BUSINESS ETHICS CHALLENGES AND A COMPREHENSIVE UNDERSTANDING OF TACKLING CHILD LABOR IN TURKEY

: Until a few decades ago, ethical issues were defined in terms of fraudulent behavior and workplace working conditions, but the concept itself has now expanded. Child labor, considered cheap and easily manageable labor in the globalizing world, is one of the chronic problems of many developing and developed countries, including Turkey. Recent studies show that child's work areas are increasingly diversified, and child labor has become an important factor in the global cheap labor market. For whatever reason, it is a universal ethical issue that children work as young adults in many different sectors, from mining to textiles and agriculture to the automotive industry. In this context, the issue of child labor in Turkey has been discussed in the context of business ethics principles. In the study, three main factors were determined as demographic, educational life, and business life variables as the determinants of the reasons for the involvement of children in work life. The multinomial logit model was used to test the relationships between the variables. The research used data from 1128 children in employment obtained from the Child Labor Force Research Micro Data Set carried out by the Turkish Statistical Institute in 2019. According to the results of the research, it has been determined that the variables related to the education life (educational status of the child, family education status, apprenticeship education status) and the variables related to the business life (working sector, working time, income and family working status) affect the working reasons of children. As a result, in an environment of increasing cooperation and competition between countries with the effect of globalization, each country's reflection of a common definition of child labor in its own national legislation is one of the most important steps to be taken in preventing child labor. In addition, in underdeveloped and developing countries, eliminating poverty, expanding school attendance conditions, and raising awareness about the existence of children's rights as individuals will be important steps. It must be admitted that it will not be possible to end child labor in all its aspects without developing a society, and public and private sector management that acts with a sense of ethical responsibility.

Introduction. The competitive structure of today's global business world forces managers to face different attendance of those who continue their education to school, provided that they make a written contract and get separate permission for each activity (Nagy, 2003).
A total of 160 million children, 63 million girls and 97 million boys, in other words, one out of every 10 children, are currently working as child laborers. According to the report published by the International Labor Organization (ILO) and the United Nations Children's Fund (UNICEF) on 10 June 2021, the number of children working as child laborers has increased by 8.4 million in the last four years to 160 million worldwide; Another 9 million children are at risk due to the effects of the COVID-19 pandemic (www.ilo.org 2 ). It would be useful to approach the issue from two different perspectives when identifying the causes that lead to child labor. The first of these is to determine the factors that force the child to supply labor, and the other is to clarify why employers direct their labor demands to children (Tunccan, 2012). So much so that child labor is largely caused by structural problems such as the income level of countries, degrees of industrialization and globalization, population growth rate, and approach to education. Child labor has many short-and long-term negative effects, such as decreased human capital growth and contrary effects on future labor market values. Such negative effects will obstruct individual and national well-being (Tang and Zhao, 2022). It should be noted that the inadequacies in the physical conditions of education opportunities, especially in economic conditions, and the parents' perception of education are among the determining factors in children leaving education life at an early age and turning to business life (Tunccan, 2012). In other words, concrete data obtained from studies conducted on different continents show that as the school enrollment and school attendance rates of individuals in the child worker age group increase, their tendency to work decreases (Kozhaya and Flores, 2022). In this sense, it can be said that the social order-providing role of education in school is not to be underestimated.
Methodology and research methods. In the study, three main factors were determined as demographic, educational life, and business life variables as the determinants of the reasons for the involvement of children in working life. The model of the research is shown in Figure 1.  What is your main reason for Working in a Business? (1: Contributing to household income and helping with their economic activities, 2: learning a job, having a profession, 3: Meeting their own needs) S13 Household Status Employment status of the household head (1: employed, 2: unemployed, 3: not included in the labor force) S14 Mother Status Employment status of mothers (1: employed, 2: unemployed, 3: not in the labor force) S15 Father Status Employment status of fathers (1: employed, 2: unemployed, 3: not in the labor force) Sources: developed by the authors.
Information on the descriptive statistics (frequency and percentage) of the variables is given in Table 2. According to Table 2, 4.3% of the employed children stated that they did not finish any school, 14% completed primary school, 80% secondary school, 1% general high school, and 1.1% vocational high school. Sources: developed by the authors.
To the question of whether or not they received apprenticeship training, 93% of the children answered «no» and 7% answered «yes». 68% of the employed children stated that they continued their education, and 32% did not continue their education. 23% of mothers of employed children are illiterate, 72% are educated below high school, 2% are high school graduates, 2% are vocational high school graduates and 0.4% are university graduates. 4.1% of the fathers of employed children are illiterate, 84% have less than a high school education, 5% are high school graduates, 6% are vocational high school graduates, and 2% are university graduates. 37% of children are employed in the agricultural sector, 20% in industry, and 44% in the service sector. Within a month, 4% of children are earning 200 Turkish liras (TL) or less, 13% are earning 201-400 TL, 31% are earning 401-600 TL, 10% are earning 601-800 TL, 8% are earning 801-1000 TL, % 4 of them earn 1001-1200 TL and 19% earn 1201 TL or more.
Considering the reasons for the children working, which is the dependent variable of the study, 62% of the children stated that they work to contribute to the household income and help their economic activities, 33% to learn a job and have a profession, and 6% to meet their own needs. Children stated that 54% of their mothers are employed, 4% are unemployed, and 43% are not in the labor force. Children stated that 86% of their fathers are employed, 5% are unemployed, and 10% are not in the labor force.
It is possible to talk about preferences in every condition of human existence. People usually continue life by choosing the best alternative for themselves. At this point, categorical variables are observed in many areas. Situations where categorical variables are observed as dependent variables, are also frequently encountered. Multiple models are divided into two unordered qualitative preference models, which are used to predict dependent variables consisting of more than two categories with nominal scale, and ordinal qualitative preference models, which predict dependent variables with more than two categories with ordinal scale (Davidson and Mackinnon, 2004). Unordered qualitative preference models are used when the dependent variable has more than two alternatives, and there is no ordering between these alternatives. Unordered qualitative preference models were preferred since the reasons for the study of the children, which constitute the study's dependent variable, also fit this situation. The multinomial logit model is an extended version of the binary logit model with respect to the J alternative, and the dependent variable has a multinomial distribution. In the multinomial logit model, the categories that make up the dependent variable should be independent, and there should be no ordering between these categories. The individual will choose the best alternative for himself/herself, or he/she will prefer the category that gives him/her the most benefit. The random utility theory underlies this choice to be made. According to this theory, the individual will make a good choice and it is assumed that they will provide the highest benefit (Amemiya, 1981;Bartels et al., 1999).
In multinomial logit models, linearity and normal distribution assumptions must not be satisfied. The dependent variable is not an ordered categorical variable; the dependent variable must be classifiable (Long, 1997) among M categories constituting the dependent variable. Probabilities are calculated to determine the extent to which the option would be realized. The probability of m option occurring for a determined value of the explanatory variable of the dependent variable with category M is ( = ⁄ ). This function is a linear function of . According to the multinomial logit model, the probability of individual i choosing the mcategory among M categories is shown in Equation 1. (1) The multinomial logit models are estimated using the maximum likelihood method. Considering the situation in which there are n individuals in the study and the children who are the dependent variable of the ith individual are faced with 3 categories of reasons for working (contributing to household income and helping their economic activities, learning a job, having a job and meeting their own needs), the log-likelihood function can be found in the equation. It is expressed in 2.
The log-likelihood function is maximized using the Newton-Raphson method, and the parameters are estimated. The estimators obtained from this model are consistent, asymptotic normal, and asymptotic efficient (Long, 1997). In Equation 2, Y is the dependent variable, X is the vector of the independent variables, and β is the parameter vector.
In the multinomial logit model, the assumption of «independence of irrelevant alternatives (IIA)», which states that adding a category to the dependent variable or changing the properties of existing categories, has no effect on the different rates (Hensher et al., 2005). The Hausman test developed by Hausman and McFadden (1984) is used to test the validity of the IIA assumption. In the Hausman test, in the first stage, the model, including all the dependent variable categories, is estimated. In the second stage, the limited model is estimated by removing one or more categories. If the difference between the two models is insignificant, it is decided that the IIA assumption is valid (Long and Freese, 2006). The Hausman test statistic conforms to the χ^2 distribution with k degrees of freedom. The basic hypothesis is «the assumption of independence of irrelevant alternatives is valid». If the basic hypothesis is rejected, the multinomial logit model is not used. Alternatively, analyzes are carried out with the multinomial probit model (Cheng and Long, 2007).
If the IIA assumption is met, goodness-of-fit criteria should be used to determine whether the model estimation obtained in the multinomial logit model is statistically appropriate and which of the alternative models gives better results (Maddala, 1986). McFadden's Adjusted R2, Pseduo-R2, and log-likelihood function value are used to determine the model's goodness of fit, and Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) criteria are used for model selection. The model that minimizes the AIC and BIC criteria is the preferable model (Raftery, 1995).
Results. Alternative independent variables were included in the model and it was determined that the model, considering the AIC and BIC criteria, are the variables that affect the preferences of the children's reasons for working. The results are given in Table 3. Also, in Table 3, the coefficients (β) obtained for each coefficient of the dependent variable in the first column and the marginal effects (dy/dx) in the second column are given. The variables S1, S4, S7, S9, S13 were excluded from the model because they were highly correlated with some other independent variables in the model.
According to the results of the Hausman-McFadden test statistics given in Table 3, it was determined that the IIA assumption was met for all categories of the dependent variable. According to the result of Wald test statistics (Wald X 2 (14) = 6122.80; Probability Value = 0.0000), the model as a whole was determined to be statistically significant.
According to the marginal impact values, children who continue their education are 14.6% less likely to work to contribute to the household income and provide economic benefits to their families, while 10.2% more likely to learn a job and have a job, and 4.5% more to meet their own needs, than children who do not continue their education found to be excessive.  Based on the model test, it was determined that children who received apprenticeship training were 142% less likely to work to contribute to household income and provide economic benefits to their families, 49% less likely to work to meet their own needs, and 191% more likely to work to learn a job and have a job, compared to children who did not receive apprenticeship training. It has been determined that the probability of working to contribute to the household income of children whose mothers are less than high school degrees is 35.4%, and the probability of working to meet their own needs is 28.5% more compared to children whose mothers are illiterate. It has been determined that children whose mothers are graduates of higher education are 35.3% less likely to work to contribute to the family economy, 9.3% less to meet their own needs, and 44.7% more to learn a job and have a profession, compared to children whose mothers are illiterate.
Children working in the industry sector are 29.8% less likely to contribute to the family economy than children working in the agricultural sector. In comparison, the probability of working on learning a job and having a profession is 29.7% higher. It has been determined that children working in the service sector are 31.6% less likely to work to contribute to the family economy than children working in the agricultural sector. In comparison, the probability of working on learning a job and having a profession is 27.7% more. It has been determined that children with a monthly income of 1001-1200 TL are 38.8% more likely to work to contribute to the family economy than children with a monthly income of 200 TL or less. Children whose fathers are not in the labor force are 6.8% less likely to work to meet their needs than those whose fathers are employed.
Conclusions. It can be observed that social resistance has started in recent years against the working life conditions that encourage the endless production and consumption of globalization. Illegal and unethical practices are among the main motivations that cause this resistance, and child labor is one of the sources of motivation (Asad et al., 2018). Firms that present themselves as ethical company to society claim that they do not use harmful products for consumers and declare that they are committed to labor and labor law and universal ethical values (Ferrell et al., 2019). In this sense, international companies especially try to determine their position in the market based on the perception that they are sensitive to issues such as children's rights and health because children have always been the most vulnerable members of any society. Since humans started to produce, children also helped their elders in their families. Although this helping work contributes to their physical and mental development from time to time, exploitation, starvation, being forced to work for long hours, injury, and death of children are also situations encountered from past to present (Tat, 2016). Child labor has become a conscious choice in many countries, especially since the industrial revolution, with pioneering reasons such as reducing labor costs and easy control of the workforce. At the same time, business ethics aims to have value judgments in line with common goals and purposes in the business world by covering the whole world universally (Atacan and Genc, 2020). The extent to which this goal has been achieved is debatable. Because the religious beliefs, cultural codes, and value judgments of the society to which the companies belong can also affect the perception and practice of business ethics (Gallego-Alvarez et al.,2020).
Since the children constituting the study's dependent variable had more than two reasons for working, the multinomial logit model was used to test the relationship between the variables. The research used data from 1128 children in employment obtained from the Child Labor Force Research Micro Data Set carried out by TURKSTAT in 2019. According to the research results, it has been found that children who continue their education are more likely to work to learn a job, has a profession, and meet their own needs. In contrast, the children who do not continue their education are more likely to work to contribute to the household income and provide economic benefits. In this case, it is seen that the socio-economic factors that children sustain their lives directly affect the working preferences of children.
On the other hand, the probability of working for the purpose of learning a job and having a profession was found to be significantly higher for the children who received apprenticeship training. It has been determined that children who do not receive apprenticeship training have a high probability of working to contribute to the family income and meet their needs. In this sense, the importance of evaluating and disseminating apprenticeship training organized for the training of intermediate staff in needed sectors as a human resource training policy emerges.
When we look at the education-oriented findings of the study, it was found that the children of mothers who graduated from high school or lower were more likely to work for their own needs than those of illiterate mothers. At the same time, they were less likely to work to contribute to the family economy. Children whose mothers are higher education graduates are more likely to learn a job and work for a profession than children whose mothers are illiterate. On the other hand, children whose fathers are not in the labor force are less likely to work to meet their own needs than children whose fathers are employed. In another study on child labor in Turkiye, it was determined that child labor is higher in rural areas and parental education levels are lower in the same regions (Acaroglu, 2010;AvSar and Ogutogulları, 2012).
As a result, in an environment of increasing cooperation and competition between countries with the effect of globalization, each country's reflection of a common definition of child labor in its own national legislation is one of the most important steps to be taken in preventing child labor (Sahin, 2010). In addition, in underdeveloped and developing countries, eliminating poverty, expanding school attendance conditions, and raising awareness about children's rights as individuals will be important steps. It must be admitted that children have been working or are being made to work for a reason since human beings existed. On the other hand, some children must work due to their family conditions and provide for their families. In addition to the fact that children in this obligation, which should be questioned in terms of ethical principles, are forced to work, they should be preferred, especially as cheap labor. In this sense, besides the national and international legal regulations, this situation is the ethical responsibility of the employers. It must be admitted that it will not be possible to end child labor in all its aspects without developing a society, and public and private sector management that acts with a sense of ethical responsibility. Ultimately, it is obvious that this issue will remain unresolved as a universal violation of business ethics unless the ties of any child who has to work with the labor market are cut by the hands of the states.
First of all, it should be noted that there is no complete consensus in the existing literature on the definition of child labor, even though there are similar reasons that encourage children to work based on countries, as well as legal and social differences at the national or regional level. In this sense, the study has some limitations when discussing a concept that has not yet completed its conceptualization stage. In addition, although the data from a national survey conducted in Turkey were used, the fact that employers were not consulted in the survey is another limitation. In addition, there are some future suggestions for the study. First of all, it is seen that the measures taken by supranational organizations on child labor are not sufficient. The advisory decisions of these institutions should be supported by legal regulations with high sanction power to prevent child labor.
Furthermore, steps should be taken to consider child labor as an ethical violation through employers' organizations from a comprehensive human resources perspective. Finally, although many countries have compulsory school attendance up to a certain age or education level, it is clear that there are areas where this obligation cannot be applied. Social policies should be developed and broadened, especially for those children who must make a living or support their families. Author