“Job satisfaction of Indonesian workers: An analysis and forecasting using STAR model”

This study examines and forecasts job satisfaction of Indonesian workers from 2000 to 2022 using the happiness index and its influencing factors. The Smooth Transition Autoregressive (STAR) method is employed to analyze the non-linear relationship be-tween the happiness index and worker welfare, performance, and motivation, which are measured by per capita consumption, GDP per capita, and labor force participation. The inflation rate serves as the transition variable. The findings reveal positive effects of worker welfare and performance and a negative effect of work motivation on the happiness index. A significant threshold effect is also observed, which varies with the inflation rate. The study predicts an increase in the happiness index from 2023 to 2026, indicating improved job satisfaction post-COVID-19. This study contributes to the literature by employing a novel method and providing empirical evidence from Indonesia, a developing country with a large and diverse workforce, before and after the COVID-19 pandemic. The study acknowledges some limitations and implications for future research, such as the use of aggregate data, the linear assumption, and the lack of control variables. The paper underscores the need for policymakers and practitioners to enhance worker welfare and performance and to mitigate the negative impact of work motivation. It also highlights the need for workers and society to elevate the happiness index as a measure of job satisfaction and well-being and to address the economic and social challenges and opportunities that affect workers’ quality of life.


INTRODUCTION
Economic conditions have a significant effect on the job satisfaction and performance of Indonesian workers, which requires strategies to overcome the challenges of the economic crisis.Job satisfaction and performance are influenced by internal and external factors, such as motivation, competence, commitment, work environment, organizational culture, leadership, and compensation (Baquero, 2022).
Indonesia, which has a large territory and population as well as a diverse and dynamic economy, experienced a deep economic crisis in 1997-1998, which had an impact on the decline in the rupiah exchange rate, inflation, negative economic growth, unemployment, welfare, health, and quality of life of workers (Bawono et al., 2019).The economic crisis also reduces the level of job satisfaction and performance of Indonesian workers who experience stress, fatigue, distrust, conflict, motivation, loyalty, productivity, and creativity at work (Winasis et al., 2020).Two other major crises, namely the global economic crisis in 2008-2009 and the COVID-19 crisis, have reduced economic growth, trade, investment, and tourism in Indonesia, as well as job satisfaction and performance of Indonesian workers (Adenan, 2021).
Indonesian workers face various challenges at work, including fear, anxiety, depression, and social isolation.Although the Worker Happiness Index, developed by the OECD, shows a fairly high level of happiness, it is still below the OECD average (Khairani et al., 2022).Workers also face challenges related to low levels of occupational health and safety, high unemployment rates, and low political participation.Therefore, the government and other stakeholders need to take steps to improve workers' welfare and happiness, such as enhancing the quality and accessibility of health services, creating more inclusive work opportunities, and encouraging civil society participation in decision-making processes (Maksum, 2021).The happiness index and consumption indicators are tools for measuring worker welfare and happiness, which are influenced by various variables, such as social interactions, government policies, income, education, health, and the environment (Widarni & Bawono, 2022).Job satisfaction is an essential determinant of worker well-being, performance, and motivation, which can be influenced by various internal and external factors (Dorta-Afonso et al., 2021).
The relationship between the happiness index and consumption indicators and the performance and work motivation of Indonesian workers is positive and complex and is influenced by various economic, social, and cultural factors (Sasongko et al., 2021).There is still little research on this topic in Indonesia, which has its own characteristics and challenges in economic and social development compared with other developed or developing countries, such as the UK, China, the United States, India, and Brazil (Wijaya et al., 2021).

LITERATURE REVIEW AND HYPOTHESES
Consumption is an indicator that is often used to measure the level of social welfare.The study used the ML and NLS methods to estimate the parameters of the ESTAR model.The ML method was found to be more effective due to its higher log-likelihood value, indicating a better fit to the data.The results showed that consumption and GDP per capita positively affect the happiness index of Indonesian workers, while work participation negatively affects it.The ESTAR model operates under two conditions: linear when the inflation rate is low and non-linear when the inflation rate is high.The transition from linear to non-linear conditions, or vice versa, is determined by the transition parameter, with larger values indicating faster changes.Based on the research results, the study accepts H1 and H2 but rejects H3.

RESULTS AND DISCUSSION
The significance test on the estimated coefficients was performed using the t-statistical test, and the Wald test showed that all coefficients were statistically significant (Table 5).6).This means that the ESTAR model created is valid and reliable to use.This paper also carries out forecasting using in-sample or out-of-sample data, using recursive methods and direct methods (Table 7).Table 7 shows that the direct method provides more accurate forecasting results than the recursive method, both for in-sample and out-of-sample data.This can be seen from the smaller error value for the direct method.This means that the direct method is more suitable for use in forecasting the Indonesian happiness index using consumption variables, GDP per capita, labor force participation and inflation rate.Table 8 shows the results of ESTAR estimation forecasting using the direct method.9 shows an evaluation of ESTAR's forecasting performance.Table 9 shows that the ML method has lower MAE, MSE, RMSE, and MAPE values than the NLS method.This means that the ML method is more accurate and better at forecasting Indonesia's happiness index using consumption variables, GDP per capita, labor force participation, and inflation rate.The lower the MAE, MSE, RMSE, and MAPE values, the smaller the forecasting error that occurs.Table 10 shows ESTAR model estimation results using the maximum likelihood method.
The study estimates the ESTAR model coefficients and tests their significance using t statistical and Wald tests.Findings reveal that consumption positively and significantly affects the happiness Consumption and the worker happiness index in Indonesia have a close relationship.In other words, the higher a worker's level of consumption, the higher his/her level of happiness.This is possible because consumption reflects a worker's ability to fulfill needs and desires, which in turn can improve the quality of life and worker satisfaction.This supports the findings of Patria (2022) and Wijaya et al. (2021).
The worker happiness index in Indonesia is also directly proportional to GDP per capita.This means that the level of worker happiness will increase as GDP per capita increases.GDP per capita reflects the level of prosperity and wellbeing of a country, which of course has an imon workers' income, purchasing power, and quality of life.This supports the outcomes of Kadir and Ismail (2020).
However, there is one interesting thing.Even though consumption and GDP per capita are directly proportional to the worker happiness index, the work participation rate is actually inversely proportional to the worker happiness index in Indonesia.In other words, the higher the level of work participation, the lower the level of worker happiness.This may be caused by excessive workload and high work pressure, which can reduce free time and balance between work and personal life, which in turn can reduce workers' happiness levels.This supports Anwar et al. (2021).

CONCLUSION
Indonesian workers' job satisfaction is influenced by worker welfare, performance, and motivation.Worker welfare is measured by per capita consumption, which increases Indonesia's happiness index.This means that the higher the per capita consumption, the higher the job satisfaction of Indonesian workers.Worker performance is measured by GDP per capita, which also improves Indonesia's happiness index.Workers in Indonesia feel more satisfied in their jobs when the country's GDP per capita is higher.Indonesian workers' work motivation is measured by labor force participation, which lowers Indonesia's happiness index.This means that the higher the labor force participation, the lower the job satisfaction of Indonesian workers.These findings indicate that increasing work motivation is not always accompanied by increasing job satisfaction.The inflation rate acts as a transition variable that determines changes in conditions from linear to non-linear or vice versa as a crisis indicator.When the inflation rate reaches a value declared as a crisis, there is a significant change in the relationship between the independent variables and the Indonesian happiness index.This means that when an economic crisis occurs, every indicator that influences job satisfaction will change significantly.
(Jumady & Lilla, 2021))s in Indonesia is influenced by various factors, both from within and outside the worker(Sutarto et al., 2021).High worker happiness can have a positive impact on labor force participation, which is an important indicator of economic and social development(Pilipiec et al., 2021).Therefore, there needs to be efforts from the government, companies, and society to create conditions that support the happiness of workers in Indonesia(Jumady & Lilla, 2021).
(Kadir & Ismail, 2020)2;Kanto & Sjahruddin, 2020)ra & Anoraga, 2021)desires.It also influences a person's quality of life and happiness(Spielman et al., 2020;Widarni & Bawono, 2022;Zahra & Anoraga, 2021).Happiness is a feeling of satisfaction and pleasure with the life circumstances experienced.The worker happiness index in Indonesia is used to determine the level of worker satisfaction with work and living conditions(Kustiawan et al., 2022;Kanto & Sjahruddin, 2020).This index is based on several aspects, such as salary, health, work environment, social relationships, and others.This index can be used Consumption can increase workers' self-confidence and self-esteem because they feel able to buy the goods and services they want or need.It can provide opportunities for workers to enjoy a variety tion, and stress in the workplace and reduce worker motivation, loyalty, and performance (Di Stefano et al., 2019; Lee & Liu, 2021).The happiness index of Indonesian workers is influenced by GDP per capita, which reflects economic conditions and welfare and impacts income and quality of life (Dahliah & Nur, 2021; Audi & Ali, 2023).It is also affected by work culture, which influences the work atmosphere and performance(Chien et al., 2020).Efforts to increase GDP per capita and foster a positive work culture are needed to enhance this index(Kadir & Ismail, 2020).The happiness index has a significant relationship with labor force participation, indicating that happier workers are more likely to be working or seeking work(Nazah etal., 2021; Pilipiec et al., 2021).Work environment conditions, including atmosphere, facilities, safety, and relationships, influence worker happiness.A good environment can boost motivation, performance, and loyalty, while a poor one can lead to dissatisfaction or feelings of abuse (Dziuba et al., 2020; Sukawati & Suwandana, 2021; De Clercq et al., 2020).Apart from the work environment, other factors that influence worker happiness are income level, career opportunities, balance between work and personal life, and social recognition (Bhende et al., 2020).Workers who have sufficient income, opportunities for development, balanced time for work and leisure, and a sense of respect from society tend to be happier than workers who do not have these things (Wong et al., 2021).Worker happiness is also influenced by personal factors, such as personality, attitudes, values, and expectations (Silva Munar et al., 2020).pendent variables are consumption, GDP per capita, and work participation.Economic crises that negatively affect the workers' happiness index are also considered.This model can provide insights Problems and Perspectives in Management, Volume 22, Issue 2, 2024 http://dx.doi.org/10.21511/ppm.22(2).2024.34into how economic factors influence worker satisfaction levels and how economic crises can alter the behavior of these variables.Furthermore, this model can be used to predict the workers' happiness index in the future.The analysis begins by collecting secondary data from the World Bank, including the workers' happiness index, consumption, GDP per capita, work participation, and economic crisis indicators.The inflation rate is an indicator of economic crises.Definitions, measurements, and data sources for the variables are provided in Table 1.

Table 1 .
Definition, measurement, and variables' data sources This study uses ADF and PP tests to check for unit roots in time series variables.A unit root means a variable does not revert to its average value after a shock, making it non-stationary.Stationary variables, which lack unit roots, have constant statistical properties like mean, variance, and covariance.

Table 2 .
Stationarity test STAR analysis, the ESTAR or LSTAR model must be selected.This selection requires a non-linearity test with the Lagrange Multiplier (LM) test and F-test.The LM test identifies the appropriate form of the transition function, while the F-test determines the order of the STAR model.

Table 3
presents the results of the non-linearity test.

Table 4 .
ESTAR model estimation results using the non-linear least squares (NLS) method and maximum likelihood (ML) method

Table 5 .
Wald test valueThis study uses the Durbin-Watson test, Breusch-Pagan test, Shapiro-Wilk test, and Ramsey RESET test to test the classic assumptions of linear regression in the ESTAR model.It models the happiness index of Indonesian workers as a function of consumption, GDP per capita, and labor force participation, taking into account the system conditions indicated by the inflation rate as a transition variable for the economic crisis.The test results show that the ESTAR model does not have autocorrelation, heteroscedasticity, and normally distributed residuals and is in accordance with the existing data (Table

Table 7 .
Forecasting from ESTAR estimates using in-sample data or out-of-sample data, using recursive methods and direct methods

Table 8 .
ESTAR estimation forecasting results using the direct method

Table 8
shows that Indonesia's happiness index value is expected to increase linearly from 2023 to 2026, using the variables consumption, GDP per capita, labor force participation, and inflation rate.This shows that post-COVID-19, job satisfaction is increasing.The error value shows the forecast value and the actual observed value are different.The smaller error value indicates that the forecast value is closer to the actual observed value.Next, Table

Table 9 .
ESTAR forecasting performance evaluation

Table 6 .
Diagnostic test results using autocorrelation test, heteroscedasticity test, normality test, and Ramsey RESET test

Table 10 .
ESTAR model estimation results using the maximum likelihood method