The effect of mismatch on labor income in Indonesia

 Corresponding author: Ida Ayu Made Dwi Utari. Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution License 4.0. The effect of mismatch on labor income in Indonesia Ida Ayu Made Dwi Utari *, Putu Ayu Paramitha Purwanti, Ni Made Tisnawati and I Nyoman Mahaendra Yasa Master of Economics Study Program, Faculty of Economics and Business, Udayana University, Bali, Indonesia World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 Publication history: Received on 29 May 2025; revised on 08 July 2025; accepted on 10 July 2025 Article DOI: https://doi.org/10.30574/wjarr.2025.27.1.2594 Abstract Problems that often arise in the mismatch phenomenon and the impact of mismatches received by workers tend to emphasize the wage gap or the difference in income between individuals who meet the qualifications and individuals who do not meet the qualifications. In 2018, the workforce experienced undereducation of 4.6% and overeducation of 27.9%. Meanwhile, field of study mismatch occurs in 68.4% of the workforce in Indonesia. This research aims to analyze the vertical mismatch phenomenon that occurs in Indonesia and the influence of mismatch in the workforce on differences in workforce income. This research was conducted in Indonesia because it used SAKERNAS Micro data. The independent variables in this research are education level, age, gender, working hours, work location and employment status. The dependent variables in this research are income and productivity as intervening variables. The method used in collecting this data is non-participant observation. The analysis technique used in this research is SMART PLS. The results of the analysis show that education level has a significant effect on labor income in Indonesia. Age has a significant effect on labor income in Indonesia. Gender has a significant effect on labor income in Indonesia. Working hours have a significant effect on labor income in Indonesia. Location of work has a significant effect on labor income in Indonesia. Education level, age, gender, working hours and working location together have a significant effect on labor income in Indonesia in 2021. Productivity has a significant effect on labor income in Indonesia. Productivity mediates the influence of the variables Educational Level, age, gender, working hours and location of work on labor income in Indonesia in 2021. Keywords: Mismatch; Labor; SAKERNAS; Income 1. Introduction Economic growth and development are one of the benchmarks for a country's development and are a very important concern. Therefore, economic growth and development are a priority for all countries (Todaro and Smith, 2011). The objectives of economic development include increasing national income and increasing productivity levels both in the macro and micro sectors (Adisasmita, 2013). The factors that influence a country's economic development consist of two factors, namely economic and non-economic factors. Economic factors that influence economic growth and development include natural resources, human resources, capital resources, and expertise. While non-economic factors include socio-cultural conditions in society, political conditions, institutional structures, and systems that develop and apply in society (Yannisa, 2017). Education is a basic key to forming a more productive generation of human resources to build the country's economy by contributing using the abilities possessed by the individual (Borjas, 2016). Open access to information on various sources of knowledge in Indonesia and the world can increase the number of people attending school in general. The increasing number of people who receive education will change the structure of the Indonesian workforce. Where the number of educated workers will increase (Hasibuan and Handayani, 2021). Figure 1 shows changes in the structure of the workforce in Indonesia, comparing the educational structure of the workforce in 2010 and 2022. World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 1191 Source: Badan Pusat Statistik (2022) Figure 1 Labor Force According to Highest Education Completed Through Figure 1. shows the changes in the structure of the workforce education in Indonesia. The striking difference that occurred in 2010 and 2022 lies in the level of education that did not attend school and did not finish elementary school which is decreasing and the increase in the workforce that completed junior high school to university level. This shows that the number of educated workers has increased in the last 12 years. The increase in the educational attainment of the workforce can indicate that the human capital they have is getting higher. With the increasing ability and skills of the workforce, it is expected to increase workforce productivity and the rate of return on investment in education (Becker, 1975; Schultz, 1961). Source: Badan Pusat Statistik (2024) Figure 2 Latest Education Completed by Unemployed in Indonesia, 2023 There is an issue in the structural change of the workforce that is increasing every year. The level of education possessed by workers will be a signal for employers who are looking for jobs with appropriate skills. Mismatch of educational qualifications indicates a mismatch between the availability of knowledge and skills with what is needed (Allen and de Weert, 2007). The difficulty of the process of adjusting the demand and supply of labor causes a skills gap. Mismatch is a gap between the number of workers according to the highest level of education completed with the need for workers for certain types of jobs and levels of education (skills) in the labor market. Mismatch can occur vertically or horizontally (Dilly and Papuling, 2021). Vertical mismatch is a mismatch due to the educational qualifications or skills possessed by workers being lower or higher than those required. When the educational qualifications possessed are higher, the workforce experiences overeducation, conversely if the educational qualifications possessed are lower, it is called undereducation. Facts from the Organisation for Economic Co-operation and Development (OECD) state that in most developed countries there is increasing concern about the potential for over-investment in education because individuals with high levels of education can be employed in jobs with lower requirements. World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 1192 The increase in the quality of the workforce offered is not followed by an increase in demand for workers who require high skills, which can cause a mismatch. Indications of mismatch in the labor market as in Figure 2 show that in 2024, which shows high unemployment at the high school level of 2,514,481 to universities or colleges of 787,973, where this phenomenon is a very serious problem. Fathkul (2013) stated that unemployment rates can be used as an indicator to show the needs of the labor market with the education completed by the unemployed, where Figure 1.2 shows an indication of a mismatch between the level of education and the job structure in Indonesia. Although in Figure 1 there is an increase in educated workers, it is not supported by an increase in jobs that are in accordance with qualifications. The population is a large capital to increase national production if there are sufficient jobs available, but if many people are unemployed due to the unavailability of jobs, it will result in a decrease in the level of community welfare. Problems that often arise in the phenomenon of mismatch and the impact of mismatch received by the workforce tend to emphasize the wage gap or the difference in income between qualified individuals and unqualified individuals. Research conducted by Hartog (2000) states that what happens in the phenomenon of labor mismatch shows that highly educated individuals tend to receive lower wages and are less satisfied with their jobs than their educated and matched colleagues but interest in analyzing the impact of labor mismatch goes beyond individual effects because these negative outcomes can spread to companies and the economy as a whole; in fact, some works show that lower wages or lower levels of job satisfaction can lead to lower productivity at the aggregate level (Lazear 2000; Böckerman and Ilmakunnas 2012). . Safuan and Nazara (2005) showed that overeducated workers are negatively correlated with their income, while undereducated workers are positively correlated with their income. This is in line with the OECD facts where workers who work in jobs that do not match their educational qualifications will receive lower wages (wage penalty) compared to workers with the same qualifications and have suitable jobs (Béduwé and Giret, 2011; Dolton and Silles, 2008; Nordin et al., 2010) In a study conducted in Brazil, Reis (2017) stated that vertical mismatch, which means a mismatch due to the educational qualifications or skills possessed by workers being lower or higher than those required, has different wage effects (Sitorus and Wicaksono, 2020). Workers with a higher level of education than their job requires (overeducation) will receive excess wages (wage premium), while workers with a lower level of education (undereducation) will receive a wage penalty or the effect of lower income (Friska and Damayanti, 2021). Companies that employ workers with higher educational qualifications than needed can increase the productivity of the workforce (Sabarofek and Anggraeni, 2022). However, on the other hand, workers cannot use all their abilities, causing dissatisfaction at work and resulting in decreased productivity (Allen and van der Velden, 2001; Battu et al., 1999; Sam, 2019). Dissatisfaction and wage effects experienced by workers can encourage them to try to find other jobs (Allen and van der Velden, 2001; Béduwé and Giret, 2011; Zakariya, 2017). This can increase employee turnover in the company. The high turnover rate is a loss for the company because the company will need additional costs for recruitment and retraining. In addition, new employees are not yet able to work effectively, resulting in lower productivity (Urbancova and Linhartova, 2011). Prabowo (2014) stated that mismatch within the individual can cause a decrease in productivity due to stress that arises because work does not match the individual's interests and the wage penalty effect which is quite a burden on the workforce. The micro mismatch effect does not only stop at individuals, but the company's internal will be affected if there is a mismatch between work and type of work. Wulandari (2021) stated that the effect of workforce mismatch on companies is low productivity and the risk of workforce turnover which also contributes to the decline in the company's productivity. The difference in company or sectoral human capital and low-skilled human resources will have an impact on the training costs incurred by the company (Mavromaras and Sloane, 2015). If this happens on a macro scale, it will certainly affect the productivity of a country or region, which is supported by the statement from the International Labor Organization (2018) namely that mismatch affects economic growth through the reduction or non-utilization of human capital. Research related to the mismatch phenomenon in Indonesia is still very little. Therefore, research related to mismatch, especially vertical mismatch, needs to be carried out considering that the impact greatly affects the quality of the workforce. Through the analysis of macroeconomic and microeconomic impacts caused by labor mismatch, it can reduce labor productivity due to wage penalties and feelings of displeasure in doing work that does not match academic qualifications (Ghaffarzadegan, et al, 2017). Supported by the fact that the phenomenon of qualification mismatch is quite emphasized on income differences. In 2018, the workforce experienced undereducation of 4.6% and overeducation of 27.9%. Meanwhile, the field of study mismatch occurred in 68.4% of the workforce in Indonesia. The results of research by Hasibuan and Handayani, 2021 showed that there was a wage premium of 5.24% -6.24% for workers who experienced undereducation, a wage penalty of 6.26% - 7.505 was obtained by workers who experienced overeducation, while those who experienced field of study mismatch were 5.89% -6.80%. As with the research of Enawati et al. (2021) conducted in Indonesia related to World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 1193 qualification mismatch where the results showed that there was a wage premium of 5.24% - 6.24% for workers who experienced undereducation and a wage penalty of 6.26% - 7.50% was obtained by workers who experienced overeducation. Meanwhile, Hersch and Xiao (2016) stated that mismatch has different wage impacts between genders. Different research results were also found in research conducted by Suryono and Pitoyo (2020) which stated that higher education graduates are absorbed in types of jobs that match the worker's education level. This study will further discuss the impact of mismatch on wages earned by workers in Indonesia. 2. Research methods The research design used in this study is a quantitative research design in the form of associative. This study was conducted in Indonesia because it uses SAKERNAS Micro data. The subjects of this study were from households in villages and cities from each village and city listed in the Micro Data provided by the Sakernas institution. The object of this study is mismatch, and various characteristics that affect labor mismatch in Indonesia. The Independent Variables in this study are: Variable of Interest, namely education level (X1). Control Variable (Xi), namely various characteristics that affect Mismatch, namely, Age, Gender; Working Hours; Work Location; Employment Status the Dependent Variable in this study is Income ln (Y2). The Intervening Variable in this study is Productivity (Y1). The method used in collecting this data is by means of non-participant observation. Data analysis in this study uses the Partial Least Square (PLS) approach. 3. Results and discussion Table 1 Validity Testing based on Outer Loading Working hours Gender Department of Education Work Location Labor Income Productivity Level of education Age Working hours 1.000 Gender 1.000 Department of Education 1.000 Work Location 1.000 Labor Income 1.000 Productivity 1.000 Level of education 1.000 Age 1.000 Source: Processed data, 2025 World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 1194 Figure 3 Validity Testing based on Outer Loading Based on the outer loading validity test in Table 1 and Figure 3, it is known that all outer loading values are > 0.7, which means that they have met the validity requirements based on the outer loading value. Furthermore, validity testing is carried out based on the average variance extracted (AVE) value, and reliability based on Cronbach's alpha (CA) and composite reliability (CR). Table 2 Validity Testing based on Average Variance Extracted (AVE); Reliability based on Cronbach’s Alpha (CA) and Composite Reliability (CR) Cronbach's Alpha Composite Reliability Average Variance Extracted (AVE) Working hours 1.000 1.000 1.000 Gender 1.000 1.000 1.000 Department of Education 1.000 1.000 1.000 Work Location 1.000 1.000 1.000 Labor Income 1.000 1.000 1.000 Productivity 1.000 1.000 1.000 Level of education 1.000 1.000 1.000 Age 1.000 1.000 1.000 Source: Processed data, 2025 The recommended AVE value is above 0.5. It is known that all AVE values are > 0.5, which means that they have met the validity requirements based on AVE. Furthermore, reliability testing is carried out based on the composite reliability (CR) value. The recommended CR value is above 0.7. It is known that all CR values are > 0.7, which means that they have met the reliability requirements based on CR. Furthermore, reliability testing is carried out based on the Cronbach's alpha (CA) value. The recommended CA value is above 0.7. It is known that all CA values are > 0.7, which means that they have met the reliability requirements based on Cronbach's alpha. Furthermore, discriminant validity testing is carried out using the Fornell-Larcker approach. Table 4.3 presents the results of the discriminant validity test. World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 1195 Table 3 Discriminant Validity Testing: Fornell and Larcker Working hours Gender Department of Education Work Location Labor Income Productivity Level of education Age Working hours (1.000) Gender -0.258 (1.000) Department of Education 0.038 0.023 (1.000) Work Location 0.518 -0.303 -0.009 (1.000) Labor Income 0.399 -0.221 0.096 0.227 (1.000) Productivity -0.107 0.017 -0.222 -0.005 -0.180 (1.000) Level of education 0.128 -0.008 0.407 -0.043 0.257 -0.477 (1.000) Age 0.074 0.009 0.017 0.140 0.068 0.005 -0.203 (1.000) Note: The values between “()” are the square roots of AVE.; Source: Processed data, 2025 In discriminant validity testing, the AVE square root value of a latent variable is compared with the correlation value between the latent variable and other latent variables. It is known that the AVE square root value for each latent variable is greater than the correlation value between the latent variable and other latent variables. So, it is concluded that it has met the requirements of discriminant validity. Table 4 Discriminant Validity Testing: HTMT Working hours Gender Department of Education Work Location Labor Income Productivity Level of education Gender 0.258 Department of Education 0.038 0.023 Work Location 0.518 0.303 0.009 Labor Income 0.399 0.221 0.096 0.227 Productivity 0.107 0.017 0.222 0.005 0.180 Level of education 0.128 0.008 0.407 0.043 0.257 0.477 Age 0.074 0.009 0.017 0.140 0.068 0.005 0.203 Source: Processed data, 2025 Based on the results of the discriminant validity test using the HTMT approach, it is known that all values are <0.9, which means that it is concluded that the discriminant validity requirements based on the HTMT approach have been met. World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 1196 Table 5 Path Coefficient Test and Significance of Influence Original Sample (O) Sample Mean (M) Standard Deviation (STDEV) T Statistics (|O/STDEV|) P Values RSquares SRMR Working Hours -> Productivity -0.042 -0.042 0.003 15.917 0.000 0.238 0.000 Gender -> Productivity 0.006 0.006 0.002 3.076 0.001 Education Department - > Productivity -0.024 -0.024 0.003 7.320 0.000 Work Location -> Productivity 0.010 0.010 0.002 4.234 0.000 Education Level -> Productivity -0.480 -0.480 0.003 181.233 0.000 Age -> Productivity -0.090 -0.090 0.002 46.206 0.000 Working Hours -> Labor Income 0.316 0.318 0.014 22.728 0.000 0.228 Gender -> Labor Income -0.131 -0.132 0.006 21.680 0.000 Education Major -> Labor Income -0.013 -0.013 0.003 5.125 0.000 Lokasi Kerja -> Pendapatan Tenaga Kerja Work Location -> Labor Income 0.020 0.021 0.003 6.479 0.000 Productivity -> Labor Income -0.043 -0.044 0.004 11.086 0.000 Education Level -> Labor Income 0.218 0.220 0.011 20.662 0.000 Age -> Labor Income 0.087 0.088 0.003 26.600 0.000 Source: Processed data, 2025 Based on the results in Table 5. the results obtained Working Hours have a significant effect on Productivity, with a PValue = 0.000 <0.05, this means the hypothesis is accepted, there is a significant effect of working hours on productivity. Gender has a significant effect on Productivity, with a P-Value = 0.001 <0.05, this means the hypothesis is accepted, there is a significant effect of gender on productivity. Educational Major has a significant effect on Productivity, with a P-Value = 0.000 <0.05, this means the hypothesis is accepted, there is a significant effect of educational major on productivity. Work Location has a significant effect on Productivity, with a P-Value = 0.000 <0.05, this means the hypothesis is accepted, there is a significant effect of work location on productivity. Education Level has a significant effect on Productivity, with a P-Value = 0.000 <0.05, this means the hypothesis is accepted, there is a significant effect of education level on productivity. Age has a significant effect on Productivity, with a P-Value = 0.000 <0.05, this means that the hypothesis is accepted, there is a significant effect of age on productivity. Working Hours have a significant effect on Labor Income, with a P-Value = 0.000 <0.05, this means that the hypothesis is accepted, there is a significant effect of working hours on labor income. Gender has a significant effect on Labor Income, with a P-Value = 0.000 <0.05, this means that the hypothesis is accepted, there is a significant effect of gender on labor income. Educational Major has a significant effect on Labor Income, with a P-Value = 0.000 <0.05, this means that the hypothesis is accepted, there is a significant effect of educational major on labor income. Work Location has a significant effect on Labor Income, with a P-Value = 0.000 <0.05, this means that the hypothesis is accepted, there is a significant effect of work location on labor income. Productivity has a significant effect on Labor Income, with a P-Value = 0.000 <0.05, this means that the hypothesis is accepted, there is a significant effect of productivity on labor income. Education Level has a significant effect on Labor Income, with a P-Value = 0.000 <0.05, this means that the hypothesis is accepted, there is a significant World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 1197 effect of education level on labor income. Age has a significant effect on Labor Income, with a P-Value = 0.000 <0.05, this means that the hypothesis is accepted, there is a significant effect of age on labor income. The R-Squares value of Productivity is 0.238, which means that Education Level, Age, Gender, Working Hours, Work Location, Education Major are able to explain or influence Productivity by 23.8%. The R-Squares value of Labor Income is 0.228, which means that Education Level, Age, Gender, Working Hours, Work Location, Education Major, Productivity are able to explain or influence Labor Income by 22.8%. It is known based on the results of the SRMR goodness of fit test, the SRMR value = 0.000 <0.1, so it is concluded that the model has FIT. Table 6 Mediation Test Original Sample (O) Sample Mean (M) Standard Deviation (STDEV) T Statistics (|O/STDEV|) P Values Hours Worked -> Productivity -> Labor Income 0.002 0.002 0.000 9.237 0.000 Gender -> Productivity -> Labor Income 0.000 0.000 0.000 2.929 0.002 Education Department -> Productivity -> Labor Income 0.001 0.001 0.000 6.119 0.000 Work Location -> Productivity -> Labor Income 0.000 0.000 0.000 4.021 0.000 Education Level -> Productivity -> Labor Income 0.021 0.021 0.002 11.019 0.000 Age -> Productivity -> Labor Income 0.004 0.004 0.000 10.721 0.000 Based on the results of the mediation test in Table 6, namely Productivity significantly mediates the relationship between Working Hours and Labor Income, with P-Values = 0.000 <0.05, this means that the hypothesis is accepted. Productivity significantly mediates the relationship between Gender and Labor Income, with P-Values = 0.002 <0.05, this means that the hypothesis is accepted. Productivity significantly mediates the relationship between Educational Major and Labor Income, with P-Values = 0.000 <0.05, this means that the hypothesis is accepted. Productivity significantly mediates the relationship between Work Location and Labor Income, with P-Values = 0.000 <0.05, this means that the hypothesis is accepted. Productivity significantly mediates the relationship between Education Level and Labor Income, with P-Values = 0.000 <0.05, this means that the hypothesis is accepted. Productivity significantly mediates the relationship between Age and Labor Income. Based on the results of the data analysis, it was found that the level of education has a significant effect on labor income in Indonesia, indicating that education plays an important role in determining the level of individual income in the labor market. Higher education tends to provide individuals with access to jobs with higher skills and higher incomes. Increasing skills and knowledge through education allows individuals to be more productive in their jobs, which in turn can increase income. This is in line with economic theory which states that the level of education is directly proportional to labor productivity (Becker, 1993). In addition, education also plays a role in increasing social mobility, which allows individuals to achieve higher social status and jobs with higher incomes (Sen, 1999). Therefore, the results of this study provide important implications for education policies that aim to reduce socio-economic inequality by improving the quality and accessibility of education across all levels of society. Based on the results of data analysis showing that age has a significant effect on labor income in Indonesia, this means that age or work experience plays a significant role in determining an individual's income level. In general, the relationship between age and income often describes a pattern related to the length of work experience and the responsibilities held by individuals in their jobs. Older individuals with more experience tend to earn higher incomes because they have more mature skills and broader experience, which increases their productivity value in the workplace. World Journal of Advanced Research and Reviews, 2025, 27(01), 1190-1200 1198 Based on the results of the data analysis showing that gender has a significant effect on labor income in Indonesia, this means that gender differences are still a factor that influences individual income levels in the labor market. In general, income differences based on gender, often referred to as the gender pay gap, can be explained through several social, economic, and cultural mechanisms. These findings indicate the need for more proactive policies in addressing gender gaps in the workplace, for example through strengthening fair wage regulations, implementing gender equality policies in the workplace, and a more inclusive approach in providing equal opportunities for women and men in achieving positions with higher incomes. Overall, despite progress, gender income inequality shows that there is still work to be done to achieve true gender equality in the labor market. Based on the results of data analysis showing that the level of education has a significant effect on labor income in Indonesia, this means that education has an important role in increasing the level of individual income in the labor market. Theoretically, this phenomenon can be explained through several interrelated economic and social mechanisms. Individuals with higher levels of education usually have better skills and knowledge, which makes them more competitive and able to obtain jobs with higher salaries. This study provides a clear picture that education is one of the factors that greatly influences labor income, and increasing access and quality of education can play an important role in reducing income inequality and encouraging more inclusive economic growth. Based on the results of data analysis showing that the location of work has a significant effect on the income of workers in Indonesia, this means that the geographical factor or location where someone works has a significant impact on the level of income received. This phenomenon can be explained through several structural and economic factors related to differences between one region and another, both in terms of access to the labor market, availability of jobs, and the economic conditions of the area. Location can also affect the types of jobs available. In big cities, the jobs available are more varied and more often require higher education and technical skills, which tend to be accompanied by higher salaries. Conversely, in rural areas or areas with economies that are more based on the agricultural or informal sector, job opportunities with high salaries are relatively limited. Based on the results of the data analysis showing that education level, age, gender, working hours, and work location together have a significant effect on labor income in Indonesia, this means that a combination of individual factors and work contexts play an interrelated and complex role in determining income levels. This phenomenon indicates that labor income is not only influenced by one variable, but by the interaction of various factors that influence each other. The results of this analysis show that labor income in Indonesia is not only influenced by a single factor, but by a combination of various factors that interact with each other. Therefore, to increase income evenly, the policies taken need to include the development of more inclusive education, the elimination of gender disparities, and increasing access to quality employment throughout the region. In addition, fair working hour arrangements and improving the quality of life in various regions also need to be considered to reduce income disparities. Based on the results of data analysis showing that productivity has a significant effect on labor income in Indonesia, this means that the level of individual productivity is one of the main factors that influences the level of income received by the workforce. In general, productivity refers to the ability of individuals or groups to produce output in a certain period of time, and this is closely related to how skills, knowledge, and work efficiency affect the results achieved. The results of this study highlight the importance of policies that can increase overall labor productivity. Policies such as developing education and skills training, implementing more efficient technology, and improving the work environment can contribute to increased productivity. In addition, it is important for the government and the private sector to create conditions that encourage increased productivity in all levels of the workforce, both in the formal and informal sectors. Based on data analysis showing that productivity can mediate the influence of variables of education level, age, gender, working hours, and work location on labor income in Indonesia in 2021, this means that productivity is not only a factor that directly influences income, but also acts as a mediator that bridges the relationship between individual factors and work contexts with the level of income received by the workforce. The results of this analysis show that productivity is a key factor that can explain how other factors such as education, age, gender, working hours, and work location affect labor income. Thus, productivity is an important variable in the development of labor policies. Policies that encourage increased productivity, either through skills training, improving the quality of education, or improving working conditions, will be very effective in increasing labor income in Indonesia. 4. Conclusion Based on the discussion that has been done, it can be concluded that the level of education has a significant effect on labor income in Indonesia. Age has a significant effect on labor income in Indonesia. Gender has a significant effect on labor income in Indonesia. Working hours have a significant effect on labor income in Indonesia. Work location has a