Quitters from Hospitality Industry: Misfit or Just Looking for Better Conditions?

Abstract

After the COVID-19 crisis, a considerable contraction emerged in the labor supply of the hospitality industry. This investigation aims to identify some factors that justify the intention behind an ex-worker in tourism changing their occupation. In particular, we investigate if the change is motivated by skill mismatches or by a willingness to find more stable and rewarding jobs. Several datasets were combined to obtain multilevel information on all the unemployed from the hospitality industry between September 2022 and August 2023. A probit model was used to estimate the intention to change occupation, considering different personal, regional and occupational characteristics. Our results demonstrate that, for overqualified individuals, the intention to change occupation is motivated by trying to find better conditions and not by the mismatch per se. In contrast, the underqualified are significantly affected by the mismatch, which make them more vulnerable. Generally, movers are significantly influenced by the odds of having higher job stability and better wages (especially higher overtime premium) and are more frequently younger and higher educated individuals. Therefore, employers and policymakers should promote better wages, job stability, training, and career progression opportunities to reduce turnover in the hospitality industry.

1. Introduction

According to the most recent estimates of the direct and indirect impacts of tourism provided by the Portuguese Tourism Satellite Account (INE 2023), the tourism and hospitality industry assumes a pivotal role within the Portuguese economy, accounting for a substantial 12.2% share of GDP. After the strong impact of the COVID-19 crisis on the turnover of the hospitality industry, which significantly affected labor demand in this industry (Lopes et al. 2021), and with external effects for other sectors within the same region (Lopes and Sargento 2023), an upturn was observed. As reported by the most recent data provided by Statistics Portugal, during the initial seven months of 2023, there was a notable 15% increase in the volume of overnight stays in Portugal when compared to the corresponding period in 2022. Furthermore, compared with an analogous timeframe in 2019, this variation corresponds to a 10% increase.

Despite the observed sustained growth, the worldwide tourism sector currently faces a potential bottleneck, motivated by labor shortages (Jus et al. 2022). The Portuguese case is no exception to this (Duarte and Silva 2024). According to the Portuguese Hotels’ Association, even though there was some recovery between 2021 and 2022 in the workforce within the accommodation and restaurant sectors, a comparison of 2022 with the year 2019 reveals a distressing decline of 34 thousand jobs, amounting to a reduction of 10.6%. After the forced leave from tourism to other sectors provoked by the effects of COVID-19, many workers did not return (Jus et al. 2022). Moreover, those that are currently searching for a job with previous experience in tourism reveal a higher-than-average intention to change their career, switching to other industries of the economy. According to our data, 40% of the unemployed from the hospitality industry intended to change to another occupation, while this percentage fell to 33% when all unemployed were considered. Medium-term perspectives do not look better: a recent survey of Portuguese hospitality industry workers, conducted by a recruitment and talent management firm (Eurofirms 2022), reveals that 58% of respondents do not see themselves working in the same industry in five years.

It is widely acknowledged that the tourism sector is a labor-intensive industry. Notwithstanding the technological advancements within this sector, heavily influenced by innovative digital technologies that have begun to supply certain labor-based tasks, human labor remains an indispensable aspect of the hospitality industry’s service provisions. As such, it is imperative to understand the underlying causes of labor shortages in this sector, particularly in light of its critical role in the Portuguese economy. Thus, the main objective of this study is to find out which factors lie behind this career change intention by those who were previously working in this industry. Specifically, we aim to determine whether the quitting intention is motivated by skills’ mismatch or if jobseekers are instead searching for more stable and better remunerated employment opportunities.

Drawing upon the theoretical framework of job search behavior, particularly person–job (P-J) fit theory, this study makes a noteworthy contribution to the extant body of literature and empirical research in the following ways. First, it relies on data sourced from authentic jobseekers, individuals who are unemployed and actively involved in the pursuit of a new job. This approach departs from the customary practice of relying on data stemming from surveys, which typically involves surveying samples of employees from particular firms, groups of firms, or sectors and asking them about their intention to change occupation (e.g., Duarte and Silva 2024; Erdogan and Bauer 2021; Aho 2020). Second, because of the comprehensive dataset gathered from several sources, this research employs objective metrics to measure the principal variables under scrutiny as potential catalysts for disengagement from the tourism sector (e.g., qualification mismatch, wage differentials, and job stability differentials). This approach differs from the majority of studies that rely on questionnaires, in which revealed perceptions are the primary source of data (e.g., Baquero 2022; Chen and Chen 2021; Duarte and Silva 2024; El-Sawalhy et al. 2022). Another original contribution of this study is that among the determinants evaluated as potential drivers for the intention to leave the hospitality industry, not only are the previous job characteristics accounted for (as, for example, in Manolopoulos et al. 2022), but also the differences between those characteristics and those of the newly desired jobs.

The remainder of this paper is organized as follows. The following section presents the theoretical background on a job search, person–job fit, and the intention to change careers, and indicates the hypotheses to be investigated. Section 3 describes the dataset construction, main variables, and modelling strategy. The results of the descriptive statistics and model estimations are presented and discussed in Section 4. The main conclusions, policy, and practical implications are presented in Section 5.

2. Theoretical Background

The predictors of a job search are linked to the economic assumption that employees weigh costs and benefits before deciding to search for new jobs (Wald 2005). While the costs are associated with imperfect information and the process of searching for new employment, the gains are assessed in comparison with the previous or current employment situation: Will the new job correspond better to the workers’ expectations? (Albrecht 2011; McCall 1970). Such expectations may be framed within the person–job (P-J) fit, which concentrates on the compatibility between an employee’s attributes and the attributes of the job or tasks they perform (Arvan et al. 2019; Feldman 2011; Kristof-Brown et al. 2005; Maynard et al. 2006). The P-J fit can be categorized into two subtypes: demands–abilities fit and needs–supplies fit (Edwards 1991). Demands–abilities fit concerns the match between a job’s requirements and an employee’s abilities and qualifications, encompassing skills, experience, and education (Edwards 1991; Kristof-Brown et al. 2005). Conversely, needs–supplies fit addresses how well an employee’s personal needs and preferences, including goals, values, and interests, align with the job’s characteristics that can fulfill those needs (Edwards 1991).

Following Maynard et al. (2006), P-J fit theory may be associated with the notion of under-employment, developed by Feldman (1996), who divided under-employment into five distinct dimensions: (i) possessing a level of education that surpasses the requirements of the job (overeducation); (ii) holding more skills or experience than necessary for the job; (iii) involuntarily working in a field unrelated to one’s educational background; (iv) being involuntarily engaged in part-time, temporary, or intermittent work; and (v) being underpaid, either in comparison to a previous job or in comparison to others with similar educational qualifications. Each aspect of under-employment may be viewed as an illustration of inadequate person–job (P-J) fit. While overeducation, skill surplus, and field of work misfit may be categorized as a poor demands–abilities fit, other aspects such as low-paying positions, temporary work, or part-time roles held by individuals who would prefer higher compensation, job stability, or full-time employment exemplify instances of needs–supplies incongruity.

According to the P-J fit theory, when compatibility between employees’ characteristics and the attributes of the job or tasks they perform is not met, this may lead to low job satisfaction, counterproductive behavior, and seeking new job opportunities (Maksum et al. 2021).

Among the various dimensions associated with a poor P-J fit, overqualification—either as overeducation, skill surplus, or experience—has received a great amount of attention from researchers. In today’s labor market, the quality of the workforce has been increasing, with higher levels of education and qualifications. This is a global phenomenon affecting both developed and developing countries, and has also been observed in the tourism sector, particularly in the Portuguese case. According to the Portuguese Tourism Agency (Turismo de Portugal 2023), 54% of the population employed in 2023 in the accommodation and restaurant sector had secondary or higher education (+5.0 p.p. compared to 2022 and +14p.p. compared to 2017), which confirms the upward trend in the qualifications of workers in the sector recorded in recent years. However, this increase in qualifications is not always matched by job opportunities, leading to overqualifications (Maksum et al. 2021).

Three theories have been proposed to explain the relationship between overqualification and job search behavior: matching theory, differential overqualification theory, and the career mobility hypothesis (Wald 2005). According to the matching theory, mismatches between workers’ skills and job requirements are temporary and arise due to imperfect information and the costs associated with a job search. Overqualified employees experience a sense of deprivation when their current job tasks do not align with their skills, knowledge, and abilities’ portfolio, which can trigger adverse emotions, such as reduced job satisfaction and, consequently, lead to voluntary turnover behaviors (Choi and Hur 2020; Dong et al. 2020; Erdogan and Bauer 2021; Maynard and Parfyonova 2013; Maynard et al. 2006; Marchante et al. 2007). Hence, suboptimal matches can be gradually eliminated by improving information about the labor market and job search efforts. Conversely, the differential overqualification and career mobility hypotheses posit that mismatches may be compatible with optimizing behavior. For instance, the differential overqualification theory suggests that married women who are overqualified engage in fewer job search activities than single women, which is related to geographical constraints associated with their husbands’ work (Büchel and Battu 2003). On the other hand, the career mobility hypothesis suggests that workers temporarily accept working in jobs that do not match their skills and education levels to obtain the necessary work experience and training for progression to higher-level positions (Marchante et al. 2007). However, if the expected promotions do not materialize, the tendency to quit and search for external jobs increases. The three theoretical perspectives on overqualification have different expectations regarding the effect of overqualification on job searches. The matching theory predicts a positive correlation, the differential overqualification theory predicts a positive correlation under certain circumstances (when unconstrained), and the relationship is uncertain under the career mobility hypothesis.

Wald’s (2005) empirical study, supported by data from the Canadian Changing Employment Relationship Survey, concluded that being overqualified increases the probability of a job search. Conversely, Marchante et al. (2007), in their study of the hospitality industry in Spanish Andalusia, did not find any supporting evidence for the job-matching theory or career mobility. Overqualification on entry does not work as a trigger for quicker progress in employees’ careers. A more recent study applied to hospitality and tourism confirms the matching theory, finding a positive relationship between overqualification and job search initiatives. Relying on surveys of employees in Egypt’s five-star hotels, El-Sawalhy et al. (2022) concluded that employees’ perceived overqualification has a positive and significant influence on their intention to leave their jobs.

Besides overqualification, other dimensions of P-J fit, particularly those illustrating the perspective of needs–supplies misfit, have been included in empirical studies trying to explain job search behavior. For instance, Marchante et al. (2007) found that, regarding voluntary external mobility, the primary driving force compelling employees to switch companies is the desire for enhanced working conditions, rather than seeking upward career advancement. Job-related variables have also been tested in studies applied to the hospitality industry, showing that workers with higher perceived job insecurity or less stable work contracts have a higher intention to change. This is the case, for example, for Baquero (2022) and Chen and Chen (2021), who considered the hospitality industry in the United Arab Emirates and Taiwan, respectively, concluding that perceived job insecurity is directly related to employees’ intention to quit. Manolopoulos et al. (2022), supported by the results of a survey conducted among Greek luxury hotel employees, concluded that the type of employment contract (permanent or temporary) moderates the relationship between job satisfaction and intention to change, while holding a temporary contract reinforces the impact of job dissatisfaction on turnover intention. Other demographic characteristics have also been found to influence job search, such as age and sex; older workers and females are less likely to initiate job search activities (Wald 2005).

Considering the previously reviewed theoretical and empirical contributions, the following hypotheses are established, concerning the distinct dimensions associated with a poor P-J fit, and the way these are expected to influence intention to career change, moving away from the hospitality industry:

H1: 

Over(under)qualified employees reveal a higher proclivity to career change, moving away from the hospitality industry (demands–abilities fit perspective of P-J fit).

H2: 

The perspective of having a higher paid occupation is a strong motivation for individuals to change their career path (needs–supplies fit perspective of P-J fit).

H3: 

The perspective of having a more stable work is a strong motivation for individuals to change their career path (needs–supplies fit perspective of P-J fit).

H4: 

Younger employees reveal a higher proclivity to career change, moving away from the hospitality industry.

H5: 

Male employees reveal a higher proclivity to career change, moving away from the hospitality industry.

This study’s results, further developed in the corresponding section, are supported by the P-J fit theory. Specifically, under- or over-qualified individuals reveal a higher tendency to change their career path (demands–abilities fit). Nevertheless, when controlling for the discrepancies between previous and desired occupations, this hypothesis remains valid only for underqualified individuals. For overqualified individuals, the motivation for career change is not the skill mismatch, but rather the search for more stable and better remunerated employment opportunities (needs–supplies fit).

3. Materials and Methods

The main sources of information for this study were the monthly datasets of the Portuguese Institute of Employment and Professional Training (IEFP) with information on all individuals that are formally considered unemployed in mainland Portugal. We selected all new registrations made in the IEFP centers during the one-year period, from September 2022 until August 2023, corresponding to more than 532,000 individuals. Next, considering the purpose of this investigation, only information on individuals who had been working in a payable occupation in the hospitality industry was considered, which constitutes a dataset of 46,536 individuals.

The IEFP datasets include information on personal characteristics (age, sex, schooling, residence county, whether it was their first registration in IEFP or not), related to their last job (occupation, business sector, and the motive of disruption), and the desired occupation for the future. Occupation is identified by five digit codes (PCO) defined in INE (2011). In the same document, each occupation was classified on a 1–4 scale, considering the level of specific skills that are required to be performed. The first level includes occupations that require basic skills (simple and routine tasks) and, according to INE (2011), corresponds to an educational level of five or fewer years of schooling. The second level involves executing basic calculations, the ability to acknowledge security instructions and process information, and matches educational levels between six and twelve years of schooling. The third and fourth levels require solving technical to complex problems, intermediate to top management functions, and research and development, and have a good correspondence with the higher education level. This classification and the match with the educational levels of each individual were included in our dataset.

The third source of information used in this study was Quadros de Pessoal collated by Gabinete de Estudos e Planeamento (GEP) of the Portuguese Ministry of Labor. Quadros de Pessoal provides information regarding several relevant worker-level variables, including the wage bill, hours of work, occupation, type of contract, and overtime premium. To obtain a characterization of the previous and desired occupations, we aggregated these variables at the occupational level and, by using the common PCO code, matched this dataset to our data.

Finally, following Deng et al. (2022), we included regional context variables that may influence the intention to move. By considering the residence county identification, regional variables, such as the unemployment rate and the percentage of employment in the hospitality industry (obtained from Statistics Portugal, INE), were included in the dataset as control variables. The match between the different sources, treatment of variables, and estimation of the models presented below were carried out using Stata software (version 9.2).

To identify how personal and occupational factors determine the intention to change from an occupation after a disruption, we followed Choi and Hur (2020), who apply a probit model to estimate how the propensity of turnover intention is influenced by qualification mismatch, along with variables expressing other individual and employers’ characteristics. In our model, expressed by Equation (1), the dependent variable, Y_i, corresponds to a dummy with a value of 1 if the individual i intends to change the occupation (=1 if the PCO code of the desired occupation is different from the PCO of the last occupation), and 0 if he/she intends to remain in the same occupation (=0, if the desired occupation and the previous occupation have the same PCO code). Therefore, we estimate the following equation:

$$Y_i={\alpha }_1+{\beta }_1{O}_i+{\gamma }_1{U}_i+{\delta }_1{X}_i+{\rho }_1{M}_i+{\tau }_1{R}_{c\left(i\right)}+{\mu }_i.$$

(1)

O_i stands for overqualified and denotes a dummy variable that is 1 if individual i was working in an occupation that requires a level of skill lower than her/his years of schooling (for example, individuals with higher education who were working in an occupation classified as belonging in the first or second level of skills, INE 2011). Thus, we followed Erdogan and Bauer (2021) by measuring overqualification with objective measures and considering overeducation, as it is relatively straightforward to gauge an individual’s educational level and the formal educational prerequisites of the job (instead of skill overqualification and cognitive ability overqualification). U_i—underqualified—identifies individuals who were in an occupation with a higher level of specific skills compared with their level of education (the years of schooling of these individuals are lower than those that match the level of skills required in that occupation according to INE 2011 classification).

X_i corresponds to the vector of individual characteristics such as sex, age, first unemployment (if he/she had never been registered in IEFP centers before), and disruption type (five dummies: end of temporary work, layoff from a permanent job, voluntary withdrawal from a permanent job, mutual agreement disruption, and self-employment, with other situations being the residual category). M_i identifies the month of the year in which the individual has registered in the IEFP center (capturing seasonal effects), and R_c(i) is the vector of the municipality characteristics: unemployment rate, weight of hospitality industry (in employment), and the corresponding NUT II region.

In the second model (Equation (2) below), we add the matrix P_o(i) which is composed of four variables that compare the previous occupation with the desired occupation: wage difference (difference between the average hourly wage in the desired occupation and the average hourly wage in the previous occupation), difference in overtime premium, difference in the percentage of permanent contracts, and the difference in the percentage of part-time workers.

$$Y_i={\alpha }_2+{\beta }_2{O}_i+{\gamma }_2{U}_i+{\delta }_2{X}_i+{\rho }_2{M}_i+{\tau }_2{R}_{c\left(i\right)}+\sigma P_{o\left(i\right)}+{\epsilon }_i$$

(2)

4. Results and Discussion

4.1. Descriptive Statistics

In Figure 1, we present the evolution of the hospitality industry’s registered unemployed in IEFP centers by month, for the entire period considered herein. We also separate these records into individuals who intend to remain in the same occupation (the Stayers subsample) and those who intend to move to another occupation (Movers subsample). First, we noticed that it is relevant to control for possible seasonality effects, since unemployment in this industry is clearly heterogeneous across months, with November presenting a significantly higher number of registrations (around 9000), while in August there are less than 3000 new registrations. Second, the percentage of individuals who intend to move to a new occupation is significant (near 40% considering the whole period, higher than the 33% average of all the industries), but it also changes across the year; in September, the percentage is higher and closer to 50%, while in November, it corresponds to less than 30% of total registrations.

In

Table 1. Variable descriptions and summary statistics.

Variable (1)	Description (2)	Stayers Mean/% (S.D.) (3)	Movers Mean/% (S.D.) (4)
Overqualified	Dummy: 1 if the previous occupation requires a level of skill lower than the corresponding individual schooling level (INE 2011).	28.8% (0.5)	33.9% (0.5)
Underqualified	Dummy: 1 if the previous occupation requires a level of skill higher than the corresponding individual schooling level (INE 2011).	7.4% (0.3)	6.9% (0.2)
Sex (female)	Dummy: 1 if the individual is female; 0 otherwise.	62.0% (0.5)	68.8% (0.5)
Age	Age, in years, at the moment of IEFP registration.	40.1 (12.9)	37.1 (12.2)
First unemployment	Dummy: 1 if the individual has never been unemployed before; 0 otherwise.	25.9% (0.4)	18.5% (0.4)
Schooling	Number of years of schooling.	8.99 (3.5)	10.2 (3.4)
Disruption Type:
Temporary job	Dummy: 1 if the previous job of the individual was temporary; 0 otherwise.	62.7% (0.5)	47.2% (0.5)
Permanent job (involuntary disruption)	Dummy: 1 if the individual with a permanent work contract was dismissed from his/her previous job; 0 otherwise.	11.2% (0.3)	12.1% (0.3)
Permanent job (voluntary disruption)	Dummy: 1 if the individual with a permanent work contract voluntarily decided to leave his/her previous job; 0 otherwise.	5.3% (0.2)	11.8% (0.3)
Permanent job (mutual agreement)	Dummy: 1 if the individual with a permanent work contract agrees with the employer to end his/her job; 0 otherwise.	2.7% (0.2)	1.9% (0.1)
Self-employed	Dummy: 1 if the individual was previously self-employed; 0 otherwise.	0.7% (0.1)	1.4% (0.1)
Region:	Dummy: 1 if the residence county is in the Norte/Centro/Lisboa e Vale do Tejo/Alentejo/Algarve region; 0 otherwise.
Norte		25.3% (0.4)	30.5% (0.5)
Centro		14.2% (0.3)	22.4% (0.4)
LVT		20.2% (0.4)	21.4% (0.4)
Alentejo		5.1% (0.2)	8.1% (0.3)
Algarve		35.2% (0.5)	17.6% (0.4)
Unemployment rate	Percentage of unemployed (in total population) in the residence county.	5.26 (1.54)	4.95 (1.51)
Employment in hospitality industry	Percentage of employment in accommodation and food services in the residential regional total employment.	16.5 (12.9)	11.8 (9.5)
Difference in hourly wage	Difference between the average hourly wage in the desired occupation and the average wage in the previous occupation.	0 (0)	1.04 (2.6)
Difference in overtime premium	Difference between the average hourly wage obtained during overtime in the desired occupation and in the previous occupation.	0 (0)	1.23 (3.0)
Difference in permanent contract workers percentages	Difference between the percentage of workers with a permanent contract in the desired occupation and the same variable in the previous occupation.	0 (0)	0.07 (0.2)
Difference in part-time workers percentages	Difference between the percentage of workers in part-time in the desired occupation and the same variable in the previous occupation.	0 (0)	−0.04 (0.1)
Number of observations		30,009	16,527

, we display the descriptive statistics for the main variables of the models by considering our dataset separated into two subsamples: the subsample of the individuals that intend to remain in the same occupation, “Stayers”, and those that intend to change occupation, “Movers”. According to what is presented in Table 1, around 30% of the unemployed had been working in occupations that required skills below their corresponding level of education, and only 7% were employed in occupations with job-specific skills above the individual schooling years. In addition, overqualified individuals represent a higher percentage in the Movers subsample, while the percentage of underqualified individuals is similar for both subsamples.

Comparing the two columns of Table 1, we also notice that the subsample of Movers has a higher percentage of females, younger people, more years of schooling, and of individuals that had been unemployed more often before. In relation to the motive for disruption, the end of a temporary job corresponds to a significant percentage in both subsamples, but with a higher expression in the subsample of Stayers. Movers more often come from a voluntary disruption and from the end of self-employment than Stayers. The percentages of Movers are also higher in the North and in the Center of Portugal and significantly lower in Algarve and in counties with lower unemployment or with a lower weight of hospitality employment compared with the Stayers subsample.

As expected, in Column (1), all the means of the variables corresponding to comparisons (differences) between the previous and desired occupations are zero. However, in the second column, we observe a positive value for the means of all the variables, except for the part-time percentage, which indicates that workers are attracted by occupations with better conditions in terms of wages and job stability.

4.2. Model Results

The results obtained by estimating Equation (1) are presented in

Table 2. Model estimations.

Variable	Model (1)	Model (2)
Overqualified	0.051 ***	−0.130 ***
Underqualified	0.142 ***	0.372 ***
Sex (female)	0.289 ***	0.337 ***
Age	−0.012 ***	−0.012 ***
First unemployment	−0.422 ***	−0.412 ***
Schooling	0.050 ***	0.034 ***
Disruption Type:
Temporary job	−0.223 ***	−0.221 ***
Permanent job (involuntary disruption)	−0.148 ***	−0.164 ***
Permanent job (voluntary disruption)	0.232 ***	0.205 ***
Permanent job (mutual agreement)	−0.399 ***	−0.397 ***
Self-employed	0.201 ***	0.204 ***
Region:
Norte	−0.020	−0.043 **
Centro	0.159 **	0.136 ***
Alentejo	0.193 ***	0.178 ***
Algarve	−0.157 ***	−0.123 ***
Unemployment rate	−0.012 **	−0.013 **
Employment in hospitality industry	−0.009 ***	−0.006 ***
Difference in hourly wage		0.039 ***
Difference in overtime premium		0.242 ***
Difference in permanent contract workers percentages		3.991 ***
Difference in part-time workers percentages		−1.096 ***
Number of Observations	46,536	46,536
Pseudo R2	0.0867	0.2127
Log-likelihood ratio	5252.56 (0.0000)	14,911.94 (0.0000)

Note: In both models, 11 dummy variables were included to control for seasonal effects. ** p-value < 0.05, *** p-value < 0.01.

, Column (1). The log-likelihood ratio of 5252.56 (p-value = 0.0000) reveals that the variables included in the model undoubtedly influence the decision to remain or move to another occupation. In addition, we tested for the possibility of multicollinearity problems by using Variance Inflation Factor (VIF) tests, which presented an average of 1.86, thus drawing a conclusion regarding the inexistence of this problem. In Column (2), a matrix composed of variables that compare the wages and job stability of the previous occupation with the desired one is added, which significantly increases the explanatory power of the model, suggesting the importance of needs–supplies fit factors for the intention to change occupations. Again, there was no evidence of multicollinearity among the variables.

By comparing the results of both models, it was observed that there were no significant discrepancies in the signal or the statistical significance of the coefficients of the initial variables, with the exception of the overqualified coefficient. In the first model, the coefficients for both the overqualified and underqualified were positive and statistically significant, supporting Hypothesis H1: Individuals whose qualifications do not align with the required skills for their occupation are more inclined to seek a different occupation (demands–abilities fit). However, after considering the differences between previous and desired occupations, this hypothesis holds true only for underqualified individuals. For overqualified individuals, the coefficient is negative in the second model. Consequently, it is concluded that what motivates the most qualified workers in less-qualified occupations to transition to a different occupation is not the mismatch itself, but the desire for a job with better conditions; the needs–supplies fit factors are more influential. Therefore, when controlling for differences in wages and job stability, overqualified workers prefer to remain in their current occupations. However, underqualified individuals are penalized by skill mismatches, making them more susceptible to unemployment.

Regarding personal characteristics, it was observed that individuals with more years of schooling, who had been unemployed before, females, and younger individuals seemed to be more prone to changing their occupation. This means that our results validate Hypothesis H4 but not Hypothesis H5. Clearly, individuals who decided to end their jobs (voluntary disruption and end of self-employment) were more willing to change. This means that the decision to disrupt careers is made before unemployment.

The regional context also influences the decision of the unemployed from the hospitality sector to remain in the same occupation or move to a different one. In fact, both the unemployment rate and weight of hospitality employment in the county contribute to the decision to remain in the same occupation. On the one hand, a higher unemployment rate indicates a weakened labor market, where the range of working alternatives presented to jobseekers is lower. On the other hand, individuals living in territories with a higher percentage of employment in the hospitality industry have more opportunities for employment in this industry (compared to the overall industry), thus having more motivation to stay within this industry. In addition, the NUT II region where the county is located is important for that decision, with Algarve (the most touristic region in mainland Portugal) and North Portugal having a negative impact on the decision to move. The result obtained for Algarve shows that even if the individual lives in a county where tourism has low importance, their intention to move will be low if tourism is significant in other counties of the same NUT II (neighborhood effect).

Finally, as expected, individuals are looking for better jobs (better paid and more stable), which confirms Hypotheses H2 and H3 and is in accordance with Marchante et al. (2007). Interestingly, the difference in the overtime premium seems to be more important for the decision to change than the difference in the normal hourly wage. In addition, comparing the coefficients related to job stability, it seems that the difference in permanent contract percentages is more important for motivating career changes than the percentage of part-time workers.

5. Conclusions

The COVID-19 crisis affected the hospitality industry significantly, causing an increase in unemployment in this industry above the average of the remaining industries. This might have caused a generic feeling of distress among those who lost their jobs in this industry, leading them to seek jobs in other industries. A few years after this crisis, the hospitality industry has shown an impressive recovery, but the unemployed from this industry continue to reveal more intentions to change their occupation than the overall unemployed. This deepens the gap between job demand and supply in this industry and represents a potential bottleneck for its growth. Thus, in this study, we aimed to identify some factors that might justify this quitting intention; in particular, we investigated whether it is motivated by skill mismatches or if the workers are looking for more stable and better rewarded jobs.

The research findings indicate that the inclination to change one’s occupation is significantly influenced by differences in wages and job stability between the previous and desired occupations, which is consistent with the principles of job search theory. The innovative approach used in this study, including objective measures instead of perceptions for the several variables of the model and comparing the characteristics of the previous job with those of the newly desired jobs, provides a relevant contribution to the theory that explains job search behaviour, particularly to the person–job fit theory. Specifically, our results reveal that, for individuals who previously held an occupation that required skills below their qualifications, when controlling for the differences between the previous job and the prospective one, the intention to switch is motivated by the pursuit of more favorable working conditions, rather than the mismatch itself. This implies that, among the two categories considered to measure person–job fit, the needs–supplies fit factors exert a higher influence on overqualified workers than the demands–abilities fit. In the case of underqualified individuals, the lack of a demands–abilities fit remains a relevant factor to stimulate turnover intention, even when accounting for differences between the previous occupation and the desired one. Thus, this study demonstrates that, when applying the person–job fit theory to investigate turnover intention, it is important to consider not only the influence of the current (or previous) job’s characteristics, but also the difference between those and the ones of the job that worker intends to change to.

The findings of this study are also important for managerial practice, particularly within the context of the hospitality industry, highlighted in the following paragraphs.

Firstly, the fact that individuals with lower levels of education than those required for their occupation tend to be more susceptible to unemployment (Lopes et al. 2021) and changing jobs more frequently should trigger specific actions to correct that mismatch. In particular, one important practical implication of this is that both employees and employers should invest in enhancing the qualifications of less-educated workers by providing opportunities to improve their education level or by offering on-the-job training. When implementing reskilling and upskilling programs, it is important to focus on future-proof skills such as digital literacy, which will be crucial in the face of ongoing digital transformation in the sector. Consequently, policymakers involved in the education system should collaborate with private firms operating in the sector to design vocational training plans tailored to the schedules and contents of individuals already employed in the hospitality industry.

Secondly, in terms of wage discrepancies between previous and desired occupations, our findings indicate that Movers are more motivated by differences in overtime premiums than regular wages. This is likely due to the low average wage in this industry, which makes the overtime premium particularly significant. Furthermore, we found that individuals with previous experience in the hospitality industry are drawn to occupations with higher job stability, as measured by the proportion of permanent contracts and percentage of full-time employees. These results align with the conclusions of a recent survey of Portuguese hospitality industry workers (Eurofirms 2022), which found that low average wages and low job stability are among the main factors contributing to the industry’s reduced attractiveness as a career choice. Therefore, organizations seeking to attract and retain skilled personnel should provide employees with improved compensation packages, enhanced job stability, and clear career development plans that offer opportunities for advancement. Another factor that has been identified as a deterrent to pursuing a career in tourism is the demand for schedules associated with late shifts and weekend work (Jus et al. 2022; Eurofirms 2022). However, this was not captured in our model, because the datasets used did not provide information on the work schedules. This represents a limitation of our study that could be overcome in future research by supplementing secondary data with primary data collection methods, although matching these data may be prevented by the anonymity of the information contained in the secondary data.

Finally, our results also reveal that a significant proportion of individuals who voluntarily transition from employment to unemployment and disrupt their careers tend to switch to a different occupation. Conversely, those who have been fired or whose temporary contracts have ended are more inclined to remain in the same occupation. Furthermore, higher-educated individuals and younger workers are more likely to move to another occupation, which may pose a threat to the future quality and sustainability of the labor supply in the hospitality industry. To mitigate this impact, in addition to ensuring employees’ basic needs, such as fair wages and job stability, companies in this industry must also align their strategic positioning and operations with the preferences of the new generation of workers, who typically value the social and environmental values upheld by their employers, in addition to basic working conditions.