Can Remittance Promote Tourism Income and Inclusive Gender Employment? Function of Migration in the South African Economy

: With globalisation and international trade, remittances and migration significantly influence economic activities, yet their impact on tourism income and gender-specific employment remains under-researched. This study uses autoregressive distributed lags and Granger causality to examine the effects of remittances and migration on tourism income and employment in South Africa. Three models are established as follows: for aggregate employment, male employment, and female employment, each with equations for tourism income and employment. Key findings from this study indicate that remittances significantly drive tourism income in both the short and long run across all models. Conversely, employment negatively impacts tourism income, hinting at sectoral trade-offs. Migration positively affects tourism income in the short run for male and aggregate models but is insignificant for female employment. Remittances boost male employment in both the short and long run, whereas their impact on female employment is significant only in the long run. Causality analysis shows a bidirectional relationship among employment indicators, with unidirectional causality from remittances to migration and from migration to income. This study recommends policies to support remittance inflows and their productive use in tourism, along with targeted interventions to reduce gender disparities in employment and promote equitable economic opportunities.


Introduction
One of the central topics in the migration debate is the belief that remittances from migrant workers may be tapped by developing countries to achieve sustainable development.This subject has been advocated by international financial organisations, including the International Monetary Fund and World Bank, that promote economic globalisation and the liberalisation of state government migration laws (CARAM Asia 2010).However, the usage of remittances in achieving growth and development depends on how remittances are used in the economy (Sobiech 2019).Compared with other financial flows, remittances are sent directly to the recipient's family, possibly resulting in a more equitable allocation of resources than foreign direct investment and aid, and they tend to be less unpredictable when compared with other foreign flows (Hamed 2022).Likewise, the link between labour supply and income has been established theoretically by the Classical, Neoclassical, and Keynesian schools of thought (Bortis 2023).Arguments have also been put forth that migration can influence income and employment in the empirical literature (Alderotti et al. 2023;Bossavie and Özden 2023;Diodato et al. 2023).Migration can indirectly affect income by influencing labour markets in both sending and receiving communities.The influx of migrants into a particular region can stimulate economic growth, leading to higher demand for goods and services, which, in turn, can create more job opportunities and increase incomes for both migrants and native-born residents (Alderotti et al. 2023;Bossavie and Özden 2023).On the other hand, migration can sometimes put downward pressure on wages, especially in industries where migrants are concentrated, raising worries about wage cuts and competition for jobs among home workers (Diodato et al. 2023).
Remittances and immigration usually follow a clear pattern.People relocate for better economic opportunities, find work, and transfer money back to their country of origin.Beyond the conventional roles of remittance and migration providing financial support and demographic shifts, they can help increase tourism income and promote inclusive gender employment in developing countries (Ferdous 2024;Cucovic et al. 2023).As a nation improves the tourism industry, it may have a positive effect on economic development; the nexus among remittance, migration, and tourism income can be a determining factor in achieving this goal (Tabash et al. 2023).Remittances may be channelled into investments that promote tourism infrastructure and services.
Similarly, migration patterns influence tourist flows, as diaspora communities attract visitors to their countries of origin and encourage a continuous income flow through familial ties and cultural exchange.The multiplier effects of remittance and migration may extend to gender empowerment and employment (Matz and Mbaye 2023).Remittances to low-and middle-income countries grew by 3.8 per cent in 2023 (Ratha et al. 2023).Remittance inflows to South Africa have increased because of a rise in migration; they increased by 3.5 per cent in 2022 and reached USD 872.9 million in 2023 (World Bank 2023).Also, the tourism industry continues to improve the well-being of millions of individuals by promoting economic growth, providing employment, and eliminating poverty (Wood 2007;Xuanming et al. 2024).For the eighth year, industry growth outpaces the global economy, demonstrating the sector's persistence in the face of international political instability and economic shocks.In 2019, the industry provided USD 8.9 trillion to the world economy (10.3 per cent of world GDP) and created 330 million jobs (one out of every ten on the globe) (Stellenbosch Business School 2023).
On the other hand, over 50 per cent (51.1 per cent) of South Africans are women.Additionally, two-fifths (42 per cent) of families are led by women.Women in South Africa are far more likely to be jobless than males, and they are far less inclined to engage in the employment market (Statistics South Africa 2024).From the preceding, this study investigates the effect of remittance and migration on tourism income and gender employment in South Africa.The following four reasons justified this study's motivation: (i) its alignment with the United Nations Sustainable Development Goals; (ii) issues relating to COVID-19 impacts on tourism, income per capita, and employment; (iii) the ability to promote inclusive growth; and finally, (iv) the gap identified in the literature.
Firstly, using remittances to boost tourist income is connected to SDG 8 (Decent Work and Economic Growth); it can help create job opportunities in the tourism sector, and the revenue generated may assist in promoting growth and development (Bernardo et al. 2023).Also, encouraging inclusive gender employment within the structure above is consistent with SDG 5 (gender equality), which seeks to promote equitable access to economic opportunities for men and women in the labour market.Likewise, promoting economic empowerment through remittances and tourism can indirectly assist SDG 1 (No Poverty) and SDG 10 (Reduced Inequalities) by pulling communities out of poverty and decreasing economic gaps.Economic and social sustainability studies have stressed how the labour market for souvenir production and tourism sales supports livelihoods, alleviates poverty, and addresses job quality per SDG8 and gender equality (Sousa et al. 2022;Asaleye et al. 2017).Secondly, the COVID-19 pandemic caused a global decline in international tourist arrivals, resulting in a significant loss of income for the tourism sector (World Tourism Organization 2021); this decline has led to widespread job losses and business closures within the industry.As a result of this effect, women's employment has been affected mostly compared with the males, leading to higher rates of unemployment and underemployment among women.Women constitute a portion of the tourism workforce, particularly in service-oriented roles such as accommodation, food services, and retail (International Labour Organisation 2022).In South Africa, the tourism sector contributed about 3 per cent of GDP in 2018, accounting for around 4.5 per cent of total employment.As of 2020, the number of tourists fell by 72.6 per cent, from 10.2 million in 2019 to 2.8 million.In 2019, the tourist sector directly contributed 3.7% of GDP, surpassing agriculture, construction, and utilities.In 2020, the direct contribution of the tourist sector fell to 2.2 per cent.Given the circumstances of the pandemic that witnessed economic activity plunge in most economic sectors, tourism lost ground to even the smallest industries.As tourist spending fell and economic activity stagnated, so did the total amount of direct tourism employment.The reduction in employment number was more than 320,000 in 2020, dropping from 780,096 in 2019 to 459,533 in 2020 (Statistics South Africa 2022).
Thirdly, this empirical study can help to achieve inclusive growth in the South African economy.Remittances from migrant workers play an important role in augmenting household incomes and facilitating greater participation in tourism-related activities, both as consumers and entrepreneurs, especially among marginalised communities (Khan 2024;Chowdhury et al. 2023).Moreover, as mentioned earlier, remittances may be invested in tourism infrastructure, which may enhance the attractiveness of destinations and create a platform for inclusive gender employment (World Travel and Tourism Council 2024); this, in the long run, may provide opportunities for women in hospitality and tour guiding (International Labour Organisation 2022).The Department of National Treasury (2023) has stressed the need for inclusive growth.According to the report, reforms are needed to pave the way for inclusive economic growth and development to fulfil the constitution's objective, as highlighted by the South African government.Growth without considering inclusive growth will merely worsen the inequitable economic patterns inherited from the past.Reforms that do not include inclusive growth will be limited and unsustainable (Asaleye et al. 2023b).Lastly, studies have worked on migration and remittance (Rahman et al. 2023;Santos 2023;King and Vullnetari 2010) but our study can be distinguished from the previous studies by examining the effect of migration and remittance on tourism income and gender employment.
From the preceding, the main objective of this study is to examine the impact of remittances and migration on tourism income and employment in South Africa.The research questions are as follows: i.
What are the short-and long-run effects of remittances and migration on aggregate employment?ii.What are the short-and long-term effects of remittances and migration on tourism income?iii.What are the short-and long-term effects of remittances and migration on genderspecific employment?iv.What is the causal relationship among remittances, migration, employment, and tourism income?
2. Literature Review 2.1.Review of Studies Regarding Remittance, Tourism, and Employment Remittances have been recognised in the economic literature as a significant source of income for recipient households, potentially leading to increased consumer spending and economic activity (Wang et al. 2021).Remittances can act as a form of income transfer that boosts the purchasing power of households, thereby stimulating demand for goods and services, including those in the tourism sector (Kim et al. 2024); this additional income may contribute to higher levels of tourism expenditure, as families may allocate a portion of remittance funds towards travel, accommodation, and recreational activities, bolstering tourism income.Theoretically, the New Economics of Labour Migration (NELM) theory suggests that migration is an individual decision and a family strategy to maximise income and diversify sources of risk; this migration pattern can influence tourism income, as migrants often promote tourism in their home countries, either through direct visits or by encouraging others to visit (Stark and Bloom 1985;Williams and Hall 2000).Nevertheless, the NELM connects migration, remittance growth, and the labour market (Russell 1986).Wage disparities among countries are not required for migration decisions in the model, which may be limited in applications but have implications on income and employment outcomes.However, migration is viewed as a way to expand a family's alternate sources of income through remittance.
Consequently, the Harris-Todaro migration model also explains migration with remittances but does not economically benefit non-migrant family members (Todaro and Smith 2003).However, the study by Adams and Page (2005) posits that remittances may play a crucial role in stabilising and growing economies as their source of foreign exchange.Remittances increase household disposable income, which can be spent on various goods and services, including tourism (Adams and Page 2005); this increased spending can stimulate the local tourism industry by creating demand for tourism services and products, eventually leading to higher tourism income.
Regarding gender-specific employment in the tourism sector, the Feminist Economics perspective emphasises the importance of understanding how remittances and migration impact male and female labour participation differently (Scogin 2024;Kabeer 2000); this perspective shows that women and men experience the benefits and burdens of migration and remittances in difference ways, which can lead to varying outcomes in their participation in the tourism labour market.Studies have shown that remittances can empower women by providing them with financial resources, but they can also increase their domestic responsibilities if male family members migrate (Chant 1998;Kabeer 2000;Posel 2004).Conversely, men may benefit from employment opportunities through migration than women, but they may not experience the same level of empowerment as women receiving remittances (Posel 2004;Chant 1998).The importance of remittances in promoting women's empowerment is also stressed by Pfeiffer et al. (2008).According to these scholars, remittances may provide women with financial resources that enhance their economic independence and allow them to invest in education and entrepreneurial ventures (Pfeiffer et al. 2008).As women gain financial independence, their labour participation will increase, increasing aggregate welfare, including in the tourism sector, thus promoting inclusive gender employment.
Empirically, the relationship between remittances and tourism income has been examined.Hossain (2020) investigate the impact of remittances on economic activities in Malaysia.The study involved in-depth interviews with 20 first-cycle and repeat-cycle migrants from Madhupur Village, Bangladesh.Hossain (2020) found a distinct difference in the impact of social remittances on economic activities between these two groups.Social remittances did not affect the economic activities of first-cycle migrants, whereas the economic activities of repeat-cycle migrants were somewhat influenced.A study by Noushad et al. (2022) examines recent trends, changing patterns, and determinants of low-skilled emigration from India to the Gulf Cooperation Council.It also investigates the developmental impacts of remittances in India.Noushad et al.'s (2022) study finds that the Indian diaspora is growing globally, shifting from less-developed Asian regions to more advanced areas in North America, Europe, and Oceania.In India, states with lower levels of development and well-being tend to have higher rates of low-skilled emigration, which decreases as the state advances.Poverty and unemployment are major drivers of low-skilled emigration in India.According to Noushad et al. (2022), remittances do not directly impact GDP growth; they significantly influence development and well-being.Tabash et al. (2023) use the Fixed Effect Model and Fully Modified Ordinary Least Squares technique to study the relationship among tourism, remittances, and foreign investment in selected Asian countries.Their findings indicate that international tourism activities positively and significantly impact GDP growth rates and GDP per capita.Tabash et al. (2023) recommend that governments and host countries prioritise tourism activities and focus on tourism operations' dynamic role, importance, and sensitivity.
More so, Mora-Rivera and García-Mora (2021) provide evidence from Mexico using Probit and Tobit regressions with instrumental variables.Mora-Rivera and García-Mora (2021) demonstrate that international migrants' remittances can enhance domestic tourism expenditures among Mexican households.Mora-Rivera and García-Mora (2021) suggest that international remittances help alleviate the liquidity constraints that limit access to tourism activities for recipient households.Also, Cohen et al. (2012) indicate that during economic downturns, remittances can help sustain household consumption, including spending on domestic tourism, supporting the tourism sector even during financial instability.Pfeiffer et al. (2008) emphasise that remittances can enhance women's economic independence, allowing them to participate more fully in the labour market, including tourism.King and Vullnetari (2010) report the impact of remittances on gender roles in Albania, showing how increased financial resources from remittances can shift traditional gender norms and promote greater gender equality in employment.As women become more financially independent, they are more likely to pursue employment opportunities in various sectors, contributing to a more inclusive economy.From the preceding, numerous studies have examined the connection among remittance, tourism, and employment.Some studies focus on micro-analysis (e.g., Hossain 2020; Mora-Rivera and García-Mora 2021), while others employ panel studies (e.g., Tabash et al. 2023).Additionally, some studies limit their analysis to growth (e.g., Noushad et al. 2022;Tabash et al. 2023).Research on the implications of remittance on tourism income and employment is still developing.

Review of Studies Regarding Migration, Tourism, and Inclusive Employment
Theoretically, the Human Capital Theory shows the connection between migration and employment (Becker 1964).According to this theory, migration involves transferring skills, knowledge, and experiences from migrants to their home countries.In this process, the migrants often acquire new skills abroad, which can be moved back home, benefiting the local economy and increasing employment opportunities (Becker 1964).Likewise, when remittances gained through migration processes are invested in education and skill development, they enhance human capital, indirectly benefiting sectors such as tourism.The increase in human capital can lead to improved service quality and the creation of new tourism-related businesses, thereby boosting tourism income (Docquier and Rapoport 2012).
Empirically, Privarova et al. (2022) examine the relationship between migration and tourism in European countries.Privarova et al.'s (2022) study indicates that migration and labour migration play crucial roles in international tourism and the labour market.Immigration positively impacts both the tourism sector and the labour market.In contrast, emigration negatively affects these areas, primarily because of the potential "brain drain" in the countries of origin.According to Privarova et al.'s (2022) findings, international tourism significantly benefits the labour market, offering substantial evidence for migration-led tourism in certain European Union member states.Another study by Hutsaliuk et al. (2020) documents that external labour migration benefits the economy through the stimulation of entrepreneurial activity among migrants, leading to the development of small and medium-sized enterprises and the generation of employment opportunities.Conversely, Hutsaliuk et al. (2020) also highlight several negative repercussions, which include the lack of profitability and potential losses associated with investments in the education and training of specialists, the emigration of highly skilled personnel, declines in birth rates, demographic ageing, increased demographic pressures on the domestic labour market, reduced tax revenues and contributions to social funds, and inflationary pressures on domestic goods and services, particularly notable in situations characterised by limited purchasing power among the majority of the population.Dempster and Zimmer (2020) and Withers et al. (2022) investigate the impact of COVID-19 on migration and tourism.Dempster and Zimmer (2020) show that in the short term, governmental bodies and civil society must prioritise ensuring access to healthcare and social safety nets for all migrants, mitigating xenophobia, facilitating repatriations, and dismantling barriers to mobility.According to Dempster and Zimmer (2020), it is crucial to utilise this period for a comprehensive revaluation of the operational dynamics of the tourism industry, with a particular emphasis on sustainability and improving working conditions for its key personnel.Dempster and Zimmer (2020) point out that actions should be directed towards bolstering tourism flows, especially to low-income countries.More so, Withers et al. (2022) state that the COVID-19 pandemic has disrupted the international remittance flow crucial for several South Asian economies; this disruption, termed "remittance shock", has triggered a decline in foreign exchange earnings, worsened structural unemployment, and posed a threat to the well-being of numerous low-income households.Withers et al. (2022) investigate the economic repercussions on three remittance-dependent economies including India, Nepal, and Sri Lanka.They concluded that a combination of short-term measures can be used to address the economic consequences of reduced migration and remittances, alongside long-term strategies promoting inclusive and sustainable development.
In addition, studies by Sevencan (2023), Chowdhury et al. (2023), and Mehedintu et al. (2020) focus on the relationship between remittances and economic growth.Sevencan (2023) investigates short-run and long-run behaviour among remittances, unemployment, economic growth, and human development in developing nations by applying a Vector Autoregressive Model, Fully Modified and Dynamic Ordinary Least Squares.According to Sevencan (2023), there is a causality between GDP and remittances in low-income economies in the short run.The study also reveals that the unemployment rate positively impacts the Human Development Index in the long term.While remittances significantly influence human development in low-income economies more than in other developing nations, there is no notable link between unemployment in the home country and the relationship between remittances and development in low-income countries.Chowdhury et al. (2023) use pooled ordinary least squares, fixed effect, and random effect models to assess the overall impact of remittances on economic development.They employ the Vector Error-Correction Model and Granger causality to examine countryspecific effects.Chowdhury et al.'s (2023) findings show a significantly negative impact of remittances on economic development.In Bangladesh, there is no short-term or longterm relationship between remittances and economic development; in Vietnam, there is a short-term relationship but no long-term one.In Sri Lanka, short-run causality runs from remittances to GDP per capita, and vice versa.Chowdhury et al. (2023) further note that excessive consumption and investment in unproductive sectors with transferred money negatively correlate with economic development.
More so, Mehedintu et al. (2020) analyse the effects of migration on remittances using polynomial-time regression and difference equation models.Their findings indicate a continuous rise in GDP and GDP per capita, reflecting an improved standard of living in Romania.Other indicators initially declined sharply because of the global crisis but later experienced gradual growth.Mehedintu et al. (2020) conclude that remittances have been, and remain, a relatively stable financial resource for Romania, similar to other emerging European countries, positively impacting citizens' standard of living despite a downward trend in their value.However, the study by Mehedintu et al. (2020) similarly highlights the negative aspects of remittances, such as dependency on migrant money and the emigration of skilled workers, underscoring the need for government policies to utilise remittances better.Ghani and Morgandi (2023) explore the impact of return migration on labour market outcomes, specifically, wage, consumption, and welfare outcomes for South Asian workers.Ghani and Morgandi's (2023) findings varied in labour returns depending on skill levels and industry.They note a shift in sectoral demand, with manufacturing and services experiencing increased demand compared with agriculture and primary industries.
From the preceding, while studies have examined the impact of migration and labour on the labour market (Privarova et al. 2022;Hutsaliuk et al. 2020), the focus has varied.Privarova et al. (2022) conduct a panel study analysis without addressing gender-specific employment, whereas Hutsaliuk et al. (2020) concentrate on entrepreneurial activity.Other research has explored the effects of COVID-19 on the labour market (Dempster and Zimmer 2020;Withers et al. 2022).Additionally, some studies have investigated economic growth (Sevencan 2023;Chowdhury et al. 2023;Mehedintu et al. 2020).However, the effects of migration on tourism income and inclusive employment remain under-researched.

Gap Identified in the Empirical Literature
Despite extensive research on the effects of remittances and migration, there remain significant gaps in the empirical literature, particularly regarding their impacts on tourism income and gender-specific employment.While several studies have examined the effects of remittances on employment, there is a lack of comprehensive analysis that considers both the short-and long-run effects.The channels in which migration and remittance can affect the economy documented in the empirical literature include short-run, long-run, and causal effects (Sevencan 2023;Chowdhury et al. 2023).The existing research often focuses on specific regions or sectors, leaving a gap in understanding the short-and long-run implications for employment and tourism income (Ghani and Morgandi 2023;Noushad et al. 2022;Mora-Rivera and García-Mora 2021).
Additionally, there is a need to investigate the impacts of differential gender employment.The literature on the gender-specific effects of remittances and migration remains scarce (Rahman et al. 2023;King and Vullnetari 2010).However, some studies focused on women's employment; for example, Rahman et al.'s (2023) study investigates tourism's influence on female employment within the agricultural sector.Another study by Santos (2023) assesses the impact of the return on tourist development in rural Portugal.Our study can be distinguished from the previous studies by examining the effect of migration and remittance on tourism income and gender employment.Likewise, identifying the causal relationships among remittances, migration, employment, and tourism income is critical for effective policy formulation.The existing research, such as by Sevencan (2023) and Chowdhury et al. (2023), tends to analyse these variables but focus on growth; this study aims to fill these gaps by addressing the following research hypotheses: i.
There is no significant relationship among remittances, migration, and aggregate employment in the short and long run.ii.There is no significant relationship among remittances, migration, and tourism income in the short and long term.iii.There is no significant relationship among remittances, migration, and gender-specific employment in the short and long term.iv.There is no causal relationship among remittances, migration, employment, and tourism income.
Based on the gap identified in the empirical literature and the channels of implication, this study uses the autoregressive distributed lag (ARDL) model.The ARDL model is used in this study because of its ability to handle variables integrated into different orders (I(0) and I(1)); this flexibility is crucial as the variables such as remittances, migration, employment, and tourism income may have different integration orders (Pesaran et al. 2001).Additionally, the ARDL model estimates both short-and long-run effects simultaneously, which aligns with this study's objectives to assess these impacts over different time horizons (Ogunwole et al. 2024).The model is also advantageous for small sample sizes, making it ideal for economic data from South Africa, where long-time series data might be limited (Harris and Sollis 2003).Furthermore, including lagged variables captures the dynamic relationships among remittances, migration, employment, and tourism income, providing a comprehensive understanding of these interactions (Pesaran and Shin 1995).This study also used Granger causality tests to complement the ARDL model by establishing the direction of causality among variables; this is essential for analysing the causal relationships among remittances, migration, employment, and tourism income, which is a vital objective of the study (Granger 1969).The tests identify whether one time series can predict another, clarifying the lead-lag relationships.The causal directions are crucial for policy formulation, as they inform whether interventions should focus on enhancing remittance flows to boost employment and tourism, or vice versa (Engle and Granger 1987).
Although this study uses ARDL, its analysis of the asymmetric effect is limited compared with non-linear ARDL (NARDL) (Shin et al. 2014;Niklas and Sadik-Zada 2019;Sadik-Zada and Niklas 2021;Asaleye et al. 2023b).The ARDL model is chosen for its robustness in handling small sample sizes and its ability to distinguish between short-term and long-term relationships among variables.However, ARDL assumes symmetry in the response of the dependent variable to changes in the independent variables.Future research could benefit from using NARDL, as it allows for examining potential relationship asymmetries (Shin et al. 2014).

Model Specification and Estimation Techniques
Following the studies by Asaleye et al. (2023a) and Pal et al. (2022), the income equation is given as follows: In Equation ( 1), I NC is tourism income.Also, REM, EMP, GFC, TRO, I NT, I NF, and MIG are remittances, employment, capital formation, trade openness, interest rate, inflation, and migration.Likewise, the employment equation is given as: In Equation ( 2), all other variables are as defined previously, where EMP is employment and GDP is total output.
Explicitly, Equations ( 1) and ( 2) can be written as follows: Equations ( 3) and ( 4) are in logarithm form; α 0 and β 0 are intercepts or constants, respectively.Also, α 1 , α 2 , α 3 , . . ., α 7 and β 1 , β 2 , β 3 , . . ., β 7 are respective parameters for Equations ( 3) and (4).The respective error terms are µ 1t and µ 2t .(t) is the period of observation.For this study to achieve its objective of investigating the effects of remittance and migration on inclusive gender employment, Equations (3) and ( 4) are modified with the inclusion of male and female employment as follows: Equations ( 5)-( 8) are employment gender-specific equations, and λ 0 , χ 0 , ϕ 0 , and σ 0 are the respective intercepts.Also, λ 1 , λ 2 , λ 3 , . . ., λ 7 , χ 1 , χ 2 , χ 3 , . . ., χ 7 , ϕ 1 , ϕ 2 , ϕ 3 , . . ., ϕ 7 , and σ 1 , σ 2 , σ 3 , . . ., σ 7 are respective parameters.The hypotheses (i), (ii), and (iii) stated in the literature review were achieved by estimating Equations ( 3)-(8) using autoregressive distributed lags (ARDLs).The ARDL equation is given as follows: In Equation ( 9), ∆A t is the first difference of the dependent variable and ∆C t is the first difference of the independent variables.Likewise, ∆A t−1 and ∆C t−1 are the lags of the dependent variable and independent variables, respectively.The parameters b 0 , b 1 , b 2 ,. .., b n capture the short-run dynamics, whereas a 0 captures the long-run behaviour.The speed of adjustment to long equilibrium or error correction term is ψ, which must be negative and less than one.The disturbance term is represented by ε t .Firstly, the lag selection for each model was determined using the Akaike Information Criterion (AIC).The AIC is preferred to other criteria because of the sample size.According to Liew (2004), the most appropriate lag criterion for a small sample size (less than sixty population) is the AIC.The F-statistic with the bound test is used to establish the long-run relationship; when the F-statistic is above the upper bound value, there is evidence of cointegration.The presence of cointegration cannot be decided when the F-statistic is in between the lower and upper bound values.There is no cointegration when the F-statistic is below the lower bound value.Afterwards, diagnostics are carried out to determine if the models are correctly specified; this includes the stability test, histogram normality test, serial correlation test, and heteroscedasticity test.The fourth hypothesis is achieved by examining the causal relationship among the variables.Hypothesising two variables as "X" and "Y", for the variables used in this study, the equation for the causality is given as follows: In Equations ( 10) and ( 11), a 0 and d 0 are the intercepts.b i and c i are the parameters of the lagged variables in Equation ( 10) and h i and g i are the parameters of the lagged variables in Equation ( 11).The order of the lags is represented by "n", and the error terms are e t and v t .Four outcomes are possible in the causality test.Firstly, the bi-directional relationship between "X" and "Y", when c i and g i are statistically different from zero.Secondly, unidirectional from X to Y, when g i is statistically different from zero and c i equals zero.Thirdly, unidirectional from Y to X, when c i is statistically different from zero and g i equals zero.Finally, independence, when c i and g i are not statistically different from zero.Before the ARDL and causality tests, a preliminary analysis was carried out, including summary statistics of the variables and stationarity test.

Data Sources and Definition
The information about the series used in this study is provided in Table 1.The analysis covers 1991 to 2022, chosen based on data availability.

Empirical Justification
This study investigates the effects of migration and remittances on employment, with a particular focus on gender-specific employment and tourism income in South Africa.Migration is a proxy using net migration, which accounts for immigration and emigration, as detailed in Table 1.The population inflows and outflows are essential for a comprehensive understanding of migration's economic impacts.Remittances are measured through personal remittances, including employee compensation and personal transfers.The analysis centres on international tourism income reflected in tourism receipts.Two equations are estimated in this study including the income equation and the employment equation.The employment analysis is conducted in three stages as follows: the aggregate employment equation, the male employment equation, and the female employment equation.This study incorporates respective control variables based on their relevance in the empirical literature, including investment in capital formation (proxied by gross fixed capital formation), trade openness, inflation, and interest rates, as shown in Table 1.The following studies justify the inclusion of the variables in the tourism income equation: Hussain (2023), Khalid et al. (2020), Ghosh (2022), Seetanah and Fauzel (2023), Kumar et al. (2020), Salazar (2022), and Mora-Rivera and García-Mora (2021).The employment equation is supported by Asaleye et al. (2021), Holland (1986), Bailey et al. (1980), Amuedo-Dorantes and Pozo (2006), Oliinyk et al. (2021), Conard (2023), Akkemik (2007), and Asaleye and Strydom (2023).
This study uses trade openness to capture the effect of external forces.Trade openness refers to the extent to which a country or region engages in and depends on international trade.It was initially defined as the degree of reliance on foreign trade (Ding et al. 2022); this study computes trade openness as the summation of import and export divided by the GDP following the approach used in the empirical literature (Dorn et al. 2022;Ding et al. 2022).Exchange rate pass-through (ERPT) describes how changes in exchange rates influence domestic prices of imported goods and services; this occurs through channels like producer currency pricing, local currency pricing, and pass-through of imported inputs (Campa and Goldberg 2005).ERPT varies across countries and over time because of factors like exchange rate flexibility and the share of imports in the economy (Gopinath and Itskhoki 2010).ERPT often pass through to inflation (Choudhri and Hakura 2006).The exchange rate was not included as a control variable.This study included inflation because of its ability to serve as a crucial determinant in employment and tourism income models (Khalid et al. 2020;Ghosh 2022).High inflation may lead to higher nominal wages and impact labour demand while also affecting investment decisions and overall economic growth (Baxa et al. 2015).Similarly, inflation influences tourism income through its impact on exchange rates and domestic demand, making it essential to include it in tourism income models (Gopinath and Itskhoki 2010).
The choice of type of interest rate is subjective in econometric modelling, particularly in structure dynamics and macroeconomic variables (Ang and Piazzesi 2003;Sims 2002;Hamilton 1996;Oladipo et al. 2019).This study advocates its focus on interest payments as a mechanism to influence employment and tourism income; it is imperative to consider the broader macroeconomic implications, particularly concerning inflation changes over time (Roubini and Backus 1998;Ascari and Sbordone 2014).While real interest rates are traditionally used as a monetary policy tool to manage inflation, controlling for inflation reveals a relationship between interest rates and employment outcomes (Blanchard and Fischer 1989).High inflation can pressure borrowers, reducing their purchasing power and making it increasingly challenging to meet debt obligations (Romer 2019).By reducing interest payments, policymakers can provide immediate relief to borrowers, mitigating the adverse effects of inflation on employment stability (Krugman and Wells 2015); this approach stresses the importance of addressing inflation and interest payments in this regard, creating collaborative effects that contribute to growth and employment creation.Furthermore, by stabilising inflation expectations and reducing uncertainty, policymakers can promote a conducive investment and employment expansion environment, resulting in overall economic stability and social welfare (including improvement in tourism income) (Bernanke 2007).The information and empirical justification for the series are presented in Table 1.

Results and Discussion
This section presents the preliminary analyses and the ARDL results.

Preliminary Analyses
Table 2 shows the descriptive statistics and correlation analysis.Evidence from the results indicates that GDP has the highest mean value, which stands at 11.406, and REM has the lowest mean value, with a figure of −0.759.The mean values for EMP, FEMP, GFC, INC, INF, INT,MEMP,MIG,and TRO are 1.666,1.590,10.617,3.707,0.756,10.822,1.739,1.104,and −0.309, respectively.The standard deviation of the variables is as follows: EMP (0.030), FEMP (0.020), GDP (0.195), GFC (0.218), INC (0.155), INF (0.246), INT (0.349), MEMP (0.044), MIG (4.721), REM (0.225), and TRO (0.076).The correlation value among the variables is less than 70 per cent except for the value between EMP and FEMP, EMP and MEMP, FEMP and MEMP, and MEMP and TRO.The employment variables EMP, FEMP, and MEMP were analysed using different models.The correlation analysis's outcome did not show a suspicion of multicollinearity in the models.Table 3 presents the unit root result of the variables; two stationarity approaches were used, the augmented dickey fuller (ADF) and the Phillips-Perron (PP).In the ADF outcome, all series are integrated of order one except INF and REM, which are stationary at the level, including intercept only and intercept and trend, respectively.The PP outcomes show similar results to the ADF, except that INF is stationary at the level with the inclusion of intercept and trend.Based on the outcomes, it is concluded that the variables used in this study are integrated of order one and zero; this helps to choose the ARDL estimation above other approaches for long-run analysis.a, b and c indicate significance @ the levels of 1 per cent, 5 per cent and 10 per cent, respectively.Source: Authors' computation.
Table 4 presents the bound test and cointegration result for the three models.Model 1 is for the aggregate employment indicator.Models 2 and 3 are for male and female employment indicators.LB represents the lower boundary, and UB represents the upper.If the F-statistic value is greater than the UB, there is cointegration.However, it cannot be decided when the F-statistic value is between the LB and UB.And, finally, when the F-statistic value is lower than the UB, there is no cointegration.In the case of models 1, 2, and 3, the F-statistic values are 9.261418, 9.280848, and 4.360929, respectively.Meanwhile, the UB at 10 per cent, 5 per cent, 2.5 per cent, and 1 per cent are as follows: 2.89, 3.21, 3.51, and 3.9, respectively.Given this, the values of the F-statistic in the three models are greater than the UB; hence, the presence of cointegration is established in the three models.Appendix A presents the model selections in Figures A1-A6.

ARDL Short-and Long-Run Results
Table 5 presents the short-and long-run results for model 1.When tourism income (INC) is used as the dependent variable, remittance (REM), employment (EMP), gross capital formation (GFC), trade openness, and (TRO) are statistically significant in the long run (LR).Interest rate (INT), inflation rate (INF), and migration (MIG) are not statistically significant.REM and GFC positively impact INC in the LR, meaning a one percentage increase in REM and GFC will increase INC by 0.09 per cent and 0.70 per cent, respectively.The findings align with the study of Abdulai (2023), who showed that the GDP growth rate has a long-run relationship with remittance inflows in Ghana.EMP and TRO negatively impact INC in the LR, meaning a one percentage increase in EMP and TRO will lead to a decrease of 1.2 per cent and 0.82 per cent, respectively, in INC.The findings contradict the study by Balcilar et al. (2021), which documented no significant relationship between employment and tourism income.The authors further stated that failure for employment not to be significant might be attributed to economic and demographic considerations in OECD nations and employment business cycles.In the short run (SR), REM, EMP, GFC, TRO, and MIG are statistically significant, while INT and INF are insignificant.In addition, REM, GFC, and MIG are positively related to INC, meaning a one percentage increment in REM, GFC, and MIG will cause an increase of 0.08 per cent, 0.89 per cent, and 0.001 per cent in INC, respectively.At the same time, one percentage increase in EMP and TRO will cause a decrease of 1.01 per cent and 0.35 per cent in INC, respectively.
Likewise, in Table 5, when EMP is used as the dependent variable, in the LR, REM, GDP, GFC, TRO, INT, and INF are statistically significant.REM, GFC, and INT have a positive relationship with EMP, meaning a one percentage increase in REM, GFC, and INT will lead to a 0.06 per cent, 0.74 per cent, and 0.10 per cent increase in EMP.The outcome of remittance on employment contradicts the findings by Saani et al. (2023), who reported that remittance negatively affects employment in the LR.At the same time, GDP, TRO, and INF have a negative relationship with EMP, meaning a one percentage increase in GDP, TRO, and INF will cause a 1.03 per cent, 0.54 per cent and 0.04 per cent decrease in EMP, respectively.The outcome of the negative relationship between employment and GDP is in line with the study of Asaleye et al. (2021), who stressed the need for board-based growth in developing economies.GDP, GFC, TRO, INT, INF, and MIG are statistically significant in the SR.REM is not significant, which aligns with the work of Hossain (2020), who found that remittances do not affect economic activities in Malaysia.The outcome indicates that one percentage increase in GFC, INT, and MIG will cause an increase of 0.67 per cent, 0.09 per cent, and 0.0009 per cent in EMP, respectively.Also, one percentage increase in GDP, TRO, and INF will cause a decrease of 0.75 per cent, 0.18 per cent, and 0.02 per cent in EMP, respectively.ECM measures the speed of adjustment to the long-run equilibrium; when INC and EMP are used as dependent variables, the ECM values are 0.20 and 0.90 per cent, respectively.The outcome of the diagnostic checks for both equations shows that the models are specified correctly.Likewise, the R-squared of more than 50 per cent indicates good fitness, and Durbin Waston (DW stat) is close to 2, which signifies no autocorrelation in the models.Table 6 presents the short-and long-run results for model 2. When income (INC) is used as the dependent variable, remittance (REM), employment (EMP), gross capital formation (GFC), trade openness (TRO), the interest rate (INT), and the inflation rate (INF) are statistically significant in the long run (LR).Migration (MIG) is not statistically significant.This finding contradicts the study by Simionescu et al. (2016); the authors documented a negative correlation between net migration and income in the SR.In the LR, they reported a positive relationship.REM and GFC positively impact INC in the LR, meaning a one percentage increase in REM and GFC will increase INC by 0.03 per cent and 0.69 per cent, respectively.The impact of remittance on income aligns with the study by Abdulai (2023).MEMP, TRO, INT, and INF negatively impact INC in the LR, meaning a one percentage increase in MEMP, TRO, INT, and INF will lead to a decrease of 1.06 per cent, 0.64 per cent, 0.03 per cent, and 0.03 per cent, respectively, in INC.This finding with respect to income and employment aligns with the findings of Asaleye et al. (2021).In the short run (SR), REM, EMP, GFC, TRO, INT, INF, and MIG are statistically significant.In addition, REM, GFC, and MIG are positively related to INC, meaning a one percentage increment in REM, GFC, and MIG will cause an increase of 0.05 per cent, 0.87 per cent, and 0.001 per cent in INC, respectively.At the same time, one percentage increase in MEMP, TRO, INT, and INF will cause a decrease of 1.30 per cent, 0.18 per cent, 0.04 per cent, and 0.03 per cent in INC, respectively.The diagnostic checks show that the model is correctly specified; the stability test is in the Appendix B section in Figures A7 and A8.
Likewise, in Table 6, when MEMP is used as the dependent variable, in the LR, REM, GDP, GFC, TRO, INT, and MIG are statistically significant.REM, GFC, INT, and MIG have a positive relationship with MEMP, meaning a one percentage increase in REM, GFC, INT and MIG will lead to a 0.03 per cent, 0.76 per cent, 0.09 per cent, and 0.001 per cent increase in EMP.At the same time, GDP and TRO have a negative relationship with MEMP, meaning a one percentage increase in GDP and TRO will cause a 1.06 per cent and 0.57 per cent decrease in MEMP, respectively.REM, GDP, GFC, TRO, INF, and MIG are statistically significant in the SR.INT is not significant.This outcome indicates that one percentage increase in REM, GFC, INT, and MIG will cause an increase of 0.02 per cent, 0.62 per cent and 0.0009 per cent in MEMP, respectively.Also, one percentage increase in GDP, TRO, and INF will cause a decrease of 0.66 per cent, 0.10 per cent, and 0.02 per cent in MEMP, respectively.The ECM measures the speed of adjustment to the long-run equilibrium; when INC and MEMP are used as dependent variables, the ECM values are 0.01 and 0.009 per cent, respectively.The outcome of the diagnostic checks for both equations shows that the models are specified correctly.Likewise, the R-squared of more than 50 per cent indicates good fitness, and Durbin Waston (DW stat) is close to 2, which signifies no autocorrelation in the models.
Table 7 presents the short-and long-run results for model 3.When income (INC) is used as the dependent variable, remittance (REM), female employment (FEMP), gross capital formation (GFC), and trade openness (TRO) are statistically significant in the long run (LR).Interest rate (INT), inflation rate (INF) and migration (MIG) are not statistically significant.REM and GFC positively impact INC in the LR, meaning a one percentage increase in REM and GFC will increase INC by 0.23 per cent and 0.66 per cent, respectively.The findings align with the study by Meyer and Shera (2017), which showed a positive relationship between remittance and income in six high-remittance-receiving countries.The diagnostic checks show that the model is correctly specified; the stability test is in the Appendix B section in Figures A9 and A10.
FEMP and TRO negatively impact INC in the LR, meaning a one percentage increase in FEMP and TRO will lead to a decrease of 1.36 per cent and 0.08 per cent, respectively, in INC.In the short run (SR), REM, FEMP, GFC, and TRO are statistically significant.INT, INF, and MIG are not significant.In addition, REM and GFC are positively related to INC, meaning a one percentage increment in REM and GFC will cause an increase of 0.12 per cent and 0.92 per cent in INC, respectively.At the same time, a one percentage increase in FEMP and TRO will cause a decrease of 0.92 per cent and 0.47 per cent in INC, respectively.a, b, and c indicate significance @ levels of 1 per cent, 5 per cent, and 10 per cent, respectively.Source: Authors' computation.
Likewise, in Table 7, when FEMP is used as the dependent variable, in the LR, REM, GDP, GFC, TRO, INT, and INF are statistically significant.MIG is insignificant.REM, GFC, and INT have a positive relationship with FEMP, meaning a one percentage increase in REM, GFC, and INT will lead to a 0.09 per cent, 0.65 per cent, and 0.10 per cent increase in FEMP.At the same time, GDP, TRO, and INF have a negative relationship with FEMP, meaning a one percentage increase in GDP, TRO, and INF will cause a 0.88 per cent, 0.48 per cent, and 0.05 per cent decrease in FEMP, respectively.GDP, GFC, TRO, and INT are statistically significant in the SR.REM, INF and MIG are not significant.The outcome indicates that one percentage increase in GFC and INT will cause an increase of 0.69 per cent and 0.09 per cent in FEMP, respectively.Also, one percentage increase in GDP and TRO will cause a decrease of 0.76 per cent and 0.24 per cent in FEMP, respectively.The ECM measures the speed of adjustment to the long-run equilibrium; when INC and FEMP are used as dependent variables, the ECM values are 0.78 and 0.86 per cent, respectively.The outcome of the diagnostic checks for both equations shows that the models are specified correctly.Likewise, the R-squared of more than 50 per cent indicates good fitness, and Durbin Waston (DW stat) is close to 2, which signifies no autocorrelation in the models.The diagnostic checks show that the model is correctly specified; the stability test is in the Appendix B section in Figures A11 and A12.
Investing in tourism is important for economies as it serves as a key driver of economic growth and recovery.Several studies have revealed that tourism can add to higher levels of economic activity, especially when it is a major contributor to a country's Gross Domestic Product (GDP).Using Granger causality tests and cointegration, Balaguer and Cantavella-Jorda's (2002) examined the role of tourism in the long-run economic development of Spain.Their study confirmed the tourism-led growth hypothesis, revealing that tourism plays an important role in long-term economic growth.The vastness of tourism means that the potential for employment and income can be explored in several ways.
Tourism creates both direct and indirect employment opportunities across different sectors.Direct employment is created through jobs that are directly involved in offering tourism experiences, for example, tour operators and travel agencies.Indirect employment, on the other hand, involves extra jobs generated by the tourism-related economy.For example, marketing agencies that market tourism services and producers of souvenirs.The industry's ability to create numerous jobs in related sectors highlights its value as a key contributor to economic development in countries.Evidence has been established on the various ways that income is generated through tourism.Through foreign exchange earnings, reserves are grown to finance imports.Where jobs are created, the earnings of populations increase, and this spending can further contribute to income growth.
Concerning the sources of income for tourism activities, remittances have played a significant role in facilitating tourism-related expenditure.For example, for family reunions in locations outside their home countries.This allocation of remittances towards tourism activities benefits the tourism industry of respective countries and has the potential to contribute positively to overall economic growth by providing households with extra income to use for investment, consumption, or starting a business.While the influx of funds is not consistent, it is continuous as it stimulates economic activity such as participation in tourism activities.With regard to employment, while tourism is said to create employment opportunities, remittances have also been found to have an impact on employment, albeit inconsistently.Some positive impacts include job development and reduced unemployment in the long run, while some negative impacts comprise a decline in labour market participation of certain groups and reduced labour supply.The impact varies depending on several factors such as the context and characteristics of the receiving countries.

Granger Causality Result
Table 8 presents the causality result.The chosen variables are indicators of employment, remittances, and migration.The result shows a bi-directional relationship between the following variables: FEMP and EMP, MEMP and EMP, and MEMP and FEMP.Likewise, there is a unidirectional relationship between REM and MIG, and MIG and INC, flowing from REM to MIG and from MIG to INC, respectively.Among the other variables tested, there is evidence of independence.a, b, and c indicate significance @ levels of 1 per cent, 5 per cent, and 10 per cent, respectively.Source: Authors' computation.

Summary of This Work
This study investigates the effect of remittance and migration on tourism income and gender-specific employment in the South African economy, using ARDL and Granger causality.Three models were established in this study, each with two equations.Each model comprises a tourism income equation and an employment equation.The first model is for aggregate employment.The second model is for male employment, and the third is for female employment.The key findings are as follows: Firstly, remittance appears as a significant driver of tourism income, showing a positive influence in both the short and long run across all three models; this underlines the importance of remittance inflows in stimulating economic activity within the tourism sector and its potential to promote growth and development.Additionally, migration exhibits a mixed relationship with tourism income, being statistically insignificant in the long run for female employment but positively affecting tourism income in the short run for male and aggregate employment.
Furthermore, the analysis of gender-specific employment reveals differential effects of remittance and migration.Remittance positively impacts male employment in both the short and long run, while its effect on female employment is significant only in the long run.On the other hand, migration positively influences male employment in both the short and long run.Still, it exhibits a significant relationship with female employment in the short and long run.The Granger causality results show a bi-directional relationship among all employment indicators: the aggregate and male and female employment.A unidirectional relationship is observed between remittance and migration, as well as migration and income, suggesting the flow of causality from remittance to migration and migration to income, respectively.
As indicated by the result, remittance positively influences male employment in both the short and long run across all models.Its effect on female employment is only significant in the long run; these observed differences in the impact of remittances on male and female employment are likely from existing gender roles and labour market structures.Traditionally, men are likely to engage in immediate employment opportunities because of societal expectations of being the primary source of income, hence the positive short-term effect.Conversely, women might face barriers such as limited access to education and childcare responsibilities, which lower their economic participation.Over the long term, remittances could provide resources that gradually alleviate these barriers, allowing more women to seek employment.This gendered discrepancy highlights the need for policies addressing structural inequalities, ensuring that remittance benefits can equitably support both men and women in the labour market.

Appendix A. Model Selection
Akaike Information Criteria (top 20 models)  When EMP is used as the dependent variable.
Akaike Information Criteria (top 20 models)  Model 2 (Male Equation) When INC is used as the dependent variable.
Akaike Information Criteria (top 20 models)  When MEMP is used as the dependent variable.
When FEMP is used as the dependent variable

Appendix B. Stability Result
Model 1 (Aggregate Equation) When INC is used as the dependent variable.When EMP is used as the dependent variable.
When EMP is used as the dependent variable   Model 2 (Male Equation) When INC is used as the dependent variable.When MEMP is used as the dependent variable.
When MEMP is used as the dependent variable  When FEMP is used as the dependent variable.
When FEMP is used as the dependent variable

Figure A1 .
Figure A1.Model selection when INC is used as the dependent variable for model 1.Source: Authors' computation.

Figure A1 .
Figure A1.Model selection when INC is used as the dependent variable for model 1.Source: Authors' computation.

Figure A3 .
Figure A3.Model selection when INC is used as the dependent variable for model 2. Source: Authors' computation.

Figure A2 .
Figure A2.Model selection when EMP is used as the dependent variable for model 1.Source: Authors' computation.

Figure A3 .
Figure A3.Model selection when INC is used as the dependent variable for model 2. Source: Authors' computation.Figure A3.Model selection when INC is used as the dependent variable for model 2. Source: Authors' computation.

Figure A3 .
Figure A3.Model selection when INC is used as the dependent variable for model 2. Source: Authors' computation.Figure A3.Model selection when INC is used as the dependent variable for model 2. Source: Authors' computation.

Figure A5 .
Figure A5.Model selection when INC is used as the dependent variable for model 3. Source: Authors' computation.

Figure A4 .
Figure A4.Model selection when MEMP is used as the dependent variable for model 2. Source: Authors' computation.

Figure A5 .
Figure A5.Model selection when INC is used as the dependent variable for model 3. Source: Authors' computation.Figure A5.Model selection when INC is used as the dependent variable for model 3. Source: Authors' computation.

Figure A5 .
Figure A5.Model selection when INC is used as the dependent variable for model 3. Source: Authors' computation.Figure A5.Model selection when INC is used as the dependent variable for model 3. Source: Authors' computation.

Figure A6 .Figure A7 .
Figure A6.Model selection when FEMP is used as the dependent variable for model 3. Source: Authors' computation.Appendix B. Stability ResultModel 1 (Aggregate Equation)When INC is used as the dependent variable

Figure A6 .
Figure A6.Model selection when FEMP is used as the dependent variable for model 3. Source: Authors' computation.

Figure A6 .Figure A7 .
Figure A6.Model selection when FEMP is used as the dependent variable for model 3. Source: Authors' computation.Appendix B. Stability ResultModel 1 (Aggregate Equation)When INC is used as the dependent variable

Figure A7 .
Figure A7.Stability when INC is used as the dependent variable for model 1.Source: Authors' computation.Figure A7.Stability when INC is used as the dependent variable for model 1.Source: Authors' computation.

Figure A8 .Figure A9 .
Figure A8.Stability when EMP is used as the dependent variable for model 1.Source: Authors' computation.Model 2 (Male Equation)When INC is used as the dependent variable

Figure A8 .
Figure A8.Stability when EMP is used as the dependent variable for model 1.Source: Authors' computation.

Figure A8 .Figure A9 .
Figure A8.Stability when EMP is used as the dependent variable for model 1.Source: Authors' computation.Model 2 (Male Equation)When INC is used as the dependent variable

Figure A9 .
Figure A9.Stability when INC is used as the dependent variable for model 2. Source: Authors' computation.Figure A9.Stability when INC is used as the dependent variable for model 2. Source: Authors' computation.

Figure A10 .Figure A11 .
Figure A10.Stability when MEMP is used as the dependent variable for model 2. Source: Authors' computation.Model 3 Female Equation When INC is used as the dependent variable

Figure A10 .Figure A11 .
Figure A10.Stability when MEMP is used as the dependent variable for model 2. Source: Authors' computation.Model 3 Female EquationWhen INC is used as the dependent variable.

Figure A12 .
Figure A12.Stability when FEMP is used as the dependent variable for model 2. Source: Authors' computation.

Table 1 .
Information about the series.

Table 2 .
Descriptive statistics and correlation analysis.

Table 3 .
Unit root test results.

Table 4 .
Bound test and cointegration result.

Table 5 .
Short-and long-run results for model 1.

Table 6 .
Short-and long-run results for model 2.

Table 7 .
Short-and long-run results for model 3.
Model selection when MEMP is used as the dependent variable for model 2. Source: Authors' computation.
Model 3 Female EquationWhen INC is used as the dependent variable.