Applying the Theory of Planned Behaviour (TPB) in Saving Behaviour of Pomak Households

The following study uses the Ajzen’s Theory of Planned Behaviour as a theoretical framework with the purpose of extending a model which examines the saving behaviour of the Pomak households. The aim of this research is to identify factors that influence the saving behaviour of Pomak households in Greece. For the collection of data from households, a structured questionnaire was designed and distributed. The data of the study were obtained from a survey applied in 600 households in the prefectures of Xanthi, Rodopi and Evros. A path analysis was applied in order to confirm the proposed model and the results of the hypothesis tested. Findings show that attitude, subjective norm and perceived behavioural control have a direct positive effect on intention towards saving and final saving behaviour. The understanding of the Pomaks’ saving behaviour can have a significant contribution in the pursuit of a healthy household’s financial life.


Introduction
Saving can be considered the result of a decision-making process and to save as the act of regularly setting aside resources for a goal (Lewis et al., 1995;Warneryd, 1999). Household saving is essential because it affects a family's level of living, emergency reserves, and the ability to meet financial goals and expectations (Anong & DeVaney, 2010;Hira, 1987;Lee et al., 2000). Since, savings by families flow into financial institutions and that assets are essential for their own functioning as organizations that support the economy, it is clear that decisions upon saving made by households have a powerful effect on the economic performance.
The wealth of minority groups today is important in considering the wealth of country households in the future (Hanna & Lindamood, 2008). The society in the region of Thrace in Greece is multicultural. Thrace is acknowledged as a canvas depicting interesting social, cultural and religious aspects of a community which is consisted of Christians, Pomaks (Muslims), Turkish and Roma citizens. The Pomaks is a religious minority, approximately 36.000 settled in Greece and more specifically, 23.000 in the Prefecture of Xanthi, 11.000 in the Prefecture of Rodopi and 2.000 in the Prefecture of Evros (ELSTAT, 2011). A diverse religious environment may drive to different cultures, values, and norms, and may negative affect saving cost.
This study focused on saving behaviour of Pomak households, with particular emphasis on the relationship among attitude, subjective norm, perceived behavioural control and intention on the final saving behaviour of them. To achieve this goal, SEM techniques were applied in the light of evaluating and estimating the reliability and validity of the TPB scenario through the medium of confirmatory factor analysis. The above approach enabled the identification of some issues in order to evaluate the TPB constructs.

Theoretical Framework
The Theory of Planned Behaviour provides a useful conceptual framework for dealing with the complexities of human social behaviour (Ajzen, 1991) and is frequently used to explain behavioural patterns and better understand how individuals make behavioural decisions (Xiao & Wu, 2008). Ajzen indicated that the precedent of any behaviour is the formation of an intention towards that behaviour. TPB is applied in order to investigate the deeper beliefs that influence a person's financial behaviour (Zocchi, 2013). Consequently, it is vital that strategies be researched and designed to help individuals in adopting positive financial behaviours. Generally, individuals are motivated to save (Abrahamse & Steg, 2009;Karlsson et al., 2004). Nevertheless, a participant's capability to save was influenced by the lack of opportunity, lack of knowledge, lack of willpower, and their attitudes toward savings and financial institutions (Abrahamse & Steg, 2009;Karlsson et al., 2004). The TPB has constantly been applied in the social psychology literature and has been applied to investigations in the field of credit counselling, personal finance and private money management both domestically and internationally (Ajzen, 1991;Bobek et al., 2007;Francis et al., 2004;Kidwell & Turrisi, 2004;Ramayah et al., 2009;Rutherford & DeVaney, 2009;Xiao & Wu, 2008). Financial knowledge or literacy has been found to be positively associated with improved financial behaviours, including higher rates of savings (Huston et al., 2012;Lusardi et al., 2010;Lyons et al., 2006). The usage TPB based models on financial and saving behaviour displayed positive results, with 51% financial budget, keeping (Kidwell & Turrisi, 2004) to 72% retirement saving (Croy et al., 2010) of the variance in intention. The model interpreted 41% of variance in self-reported saving deposits (Loibl et al., 2011) and further anticipates negative economic behaviours (e.g., not paying bills, using payday loans) (Xiao et al., 2011). This study will reveal if attitude, subjective norm and perceived behavioural control will have a correlation and interconnection between them with saving behaviour. The direct positive relationship of attitude, subjective norm and perceived behavioural control underlined in studies conducted by Ajzen (1991;2002), Benk et al. (2011), Dowd & Burke (2013, Faye et al. (2013), Jen-Ruei et al. (2006), Litvine & Wüstenhagen (2011), Minton et al. (2015, Taylor & Todd (1995), Tsourgiannis et al. (2014), Zocchi (2013).

Research Model and Hypotheses
The model of TPB in this study was applied to explore the extent to which attitude (ATT), subjective norm (SN) and perceived behavioural control (PBC) influence Pomak households' intention (INT) to saving. The conceptual model is presented in Figure 1: Attitudes towards saving are supposed to be potent associated and rooted with upbringing and lifestyle. In a similar case, both Keynes (1936) and Katona (1975) claimed that most persons have positive attitudes towards saving. Generally, while most individuals are positive towards saving, they might vary in their attitudes towards concrete acts of saving and on private saving. Many studies have found the significant effect of attitudes on intention (Ing-Long & Jian-Liang, 2005;Lu et al., 2003;Ramayah et al., 2003;2009;Rhodes & Courneya, 2003;Taylor & Todd, 1995) and that a behavioural intention reflects a person's decision to perform the behaviour (Ajzen, 1991). Consequently, attitude can be regarded as an essential part of predicting and describing human behaviour (Ajzen, 1987). Therefore, the next hypothesis is proposed: H1: There is a significant and positive relationship between attitude and intention toward saving. Ajzen & Driver (1992) proposed that the subjective norm can predict behavioural intention. It is the most significant predictive factor of behavioural intention (Tsai, 2010). Subjective norms of individuals can affect behavioural intention (Bock & Kim, 2002;Kuo & Young, 2008). Moreover, as subjective norm becomes more positive, behavioural intention to participation also becomes more positive (Kuo & Dai, 2012). Further, studies have shown no significant relationship between SN and intention (Chau & Hu, 2001;Lewis et al., 2003) whereas other studies depict the opposing outcome (Chan & Lu, 2004;Jen-Ruei et al., 2006;Ramayah et al., 2004;Taylor & Todd, 1995;Venkatesh & Davis, 2000). It is with these findings that this study states the subsequent hypothesis: Ajzen (1985) identified perceived control behaviour as influencing behavioural intention and directly influencing actual behaviour. Perceived behavioural control have an important contribution in predicting behavioural intention and have a greater influence than attitude (Hsu & Huang, 2012). Between the factors affecting the behavioural intention of individuals towards participation, the current construct, has the most impact (Yang, 2013). As the PBC of an individual gets more positive, the behavioural intention to participate also transforms as a more positive (Kuo & Young, 2008). Godin et al. (2005) stated that perceived behavioural control can increase the explained variance towards behavioural intention. As a result, the following research hypotheses have been constructed: H3: There is a significant and positive relationship between perceived behavioural control and intention towards saving.
H4: There is a significant and positive relationship between perceived behavioural control and final saving behaviour.
Intention is a basic construct of the TPB and it is supposed to directly affect a given behaviour because it is an indication of how willing individuals are to perform the action (Ajzen, 1991;Armitage & Conner, 2001). According to the TPB, the amount of volitional control specifies, to what degree intentions are transformed into behaviour (Ajzen, 1991;Armitage & Conner, 2001). A behaviour such as saving money is not under full volitional control, because the availability of opportunities and resources affects the competence to execute the behaviour (Ajzen, 1991). Further, individuals with bigger incomes save more (Davis & Hustvedt, 2012;Hershey et al., 2008;Lusardi, 2008). In contrast, perceptions of barriers affect saving behaviour of persons who believe that they have not much money for saving and make them unwilling (Lusardi et al., 2009). Hence, the hypothesis is: H5: There is a significant and positive relationship between intention towards saving and final saving behaviour.

Sample and Procedure
The survey was given to 600 respondents in Greece and specifically in the region of Thrace in three different prefectures of Xanthi, Rodopi and Evros from April to October 2016. The self-completion questionnaire was administered face to face in several physical locations in this region and analyse to the respondents the aim of the nature of research. The sample included 190 women and 410 men. The adult head of the household was questioned in order to retrieve information for the household. Households in the survey were selected through the snowball sampling procedure. Respondents of different ages, genders, educational levels and marital status were chosen. The data collection process was accurate and careful.

Measures
All items intended to measure the variables in this study were adopted from previously validated instruments (Table  1).

Reliability
Cronbach's alpha was calculated for all the scales of the questionnaire (Table 2). For all academic scales that was used the internal reliability was good or very good.

Principal Component Analysis
Principal components analysis was applied, regarding to confirm factor structure of the structure participated in TPB model. The method was applied for each one part of the survey in order to confirm the univariate nature of the primary constructs as well as for the total questionnaire to confirm that the four constructs are separated.
The results of the PCA confirmed the one-factor solution for all the constructs that was selected to participate in the TPB model (   Table 5). Moreover, the application of the method for all questionnaire items suggested a four components solution explaining 54,6% of the total variance (Table 6). Thus, the desired structure was confirmed and the questionnaire was proved to be adequate for the research purpose. There is little I can do to change many of the important things in my life.

,655
I always try to pick saving schemes that yield high profits.

,671
It is important always to save as much money as possible.

,696
It is important to have some money left at the end of the month. ,722 Saving should be encouraged in today's society.
Rotation Method: Varimax with Kaiser Normalization.
a. Rotation converged in 5 iterations.

Confirmatory Factor Analysis
Structural equation modelling was applied in order to confirm the structure that was suggested from the principal components analysis (Figure 2 The results of composite reliability (CR) for all constructs exceeded the 0.7 and the average variance extracted (AVE) were closely on the recommended level 0.5. Consequently, these results indicated that the convergent validity for all constructs was adequate (see Table 7).

Path Analysis
The hypothetical model was assumed to be the model suggested by the theory. All 600 observations were used in order to estimate the coefficients. The model was converged after 57 iterations (Table 8). Chi square test (c 2 (2) = 0.954, p = 0.621) implies that the assumed model is adequate for the data, that is, the model and the data are not statistically significantly different. The model was statistical significant (c 2 (10) = 280.661, p < 0.001). The model was well fitted to the data (CFI = 1.000, TLI = 1.000, RMSEA = 0, 90% C.I. 0 -0.065, SRMR = 0.009, NFI = 0.997, NNFI = 1.00, GFI = 0.999, AGFI = 0.995).
The standardized regression coefficients are presented in Figure 3, while the covariance matrix is presented at Table  9.   The direct and indirect effects of the constructs on saving behaviour are presented in Table 10.

Discussion
The study showed that attitude had a significant and positive effect on saving behaviour of the Pomaks households, a hypothesis that was suggested by various authors in the literature (Ing-Long & Jian-Liang, 2005;Lu et al., 2003;Ramayah et al., 2003;2009;Rhodes & Courneya, 2003;Taylor & Todd, 1995). In particular, the households' heads with higher positive attitudes founded to have a greater intention to save. This finding agrees with previous studies (Choo et al., 2004;Fisher & Montalto, 2010;Lea et al., 1995;Warneryd, 1999;Webley & Nyhus, 2001) which supported the TPB theory concerns of attitude as predictor of intention towards saving.
In contrast with the previous constructs of TPB, the results concerning the perceived behavioural control indicated a positive statistically effect on intention towards saving, while it provided a direct negative effect on saving behaviour. The correlation between PBC and intention is positive, which means that the effect of self-control in explaining variability in behaviour is not rare. This is related to previous researches (Gopi & Ramayah, 2007;Ing-Long & Jian-Liang, 2005;Jen-Ruei et al., 2006;Lee, 2009;Taylor & Todd, 1995). However, Ryu et al. (2003) and Albarracin et al. (2001) demonstrate that the influence of PBC on intention towards saving, is moderate. In particular, the findings on the direct effect of intention on final saving behaviour were negative, which means that as self-mastery is increased, the final actual saving behaviour is reduced.

Implication of the Findings
In this study, the proposed framework model provides emphasis upon how the saving behaviour is influenced by the elements of TPB model, with the purpose of understanding the Pomak households' saving behaviour. The current results enhance the comprehensive of the relationship between Pomaks' characteristics and their effect on the intention towards saving. Furthermore, these findings will help banks to gain better understanding of the saving behaviour among Pomak households and enlighten them to create new opportunities for considering this minority group as part of their valuable customers. In particular, they can promote financial products or services according to their needs and preferences.

Directions of Future Research
This study was limited to Pomak households in Thrace (Greece) but there are Pomak populations in Bulgaria area as well. To improve the generalized ability of future findings, future studies could expand in order to include a sample of the neighbouring country. Therefore it suggests that the TPB conceptual, theoretical framework should be empirically validated in other regions of the world and the findings being compared to these ones. A more global, holistic understanding of minority population's attitudes, views and intention towards saving on saving behaviour can emerge. Deeper investigation is required for testing of TPB in Western and Muslim nations.