Corporate budget governance through the operating managers engagement: Does locus of control matter?

In the past decades, budgetary participation has attracted increasing attention (Brownell, 1981; Amir et al., 2021; Brownell & Hirst, 1986; Chenhall & Brownell, 1988; Murray, 1990; Brownell & Dunk, 1991; Krenl, 1992; Baccouri & Hassouna, 2017; Her, Shin, & Pae, 2019; Lau, Scully, & Lee, 2018; Lleo, Viles, Jurburg, & Santos, 2020; Sofyani, Santo, Najda, & Al-maghribi, 2020; Apriwandi & Supriyono, 2021; Santos, Beuren, & Marques, 2021). Budgetary participation a as defined is whereby process subordinates are given the opportunity to get involved in and have an influence on the budget setting process (Brownell, 1992). It refers to the practice of allowing managers to participate in the decisions by which budgets are established to measure and evaluate their performance (Swieringa & Moncur, 1975; Shields & Shields, 1998). According to the behavioral approach in accounting, budgetary participation the attitudes of influence can individuals at work (Chong, Eg & Leong gleton, , the work, and with 2005), including satisfaction the (Viseu company feeling of being part of the , Pinto, Borralha, & de Jesus, 2020). The relationship between budgetary participation and job satisfaction has attracted many researchers (Nguyên, 2019; Brownell, 1981; Cheng, Chen, & Shih, 2014; Kahar 2016; Chariti, & Rohman, , Becker & Green, Maurice, Bonach, 1962; Moris & , 2012; Abstract


INTRODUCTION
In the past decades, budgetary participation has attracted increasing attention (Brownell, 1981;Amir et al., 2021;Brownell & Hirst, 1986;Chenhall & Brownell, 1988;Murray, 1990;Brownell & Dunk, 1991;Krenl, 1992 whereby process subordinates are given the opportunity to get involved in and have an influence on the budget setting process (Brownell, 1992). It refers to the practice of allowing managers to participate in the decisions by which budgets are established to measure and evaluate their performance (Swieringa & Moncur, 1975;Shields & Shields, 1998 The relationship between budgetary participation and job satisfaction has attracted many researchers (Nguyên, 2019;Brownell, 1981 (Milani, 1975;Chenhall, 1986 2020), several studies concluded that there is no significant relationship between budgetary participation and job satisfaction (Kenis, 1979;Brownell, 1982;Harrisson, 1992;Ginting, 2021). Also, there are some researchers who proposed a negative relationship between these two variables (Otley & Pollanen, 2000; Leach-López, Stammerjohan, & McNair, 2008;Lau et al., 2018). Yuliansyah, Inapty, Dahlan, and Agtia (2018) say that these mixed results indicate that no universalistic relationship exists between budgetary participation and job satisfaction, and suggest that there could be other variables involved. In this vein, several studies adopted a contingency approach and attempted to advance some contingency variables that greatly influence budget behaviors (Brownell, 1982;O'Connor, 1995;Chow, Shields, & Wu, 1999). They considered that how the budgetary system is developed and used as a complex outcome of some contingent variables (Hormati, Laduna, Mahdi, & Kahar, 2017). They demonstrated the existence of links between structural and contextual characteristics of the company (size, strategy, structure, perceived environmental uncertainty, etc.) and its use of budgetary participation (Chenhall, 2003;Komarev, 2007). Other authors added behavioral factors (authoritarianism, locus of control, etc.) related to the actors in their links to the budget system. They argued that budget control practices differ according to the context and profiles of the actors involved (Brownell, 1982;Seiler & Bartlett, 1982;Chenhall, 1986;Bonache et al., 2012;Viseu et al., 2020). This study proposes to extend this line of research by investigating whether the locus of control, as a personality variable, moderates the budgetary participation-job satisfaction relationship. We aim to provide answers to the following research questions: RQ1: How does budgetary participation affect managerial job satisfaction? RQ2: How does the locus of control moderate the relationship between budgetary participation and job satisfaction?
Answering these research questions implies using a theoretical framework that proposes certain factors as well as an indication of the direction of the impact of these factors on the budgetary process. We will approach it from the perspective of contingency theory. This perspective has -the merit of presenting management control as eminently adaptable to different types of organizations‖ (Pariente, 1998, p. 4). Our study claims originality insofar as it addresses the hotel industry. Although budgetary participation is well documented in manufacturing, it is not so in the service sector, particularly in hospitality (Viseu et & Mia, 2005). In fact, managers in the hotel industry, compared to their counterparts in manufacturing industries, face a more uncertain and complex work environment because of the hotel industry's exceptional service characteristics. Also, the business environment in the hotel industry is highly competitive (Yuliansyah et al., 2018;Mia & Patiar, 2002). In the hotel industry, good or bad service has many facets (Darvishmotevali, Arasli, & Kilic, 2017), including cycles of internal service quality (Lam & Chen, 2012;Scanlon, 2007). Unless employees are happy, it is quite difficult for an organisation to provide good service to its customers (Gouthier & Rhein, 2011). Job satisfaction, therefore, is an organization's responsibility (Jermias & Setiawan, 2008). It is very likely that the research findings on the budgetary participation-job satisfaction relationship may not be applicable in the hotel industry (Yuliansyah et al., 2018;Mia & Patiar, 2001). Thus, more research on such issues is needed to fill this gap in the hotel industry. For this purpose, the current study compiled data that is gathered via a questionnaire administered to 75 middle-level managers from Tunisian hotels. Our results confirm the contingent aspect of budgetary participation and show that the locus of control moderates the budgetary participation outcomes.
The remainder of the paper is structured as follows.
Section 2 presents the theoretical framework. Section 3 describes the research methodology, followed by the results in Section 4. Section 5 discusses the findings in connection with prior studies and concludes.

Budgetary participation and contingency theory
How the budgetary system is developed and used was considered as a complex outcome of some contingent variables (Komarev, 2007;Govindarajan, 1986). Some authors retained the contingency theory as a theoretical basis to study the issues of budgetary participation. This theory has -the merit of presenting management control as eminently adaptable to different types of organizations‖ (Pariente, 1998, p. 4). Several studies have demonstrated the existence of links between the structural and contextual characteristics of the company (size, ownership structure, type of strategy, environmental perceived uncertainty, culture, etc.) and its use of budgetary participation (Chapman, 1997 Krenl, 2003;Merchant, 1981). According to Adler (1983), culture influences organizations through societal structures such as laws and political systems and also through the values, attitudes, behavior, goals, and preferences of participants. Thus, the view of -reality‖ held by members of a group is expected to affect their perception of the budgetary process and the values that they place on participation in that process (Hofstede, 1987;Leach-López et al., 2008;Dakhli, 2009). Using the work of Hofstede (1980) for hypothesis development, Frucot and Shearon (1991) tested the interrelationship of budget participation and locus of control (a personality variable) and their impact on job performance and job satisfaction. They assumed that an organizational system such as budget participation would have different effects in different cultures. They concluded that the lack of significance in their variables was due to cultural differences. Following the methodology employed by . Power delegation is rarely practiced (Zghal, 1992) and the participants generally concern only the managers who belong to the second level of the hierarchy (Lassoued, 2008). Therefore, it would be worthwhile to examine this issue in Tunisia, a sociocultural context with very little explored (Mansour, 2015;Baccouri & Hassouna, 2017). In fact, most budgetary participation research has been conducted in American, British, and Australian contexts (Brownell, 1981;Krenl, 1992 (Brownell & McInnes, 1986). Authors of human relations school propose that participation can fulfill some of the primary needs of the subordinate such as self-expression, self-esteem, and self-accomplishment and improving, as a result, his morale and job satisfaction. In this regard, it is important to refer to the study of Swieringa and Moncur (1975), which is a first attempt to describe budgeting in behavioral terms. Swieringa and Moncur (1975) reproach budget research for being limited to a purely technical vision of budgeting and ignoring the mechanisms through which it influences managerial behavior. They concluded that the manager who is involved in the budgeting process shows favorable attitudes at work. These results corroborate Milani's (1975) findings, which already established a positive association between participation and managerial attitudes. Boujelbene and Affes (2015) note that budgetary participation motivates subordinates by encouraging them to accept and be committed to the budget targets and thus improving their job satisfaction. Finally, Baccouri and Hassouna (2017) find that budgetary participation intensity is antecedent to job satisfaction which in turn affects innovation. This result was confirmed recently with Costa da Silva Zonatto et al.'s (2020) study which concluded that budgetary participation is positively and significantly associated with managerial job satisfaction in Brazilian companies. More recently, Raucci and Paolini (2021), using a sample of clinical managers of Italian PHOs, show that budgetary participation has a direct positive effect on job satisfaction. However, other previous studies concluded that there is no significant relationship between budgetary participation and job satisfaction (Kenis, 1979;Harrisson, 1992;Ginting, 2021). In this vein, Brownell (1982) found weak evidence for a positive association between budgetary participation and job satisfaction. Most recently, Özer, Özcan, and Akpınar (2012) concluded that budgetary participation has not direct and significant association with job satisfaction. Seeking to investigate the antecedent, mediating, and outcome variables of budgetary participation in a developing country, namely Turkey, Jermias and Ygit (2013) found that budgetary participation alone does not improve job satisfaction and performance. Also, there are some researchers who propose a negative relationship between these two variables As presented above, the results of studies into the association between budgetary participation and job satisfaction are equivocal. An advanced explanation for these mixed results is that no universalistic relationship exists between budgetary participation and job satisfaction (Yuliansyah et al., 2018). In this vein, several studies adopted a contingency approach and attempted to advance some contingency variables that greatly condition budget behaviors (Brownell, 1982 Soleiman & Thalib, 2021). They considered that how the budgetary system is developed and used as a complex outcome of some contingent variables (Hormati et al., 2017). They demonstrated the existence of links between structural and contextual characteristics of the company (size, strategy, structure, perceived environmental uncertainty, etc.) and its use of budgetary participation (Chenhall, 2003;Komarev, 2007 Indriantoro, 2000) proposed to broaden this research framework to the behavioral contingency theory by integrating behavioral factors related to the actors in their links to the budget system. They argued that budget control practices differ according to the context and profiles of the actors involved (Brownell, 1982;Chenhall, 1986;Bonache et al., 2012;Viseu et al., 2020). In this vein, locus of control was presented as a relevant variable that can moderate the impact of budgetary participation on job attitudes and behaviors. The locus of control construct can be defined as a psychological construct that captures individuals' perceptions towards the casual relationship between their own behavior and its consequences (Xia & Ma, 2020). The locus of control is conceptualized as referring to a unidimensional continuum, ranging from external to internal. Internal-individual believes that his/her behavior is guided by his/her personal decisions and efforts. While, external individual believes that his/her behavior is guided by fate, luck, or other external circumstances (Rotter, 1966). Brownell (1982) examined the effects of participative budgeting and locus of control on performance and job satisfaction in a field setting. Brownell

RESEARCH METHOD
The methodology used in this research is characterized as a descriptive study, conducted through survey and quantitative data approach.

Measurement of variables
Our variables were measured by scales used by previous relevant studies and showed satisfactory validity and reliability. Milani's (1975) six-item measures were deployed to assess the middle-level managers' degree of participation in the budget process. This instrument has been widely used in previous budgeting studies (Krenl, 1992 To treat the locus of control, we used the scale of Rotter (1966). This scale contains 19 items (statements) for which the respondent must check -True‖ or -False‖ that best fit his/her own belief. If the respondent marked a statement true, he/she gets 1 point and if he/she marked it false he/she gets -1. Based on his/her score, the respondent will be qualified as an internal locus of control (if he/she obtains a positive score) or an external locus of control (if he/she obtains a negative score). This instrument is widely deployed in psychological and accounting researches and several extensive literature reviews attest to its validity and reliability (Swieringa & Moncur, 1975;Brownell, 1981Brownell, , 1982Licata et al., 1986;Frucot & Shearon, 1991;Krenl, 1992). In this study, the Cronbach's alpha coefficient was 0.796 which indicates acceptable internal reliability for the scale.
Job satisfaction was measured by a six-item scale developed by Dewar and Werbel (1979) and largely used by prior accounting studies (Chenhall & Brownell, 1988 The extent to which a person perceives events as being a consequence of his or her own behavior and therefore potentially under personal control (Rotter, 1966).

Sample and data collection
The population of this study is hotel middle-level managers. We randomly selected from the database of the Tunisian federation of hotels four and five-star hotels located in four tourist areas (Hammamet, Sousse, Mahdia, and Djerba). Moreover, hotels chosen do not pertain to any hotel chain to permit comparative analysis (Uyar & Bilgin, 2011). Our sample is homogeneous, both in terms of the number of chambers and employees. Hotels selected were expected to have clearly defined budgeting systems. The instrument used for data collection is a questionnaire composed of 27 multiple-choice questions, which were elaborated from the constructs presented in Table 1, observing their original measurement scales. The survey was pilot-tested with ten budget managers from three hotels from our sample. Generally, they considered that the questionnaire was comprehensive and they proposed some modifications in order to be adopted in Tunisian hotels. To increase our response rate, we take three steps suggested by Yuliansyah et al. (2018). Those steps are pre-notifications contact, the initial distribution of the questionnaire, and follow-up. Pre-notification is a preliminary meeting with general managers of selected hotels were held. In these meetings, we explained the purpose of our research and asked permission to handle the survey in their hotels. We informed that the respondents' names would be anonymous for security and confidential purposes. These interviews helped, also, to gather general information about the hotels and their corporate planning and budgeting systems and to identify a sample of middle-level managers who were involved in budgeting. Initial distribution of the survey instrument is by hand, physically visiting each participating hotel. The last step, and arguably the most important step, is follow-up. We collect the questionnaires and replace the questionnaire if it is said to be lost. By this approach, we received 80 questionnaires from 95 distributed questionnaires, which reflect a response rate of 84.2%. Of those collected questionnaires, 75 are usable, and the others 5 are discarded due to inappropriate responses and incomplete answers. The final sample size for hypotheses testing was reduced to 75 middle-level managers. Respondents were heading the following functional departments: Human Resources Management (25.5%), Sales and Marketing (28%), Accounting (26.5%), and Purchasing (20%). Through a descriptive analysis of information on the participants in the survey, we found that they had been with their hotels for at least five years, they occupy their current position for an average of 4 years and they have a management and financial background. These findings strengthen the reliability of our sample because our respondents have adequate academic training and sufficient experience to adequately respond to the questionnaire. The collected data was stored in a database to be simply introduced in the SPSS software for statistical treatment. Three parts structure our questionnaire. The first part aimed to gather general information about the respondents. Questions were as follows: the service held, the number of years working within the hotel, his/her academic formation, and for how many years he/she holds the current position. The second part consists of eight items classified under two groups of questions using a five-point Likert scale. These items are designed for budget participation and job satisfaction. The purpose of this part of the questionnaire was to explore respondents' perceptions about these two factors, budgetary participation, and job satisfaction. The scales included in the questionnaire have been used extensively in previous studies with satisfactory results for reliability and validity. The third part aims to survey the degree to which respondents perceive events in their lives as being a consequence of their own behavior and thereby controllable (internal control), or as being unrelated to their own actions, and therefore beyond personal control (external control). It consists of 19 items. Then, by counting the internal answers and the external ones, we can qualify the respondent as an external locus of control if the external answers exceed the internal ones and vice versa.

Model specification
The hypothesis was tested using moderated regression analysis (MRA). It is a specific application of multiple linear regression analysis, in which the regression model contains an interaction term (Jeon, 2015). This procedure is well suited to the modera detection of (Stone effects ting & Hollenbeck, 1984; Memon et al., 2019; Baron & Kenny, 1986;Hayes, 2017) and is widely used to test contingency hypotheses that predict interaction effects between budgetary participation and contextual variables (Chenhall, 1986;Tsui, 2001;Dunk, 2003;Chong et al., 2005). For this purpose, we fit the model presented below: Model 1 (1) where, SAT is the job satisfaction, BP is the budgetary participation, LOC is the locus of control, BP * LOC is the interaction term, and is the error term.
It should be noted that Baron and Kenny (1986) posited a condition to admit the moderating effect. It should be established that there is a significant interaction between the independent variable (budgetary participation) and the contingent variable (locus of control) affecting the dependent variable (job satisfaction). Thus, to support the moderating hypothesis, the coefficient in the regression Model 1 should be significant and positive. Before performing the regression analysis, several tests need to be conducted. Mainly, the presence/absence of a problem of multicollinearity is tested. Table 4 reports the correlations among the independent variables. As a rule of thumb, a correlation of 0.70 or higher in absolute value may cause multicollinearity between variables (Hayes, 2017). The correlation coefficient between BP and LOC is 0.089. To better assess the multicollinearity problem, we compute the variance inflation factor test (VIF). To assert the absence of multicollinearity between the variables of the model, the value of VIF must be less than 10 (Chatterjee & Hadi, 2012). As reported in Table 4, we can confirm that no serious multicollinearity problems exist between variables of our model.  Table  for  5  provides  descriptive  statistics  the  mean,  the  including  variables,  regression  minimu deviation, and standard median, m, Cronbach's a dependent The coefficient. lpha variable, SAT, has a mean value of 7.025 and varies between 2 and 10, notably close to Boujelbene and Affes (2012) who found that job satisfaction in Tunisian hotels spans from a minimum of 4 to 10 with a mean of 7.024. BP has a mean value of 22.21. Additionally, the mean of the LOC of the sample was 0.53 with a standard deviation of 5.8 that shows a wide variety in the locus of control among Tunisian hotels.

Regression results
Results of the regression analysis are reported in Table 6. According to this table, we note that the probability associated with the correlation test and the heteroscedasticity test is higher than the 5% threshold. Thus we can conclude the absence of any problem of correlation and heteroscedasticity of the errors of our regression. As the most important statistical tests have been performed, we will interpret the results obtained from the estimation of our empirical models.  Table 6, the interactive effect of locus of control on the relationship between budgetary participation and job satisfaction is examined, and the coefficient of BP * LOC is used to test this moderating effect. As BP * LOC is significant ( = 0.233), it can be argued that the locus of BP control moderates the -SAT link. Thus, H1 is However, the moderated regression confirmed. analysis, performing alone, can detect only the presence or absence of an interactive effect. It does not reveal how this factor influences the direct relationship. As proposed by Sharma, Durand, and Gur-Arie (1981) and Baron and Kenny (1986), we complete the moderated analysis regression with a sub-group analysis in order to determine how the locus of control moderates the budgetary participation-job satisfaction association.

Additional analysis
In order to better explore how the locus of control moderates the budgetary participation's impact on job satisfaction and as proposed by Baron and Kenny (1986), sub-group analysis was performed by dividing the sample into two balanced groups. Then, a linear regression analysis is conducted within each group based on the following models:
in Table  As shown the par 7, ticipation significant. We can statistically coefficients are the that locus of control significantly conclude moderates the budgetary participation-job satisfaction relationship. Middle-level managers have different attitudes toward budgetary participation. For Internal managers, budgetary participation affects positively and significantly ( = 0.412) their job satisfaction. Because they behave as active members of the group and aim to be involved in the decisional process of the company, internals managers prefer a participative budgeting system where they will have some effect on the budget that is jointly established. As such, a participative system provides them with opportunities to influence their environment and, consequently, to be active agents.  personal traits. Internals, because they behave as active members of the group and aim to be involved in the decisional process of the company, they prefer a participative budgeting system where they will have some influence on the budget that is set jointly. Externals managers prefer receiving and executing supervisors' orders. Thus, they react negatively to budgetary participation. These results are in line with several prior studies conducted in different socio-cultural contexts, namely American, Australian and Asian contexts (Brownell, 1981(Brownell, , 1982Krenl, 1992 practices' implementation by empirically explaining how some contingent variables contribute to or impede the development and the success of budget control tools. In this, regard, this study provides a theoretical contribution to the budgetary literature by suggesting the establishment of dynamic associations between budgetary participation and job satisfaction around the personality traits of managers involved. These results have also important practical implications for human resources management in the hotel industry. They highlight the importance of managers' budget participation as a relevant driver of positive attitudes. Thus, hotel managers should seek to value the contributions of their employees and demonstrate concern about their well-being, which will develop positive perceptions regarding organizational functioning. Our results also suggest that it is necessary for top managers to focus on the broader context in which this budgetary practice should be used. This latter has two aspects: structural and behavioral. The success of budget participation certainly depends on the organizational setting in which it is used but also on the psychological willingness of actors involved to develop and succeed in such budgetary practice. Despite these contributions, this study has some limitations that could be addressed in future research studies. The first limit concerns the framework chosen to explain budget behaviors. Admittedly, the locus of control as a personality trait has been shown to have a significant impact on budgetary participation' effects (Brownell, 1981; Frucot & Shearon, 1991; Leach-López et al., 2009); but it remains insufficient, on its own, to explain managers budget behaviors. Following Brownell (1982), four groups of variables (individual, cultural, organizational, and interpersonal) might affect budget participation. Further research should pay attention to all these variables in order to develop an integrated model indicating the conditions under which budgetary participation will generate beneficial effects. The second limit is linked to sample procedure selection. The research was conducted in hotels only. Consequently, the results reported in this paper may only apply to hotels. Therefore, next studies are essential to be conducted on other industries, which certainly have different organizational cultural contexts. The third limit concerns the data collection tool. The survey approach has limitations such as the lack of control over respondents (Nouri & Parker, 1996).