001979398904200411

40.

Sorensen

(1994). Comparable worth: Is it a worthy policy? Princeton University Press.

41.

Sparrowe

R. T.

Iverson

K. M.

(1999). Cracks in the glass ceiling? An empirical study of gender differences in income in the hospitality industry. Journal of Hospitality & Tourism Research, 23(1), 4-20. https://doi.org/10.1177/109634809902300102

42.

Steinberg

R. J.

(2001). Comparable worth in gender studies. In Smelser

N. J.

Baltes

P. B.

(Eds.), International encyclopedia of the social and behavioral sciences (Vol. 4, pp. 2393-2397). Elsevier.

43.

Thrane

(2008). Earning differentiation in the tourism industry: Gender, human capital and socio-demographic effects. Tourism Management, 29(3), 514-524. https://doi.org/10.1016/j.tourman.2007.05.017

44.

Tomaskovic-Devey

(1993). Gender and racial inequality at work: The sources and consequences of job segregation. ILR Press.

45.

Webster

(2014). Wage differentials and their determinants in US tourism and tourism-associated industries. Tourism Economics, 20(4), 695-725. https://doi.org/10.5367/te.2013.0304

46.

World Tourism Organization. (2018). Tourism highlights. UNWTO.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.65 MB