Examining the Intention to Purchase Luxury Goods Based on the Planned Behaviour Theory ()
1. Introduction
The global luxury goods market has grown substantially, starting from the second half of the 20th century (Granot et al., 2013). Moderate growth in the US and European markets, together with a growing demand for luxury goods in emerging economies such as Brazil, Russia, India and China lead to substantial growth in the global luxury goods market (Hennigs et al., 2013).
Asia-Pacific now represents the largest luxury market (Lahtinen & Tuominen, 2017). Growing appetite for luxury brands in Asia’s emerging economies, especially China, has helped drive luxury market growth in recent years (Li et al., 2012). China and Chinese consumers have been playing a leading role in the growth of the luxury goods market (Yin et al., 2020).
As the demand for luxury goods has increased, luxury consumption has remained of great interest among scholars who consider it a notable cultural trend and economic activity in contemporary society (Mundel et al., 2017).
The luxury market is experiencing a new phenomenon, with the consumption of these goods being more accessible and available to a wider range of consumers than in the past (Lim et al., 2013).
In recent years, luxury consumption has penetrated different levels of society’s communities (Wang et al., 2010).
Aspects related to the search for pleasure (hedonism), status, quality, practicality and relationship within social groups appear as attributes that can determine the behavior of luxury goods consumers (Bezerra et al., 2017).
Consumer behavior is greatly influenced and shaped by cultural, social, personal, and psychological factors (Wang et al., 2010).
The desire for status drives much of consumer behavior (Eastman et al., 1999). Thus, the consumption of status products can help people in their struggle for self-respect and social approval (Eastman et al., 1999).
Despite the considerable amount of research on the consumption of luxury goods, there is still much to be understood about the determinants that influence consumers’ purchase intentions for luxury products (Zhang et al., 2020).
The Theory of Planned Behavior has been used in several studies of consumer behavior to provide an in-depth understanding of personal and social factors (Jain et al., 2017). Previous studies have used this theory to understand the factors that influence the consumption of luxury goods (Ling, 2009; Loureiro & Araujo, 2014; Sanyal et al., 2014; Jain et al., 2017; Salem & Salem, 2018).
The theory of planned behavior, has proven to be an effective tool in understanding the factors affecting consumers’ purchase intention towards luxury brands in different countries (Zhang & Kim, 2013; Jain et al., 2015; Jain & Khan, 2017).
Little research into consumer behavior has been verified in emerging markets (Canguende-Valentim & Vale, 2020), especially on the African continent (Dhaliwal et al., 2020). In this study, the focus is given from Angola’s perspective. Angola’s luxury market is most developed in the automotive sector. Glitzy luxury SUVs are part of the permanently gridlocked traffic (Schubert, 2018).
In the Angolan market of luxury goods there is a high dependence on the external market (Kanumbua, 2019). Consumers look for other luxury products in other markets. Buyers of Angolan origin have preferences for the Portuguese luxury market, representing 38% of purchases by non-European citizens (Lopes, 2021). In this market Angolans spend an average of 1800? on each purchase (Ferreira, 2016).
In Angola particularly, no research has been identified on the factors that influence consumers to consume luxury goods. Thus, it is very important to understand what motivates Angolan consumers to buy luxury goods. Therefore, the results of research on luxury consumers in Western societies like the United States and Eastern ones like the studies in China, cannot be generalized to Angolan consumers.
In this sense, this study aims to identify the factors that influence the intention to purchase luxury goods by Angolan consumers, based on Ajzen’s theory of planned behavior (1991), thus making it possible, on the one hand, to fill the gap in the literature, as no study has previously been conducted to understand the intention to purchase luxury goods using a theory of planned behaviour in an African country.
On the other hand, luxury brands will be able to generate greater value for firms and consumers. The study used “PLSpredict” to assess the predictive relevance of the study model due to its robustness. PLSpredict is the current technique for assessing predictive relevance (Shmueli et al., 2019). Most previous studies that have sought to understand the purchase intention of luxury goods based on the theory of planned behavior have not used this most effective estimator.
Among the various categories of luxury goods, this research focused on luxury goods generally.
This article is structured as follows: The article is structured as follows. The next section presents the literature review and hypothesis development followed by conceptual framework and methodology. The subsequent section describes the findings of the structural and measurement model followed by the discussion, conclusions and the implications of the study.
The last section discusses directions for future research.
2. Literature Review and Hypothesis Development
2.1. The Luxury Concept
Luxury is a difficult concept to define. Although routinely used in our daily lives to refer to products, services or a certain lifestyle, the term “luxury” does not generate a clear understanding (Wiedmann et al., 2009). Luxury has often been used to define the top category of prestigious brands (Vigneron & Johnson, 2004). Luxury goods have always been associated with exclusivity, wealth and power, and have been identified with satisfying desires that are not essential (Loureiro & Araújo, 2014). For Karatzas et al. (2019) luxury is a feeling of high quality, it is expensive, it is aesthetically pleasing and it is rare.
According to Hennigs et al. (2013) the literature illustrates that luxury goods are defined by their exclusivity and rarity. Therefore, the common denominators of luxury goods are beauty, rarity, quality and price, but also the existence of an inspirational brand endorsing the product (Godey et al., 2013).
2.2. Drivers of Luxury Consumption
According to the literature review, we found that luxury consumption is influenced by personal factors (Dhaliwal et al., 2020; Bahri-Ammari et al., 2020; Eastman et al., 2020; Zhang & Zhao, 2019; Eastman et al., 2018; Wang et al., 2010; Hudders, 2012; Amatulli & Guido, 2012; Husic & Cisic, 2009; Tsai, 2005; Vigneron & Johnson, 2004), by psychological factors (Wang et al., 2010), by social factors (Dhaliwal et al., 2020; Bahri-Ammari et al., 2020; Eastman et al., 2020; Zhang & Zhao, 2019; Eastman et al., 2018; Wang et al., 2010; Kim & Jang, 2014; Hudders, 2012; Amatulli & Guido, 2012; Husic & Cisic, 2009; Vigneron & Johnson, 2004) and by factors related to luxury goods (Dhaliwal et al., 2020; Greenberg et al., 2019; Eastman et al., 2018; Wang et al., 2010; Hudders, 2012; Husic & Cicic, 2009; Vigneron & Johnson, 2004).
2.3. Theory of Planned Behaviour
The theory of planned behaviour (TPB) is based on the Theory of Reasoned Action (TRA), which states that human beliefs influence attitudes and shape behavioural intentions (Loureiro & Araújo, 2014). These theories have played a crucial role in providing an in-depth understanding of personal and social factors when applied to various consumer behaviour studies (Ling, 2009; Sanyal et al., 2014; Jain & Khan, 2017).
Ajzen (1991) extended the Theory of Rational Action by adding perceived behavioural control. The central factor of the TPB is the individual’s intention to perform a particular behaviour, so intentions are assumed to capture the motivational factors that influence a behaviour (Loureiro & Araújo, 2014; Salem & Salem, 2018).
The Theory of Planned Behaviour highlights three independent variables for behavioural intention, namely: attitude; subjective norm and perceived behavioural control (Ajzen, 1991). Thus, the more favourable the attitude and subjective norm with respect to the behaviour, and the greater the perceived behavioural control, the stronger should be an individual’s intention to perform a particular behaviour (Ajzen, 1991).
This theory has proven to be an effective tool in understanding the factors that affect consumers’ purchasing behaviour towards luxury goods (Ling, 2009; Zhang & Kim, 2013; Sanyal et al., 2014; Loureiro & Araújo, 2014; Jain & Khan, 2017; Salem & Salem, 2018; Jain, 2020).
2.3.1. Attitude
Attitude refers to the degree to which a person has a favourable or unfavourable evaluation or appreciation of the behaviour in question (Ajzen, 1991).
According to Jain et al. (2017) attitude refers to a person’s firm belief regarding the performance of the behaviours in question.
Consumers could have a favourable attitude towards buying a product if they considered that the product reflected their identities needs (Loureiro & Araújo, 2014).
As consumers become more familiar with luxury brands, they may develop a positive attitude towards luxury brands, leading to an increase in purchase intent (Jain, 2020).
A positive or negative attitude is specifically impacted by the intensity of behaviour and beliefs regarding the likely outcome (Jiang & Miao, 2019).
Several other studies in the past have also confirmed the significant effect of attitude on purchase intentions for luxury goods (Loureiro & Araujo, 2014; Jain et al., 2017; Jain & Khan, 2017; Salem & Salem, 2018; Jain, 2020).
In the context of this research, consumers may have a favourable attitude related to purchase intention of luxury products if they meet their needs. Thus, the higher the attitude, the more likely consumers are to buy luxury products. Therefore, we formulate the following hypothesis:
Hypothesis 1. Attitude towards buying luxury goods has a positive effect on Angolan consumers’ purchase intentions towards luxury goods.
2.3.2. Subjective Norm
The subjective norm focuses on the perceived social pressure to perform or not to perform the behaviour (Ajzen, 1991). The subjective norm may place pressure regarding the performance or non-performance of the behaviour in question, regardless of the individuals’ attitude towards the given behaviour (Jain & Khan, 2017).
Subjective norm refers to the influence of significant others on the performance of an individual’s behaviour (Jain et al., 2017), as that of friends, relatives and other consumers in the social (Singh et al., 2021).
Several studies in the past have found that subjective norm is the main predictor of luxury purchase intention among consumers (Ramayah et al., 2004; Ling, 2009; Kim & Karpova, 2010; Jain et al., 2017; Jain & Khan, 2017; Jain, 2020).
In the context of this research, subjective norm favourably influences the performance of consumers’ behaviour related to purchase intention of luxury products, if they consider the behaviour of friends, family and other consumers in the social environment. Thus, the greater the influence of subjective norm, the more likely consumers are to purchase luxury products. Therefore, based on a literature review, we formulate the following hypothesis:
Hypothesis 2. Subjective norm has a positive effect on Angolan consumers’ purchase intentions towards luxury products.
2.3.3. Perceived Behavioural Control
Perceived behavioural control is the degree of difficulty an individual perceives when performing a specific behaviour (Ajzen, 1991). Perceived behavioural control allows for the prediction of behaviours that people intend to perform, but are unable to perform due to lack of opportunities/resources (such as time, money, skills) (Jain, 2020).
In a study in China, perceived behavioural control was a key factor in determining the purchase intention of consumers of luxury goods (Ling, 2009).
Several studies have confirmed that there is a direct relationship between perceived behavioural control with luxury purchasing behaviour (Kim & Karpova, 2010; Jin & Kang, 2011; Jain, 2020).
In the context of this research, perceived behavioural control favourably influences consumer behavioural performance related to purchase intention of luxury products. Thus, the greater the perceived behavioural control, the more likely consumers are to purchase luxury products. Therefore, based on a literature review, we formulate the following hypothesis:
Hypothesis 3. Perceived behavioural control has a positive effect on Angolan consumers’ purchase intentions for luxury products.
The conceptual model of the study, is as proposed in Figure 1, derived from the above hypotheses.
3. Methodology
To empirically investigate the purchase intention of consumers of luxury goods, based on the theory of planned behaviour, Angola was selected as the study context. The convenience sampling method was adopted. Data collection was conducted over two months, in April and May 2021, by an online questionnaire. The questionnaire was made up of two sections. The first section was made up of the variables from the theory of planned behavior, using scales from authors mentioned in Section 3.1. The second section was made up of demographic variables, where among the various items the respondents were asked to indicate whether they live in the North or South of Angola. We provide a brief explanation of the concept of luxury based on Godey et al. (2013).
As suggested in previous studies (Christodoulides et al., 2008; Shukla, 2010; Jain, 2020) a screening question was probed to participants to list the name of luxury brands owned by them, to ensure that only people who owned luxury brands were part of the data set.
In this study no particular luxury product was specified. The questions considered the expression “luxury products in general”, and the aim, was to identify
the factors that influence the intention to purchase luxury products in general. A total of 153 questionnaires were answered, of which 130 were valid and included for analysis, according to Table 1.
3.1. Scale Development
The questionnaire was developed based on standard item scales in English and later translated into Portuguese because the respondents were Angolan consumers, so they were more proficient in Portuguese.
From the existing literature, established and validated scales were used to measure attitude, subjective norm, perceived behavioural control and purchase intention. The scales were adopted from several studies conducted in the area related to luxury consumer behavior. For the Subjective Norm, four items were adopted, derived from Ajzen & Fishbein (1980), Ling (2009) and Fitzmaurice (2005), namely: “Most people who are important to me think that I should purchase luxury goods (SN1)”; “Many people around me have luxury goods (SN2)”; “I feel social pressure to buy luxury goods (SN3)”; “The people who I listen to
![]()
Table 1. Survey respondents’ profiles (n = 130).
could influence me buy luxury goods (SN4)”. For Attitude, four items were adopted, derived from Suntornpithug & Khamalah (2010), Zhan & He (2012) and Loureiro & Araujo (2014), namely: “I buy luxury goods because I have many benefits (e.g., quality, designer, is fashion, status, etc.) (ATT1)”; “Luxury goods satisfy my needs (ATT2)”; “Luxury goods helps to show my social status (ATT3)”; “In general, I am happy with luxury goods (ATT4)”. For perceived behavioural control, we adopted four items derived from Suntornpithug & Khamalah (2010) and Loureiro & Araujo (2014), namely: “I feel confident about choosing when I do search for luxury products (PBC1)”; “I clearly know the right things to do (not confused) during the process of buying luxury goods (PBC2)”; “I feel comfortable during the process of buying a luxury goods (PBC3)”; “I feel I have total control over my personal information during the purchase a luxury goods (PBC4)”. For the purchase intention, we adopted four items derived from Wang et al. (2005) and Bian & Forsythe (2012), namely: “The probability that I would buy luxury goods within the next 12 months is hight (PI1)”; “I will try to buy luxury goods in future (PI2)”; “I will recommend that my friends and relatives purchase luxury products (PI3)”; “I would think about a luxury goods as a choice when buying something (PI4)”.
All other items were measured on a five-point Likert scale, where “1” denoted “strongly disagree” and “5” denoted “strongly agree”.
3.2. Data Analysis
The research data was analyzed using structural equation partial least squares modeling (PLS-SEM) using the software Smart PLS 3.3.3. The parameters were estimated using the partial least squares method. Partial least squares (PLS) is one of the structural equation modeling (SEM) approaches that has become the primary alternative to covariance-based SEM for many researchers and is deployed in fields such as behavioral science, marketing, and business strategy (Ringle et al., 2012) is considered one of the best techniques for evaluating hypothetical relationships in a complex project (Hair et al., 2016) . The ability to analyze observed and latent variables distinguishes SEM from more standard statistical techniques such as analysis of variance (ANOVA) and multiple regression, which analyze only observed variables (Kline, 2016).
4. Results
First, we report the results of evaluating the measurement model to assess the reliability of the indicators, reliability of the constructs, convergent validity, and discriminant validity. Next, we provide the results after testing the proposed structural model.
4.1. Measurement Model
The first step in assessing the results of PLS-SEM involves examining the measurement models (Hair et al., 2019).
First the indicator loadings were examined. Loadings above 0.7 are recommended because they indicate that the construct explains more than 50 per cent of the variance of the indicator, providing acceptable item reliability (Hair et al., 2010). 15 items were finally accepted after the exclusion of one item (SN1), with factor loadings below the recommended level. The factor loadings of all observed variables included to be estimated for PLS-SEM were higher than the recommended level, as shown in Figure 2.
The second stage assessed the reliability of the internal consistency of the constructs. The consistency of the constructs was assessed using Cronbach’s alpha, Dijkstra-Henseler’s individual reliability assessment (rho_A), and composite reliability (CR). These values help to assess the consistency of a construct based on its indicators (Götz et al., 2010). The calculated values can be between 0 and 1. The lower limit established for acceptance of construct consistency and reliability is usually between 0.6 and 0.7 (Hair et al., 2010). Table 1 shows that all variables exceeded the minimum values for acceptance, and as all constructs had Cronbach’s alpha, Dijkstra-Henseler rho and CR values greater than 0.7, they are considered reliable (Henseler et al., 2009).
The third stage of the measurement model assessment addresses the convergent validity of each construct measure. Convergent validity is the extent to which the construct converges to explain the variance of its items (Hair et al., 2019). The metric used to assess the convergent validity of a construct is the average variance extracted (AVE) for all items in each construct. Convergent validity is given by the AVE value, which must be at least 0.5 to be considered sufficient and explain, on average, more than half of the variance of the indicators (Henseler et al., 2009). Table 2 shows that the AVE values for all constructs were above 0.5, indicating convergent validity in all cases.
![]()
Figure 2. Smart-PLS 3.3.3 Output without bootstrapping.
![]()
Table 2. Factor loadings and indicators of internal consistency and reliability.
CAa = Cronbach’s alpha; rho_Ab = Dijkstra and Henseler’s composite reliability; CRc = Composite Reliability; AVEd = Average Variance Extracted.
The fourth step is to assess discriminant validity. Discriminant validity represents the extent to which a construct is empirically distinct from other constructs in the structural model. The study used the Fornell & Larcker (1981) and the Heterotrait-Monotrait Ratio (HTMT) criteria of Henseler et al. (2015). According to Fornell & Larcker (1981) criterion, discriminant validity is present when the shared variance for all constructs in the model is not greater than their AVEs. According to Table 3, the shared variance for all constructs in the model is less than their AVEs. The Heterotrait-Monotrait Ratio (HTMT) criterion according to Henseler et al. (2015) the acceptable value of the HTMT ratio should be less than 0.90 in order to show discriminant validity. According to Table 4, the HTMT ratio was less than 0.90. All constructs met discriminant validity according to both criteria, thus supporting the good discriminant validity of the constructs in this study, as shown in Table 3 and Table 4.
In order to evaluate the fit of the proposed model, the standardised root mean square residual (SRMR) was calculated (Henseler et al., 2016). The SRMR measures
![]()
Table 3. Fornell-Larcker criterion matrix (discriminant validity).
Note: Values on the diagonal (bolded) are the AVE’s square root, while the off-diagonals are correlations.
![]()
Table 4. Heterotrait-Monotrait ratio (HTMT) for discriminant validity.
Note: Shaded boxes are the standard reporting format for the HTMT procedure.
the difference between the observed correlation matrix and the correlation matrix implied by the model. Hu & Bentler (1998) argue that a model has a good fit when SRMR is less than 0.08. In this study, the SRMR value was 0.075, indicating a good model fit.
When the evaluation of the measurement model is satisfactory, the next step in evaluating the results of PLS-SEM is to evaluate the structural model (Hair et al., 2019).
4.2. Structural Model
After checking the reliability and validity of the measurement model, the proposed structural model is examined. The explanatory power of the model is assessed using R square and Q square (Hair et al., 2019). The R2 ranges from 0 to 1, with higher values indicating greater explanatory power (Hair et al., 2019). Falk & Miller (1992) state that R2 values below de 0.1 values below 0.1 mean that the relationships formulated with the hypotheses have low explanatory power, although they may be statistically significant. The results of this study show that the structural model explained 27.9% of the total variance between SN, ATT, and PBC to PI. Chin et al. (2008) state that Q2 is the predictive sample reuse technique, it is evaluated along with both R2, they effectively present the predictive relevance of the model. Q2 is generated through a vending technique and reveals the usefulness of the data in terms of reassembling it in practice by means of the study model and PLS features, therefore, redundancy measures with cross-validation. According to Chin et al. (2008), when the value of Q2 is greater than zero (0), the model is classified as having high predictive power. Then the value of Q2 = 0.232 presented in Table 5, proposes that the study model has high predictive power.
Collinearity was examined to ensure that it does not influence the regression results (Hair et al., 2019). Multicollinearity was assessed by examining the VIF (variance inflation factor) values. VIF values should be close to 3 and lower (Hair et al., 2019). This study showed no serious multicollinearity as illustrated in Table 6.
The statistical significance and relevance of the path coefficients was assessed. We analysed the hypotheses for purchase intention, as shown in Table 7. Of the three constructs predicted to influence purchase intention for luxury goods, the results showed the Subjective Norm (H1: β = 0.371; p = 0.000) and Attitude (H2: β = 0.246; p = 0.002) was supported, with a confidence interval of 99%. However, Perceived Behavioral Control (H3: β = 0.137; p = 0.090), was not supported.
![]()
Table 7. Model resolution using PLS algorithm and bootstrapping.
ns = no significant; significant* = (p < 0.01); R2 (Purchase Intention) = 0.279.
![]()
Table 8. PLS assessment of manifest variable (Original Model).
Note: RMSE, root mean squared error; LM, linear model.
Following the guidelines recommended by Shmueli et al. (2019), the PLS prediction of the model was evaluated. The PLS prediction indicates a highly symmetric distribution in prediction errors when the Q2 values are greater than zero (0). Table 8 illustrates that the Q2 predictive value are greater than zero, suggesting that the PLS-RMSE values should be compared with the LM RMSE (Shmueli et al., 2019). After the comparison, we noticed that the PLS-SEM analysis produced a lower forecast error for all indicators as seen in the PI1, PI2, PI3 and PI4 with 0.982, 0.875, 0.924 and 1.052, respectively. In contrast to LM (Linear Model) produced values RMSE de 0.998, 0.924, 0.948 and 1.073 respectively, for model estimation using PLS-RMSE. Thus, the negative values obtained in Table 6, after deducing the PLS-RMSE values from the LM-RMSE values indicated a high predictive power of the model (Hair et al., 2019; Shmueli et al., 2019).
5. Discussion
The hypothesized relationship between subjective norm and purchase intention was supported. The results of this study corroborate with the theory of planned behaviour, which also reflects a significant relationship between subjective norm and purchase intentions (Jain et al., 2017). The results of the present research indicate that subjective norm is the most important predicator of purchase intention, with subjective norm having a greater impact on purchase intentions than attitude. This result is consistent with the study of Jain et al. (2017) in India, where survey results indicated that subjective norm has a greater impact than attitude. In many other studies, subjective norm has also been found to be a significant predictor of purchase intention (Sanyal et al., 2014; Phau et al., 2015; Salem & Salm, 2018).
The hypothesized relationship between attitude and purchase intention was supported. Attitude was found to be the second most significant factor influencing Angolan consumers’ purchase intentions. This result is consistent with previous studies (Zhang et al., 2007; Jain et al., 2017). Most studies have found positive and significant relationships between attitude and purchase intention (Ling, 2009; Jain et al., 2017).
The hypothesised relationship between perceived behavioural control and purchase intention is not supported. Although most previous studies have found positive and significant relationships between perceived behavioural control and purchase intention (Ling, 2009; Jain et al., 2017).
6. Conclusion
This study seeks to understand the factors that influence the intention to purchase luxury goods based on the theory of planned behaviour in the Angolan context. Specifically, this study consists of the variables attitude, subjective norm, perceived behavioural control and behavioural intention from Ajzen’s (1991) theory of planned behaviour.
This study revealed that attitude and subjective norm are the main factors in purchase intention of luxury goods. Subjective norm has a greater impact on purchase intentions than attitude. This result will help companies to formulate better marketing strategies to position their luxury brand in Angola and communicate with target consumers.
7. Implications
7.1. Theoretical Implications
This study developed a conceptual framework that can help us understand purchase intention toward luxury products, based on the theory of planned behavior, in the Angolan context. This study considered the variables of the theory of planned behaviour, namely attitude, subjective norm and perceived behavioural control and their influence of consumers’ purchase intention towards luxury products.
The examination of this relationship allows for a better understanding of consumers’ disposition in purchase intention. This research contributes to the literature by highlighting that attitude and subjective norm significantly influence consumers’ decision to purchase luxury products, in the Angolan context.
7.2. Practical Implications
Based on the findings of this study, understanding the strong relationship between attitude and subjective norm in purchase intention would help companies formulate better marketing strategies to position their luxury brand in Angola and communicate with target consumers. Marketers need to focus on the social significance of their products and clearly communicate how their products can benefit consumers in a way that reflects their social status in society (Salem & Salem, 2018).
Limitation and Direction for Future Research
Future research should further examine the issue of gender, age and income as moderating variables in the relationship between the variables in the current research model.