Household behavior in practicing mental budgeting based on the theory of planned behavior

“We can enrich our understanding of financial markets by adding a human element.”

Thaler (1999)

As every student of behavioral finance knows, explaining mental accounting (MA) in all of its complexities is a difficult task. Different researchers have approached mental accounting in different dimensions. For example, Thaler (1985) and Thaler and Johnson (1990) approached it in terms of hedonic editing. Henderson and Peterson (1992) studied gains and losses categorization, Shefrin and Thaler (1988) investigated income and assets labeling and earmarking, and Thaler (1985) studied borrowing and saving simultaneously. Thaler (1999), moreover, approached mental accounting in terms of transaction utility and acquisition utility, while Kivetz (1999) and Prelec and Loewenstein (1998) studied payment decoupling. Arkes and Blumer (1985), Shafir and Thaler (2006), Soman (2001), and Thaler (1999) studied the sunk cost phenomenon, and Read, Loewenstein, Rabin, Keren, and Laibson (1999) and Thaler (1999) studied choice bracketing. Finally, Heath and Soll (1996) approached mental accounting in the dimension of budgeting and provided a theory of mental budgeting.

Mental budgeting (Heath & Soll, 1996) is the psychological separation of the household budget. Within specific mental budget accounts, negative balances are avoided via the process of mental budgeting. Individuals avoid having a negative balance in any mental budget account. For example, households may keep three different mental budgets: food budget, outing budget, and savings budget. They allocate a certain amount/portion of money to each budget depending on the objective of developing that budget. Sources of money also affect the allocation of money to mental budgets, which are kept mentally separate from each other. Based on laboratory experiments where respondents were presented with scenarios, Heath and Soll (1996) reported that when a particular household budget becomes exhausted, people say they will spend less in that particular category at the end of the month. Meanwhile, Soman (2001), Soman (2003), and Soman and Lam (2002) found that an upcoming purchase in a particular budget category becomes unlikely when the previously paid amount was in the form of paper money, not plastic.

Mental budgeting is important for households for a couple reasons. First, it affects consumer decisions (Heath and Soll, 1996) such that a consumer may underconsume in a specific category if he or she made an unplanned purchase that is highly typical of that category. Also, people might often engage in mental budgeting processes to control their spending behavior and avoid debt (Bénabou & Tirole, 2004; Chattoe & Gilbert, 1999; Thaler, 1999). Similarly, households might perform mental budgeting to acquire savings. As Thaler (1985, 1999) noted, low-income consumers might be especially likely to make mental budgets, which in turn can affect whether savings are accrued.

Prior literature has shown how mental budgeting is related to household finances, but a question remains about which factors predict mental budgeting behavior. This study, therefore, aimed to explain the household behavior of managing finances while practicing mental budgeting based on the theory of planned behavior (TPB). Mental budgeting is involved in household money (finance) management (Antonides, de Groot, & van Raaij, 2011), which is the most important factor in financial capability (Atkinson, McKay, Kempson, & Collard, 2006). To investigate the determinants of mental budgeting behavior, we first present a brief overview of the theory of planned behavior (Ajzen, 1985, 1987, 1991). This study also extends the literature on TPB by empirically testing whether a modified TPB can explain mental budgeting behavior. Hypotheses are developed in next section. We then present the data collection tools, describe the data collection, and report the empirical results. The final section concludes the paper with some recommendations and future research directions.

Structural equation modeling (SEM) is a second-generation multivariate analysis technique. It is used to test the linkages between constructs (Gefen, Straub, & Boudreau, 2000; Hair Jr, Hult, Ringle, and Sarstedt, 2013). Two different but complementary statistical methods for estimating structural models are partial least squares (PLS) SEM (which is variance based) and covariance-based (CB) SEM (Hair Jr et al., 2013). CB-SEM is used for confirmatory research with a sufficient theoretical base to test the theory when the sample size is large (200–800). Meanwhile, for exploratory research (also for confirmatory research with an insufficient theoretical base), for prediction purposes, when the sample size is small, PLS-SEM is applied. This study met all of the criteria for using PLS-SEM.

The model (Fig. 1) used for this research involves multiple constructs. PLS-SEM is more rigorous compared to regression analysis for testing measurement and structural components simultaneously (Hair Jr et al., 2013; Henseler et al., 2009). Further, PLS-SEM seems more appropriate because some CB-SEM assumptions cannot be observed (Hulland, Ryan, & Rayner, 2010).

PLS-SEM is more appropriate if models and measures in any field are in the early stages of development (Fornell & Bookstein, 1982; Hair Jr et al., 2013). Current study is intended to explain TPB in a new context named as mental budgeting behavior. As such, the relationships between mental budgeting attitude, mental budgeting past behavior, mental budgeting intention, and mental budgeting behavior were tested for the first time; therefore, PLS-SEM was used.

Since this study used PLS-SEM (Hair, Sarstedt, Ringle, & Mena, 2012), the minimum sample size required for the conceptual model through SmartPLS was 30. The “rule of ten” was applied, as suggested by Hair et al. (2011), to determine the sample size and test the model using PLS-SEM. According to this rule, “the minimum sample size should be ten times the highest number of structural paths directed at a latent construct at a time” (Hair et al., 2011, p. 144). The sample for this study consisted of 275 respondents.

Univariate analysis was conducted to determine the central tendency and dispersions in the data set. Skewness and kurtosis were also tested for normal data distribution. While PLS-SEM modeling does not consider assumptions about data distribution, it is worthwhile to examine distributional properties (Hair Jr et al., 2013).

Further, bootstrapping was applied to test the statistical significance of various PLS-SEM results, including path coefficients. PLS-SEM does not assume data normality, which implies that parametric testing cannot be used to test statistical significance; rather, nonparametric tests are applied—that is, bootstrapping. In this procedure, subsamples are created with observations randomly drawn from the original data set with replacement. To ensure the stability of results, the number of subsamples should be large; in our case, the number of subsamples was 500.

The results of this study suggest that attitude influences households’ intention to engage in mental budgeting. Thus, H1 is supported. This finding aligns with many earlier studies that found that attitude is an important determinant of intention (e.g., Ajzen & Fishbein, 2000; Armitage & Conner, 2001; Cameron, 2010; Close et al., 2018; Kashif et al., 2018; Tariq et al., 2017).

The present study’s findings reveal that when people have positive, satisfactory, good, or favorable feelings about mental budgeting, they are more likely to have the intention to make mental budgets. Meanwhile, when people have negative, unsatisfactory, bad, or unfavorable feelings about mental budgeting, they are less likely to have the intention to make mental budgets.

The results also suggest that favorable and satisfactory mental budgeting attitudes can enhance mental budgeting behavioral intention, thus bringing about actual behavior toward mental budgeting among households. Attitude was shown to be positively related to intention, but this relation was weaker than those of past behavior-intention relations, which is in line with Honkanen et al. (2005) and Verbeke and Vackier (2005). Trafimow and Borrie (1999) found that past behavior might have a priming effect on future behavioral intention, making past behavior a predictor of intention to perform behavior later.

Trafimow and Borrie (1999) reasoned that performing a behavior repeatedly increases its cognitive accessibility, which again affects the intention to perform that behavior. The present study obtained the same finding since a strong relationship was found between mental budgeting past behavior and mental budgeting intention. Thus, H2 is supported—that is, in the case of household mental budgeting, when the past frequency of making mental budgets increases, it positively affects the intention toward mental budgeting. Increased past frequency increases cognitive accessibility, which affects the intention to perform mental budgeting behavior.

The results also demonstrated a strong relationship between mental budgeting intention and mental budgeting behavior. This aligns with previous studies showing that intention predicts later behavior (e.g., Athiyaman, 2002; Sheeran & Orbell, 2000; Tariq et al., 2017; Warshaw & Davis, 1984). When households have a strong intention toward mental budgeting, they ultimately perform it. Thus, we can conclude that the theory of planned behavior can predict mental budgeting behavior among Pakistani households. Further, we provide evidence showing that mental budgeting intention is a mediator between mental budgeting attitude and mental budgeting behavior, and mental budgeting past behavior and mental budgeting behavior.

This study aimed to explain the behavior of households when practicing mental budgeting behaviors. The explanation was based on the theory of planned behavior. We found that attitude and past behavior toward mental budgeting behavior significantly influenced mental budgeting behavioral intention. The results also showed that mental budgeting occurs frequently among young households in the Southern Punjab.

Most households showed positive attitudes toward mental budgeting. Mental budgeting has long entered into the daily lives of households. The findings also showed that households generally had satisfactory, positive, and favorable attitudes toward the intention to make budgets. We can conclude that when households have favorable feelings about mental budgeting, they instruct themselves to make mental budgets. When people have unfavorable feelings about mental budgets, they do not instruct themselves to make mental budgets. In this study, overall attitude was taken as a measure. Future research could expand attitude as per its bipolar attributes.

This study suggests that households in the Southern Punjab in Pakistan actively manage their budgets in the performance of consumer behavior. This finding is more suitable for young households. This study also contributes to the literature by adding the determinants of household mental budgeting behavior in light of TPB (Ajzen, 1991). This study generalized the factors of TPB on mental budgets. Previous studies, meanwhile, have determined the factors affecting money management behavior among college students (Kidwell, Brinberg, & Turrisi, 2003). The present study also contributes to the literature by investigating household mental budgeting behavior. Moreover, this study found that households might do mental budgeting because they are more interested in generating savings (Fisher & Anong, 2012; Xiao & Fan, 2002).

This study does have some limitations. The first is related to sample size. Although the sample size (n = 275) was large enough to use PLS-SEM (Hair Jr. et al., 2013), employed households were underrepresented. Second, this study was a first attempt to predict mental budgeting behavior based on TPB; thus, the PLS-SEM approach was used because the main purpose was to predict the determinants of mental budgeting behavior (Hair et al., 2011; Hair Jr. et al., 2013). PLS-SEM is based on nonparametric measures and does not consider distributional assumptions for parametric estimation (Chin, 1998; Hair Jr. et al., 2013). Future research should begin with a scoping study as the basis to form a theoretical framework and use a large-scale survey utilizing the CB-SEM approach to address the limitations of PLS-SEM, thereby confirming (or rejecting) the model proposed for determining mental budgeting behavior. Furthermore, younger respondents (21–30 years old) were overrepresented in the sample. For more comprehensive results, future research should collect data from respondents over age 30. In addition, stratified and cluster sampling are also recommended for future studies.