Behavioural biases among SME owners

Int. J. Management Practice, Vol. 11, No. 3, 2018 Behavioural biases among SME owners H. Kent Baker* Kogod School of Business, Department of Finance and Real Estate, American University, 4400 Massachusetts Avenue, NW, Washington, DC 20016, USA Email: [email protected] *Corresponding author Satish Kumar Department of Management Studies, Malaviya National Institute of Technology (MNIT), Jaipur 302017, India Email: [email protected] Harsh Pratap Singh G.L. Bajaj Institute of Management and Research, Knowledge Park III, Greater Noida, India Email: [email protected] Abstract: People are not fully rational and their decisions suffer from errors and biases. Because behavioural finance research focuses on investor irrationality, additional attention should examine managerial decision-making. This study helps to fill this gap by addressing three objectives: (a) to identify whether owners of small and medium-sized enterprises (SMEs) in India are prone toward behavioural bias; (b) to assess the impact of these biases on their decision-making; and (c) to determine whether age, experience, education, and gender affect their propensity to exhibit behavioural biases. Using responses from 154 Indian SME owners, the study uses logistic regression to assess how demographic variables affect behavioural biases. The evidence shows that SME owners are prone to self-attribution, overconfidence, and loss aversion, but not anchoring. Differences exist in the working capital management decisions of owners prone to behavioural biases. Gender, age, and experience significantly affect the propensity to exhibit behavioural biases. Keywords: behavioural biases; overconfidence bias; self-attribution bias; loss aversion bias; anchoring bias; working capital management. Reference to this paper should be made as follows: Baker, H.K., Kumar, S. and Singh, H.P. (2018) ‘Behavioural biases among SME owners’, Int. J. Management Practice, Vol. 11, No. 3, pp.259–283. Copyright © 2018 Inderscience Enterprises Ltd. 259 260 H.K. Baker, S. Kumar and H.P. Singh Biographical notes: H. Kent Baker is University Professor of Finance at the Kogod School of Business, American University in Washington, DC. He has published 29 books and more than 170 refereed journal articles. He is a leading finance academic in survey research and co-authored Survey Research in Corporate Finance (Oxford University Press). Satish Kumar is Assistant Professor in the Finance area at the Malaviya National Institute of Technology Jaipur. He has more than 13 years of teaching and research experience. He has published 24 refereed journals articles. His research interests include corporate finance, behavioural finance, and survey research in finance. Harsh Pratap Singh is Assistant Professor of Finance and Accounting at the G.L. Bajaj Institute of Management & Research, Greater Noida, India. His areas of interest include behavioural decision making, SME financing and corporate finance. 1 Introduction Classical finance theories assume that individuals are rational and make decisions based on expected utility maximisation (Singh et al., 2016). However, reality does not match these assumptions as in practical situations individuals are not fully rational. There is growing literature in the field of experimental psychology to indicate that people generally deviate from this traditional paradigm of rationality (Hackbarth, 2008). Their decision making is influenced by various behavioural factors such as moods, emotions and personality traits (De Bondt and Thaler, 1987; Todd and Gigerenzer, 2003). Although much behavioural research focuses on the irrationality of financial decision makers, a largely neglected area concerns financial decisions involving working capital management (WCM) (Modi, 2012). Our study attempts to help fill this gap by examining small and medium-sized enterprises (SMEs) in India. The objective of this study is three folded (1) to identify whether owners of small and medium-sized enterprises (SMEs) in India are prone toward behavioural bias namely overconfidence, self-attribution, anchoring and loss aversion; (2) to assess the impact of these biases on WCM decision-making of Indian SME owners; and (3) to determine whether age, experience, education, and gender affect the propensity of Indian SME owners to exhibit these behavioural biases. Our study contributes to the finance literature by being the first study to identify and assess behavioural biases in the decision-making of Indian SME owners related to WCM. It also extends Ramiah et al. (2014) by examining whether demographic variables affect specific behavioural biases. Ramiah et al. are the first to explore the behavioural aspects of WCM by identifying biases in the WCM practices adopted by corporate treasurers of large Australian firms. However, their findings may not apply to SMEs for several reasons. First, SMEs may differ from large firms in terms of formal processes used in WCM. Second, SME owners may not have the same level of financial sophistication as corporate treasurers in large firms. Third, cultural and other differences may exist between managers of Australian and Indian firms. Behavioural biases among SME owners 2 261 Literature review Behavioural finance integrates traditional finance theories, psychology and sociology to explain human behaviour in financial decisions (Ricciardi and Simon, 2000). Traditionally, behavioural factors of financial agents are not incorporated in theoretical and empirical research in the field of finance (Mendes-da-Silva et al., 2015). Classical finance theories assume that individuals are rational and make decisions based on expected utility maximisation (Singh et al., 2016). However, reality does not match these assumptions as in practical situations individuals are not fully rational. There is growing literature in the field of experimental psychology to indicate that people generally deviate from this traditional paradigm of rationality (Hackbarth, 2008). Because decision makers are not fully rational, they are subject to various biases (Ritter, 2003). Research on behavioural biases can provide important insights about decision-making processes (Sahi and Arora, 2012). Over the years, extensive literature has been developed on behavioural biases and researchers have identified a long list of biases which mainly includes representativeness, overconfidence, anchoring, loss aversion, self-attribution, mental accounting, overreaction, and herding. These biases influence the behaviour and choice of decision makers and thus need to be examined to get important insights into the decision-making process (Sahi and Arora, 2012).The literature on behavioural finance advocates two approaches. The first approach deals with the behaviour of investors and the second approach deals with the behaviour of corporate managers (Baker et al., 2004). The major development in the field of behavioural finance incorporates the behaviour of investors in their decision making (Odean, 1999; Barber and Odean, 2001; Bhandari and Deaves, 2006; Cheng, 2007; Sahi and Arora, 2012; Prosad et al., 2015). On the contrary, the literature related to irrationality of corporate finance managers is less developed (Baker et al., 2004). The literature on behavioural corporate finance is primarily dominated by studies on the effect of behavioural biases like optimism and overconfidence on decision making in the context of investment decisions and decisions related to structure of financing (Baker et al., 2004). However, evidence related to the effect of behavioural biases on short-term financial decisions especially those related to WCM is non-existent except for Ramiah et al. (2014) despite the fact that it is an important yardstick to measure a firm’s operational and financial efficiency. Adding to this literature, we examine whether behavioural biases affect WCM decisions specifically focusing on self-attribution, overconfidence, loss aversion, and anchoring bias. 2.1 Self-attribution bias Self-attribution bias is the tendency to attribute positive outcomes to one’s own ability and negative outcomes to outside forces or “bad luck” (Shefrin, 2007). People exhibit self-attribution bias to maintain high self-esteem and positive feelings about themselves. In terms of WCM decisions, Ramiah et al. (2014) find that corporate treasurers with selfattribution bias are more aggressive in making financing decisions. Studies reveal that men have a higher locus of control than women and high self-esteem, which contributes to men exhibiting self-attribution. Deaux (1979), Rosenthal et al. (1996), and Shuch Mednick and Weissman (1975) find that men have a greater tendency than women to attribute success to ability and failure to luck. Billett and Qian (2008) focus on the self-attribution bias of CEOs and its effect on mergers and acquisition deals and concluded that high-order deals are motivated by previous positive acquisition 262 H.K. Baker, S. Kumar and H.P. Singh experience. In terms of WCM decisions, Ramiah et al. (2014) find that corporate treasurers with this bias are more aggressive in financing and in techniques which are under their own control in managing working capital. 2.2 Overconfidence bias Overconfidence bias is when someone’s subjective confidence is greater than the objective accuracy of the judgments. Thus, overconfident people usually overestimate their ability and ignore the actual risk involved in decision-making. Because overconfident managers rely too much on their own judgment due to feelings of superiority (Agrawal, 2012), they tend to be slow in combining additional information about any decision-making situation. In addition, overconfident investors tend to trade excessively which usually results in poor returns (Barber and Odean, 2001). Although both men and women are prone to overconfidence bias (Lundeberg et al., 1994), men exhibit more confidence than women especially in financial matters (Prince, 1993). Studies such as Barber and Odean (2001), Jaiswal and Kamil (2012), Mittal and Vyas (2011), and Singh et al. (2016) find that male investors are more overconfident in their decision-making. Similarly, in case of financial experts, Beckmann and Menkhoff (2008) report that female financial experts are more risk averse and less overconfident than their male counterparts. Frascara (1999) and Heath and Tversky (1991) conclude that experienced people are more likely to be overconfidence. In general, past experience in making successful decisions leads to greater overconfidence. Kirchler and Maciejovsky (2002) find that the degree of overconfidence increases as the length of professional experience increases. The degree of overconfidence also increases with increases in level of education. For example, Bhandari and Deaves (2006) show that highly educated males are more likely to show overconfidence in their decision-making. 2.3 Loss aversion bias Loss aversion is an important concept in behavioural finance. According to Tversky and Kahneman (1991), people fear losses more that they value gains. Because they feel more pain in the case of a loss compared to the joy resulting from an equal amount of gain, people are willing to undertake higher risks to avoid losses but become risk averse in the case of gains. Ramiah et al. (2014) find loss-averse corporate treasurers do better at controlling bad debts, are less likely to use a leading and lagging approach in cash management, and focus more on the cash conversion cycle to monitor WCM effectiveness. Evidence shows demographic characteristics affect loss aversion. Rau (2014) finds that women are more risk averse than are men in their investment decisions and also trade less frequently. Similarly, Johnson et al. (2006) distinguish the tendency to exhibit loss aversion on the basis of age and education level. Johnson et al. (2006) find that older and less educated people are more prone to loss aversion biases as compared to young and highly educated people. Loss version bias also affects the decision making of corporate managers especially in the case of decision making related to a firm’s financing (Jarboui and Ali, 2012). Kisgen (2006) show that managers with loss aversion bias avoid debt financing as it increases the chances of bankruptcy. Similarly, in the field of WCM, Ramiah et al. (2014) found loss-averse corporate treasurers do better in bad debt control and keep bad debt under the 1% level. Behavioural biases among SME owners 263 They also find that corporate treasurers with loss aversion bias are less likely to use the leading and lagging approach in cash management and focus much on the CCC to monitor the effectiveness of WCM. 2.4 Anchoring bias Anchoring bias is the tendency to rely too heavily on the first piece of information offered when making decisions. Decision makers often consider past events and trends as anchors. Thus, anchoring biases are related to the human tendency to attach or "anchor” any thought with some reference point without logically evaluating the relevance to the decision under consideration. People tend to consider past events and trends as anchors. In the case of investment decisions, share price that investors use to compare the current share price is called the reference point. Usually, the purchase price of a security serves as a reference point for decision making of investors (Baker and Nofsinger, 2002). In studying the effect of anchoring bias on WCM decision-making, Ramiah et al. (2014) find that treasurers are more likely to use term sheets but less likely to use a bill of exchange for financing working capital needs. Researchers mainly study anchoring bias from an investor’s viewpoint because the effect of anchoring bias in managerial decisionmaking is limited. 3 Research methodology The following sections discuss our research questions and hypotheses, sample and limitations, questionnaire design, and data analysis. 3.1 Research questions and hypotheses People often deviate from the traditional paradigm of rationality because various behavioural factors such as moods, emotions, and personality traits influence their decision-making (Todd and Gigerenzer, 2003). According to Tversky and Kahneman (1974), people use cognitive heuristics in complex decision-making situations resulting in behavioural biases. Many researchers find that behavioural biases influence the decision-making process of investors (Bhandari and Deaves, 2006; Cheng, 2007; De Bondt and Thaler, 1987; Sahi and Arora, 2012; Peteros and Maleyeff, 2013; Prosad et al., 2015; Rzeszutek et al., 2015; Tarim, 2016). Regarding corporate managers, Fairchild (2005), Malmendier and Tate (2008), and Ramiah et al. (2014) advocate the irrationality in decision-making. Consistent with this stream of research, we formulate the following hypothesis. H1a: Indian SMEs owners exhibit self-attribution bias when making WCM decisions. H1b: Indian SMEs owners exhibit overconfidence bias when making WCM decisions. H1c: Indian SMEs owners exhibit loss aversion bias when making WCM decisions. H1d: Indian SMEs owners exhibit anchoring bias when making WCM decisions. Demographic factors can also influence how susceptible decision makers are to behavioural biases (Acker and Duck, 2008; Kourtidis et al., 2009; Kuo et al., 2005; Lin, 2011). In terms of overconfidence, Bhandari and Deaves (2006) and Barber and Odean 264 H.K. Baker, S. Kumar and H.P. Singh (2001) find that males show greater overconfidence compared to females in investment decisions. Similarly, Frascara (1999) and Heath and Tversky (1991) find a positive association between overconfidence and experience. According to Mishra and Metilda (2015), self-attribution bias increases with the level of education. Deaux (1979) and Rosenthal et al. (1996) show that males are more prone to self-attribution bias than females. Regarding loss aversion, Johnson et al. (2006) observe that older people are more likely to exhibit loss aversion bias than their younger counterparts. Rau (2014) also finds that females are more loss averse than males. Based on these arguments, we examine whether gender, age, education, and experience of SMEs owners are associated with behavioural biases based on the following hypothesis. H2a: Demographic variables − gender, age, education, and experience significantly affect the tendency of Indian SME owners to exhibit self-attribution bias. H2b: Demographic variables − gender, age, education, and experience significantly affect the tendency of Indian SME owners to exhibit overconfidence bias. H2c: Demographic variables − gender, age, education, and experience significantly affect the tendency of Indian SME owners to exhibit loss aversion bias. H2d: Demographic variables − gender, age, education, and experience significantly affect the tendency of Indian SME owners to exhibit anchoring bias. 3.2 Sample and limitations Given the impracticality of surveying all Indian SMEs, we focus on manufacturing firms operating in Rajasthan, which is among the top 10 states in terms of the number of registered enterprises and among the top five states in terms of exports. We study manufacturing SMEs because their WCM decisions are more important than services firms (Padachi et al., 2012). WCM is vital to manufacturing firms because they hold larger inventories and accounts receivable. The business directory of Sunrise Consultancy Services (SCS) provides the names, addresses, and contact details of business firms and contains data on 481 manufacturing SMEs operating in Rajasthan. Of these firms, 42 SME owners agreed to participate in the survey. To increase the sample size, we used a database including 1,493 manufacturing firms from Vishwakarma Industries Association (VKI) Jaipur, one of the largest industrial associations in Rajasthan. Our final sample excludes 160 firms due to a lack of contact information and 32 firms previously provided by SCS. Of the 1,301 remaining firms, 112 agreed to provide data. Thus, 154 SME owners participated in our survey resulting in an 8.7% response rate, which is similar to Graham and Harvey (2001) and Ramiah et al. (2014). To test the consistency between the 42 SMEs from the SCS business directory and the 112 SMEs from VKI’s database, we use chi-square tests involving three firm characteristics (firm size, export, and ownership structure) and four owner characteristics (gender, age, education, and experience). Finding no statistically significant differences at the 0.05 level between the two samples for any of the characteristics reduces our concern about the sample consistency. These results are available from the authors. A potential limitation of survey research is non-response bias. Because firm- and owner-specific data are unavailable on non-respondents, we use an approach suggested Behavioural biases among SME owners 265 by Wallace and Mellor (1988) to test for non-response bias. The basis of this approach is the premise that late responders tend to resemble non-respondents more than do early responders. We categorised the 68 respondents who initially answered our survey without further follow up as early responders and the remaining 86 respondents as late responders. To test for non-response biases, we compare the same three firm characteristics and four owner characteristics as previously discussed between early and late responders using chi-square tests. Finding no statistically significant differences at the 0.05 level between the two samples for any of the characteristics lessens our concern about non-response bias. These results are available from the authors. 3.3 Questionnaire design To develop a questionnaire, we started with the WCM surveys of Belt and Smith (1991) and Ramiah et al. (2014) and then modified our instrument to apply to SMEs. Next, we asked subject experts and officials of the Ministry of Micro Small and Medium Enterprises (MSME) to review the initial draft of the questionnaire for content validity. After incorporating various suggestions, we prepared the final questionnaire and administered it to 154 SME owners. The questionnaire is available from the authors. To minimise time and effort in completing the questionnaire, we used only closeended questions, which managers prefer (Greer et al., 2000). The questionnaire has two parts: the first part identifies the WCM practices of SMEs and the second part identifies potential behavioural biases in the decision-making of SME owners. We discuss the question design for each behavioural bias below. 3.3.1 Self-attribution bias People often attribute success to their personal capabilities and blame external factors for their failures (Miller and Ross, 1975). Based on this proposition, we design two questions to capture self-attribution bias. Specifically, the questionnaire asks SME owners to consider two situations: (1) when their firm is in financial distress and (2) when it is experiencing good financial performance. They are asked to attribute each situation to their own financial policy and the economic environment using a five–point scale where 1 = not at all, 2 = somewhat, 3 = moderately, 4 = highly, and 5 = extremely. We identify self-attribution bias based on the combination of responses to these two questions. If respondents rate both questions a 4 or 5, we classify them as prone to selfattribution bias. Otherwise, we consider them as not being prone to self-attribution bias. However, if respondents give a 1, 2, or 3 to one question and a 4 or 5 to the other question, we label the situation as “other.” 3.3.2 Overconfidence bias We design questions to capture overconfidence bias by asking SME owners to rate their confidence in cash management in two situations: (1) when their firm’s performance is strong and (2) when their firm’s performance is poor using a five-point scale where 1 = not at all confident, 2 = somewhat confident, 3 = moderately confident, 4 = highly confident and 5 = extremely confident. Overconfidence bias is the combination of responses to these two questions. We classify respondents who give a rating of 4 or 5 for 266 H.K. Baker, S. Kumar and H.P. Singh both questions as prone to overconfidence bias. Otherwise, we consider them not prone to overconfidence bias. We classify respondents with a 1, 2, or 3 on one question and a 4 or 5 on the other question as “other.” 3.3.3 Loss aversion bias The questionnaire contains two questions designed to identify the tendency of loss avoidance. One question asks SME owners to rate their disappointment with a total annual bad debt of 5% and 10% of sales revenue on a five-point scale where 1 = not at all disappointed, 2 = somewhat disappointed, 3 = moderately disappointed, 4 = highly disappointed, and 5 = extremely disappointed. The other question asks them to rate their satisfaction with a total annual profit of 5% and 10% of sales revenue on a fivepoint scale where 1 = not at all satisfied, 2 = somewhat satisfied, 3 = moderately satisfied, 4 = highly satisfied, and 5 = extremely satisfied. If a SME owner’s rating for disappointment is higher than rating for satisfaction at both 5% and 10% sales level, we classify them as loss averse. If their rating in case of disappointment is equal to or less than the rating of satisfaction at both 5% and 10% sales level, we view them as not loss averse. We consider any other combination as “other.” 3.3.4 Anchoring bias To capture anchoring bias, the questionnaire includes a situation about making credit sales to a low credit rated company A that has paid on time. The questionnaire asks SME owners to rate their chances of making credit sales to company A and another low rated company B in the future using a five-point scale where 1 = not at all likely, 2 = somewhat likely, 3 = moderately likely, 4 = highly likely, and 5 = extremely likely. We identify anchoring bias based on the responses to these two options. If SME owners rate both questions a 4 or 5, we consider them prone to anchoring bias because they ignore the credit rating and give credit based on a good recent payment history. If SME owners select 1, 2 or 3 for both situations, we do not consider them prone to anchoring bias. Finally, we categorise any other combination as “other.” Table 1 presents the operational definitions of each of all four behavioural biases analysed in the present study. Table 1 Operational definition of research variables. This table describes operational definition of research variables used in the present study Research variable Operational definition Self-Attribution bias To identify the behavioural bias among SME owners, situational question related to financial decision-making on a Likert type scale have been asked. Self-attribution is identified in this study by the tendency of SMEs owners to attribute their good financial performance to internal factors and blame of financial distress to the external factors. Overconfidence bias The overconfidence bias is identified by the tendency of SME owners to overestimate their capabilities. In this study overconfidence owners are those who give a rating of 4 and 5 when they asked to rate their confidence level in cash management in good times and bad times on a scale of 1 to 5. Behavioural biases among SME owners Table 1 267 Operational definition of research variables. This table describes operational definition of research variables used in the present study (continued) Research variable Operational definition Loss Aversion bias Loss Aversion Bias is identified by the tendency of SME owners to feel more pain in the case of a loss compared to the joy resulting from an equal amount of gain. In this study loss averse owners are those whose rating for disappointment is higher that the rating for satisfaction for the same amount of loss and profit respectively. Anchoring bias Anchoring bias is the tendency to rely too heavily on the first piece of information offered when making decisions. In this study owners with anchoring bias are those who give credit to low rated Company B on the basis of credit history of similar low rated Company A 3.4 Data analysis To determine the effect of behavioural biases, we group responses based on selfattribution bias (yes or no), overconfidence bias (yes or no), anchoring bias (yes or no), and loss aversion bias (yes or no). To test the difference between the WCM practices of groups based on the above factors, we use an independent sample t-test and a chisquare test. To compare the WCM practices measured on a nominal scale, we use a chi-square test for independence. Additionally, we use an independent t-test to compare WCM practices measured on five-point Likert’s scale between groups. Finally, to examine the third research objective of assessing the effect of demographic variables on the tendency to exhibit behavioural biases (self-attribution, overconfidence, loss aversion, and anchoring) involves using a binary logistic regression. Instead of employing continuous independent variables, this study includes categorical independent variables. Similar to independent variables, the dependent variable is also a dichotomous categorical variable (Field, 2009). We use binary logistic regression model to model the relation between the tendency to exhibit behavioural biases and demographic characteristics of SME owners (i.e., age, gender, education, and experience of SME owners). We estimate a respondent’s probability being prone to self-attribution bias. However, due to th e limited value of probability, we cannot use these probabilities directly in the regression models and instead use the odd [P(1 ‒ P)]. Further, we calculate the natural log of the odds so that the relation can be linearised and treated as in multiple linear regressions. Finally, the logistic model used can be expressed as follows: Log (P/1 ‒ P) SEB = B0 + B1(GEN) + B2(AGE) + B3(EDU) + B4(EXP) + ei Model 1 Log (P/1 ‒ P) OB = B0 + B1(GEN) + B2(AGE) + B3(EDU) + B4(EXP) + ei Model 2 Log (P/1 ‒ P) LAB = B0 + B1(GEN) + B2(AGE) + B3(EDU) + B4(EXP) + eiModel 3 Log (P/1 ‒ P) AB = B0 + B1(GEN) + B2(AGE) +B3(EDU) + B4(EXP) + ei Model 4 where SEB = self-attribution bias; OB = overconfidence bias; LAB = loss aversion bias; AB = anchoring bias; P = probability of a respondent with self-attribution bias; Gender = 1 if the respondent is male and 0 otherwise; AGE = 1 if the respondent is less than 40 years of age and 0 otherwise; EDU = 1 if the respondent has a secondary education and 0 otherwise; and EXP = 1 if the respondent has less than 10 years of experience and 0 otherwise. H.K. Baker, S. Kumar and H.P. Singh 268 4 Research findings Table 2 presents a firm and respondent profile. Regarding firm characteristics, SME owners use a low level of leverage (73.4%), do not engage in export (76.6%), are microsized (55.8%), and are sole proprietorships (57.8%). They are typically male (88.3%), are more than 40 years old (51.3%), have at least 10 years of experience (55.2%), and have engaged in higher education (78.6%). Table 2 Firm and respondent profile of Indian SMEs. This table describes both firm and respondent characteristics of the sample of 154 Indian SMEs Firm characteristics % Level of leverage Respondent characteristics % Gender Low 73.4 Male 88.3 High 26.6 Female 11.7 Export Age No 76.6 Young (< 40 years) 48.7 Yes 23.4 Old (≥ 40 years 51.3 55.8 Low (< 10 years) 44.8 Small 37.7 High (≥ 10 years 55.2 Medium 6.5 Size of Firm Micro Experience Ownership Structure Education Sole proprietorship 57.8 Senior secondary 21.4 Partnership 18.2 Higher education 78.6 Private limited company 24.0 4.1 Self-attribution bias Consistent with the argument presented by Miller and Ross (1975), the questionnaire poses two questions: (1) When your firm is in financial distress, to what extent do you blame your own financial policy and the external environment respectively? and (2) In times of good financial performance, to what extent do you think your own financial policy and the external environment have contributed? As Figure 1 shows, during times of good performance, SME owners attribute success more often to internal (3.77) than external factors (3.19). During times of poor performance (i.e., financial distress), they blame external factors (3.74) more than internal factors (3.23), when measuring the means on a 1 to 5 scale. Table 3 shows statistically significant differences between the rating of both internal (i.e., the firm’s own financial policy) and external (i.e., the economic environment) factors in the case of poor and good performance. The evidence supports H1a suggesting that SME owners are generally prone to selfattribution bias when making WCM decisions. These results imply that SME owners with self-attribution bias tend to focus more on things within their control than external factors. As a result, they tend to focus on internal funding for working capital financing and avoid using such techniques as factoring and outsourcing to manage their Behavioural biases among SME owners 269 receivables. However, having self-attribution bias many not always negatively affect WCM. In some instances, such managers may be more aggressive in trying to reduce the cost of financing working capital components. Figure 1 Self-attribution bias. This figure shows the means of internal (own financial policy) and external (the economic environment) factors in the case of strong and poor firm performance among Indian SME owners using a scale from 1 to 5 Table 3 Results of the paired t-tests for self-attribution bias of Indian SMEs. This table shows the results of paired t-tests for the mean difference between internal (i.e., the firm’s own financial policy) and external (i.e., the economic environment) factors in describing of poor and good performance Pair Mean difference t-statistic Degrees of freedom p-value 1 Blame poor performance to internal factors – Attribute strong performance to internal factors −0.532 −5.027 153 0.000 2 Blame poor performance on the economic environment – Attribute strong performance to the economic environment 0.552 6.145 153 0.000 As Table 4 shows, 63 (40.9%) SME owners exhibit self-attribution bias in their decisionmaking and 52 (33.8%) do not. Because we cannot determine whether 39 (25.3%) SME owners exhibit self-attribution bias, we exclude them from further analysis. We group the remaining respondents into SME owners with and without self-attribution bias. Table 4 Identification of behavioural biases in SME owners. This table shows the percentage of the 154 SME owners with or without a specific behavioural bias Behavioural bias Yes (%) No (%) Other (%) Self-attribution 40.91 33.77 25.32 Overconfidence 65.58 22.73 11.69 Loss aversion 57.14 26.62 16.23 Anchoring 19.48 53.25 27.27 We use a binary logistic regression to test H2a related to the effect of demographic variables on self-attribution bias. As Table 5 shows, the level of self-attribution bias varies among the demographic characteristics and that Model 1 correctly classifies 270 H.K. Baker, S. Kumar and H.P. Singh 77.4% of the observations, which shows a strong goodness of fit. The Hosmer and Lemeshow test is significant and provides another indicator of the model’s goodness of fit. The coefficients of age (1.873) and experience (1.227) are positive and significant at the 0.01 and 0.05 levels, respectively, which support H2a for age and experience. The positive signs show older and highly experienced owners are more prone to selfattribution bias. The coefficient for gender is also positive and statistically significant at the 0.10 level, supporting H2a for gender and indicating that males are more likely to exhibit self-attribution bias than females. These latter results are consistent with Deaux (1979) and Rosenthal et al. (1996) who report that women managers attribute their success less strongly to ability than do men. Because the results do not support H2a, owner education does not appear to affect the propensity to exhibit self-attribution bias. Logistic regression model for self-attribution bias. This table shows the results of logistic regression Model 1 for self-attribution bias. The reference categories are: female (gender), young (age), secondary (education), and low (experience) Table 5 Demographic variable Gender β Standard error Wald Degrees of freedom p-value Exp(β) 1.559* 0.846 3.396 1 0.065 4.755 1.873*** 0.581 10.403 1 0.001 6.506 Education 0.604 0.574 1.104 1 0.293 1.829 Experience 1.227** 0.578 4.501 1 0.034 3.411 Constant ‒3.331 1.057 10.168 1 0.001 0.034 R2 0.328 Age Hosmer and Lemeshow test 1.767 (0.778) Correct classification 77.4% Note: *’**’*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. Finally, we compare WCM practices adopted by SME owners with and without selfattribution bias. As Table 12 shows, self-attribution bias affects the working capital financing preferences of SME owners. Owners with self-attribution bias show a higher preference for retained earnings, cash credit, leasing, suppliers’ credit, government sponsored schemes, buyers’ credit, and letters of credit. Owners without self-attribution bias express a higher preference for loans from money lenders. Table 13 indicates that owners with self-attribution bias use the cash conversion cycle to monitor and manage working capital. According to Table 14, owners with and without self-attribution bias do not significantly differ in terms of cash management approaches. According to Table 15, owners with self-attribution bias pay more attention than those without this bias to such external factors affecting cash management as currency exchange rates, the level of inflation, and interest rates. Those without self-attribution bias view market conditions as having a stronger effect on their firm’s cash management. As Table 16 shows, owners with self-attribution bias rely more on material requirement planning, inventory models, ERP systems, supply chain management, and sales forecasting compared to those without self-attribution bias. Behavioural biases among SME owners 271 4.2 Overconfidence bias Figure 2 shows how SME owners rate their confidence in cash management on a five-point scale. In times of both good and bad performance, the level of confidence exceeds 3 (moderately confident), which Table 6 shows is statistically significant at the 0.01 level. This significance difference supports H1b indicating that SME owners are generally overconfident. Figure 2 Table 6 Level of confidence. This figure shows the mean confidence rating of Indian SME owners in the case of poor and strong firm performance using a scale from 1 to 5 Results of a one sample t-test for overconfidence bias. This table shows the results of one sample t-test for comparing the level of confidence in case of strong and poor performance against a value of 3 (moderately confident) Mean difference t-Statistic Degrees of freedom p-value Level of confidence in case of strong performance 1.344 21.423 153 0.000 Level of confidence in case of poor performance 0.708 10.462 153 0.000 As Table 4 shows, 101 (65.6%) SME owners exhibit overconfidence bias in their decision-making but 35 (22.7%) do not. We place the remaining 18 (11.7%) owners in the “other” category because they provide a rating of 4 or 5 in the case of good performance and a 1, 2 or 3 in the case of poor performance to cash management. Overall, the results show that the majority of owners are overconfident, which is consistent with Ramiah et al. (2014) involving overconfident corporate treasurers. We exclude the 18 owners whose status is not confirmed from further analysis. Next, we group the remaining respondents in SME owners with and without overconfidence bias. Hribar and Yang (2015) contend that overconfidence makes individuals more optimistic about uncertain outcome. They find that overconfident managers are more likely to issue forecasts that are overly optimistic. To counter this bias, firms experiencing a high level of sales variability should be cautious about employing overconfident managers because this could lead to overestimating future sales and excessive inventories. To test H2b, we use a binary logistic regression to assess the effect of demographic variables on overconfidence bias. As Table 7 shows, the chi-square value for the Hosmer and Lemeshow test indicates a good overall fit with the model correctly classifies 75.7% 272 H.K. Baker, S. Kumar and H.P. Singh of the cases. As Table 7 shows, gender has a significantly positive coefficient, which supports H2b for gender indicating that males are more overconfident than females. This finding is consistent with Barber and Odean (2001), Mittal and Vyas (2011), Jaiswal and Kamil (2012), and Singh et al. (2016) in an investments context. Similarly, Table 7 supports H2b for experience and reveals that overconfidence bias increases with the length of professional experience of SME owners. Mishra and Metilda (2015) report similar results involving stock investors. Additionally, the coefficient of age is statistically significant at the 0.1 level supporting H2b for age. Thus, older SME owners are more overconfident than their younger counterparts. Logistic regression model for overconfidence bias. This table shows the results of logistic regression Model 2 for overconfidence bias. The reference categories are: female (gender), young (age), secondary (education), and low (experience) Table 7 β Standard error Wald Degrees of freedom p-value Exp(β) Gender 1.577** 0.654 5.819 1 0.016 4.839 Age 0.900* 0.540 2.785 1 0.095 2.461 Education 0.097 0.573 0.029 1 0.865 1.102 Experience 1.127** 0.537 4.409 1 0.036 3.087 Constant ‒1.292 0.838 2.376 1 0.123 0.275 Demographic variable R 2 0.180 Hosmer and Lemeshow test 4.857 (0.434) Correct classification 75.7% Note: *’**’*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. To determine whether overconfidence bias affects the decision-making of SME owners related to WCM, we compare the WCM practices adopted by owners with and without overconfidence bias. In terms of working capital financing, Table 12 shows that overconfident owners rely more on suppliers’ credit and government sponsored schemes such as a credit guarantee fund than those who do not display overconfidence bias. In terms of cash management approaches, Table 14 shows that overconfident owners are likely to maintain emergency liquidity reserves relative to those without overconfidence bias. In terms of external factors affecting cash management, Table 15 shows no significant differences between owners with and without overconfidence. As Table 16 indicates, overconfident owners differ in terms of using different inventory management approaches. They rely more on sales forecasting and attach greater importance to material requirement planning, ERP systems, and just-in-time inventory. 4.3 Loss aversion bias The questionnaire asks SME owners to rate their level of disappointment with total bad debts of 5% and 10% of sales revenue and satisfaction with an annual profit of 5% and 10% of sales revenue. As expected, Figure 3 shows that respondents express greater disappointment in the case of bad debts than satisfaction in the case of annual profits. Behavioural biases among SME owners 273 Loss aversion bias is also more prominent for a higher level of sales revenue. Table 8 shows statistically significant differences with the disappointment associated with bad debt and satisfaction associated with sales revenues, supporting H1c for loss aversion. In general, our results show that SME owners are prone to loss aversion bias, which is similar to the results reported by Ramiah et al. (2014) for treasurers. A beneficial implication of this tendency is that loss averse owners may be more inclined to manage bad debt efficiently and keep it to the minimum. Hence, firms with high levels of credit sales may want to consider hiring working capital managers with loss aversion bias. Table 4 shows that 88 SME owners (57.1%) are prone to loss aversion bias and 41 (26.6%) are not. We cannot classify 25 owners (16.2%) and thus categorise them as “other” and exclude them from further analysis. Figure 3 Table 8 Loss aversion bias. This figure shows the mean for the level of disappointment with total bad debts of 5% and 10% of sales revenue and satisfaction with an annual profit of 5% and 10% of sales revenue among Indian SME owners using a scale from 1 to 5 Results of paired t-test for loss aversion bias. This table shows the paired comparison between the level of disappointment with total bad debts of 5% and 10% of sales revenue and satisfaction with an annual profit of 5% and 10% of sales revenue Pair Mean difference t-statistic Degrees of freedom p-value 1 Disappointment with total bad debts of 5% of sales revenues − satisfaction with annual profits of 5% of sales revenues 0.721 11.473 153 0.000 2 Disappointment of total bad debts of 10% of sales revenues − satisfaction with annual profits of 10% of sales revenues 0.968 12.984 153 0.000 We use a binary logistic regression model to test H2c to determine if demographic variables affect the tendency to exhibit loss aversion bias. As Table 8 shows, the Hosmer and Lemeshow test indicates a good fit of Model 3 by correctly classifying 72.1% of the cases. Of the four demographic variables, Table 9 shows that only the coefficient of age is statistically significant, supporting H2c. Thus, older SME owners are more likely to exhibit loss aversion bias than their younger counterparts. These findings 274 H.K. Baker, S. Kumar and H.P. Singh are similar to Johnson et al. (2006) and Arora and Kumari (2015) but conflict with Rau (2014) who finds females are more loss averse than males. The difference between these studies could result from differing samples. Logistic regression model for loss aversion bias. This table shows the results of logistic regression Model 3 for loss aversion bias. The reference categories are: female (gender), young (age), secondary (education), and low (experience) Table 9 β Standard error Wald Degrees of freedom p-value Exp(β) ‒0.448 0.655 0.467 1 0.494 0.369 1.594*** 0.519 9.420 1 0.002 4.921 Demographic variable Gender Age Education 0.044 0.483 0.008 1 0.928 1.045 Experience ‒0.572 0.529 1.166 1 0.280 0.565 Constant 0.672 0.760 0.783 1 0.376 1.959 R2 0.090 Hosmer and Lemeshow test 6.455 (0.264) Correct classification 72.1% Note: *’**’*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. We also examine whether significant differences exist between the WCM preferences of SME owners with and without loss aversion bias. Table 12 shows that owners with loss aversion bias have a lower preference for suppliers’ credit, leasing, and factoring for working capital financing than owners without this bias. According to Table 13, owners with loss aversion use working capital turnover as a key value metric in managing and monitoring working capital. As Table 14 indicates, loss averse owners tend to use emergency liquidity reserves for managing cash. Table 15 shows that owners with loss aversion believe that the overall economic environment affects cash management. Loss aversion bias also significantly affects inventory management practices. As Table 16 reports, loss averse owners pay less attention to sales forecasting and are less likely to adopt just-in-time and supply chain management as an inventory management approach. 4.4 Anchoring bias To capture anchoring bias, the questionnaire asks SME owners to indicate their likelihood of making credit sales in the future to a low rated company A that has paid on time and to a similar low rated company B without a similar payment history. As Figure 4 shows, the means based on a five-point scale for company A and B are 2.974 and 2.260, respectively. Table 10 indicates that the difference between these means is statistically significant at the 0.01 level. However, given that the means for company A and B are lower than 3, past experience with company A does not substantially affect the decision to grant future credit relative to company B. Thus, we conclude that SME owners are not generally prone to anchoring bias. Information presented in Table 4 also supports this finding, which shows that only 30 (19.5%) of SME owners exhibit evidence of anchoring bias and 82 (53.3%) do not. Behavioural biases among SME owners 275 Figure 4 Anchoring bias. This figure shows the means for the likelihood of Indian SME owners to grant credit in the future to a low credit-rated company A that has paid on time and a similarly low-rated company B without the same repayment history using a scale from 1 to 5 Table 10 Results of paired t-test for anchoring bias. This table shows the results of a comparison between the mean rating of the likelihood of Indian SME owners granting credit in the future to low credit-rated company A that has paid on time and low-rated company B without the same repayment history using a scale from 1 to 5 1 Pair Mean difference t-statistic Degrees of freedom p-value Likelihood of granting credit sales to − Likelihood of granting credit sales to Company B 0.714 10.118 153 0.000 We use binary logistic regression to examine the effect of gender, age, education, and experience of the propensity of SME owners toward anchoring bias. As Table 11 shows, only the coefficient of education is statistically significant, supporting H2d. A negative coefficient shows that SME owners with higher education are less prone to anchoring bias. Logistic regression model for anchoring bias. This table shows the results of logistic regression Model 4 for anchoring bias. The reference categories are: female (gender), young (age), secondary (education), and low (experience) Table 11 Demographic variable Degrees of freedom p-value Exp(β) 0.818 1 0.366 2.750 0.005 1 0.943 0.960  Standard error Gender 1.012 1.118 Age ‒0.040 0.562 Education ‒1.627*** 0.490 11.034 1 0.001 0.196 Experience 0.573 0.589 0.947 1 0.330 1.774 Constant ‒1.131 1.177 0.923 1 0.337 0.323 R2 0.139 Hosmer and Lemeshow test 2.528 (0.772) Correct classification 77.7% Note: Wald *’**’*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. 276 H.K. Baker, S. Kumar and H.P. Singh Other significant differences also exist between those with and without anchoring bias. As Table 12 reveals, owners with anchoring bias have a lower preference for using loans from friends and family as well as money lenders for working capital financing but a higher preference for using suppliers’ credit. Table 13 indicates that owners with anchoring bias use the current ratio a key value metric for managing and monitoring working capital. These findings suggest that SME owners with anchoring bias are slow to incorporate new information into their decision-making processes. In the context of WCM, managers with anchoring bias could make credit-granting decisions on some unrelated past information, which could lead to higher bad debt. In terms of cash management, Table14 indicates that twice as many owners with anchoring bias focus on minimising float to speed up cash collection than owners without anchoring bias (36.7% vs. 18.3%, respectively). In terms of external factors affecting cash management, Table 15 indicates that owners with anchoring bias view the financial and banking environment, market conditions, overall economic environment, and the level of inflation as significantly affecting their firms’ cash management. Regarding inventory management, Table 16 shows that owners with anchoring bias attach more importance to sales forecasting, material requirement planning, and ERP systems. Table 12 Effect of behavioural biases on working capital financing. This table shows the comparison between the working capital financing preferences of biased and unbiased SME owners. The columns are: 1 = Retained Earnings, 2 = Cash Credit, 3 = Bank Loan, 4 = Hire Purchase/Leasing, 5 = Suppliers’ Credit, and 6 = Factoring. The numbers reported in columns 1 to 6 for Yes and No are mean scores obtained from a five-point scale where 1 = Not at all preferred, 2 = Somewhat preferred, 3 = Moderately preferred, 4 = Highly preferred, and 5 = Extremely preferred. The numbers in parentheses indicate the corresponding p-value of the t-statistics Behavioural bias Self-attribution 1 2 3 4 5 6 Yes 4.302 4.206 3.238 2.794 3.794 1.905 No 3.846 3.784 2.962 2.173 3.500 1.981 3.229*** 2.514** 1.595 3.217*** 1.971* ‒0.487 (0.002) (0.013) (0.113) (0.002) (0.052) (0.627) t-statistic Overconfidence Yes 4.000 3.970 3.040 2.505 3.743 1.921 No 4.000 3.857 3.057 2.314 3.429 1.800 t-statistic Loss aversion 0.988 ‒0.096 0.969 2.048** 0.773 (0.545) (0.932) (0.334) (0.043) (0.441) Yes 4.046 3.931 2.898 2.284 3.602 1.761 No 3.878 3.732 3.146 2.732 3.951 2.098 t-statistic Yes Anchoring 0.000 (1.000) No t-statistic 0.997 1.118 ‒1.535 (0.322) (0.266) (0.127) ‒2.322** ‒2.283** ‒2.275** (0.022) (0.024) (0.025) 4.000 4.000 3.233 2.600 3.900 1.833 4.012 3.938 3.061 2.220 3.500 1.866 ‒0.070 0.307 0.913 1.645 2.814*** ‒0.197 ( 0.950) (0.760) (0.363) (0.108) (0.006) (0.844) Behavioural biases among SME owners Effect of behavioural biases on working capital financing. This table shows the comparison between the working capital financing preferences of biased and unbiased SME owners. The columns are: 1 = Retained Earnings, 2 = Cash Credit, 3 = Bank Loan, 4 = Hire Purchase/Leasing, 5 = Suppliers’ Credit, and 6 = Factoring. The numbers reported in columns 1 to 6 for Yes and No are mean scores obtained from a five-point scale where 1 = Not at all preferred, 2 = Somewhat preferred, 3 = Moderately preferred, 4 = Highly preferred, and 5 = Extremely preferred. The numbers in parentheses indicate the corresponding p-value of the t-statistics (continued) Table 12 Behavioural bias Self-attribution 7 8 9 10 11 12 Yes 2.111 2.714 2.191 3.143 2.429 2.127 No 2.019 2.769 2.500 2.731 2.000 1.769 0.565 ‒0.357 ‒2.026** 2.412** 2.504** 1.978** (0.573) (0.721) (0.045) (0.017) (0.014) (0.045) Yes 2.079 2.792 2.416 2.990 2.317 1.951 No 1.943 2.886 2.514 2.600 2.086 2.200 0.893 ‒0.576 ‒0.589 2.222** 1.409 ‒1.280 (0.374) (0.566) (0.557) (0.028) (0.163) (0.203) 1.921 2.750 2.296 2.966 2.182 1.932 t-statistic Overconfidence t-statistic Yes Loss aversion No t-statistic Anchoring Table 13 2.220 2.902 2.512 2.805 2.268 2.171 ‒2.108** ‒0.947 ‒1.373 0.944 ‒0.476 ‒1.252 (0.037) (0.346) (0.172) (0.347) (0.635) (‒0.213) Yes 2.067 2.367 2.033 3.200 2.233 2.100 No 1.927 3.024 2.720 2.793 2.160 1.902 0.741 ‒3.735* ‒4.756* 1.068 ‒0.078 0.962 (0.463) (0.000) (0.000) (0.288) (0.938) (0.338) t-statistic Note: 277 *,**,*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. Effect of behavioural biases on key value metrics. This table shows the comparison between key value metrics considered by biased and unbiased SME owners in managing and monitoring working capital. The columns are: 1 = Return on Investment, 2 = Net Working Capital, 3 = Cash Conversion Cycle, 4 = Current Ratio, and 5 = Working Capital Turnover. Figures in parentheses indicate the corresponding p-value of the chi-square statistics given in the table Behavioural bias Self-attribution N 1 2 3 4 5 Yes 63 12.7% 28.3% 47.6% 14.3% 6.3% No 52 15.4% 23.1% 30.8% 15.4% 9.3% 0.172 0.446 3.370* 0.027 0.421 (0.679) (0.504) (0.066) (0.869) (0.516) t-statistic Overconfidence Yes 101 15.8% 24.8% 38.6% 18.8% 5.9% No 35 5.7% 25.7% 37.1% 14.3% 11.4% t-statistic 2.321 0.013 0.024 0.366 1.149 (0.128) (0.910) (0.871) (0.545) (0.186) 278 H.K. Baker, S. Kumar and H.P. Singh Effect of behavioural biases on key value metrics. This table shows the comparison between key value metrics considered by biased and unbiased SME owners in managing and monitoring working capital. The columns are: 1 = Return on Investment, 2 = Net Working Capital, 3 = Cash Conversion Cycle, 4 = Current Ratio, and 5 = Working Capital Turnover. Figures in parentheses indicate the corresponding p-value of the chi-square statistics given in the table (continued) Table 13 Behavioural bias Loss aversion N 1 2 3 4 5 Yes 88 10.2% 21.6% 37.5% 18.2% 12.5% No 41 17.1% 31.7% 34.1% 14.6% 0.00% 1.206 1.535 0.136 0.249 5.604** (0.272) (0.215) (0.712) (0.618) (0.017) t-statistic Anchoring Yes 30 3.3% 26.7% 40.0% 26.7% 13.3% No 82 11.0% 24.4% 32.9% 11.0% 4.9% t-statistic Note: 0.061 0.484 4.200** 2.376 (0.805) (0.484) (0.040) (0.124) *’**’*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. Effect of behavioural biases on cash management approaches. This table compares cash management approaches adopted by biased and unbiased SME owners. The columns are: 1 = Managing Cash through Netting, 2 = Centralisation of Cash Management Decisions, 3 = Meet Payment in a Timely Manner, 4 = Diversification of Banks, 5 = Minimise Float, 6 = Emergency Liquidity Reserves, and 7 = Management of Cash through Leading and Lagging. The figures in parentheses indicate the corresponding p-value of the chi-square statistics given in the table Table 14 Behavioural bias Self-attribution 4 5 6 7 Yes 63 N 20.6% 73.0% 39.7% 14.3% 28.6% 46.0% 12.7% No 52 11.5% 65.4% 32.7% 17.3% 26.9% 44.2% 17.3% 1.709 0.197 0.039 0.037 1.480 t-statistics Overconfidence Loss aversion 2 0.784 3 0.600 (0.191) (0.376) (0.438) (0.657) (0.844) 101 15.8% 70.3% 35.6% No 35 12.9% 22.8% 20.0% 68.6% 32.9% 8.6% 20.0% 0.320 0.463 0.116 0.237 1.942 (0.847) (0.488) 43.6% 14.9% 62.9% 22.9% 3.873** 1.186 (0.572) (0.848) (0.163) (0.496) (0.733) (0.049) (0.276) Yes 88 19.3% 75.0% 29.5% 13.6% 21.6% 56.8% 18.2% No 41 14.6% 65.9% 39.0% 17.1% 24.4% 26.8% 17.1% 0.419 0.263 0.126 t-statistics Anchoring 1 Yes t-statistics 1.163 1.144 (0.518) (0.281) (0.285) (0.608) (0.723) 10.091* 0.023 (0.001) (0.878) Yes 30 13.3% 66.7% 46.7% 10.0% 36.7% 43.3% 10.0% No 82 15.9% 68.3% 32.9% 12.1% 18.3% 46.3% 18.3% 0.108 0.103 4.160** 0.080 1.120 t-statistics Note: 1.578 (0.209) 0.027 1.787 (0.742) (0.870) (0.181) (0.748) (0.041) *’**’*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. (0.777) (0.290) Behavioural biases among SME owners Effect of behavioural biases on external factors affecting cash management. This table compares the rating score of biased and unbiased SME owners to different external factors based on their effects on cash management. The columns are: 1 = Currency Exchange Rate, 2 = Level of Inflation, 3 = Interest Rate, 4 = Financial and Banking Environment, 5 = Market Conditions, and 6 = Overall Economic Environment (Gross Domestic Product). The numbers reported in columns 1 to 6 for Yes and No are mean scores obtained from a five-point scale where 1 = Not at all, 2 = Somewhat, 3 = Moderate, 4 = High, and 5 = Extreme. The numbers in parentheses indicate the corresponding p-value of the t-statistics Table 15 Behavioural bias Self-attribution Yes No N 63 52 t-statistics Overconfidence Yes No 101 35 t-statistics Loss aversion Yes No 88 41 t-statistics Yes No Anchoring 30 82 t-statistics Note: Table 16 279 1 2 3 4 5 6 3.429 3.258 3.064 2.587 3.619 2.698 2.865 2.673 2.615 2.654 3.250 2.558 2.484** 3.202*** 2.741* ‒0.476 ‒2.943*** 1.066 (0.014) (0.002) (0.007) (0.635) (0.004) (0.890) 3.109 2.889 2.802 2.584 3.366 2.614 2.771 2.657 2.743 2.400 3.286 2.400 0.343 1.107 0.332 1.152 0.503 2.574 (0.182) (0.270) (0.741) (0.251) (0.561) (0.118) 3.148 2.837 2.773 2.602 3.386 2.534 3.220 3.146 3.049 2.756 3.537 2.781 ‒0.294 ‒1.374 ‒1.603 ‒0.864 ‒1.146 ‒1.823** (0.769) (0.174) (0.111) (0.391) (0.254) (0.071) 2.833 2.933 3.233 3.533 3.800 2.933 2.720 2.531 3.122 2.902 3.207 2.500 0.409 1.673* 0.553 3.002*** 4.283*** 2.576** (0.683) (0.097) (0.582) (0.003) (0.000 ) (0.011) *’**’*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. Effect of behavioural biases on inventory management approaches. This table compares inventory management approaches adopted by biased and unbiased SME owners. The columns are: 1 = Material Requirement Planning, 2 = Inventory Models, 3 = ERP System, 4 = Just-In-Time, 5 = Supply Chain Management, and 6 = Sales Forecasting. The numbers reported in columns 1 to 6 for Yes and No are mean scores obtained from a five-point scale where 1 = Not at all important, 2 = Somewhat important, 3 = Moderately important, 4 = Highly important, and 5 = Extremely important. The numbers in parentheses indicate the corresponding p-value of the t-statistics Behavioural bias Self-attribution N 1 2 3 4 5 6 Yes 63 3.730 2.127 2.825 1.968 2.365 3.349 No 52 3.115 1.769 2.096 1.712 1.923 2.962 3.179* 2.111** 2.727* 1.600 2.834* 2.455** (0.002) (0.037) (0.007) (0.111) (0.005) (0.016) t-statistics Overconfidence Yes 101 3.545 1.990 2.535 1.820 2.089 3.218 No 35 2.829 1.714 2.000 1.743 1.943 2.886 t-statistics 4.262* (0.000) 1.611 (0.120) 1.957*** 2.143** (0.053) 0.484 (0.035) (0.629) 2.104** (0.037) H.K. Baker, S. Kumar and H.P. Singh 280 Effect of behavioural biases on inventory management approaches. This table compares inventory management approaches adopted by biased and unbiased SME owners. The columns are: 1 = Material Requirement Planning, 2 = Inventory Models, 3 = ERP System, 4 = Just-In-Time, 5 = Supply Chain Management, and 6 = Sales Forecasting. The numbers reported in columns 1 to 6 for Yes and No are mean scores obtained from a five-point scale where 1 = Not at all important, 2 = Somewhat important, 3 = Moderately important, 4 = Highly important, and 5 = Extremely important. The numbers in parentheses indicate the corresponding p-value of the t-statistics (continued) Table 16 Behavioural bias Loss aversion N 1 2 3 4 5 6 Yes 88 3.341 1.943 2.171 1.644 1.989 2.955 No 41 3.488 2.000 2.951 2.098 2.342 3.439 t-statistics Anchoring Yes 30 No 82 t-statistics Note: 5 ‒0.774 ‒0.316 ‒3.066* ‒3.428* ‒2.265* ‒3.359* (0.441) (0.752) (0.003) (0.001) (0.005) (0.001) 3.767 1.933 2.800 2.035 2.100 3.267 3.122 1.854 2.024 1.707 2.000 2.939 3.076* 0.391 2.383* 1.538 0.537 1.841*** (0.003) (0.696) (0.005) (0.133) (0.594) (0.068) *’**’*** Significant at the 0.10, 0.05, and 0.01 levels, respectively. Summary and conclusions Growing evidence reveals that people are not always rational. Studying how heuristics and behavioural biases affect decision-making is important. This study attempts to identify the behavioural biases of SMEs owners and to assess the effect of these biases on decision-making related to WCM. It also provides empirical evidence on how demographic variables affect behavioural biases. Based on responses from 154 Indian SME owners, the evidence shows that they are prone to self-attribution, overconfidence, and loss aversion bias. Hence, our results support H1a, H1b, H1c except H1d for anchoring bias. Demographic factors including gender, age, and experience are significantly related to the propensity to exhibit these behavioural biases supporting H2a, H2b and H2c. Results of this study make it conclusive that Indian SME owners are generally prone to selfattribution bias when making WCM decisions. There propensity to exhibit selfattribution bias also affected by their age experience and gender. It is also found that selfattribution bias may not always negatively affect WCM. In some instances, such managers may be more aggressive in trying to reduce the cost of financing working capital components. In addition to this, it is also found majority of owners are overconfident and more likely to issue forecasts that are overly optimistic. This could lead to overestimating future sales and excessive inventories. 5.1 Limitation of the study and directions for future research Our findings add to the literature of behavioural finance by identifying behavioural biases when making WCM decisions. Although this study provides useful insights into the behavioural aspects of SMEs owners in managing working capital, generalising its findings requires caution because they are based on responses from only 154 SMEs Behavioural biases among SME owners 281 owners. Additional research is needed to verify the results in different countries and contexts using larger samples. Similarly, this study tested the propensity of SMEs owners to exhibit self-attribution bias, overconfidence bias, loss aversion bias and anchoring bias. However, behavioural finance literature advocated a long list of biases including confirmation bias, optimism bias, herding bias etc. Thus a future study is needed which can incorporate these additional bias and assess the effect of these biases on WCM decisions. References Acker, D. and Duck, N.W. (2008) ‘Cross-cultural overconfidence and biased self-attribution’, Journal of Socio-Economic, Vol. 37, No. 5, pp.1815–1824. Agrawal, K. (2012) ‘A conceptual framework of behavioral biases in finance’, IUP Journal of Behavioral Finance, Vol. 9, No. 1, pp.7–18. Arora, M. and Kumari, S. (2015) ‘Risk raking in financial decisions as a function of age, gender: mediating role of loss aversion and regret’, International Journal of Applied Psychology, Vol. 5, No. 4, pp.83–89. Baker, H.K. and Nofsinger, J.R. (2002) ‘Psychological biases of investors’, Financial Services Review, Vol. 11, No. 2, pp.97–116. Baker, M., Ruback, R.S. and Wurgler, J. (2004) ‘Behavioral corporate finance: a survey’, Working paper (No. w10863), National Bureau of Economic Research. Barber, B.M. and Odean, T. (2001) ‘Boys will be boys: gender, overconfidence, and common stock investment’, Quarterly Journal of Economics, Vol. 116, No. 1, pp.261–292. Beckmann, D. and Menkhoff, L. (2008) ‘Will women be women? Analyzing the gender difference among financial experts’, Kyklos, Vol. 61, No. 3, pp.364–384. Belt, B. and Smith, K.V. (1991) ‘Comparison of working capital management practices in Australia and the United States’, Global Finance Journal, Vol. 2, Nos. 1/2, pp.27–54. Bhandari, G. and Deaves, R. (2006) ‘The demographics of overconfidence’, Journal of Behavioral Finance, Vol. 7, No. 1, pp.5–11. Billett, M.T. and Qian, Y. (2008) ‘Are overconfident CEOs born or made? Evidence of selfattribution bias from frequent acquirers’, Management Science, Vol. 54, No. 6, pp.1037–1051. Cheng, P.Y. (2007) ‘The trader interaction effect on the impact of overconfidence on trading performance: an empirical study’, Journal of Behavioral Finance, Vol. 8, No. 2, pp.59–69. De Bondt, W.F. and Thaler, R.H. (1987) ‘Further evidence on investor overreaction and stock market seasonality’, Journal of Finance, Vol. 42, No. 3, pp.557–580. Deaux, K. (1979) ‘Self-evaluations of male and female managers’, Sex Roles, Vol. 5, No. 5, pp.571–580. Fairchild, R.J. (2005) ‘The effect of managerial overconfidence, asymmetric information, and moral hazard on capital structure decisions’, ICFAI Journal of Behavioral Finance, Vol. 2, No. 2, pp.34–44. Field, A. (2009) Discovering Statistics using SPSS, 3rd ed., Sage Publications, London. Frascara, J. (1999) ‘Cognition, emotion and other inescapable dimensions of human experience’, Visible Language, Vol. 33, No. 1, pp.74–87. Graham, J.R. and Harvey, C.R. (2001) ‘The theory and practice of corporate finance: evidence from the field’, Journal of Financial Economics, Vol. 60, No. 2, pp.187–243. Greer, T.V., Chuchinprakarn, N. and Seshadri, S. (2000) ‘Likelihood of participating in mail survey research: business respondents’ perspectives’, Industrial Marketing Management, Vol. 29, No. 2, pp.97–109. 282 H.K. Baker, S. Kumar and H.P. Singh Hackbarth, D. (2008) ‘Managerial traits and capital structure decisions’, Journal of Financial and Quantitative Analysis, Vol. 43, No. 4, pp.843–881. Heath, C. and Tversky, A. (1991) ‘Preference and belief: ambiguity and competence in choice under uncertainty’, Journal of Risk and Uncertainty, Vol. 4, No. 1, pp.5–28. Hribar, P. and Yang, H. (2015) ‘CEO overconfidence and management forecasting’, Contemporary Accounting Research, Vol. 33, No. 1, pp.204–227. Jaiswal, B. and Kamil, N. (2012) ‘Gender, behavioral finance and the investment decision’, IBA Business Review, Vol. 7, No. 2, pp.8–22. Jarboui, A.N.I.S. and Ali, A.M. (2012) ‘CEO emotional bias and dividend policy: Bayesian network method’, Business and Economic Horizons, Vol. 7, No. 1, pp.1–18. Johnson, E.J., Gächter, S. and Herrmann, A. (2006) ‘Exploring the nature of loss aversion’, IZA Discussion Paper No. 2015. Available online at: http://ssrn.com/abstract=892336. Sahi, S.K. and Arora, A.P. (2012) ‘Individual investor biases: a segmentation analysis’, Qualitative Research in Financial Markets, Vol. 4, No. 1, pp.6–25. Kirchler, E. and Maciejovsky, B. (2002) ‘Simultaneous over-and under confidence: evidence from experimental asset markets’, Journal of Risk and Uncertainty, Vol. 25, No. 1, pp.65–85. Kisgen, D. (2006) ‘Credit ratings and capital structure’, Journal of Finance, Vol. 61, No. 3, pp.1035–1072. Kourtidis, D., Sevic, Z. and Chatzoglou, P. (2009) ‘Investors’ trading activity: a behavioural perspective’, International Journal of Trade and Global Markets, Vol. 3, No. 1, pp.52–67. Kuo, M.H., Kuo, N.F., Chiu, Y.C. and Fan, P.H. (2005) ‘Gender and investment behavior: on Taiwanese individual investors’, Journal of Financial Studies, Vol. 13, No. 2, pp.1–28. Lin, H.W. (2011) ‘Elucidating the influence of demographics and psychological traits on investment biases’, World Academy of Science, Engineering and Technology, Vol. 5, No. 5, pp.137–142. Lundeberg, M.A., Fox, P.W. and Punćcohaŕ, J. (1994) ‘Highly confident but wrong: gender differences and similarities in confidence judgments’, Journal of Educational Psychology, Vol. 86, No. 1, pp.114–121. Malmendier, U. and Tate, G. (2008) ‘Who makes acquisitions? CEO overconfidence and the market’s reaction’, Journal of Financial Economics, Vol. 89, No. 1, pp.20–43. Mendes-da-Silva, W., Barros, L.A., Armada, M.R. and Norvilitis, J.M. (2015) ‘Behavioral finance: advances in the last decade’, Revista de Administracao de Empresas, Vol. 55, No. 1, pp.10–13. Miller, D.T. and Ross, M. (1975) ‘Self-serving biases in the attribution of causality: fact or fiction?’ Psychological Bulletin, Vol. 82, No. 2, pp.213–225. Mishra, K.C. and Metilda, M.J. (2015) ‘A study on the impact of investment experience, gender, and level of education on overconfidence and self-attribution bias’, IIMB Management Review, Vol. 27, No. 4, pp.228–239. Mittal, M. and Vyas, R.K. (2011) ‘A study of psychological reasons for gender differences in preferences for risk and investment decision making’, IUP Journal of Behavioral Finance, Vol. 8, No. 3, pp.45–60. Modi, S. (2012) ‘A study on the adequacy and efficacy of working capital in automobile industry in India’, IUP Journal of Accounting Research and Audit Practices, Vol. 11, No. 2, pp.69–90. Odean, T. (1999) ‘Do investors trade too much?’, The American Economic Review, Vol. 89, No. 5, pp.1279–1299. Padachi, K., Howorth, C. and Narasimhan, M.S. (2012) ‘Working capital financing preferences: the case of Mauritian manufacturing small and medium-sized Enterprises (SMEs)’, Asian Academy of Management Journal of Accounting and Finance, Vol. 8, No. 1, pp.125–157. Peteros, R. and Maleyeff, J. (2013) ‘Application of behavioural finance concepts to investment decision-making: suggestions for improving investment education courses’, International Journal of Management, Vol. 30, No. 1, pp.249–261. Behavioural biases among SME owners 283 Prince, M. (1993) ‘Women, men and money styles’, Journal of Economic Psychology, Vol. 14, No. 1, pp.175–182. Prosad, J.M., Kapoor, S. and Sengupta, J. (2015) ‘Behavioral biases of Indian investors: a survey of Delhi-NCR region’, Qualitative Research in Financial Markets, Vol. 7, No. 3, pp.230–263. Ramiah, V., Zhao, Y., Moosa, I. and Graham, M. (2014) ‘A behavioral finance approach to working capital management’, European Journal of Finance, Vol. 22, Nos. 8/9, pp.1–26. Rau, H.A. (2014) ‘The disposition effect and loss aversion: do gender differences matter?’ Economics Letters, Vol. 123, No. 1, pp.33–36. Ricciardi, V. and Simon, H.K. (2000) ‘What is behavioral finance?’ Business, Education & Technology Journal, Vol. 2, No. 2, pp.1–9. Ritter, J.R. (2003) ‘Behavioral finance’, Pacific-Basin Finance Journal, Vol. 11, No. 4, pp.429–437. Rosenthal, P., Guest, D. and Peccei, R. (1996) ‘Gender differences in managers’ causal explanations for their work performance: a study in two organizations’, Journal of Occupational and Organizational Psychology, Vol. 69, No. 2, pp.145–151. Rzeszutek, M., Szyszka, A. and Czerwonka, M. (2015) ‘Investors’ expertise, personality traits and susceptibility to behavioral biases in the decision making process’, Contemporary Economics, Vol. 9, No. 3, pp.237–352. Shefrin, H. (2007) Behavioral Corporate Finance: Decisions That Create Value, McGrawHill/Irwin, New York. Shuch Mednick, M.T. and Weissman, H.J. (1975) ‘The psychology of women-selected topics’, Annual Review of Psychology, Vol. 26, No. 1, pp.1–18. Singh, H.P., Goyal, N. and Kumar, S. (2016) ‘Behavioral biases in investment decisions: an exploration of the role of gender’, Indian Journal of Finance, Vol. 10, No. 6, pp.51–62. Tarim, E. (2016) ‘Situated cognition and narrative heuristic: evidence from retail investors and their brokers’, European Journal of Finance, Vol. 22, Nos. 8/9, pp.688–711. Todd, P.M. and Gigerenzer, G. (2003) ‘Bounding rationality to the world’, Journal of Economic Psychology, Vol. 24, No. 2, pp.143–165. Tversky, A. and Kahneman, D. (1974) ‘Judgment under uncertainty: heuristics and biases’, Science, Vol. 185, No. 4157, pp.1124–1131. Tversky, A. and Kahneman, D. (1991) ‘Loss aversion in riskless choice: a reference-dependent model’, Quarterly Journal of Economics, Vol. 106, No. 4, pp.1039–1061. Wallace, R.O. and Mellor, C.J. (1988) ‘Nonresponse bias in mail accounting surveys: a pedagogical note’, British Accounting Review, Vol. 20, No. 2, pp.131–139.