“Determinants of consumer purchasing behavior toward Korean cosmetic products: Evidence from Indonesia”

Korean cosmetics have become a favorite product in the Indonesian market. Therefore, it is essential to understand what stimulates Indonesian consumers to buy Korean beauty products. This study aims to examine the structural model of the determinants of purchasing behavior for cosmetic products made in Korea, applying a quantitative design. Data were collected using a questionnaire targeting a random sample of respondents (n = 250) who are social media followers of cosmetic products made in Korea. The collected data were then analyzed using a partial least square approach using Smart-PLS 4.0 software. The results indicate that consumer attitude significantly influences consumer behavior. Moreover, the findings imply that a positive attitude can build consumer behavior (make consumers proud and idolize a particular product). Furthermore, the positive image of the place of origin and the quality of the product can influence consumer attitude. Management practice can alter how individuals view a product by using brand ambassadors and the rich culture of the place the product comes from.


INTRODUCTION
Along with the development of civilization, goods for personal care, beauty, or cosmetics have become necessary for modern society (Jairath & Daima, 2021;Le et al., 2020), including in Indonesia.The need for cosmetic products has increased, so the cosmetics business has grown significantly (Ishak et al., 2020;Nadim & Mohd Jani, 2021).Korea is one of Indonesia's countries that are expansive in marketing cosmetic products (Nguyen & Özçaglar-Toulouse, 2021; Sang- Hyeon et al., 2020).Korean products are the top choice among Indonesian consumers who choose and are proud of products made in Korea and recommend them to others.This is a fascinating behavioral phenomenon.
According to the theory of planned behavior and stimulus-organism-response approach, attitudes can develop due to stimuli (Miden et al., 2018; L. T. H. Nguyen & H. A. Nguyen, 2022;Paul et al., 2016).Consumer attitudes develop due to exposure stimuli that marketers have created (Aung, 2020).Marketing goods made in or imported from other nations is challenging for marketers.Thus, developing a positive perception of the country of origin and the product itself is essential.Brand ambassadors are frequently employed in marketing to promote specific products and their nation of origin.Utilizing notable individuals and artists from the product's country of origin is generally helpful in developing a positive image (Sudaryanto et al., 2022).Besides, exposure to the culture of the product's place of origin can also promote positive impressions.The function of brand ambassadors, accompanied by the arrival of Korean culture, cannot be separated from the proliferation of Korean cosmetics in Indonesia (Jin et al., 2021;Le et al., 2020;Long & Khoi, 2020).The mental image that Indonesian customers have of Korean cosmetics shape their attitude: intentions to buy, suggest to others, remain loyal, and continue to use such products shape consumer behavior (Vergura et al., 2020).
One of the pillars of the image of Korean products is the infusion of Korean culture and the splendor of Korean artists as brand ambassadors for Korean products.Indonesian consumers of cosmetics are under a hypnotic spell that promotes the establishment of attitudes and behaviors toward Korean cosmetic products.Therefore, this study seeks to identify determinants influencing the desire of Indonesian customer to use and recommend Korean beauty products.

Attitude and behavior
The theory of planned behavior explains that attitude is one of the explanations of behavior.This framework is applied in various business and marketing studies (Hackman & Knowlden, 2014), either as a whole or adopting only one explanatory variable, especially attitude.Attitudes influence behavior.For example, in the context of Korean products, a person's attitude toward Korean products can encourage the behavior to try and repurchase these products.
Islam and Hani (2021) found that attitudes are essential in explaining Bangladeshi female consumers' purchase intention.Handriana et al. (2020) showed that the intention to buy halal cosmetics is affected by how people feel about the product and how much they know about halal.Cosmetic hor-  (Suparno, 2020).
This study aims to assess the influence of brand ambassadors, country-product image, and Korean wave on the dependent variable -behavior -by employing attitude as a mediating variable.The study proposes the following hypotheses: H1: Brand ambassadors influence country and product image.
H2: Korean wave influences country and product image.
H4: Country and product image affect consumer attitude.

METHODOLOGY
A  The t-stat was used to test the premise of the relationship between one latent variable and another, as shown in the model's path diagram.Suppose the t-statistic is more significant than the t-table, at 5% alpha, or by comparing the significance value of the path with the alpha (5%).In that case, the alternative hypothesis (Ha) is confirmed.On the other hand, the theory (Ha) is established if the path significance value (p) is less than alpha (5%).The model and the direction of the relationship between the hypothesized variables are depicted in Figure 1.

Characteristics of respondents
Two hundred fifty questionnaires met the requirements for further analysis.The questionnaires revealed the respondents' education, formal employment, age, and gender.Most respondents are female (95%), and the rest are male.This implies that Korean cosmetic products are not only for women.Even though the amount is minimal, men also use cosmetics.The majority of respondents (70%) are high school or college students.20% of respondents work as employees in private institutions.Only a tiny proportion (2%) work as teachers or lecturers at tertiary institutions; the rest work part-time or other jobs.Most respondents (59%) are 20-24 years, and as many as 24% are 17-19 years.Others have been in the group for over 25 years.This indicates that most young adults use cosmetics, especially those made in Korea.
In addition to demographic aspects, the data collected are also related to product purchase loyalty.This loyalty is seen from the average purchase frequency per month.Most respondents only buy once a month (41%).However, 30% buy cosmetic products more than twice a month.The frequency of this purchase depends on each respondent's usage.

Model evaluation
This paper investigates the structural model of consumer behavior concerning cosmetic products manufactured in Korea.The model explains the connection between several latent variables, each of which can be measured in various ways.
The consequence of the modeling carried out on multiple levels of evaluation is the final model.Furthermore, the validity and reliability of the latent variable structure and the model under consideration are put through their paces.
The study investigates the structural model of consumer behavior.The model shows how many hidden variables are connected, each tracked by several indicators.The modeling output, which has been evaluated in some different ways, is the model that is finally addressed.First, an analysis was done to determine the validity of the model and the dependability of the latent variable structure.The composite reliability and Cronbach's alpha values can be found in Table 2, which run from 0.544 to 0.918.The values exceed the threshold value of 0.600 for each latent variable.Another indicator, rho-A, has a composite dependability and a fair value above 0.70, except for attitude, which has a value of 0.681 but is still thought to be feasible, closer to 0.70.
On the other hand, the average variance extracted (AVE) value shows that the data are convergent.If the AVE is more than 0.50, a variable can be approved and considered valid.Finally, the AVE values for all variables fall within an acceptable range of 0.629 to 0.709 for the highest and lowest values, respectively.The relevance of these standards implies that all latent variables can be considered adequate or acceptable.
The Fornel-Larker Criterion (FLC) and the Heterotrait-Monotrait Ratio (HTMT) can be used to check the discriminant validity of each latent variable.These two criteria determine if two or more constructs are similar.If the value is less than 0.80 and HTMT is less than 0.90, the variables do not have multicollinearity problems.Tables 3 and  4 show the highest and lowest values of all variables.Except for the country and product image, which have a value of 0.793, the numbers in the off-diagonal cell have a value of 0.80 or more.This condition illustrates no multicollinearity problem for all latent variables.Table 4 shows HTMT between latent variable.If the HTMT value is less than 0.900, there is no multicollinearity problem.
In addition, to carry out a fit and identity evaluation of the model, it is also necessary to assess the degree of the measuring items' validity for each latent variable.The cross-loading result of this study shows that the latent variable measurement method can be used.The measuring indicator of a latent variable is allowed and valid as a measure of a latent variable if the value of the cross-loading on the variable in question is more significant and the highest among the cross-loading values on other variables.Table 5 shows the cross-loading of each item used in this study to measure the latent variable; for instance, the first column of the table is an indicator, and the first seven columns are latent variables.In column 1, there is a brand ambassador indicator.In column 1, there are ATT1 and ATT4 item symbols for the attitude variable.The ATT1 indicator's value appears to be 0.923, whereas the ATT4 indicator's worth seems to be 0.702.This value is higher and more significant than the loading value in columns 2 to 5.This fact shows that ATT1 and ATT4 are the most reliable indicators for the attitude variable.Therefore, assessing the items' validity of other variables is possible.
For example, in column 2, the loading of the behavior indicator (BH5-BH8) is 0.804-0.841,higher than the value in columns 1, 3, 4, and also 5. Column 3 is for loading the brand ambassador indicator, column 4 is for country and product image, and the last column is for the Korean wave.Cross loading values indicate that all variables in the model are formed with the valid indicator.
The goodness of fit indices help the researchers in commendable ways for assessing the model fit of both measurement and structural models; however, the models must have a substantive theoretical base.The SRMR and NFI values can be used to determine whether a model is fit.The model is deemed acceptable when the appropriate indices of SRMR are less than 0.09.The NFI yields values ranging from 0 to 1.The closely NFI reaches 1, the better the fit.NFI values above 0.9 usually reflect an acceptable fit.The SRMR model value in this investigation is 0.078, less than 0.09, and the NFI value is 0.764.
In Figure 1, the R 2 values for the country and product image (0.460), attitude (0.362), and behavior (0.357) variables fall within the moderate range.While the weight of F 2 for attitude-behavior (0.555) describes the effect of changes in exogenous latent variables on endogenous variables, the others fall into the minor category: country and product image-attitude (0.204), Korean wave-attitude (0.033), brand and product image (0.109), and Korean wave-country and product image (0.282).
The results of calculating the PLS algorithm using Smart PLS 4.0 are visualized in Figures 2 and 3, showing the construction of latent variables composed of valid indicators above 0.70.For example, the brand ambassador variable comprises BA1-BA8, with a coefficient value of 0.801 (BA6), and the highest is 0.841 (BA8).The country and product image variable was constructed using CI3, CI4, and PI1 indicators with the lowest coefficient value of 0.760 (PI1) and the highest of 0.849 (CI4).The Korean wave is built with three valid indicators (KW1, KW2, and KW3), each loading value greater than 0.700.Meanwhile, the attitude variable only has two items, namely ATT1 (0.923) and ATT4 (0.702).Behavior is also a latent variable composed of valid indicators (BH5-BH8) with loading above 0.700.In Figure 2, the value of the T-statistics is shown, as all the indicators that make up the latent variables and the structural model.The T-statistic value in the measurement model shows that all indicators are valid (the T-statistic is higher than the t-critical 2.00, and the significance level is 95%).
In addition to the measurement model, Figure 2 shows the structural model with the coefficient values of the paths related to the latent variables.Meanwhile, the T-statistic value for each route is shown in Figure 3.For example, in Figure 2, it can be seen that the attitude-behavior path has a (positive) coefficient of 0.597 and a T-statistic of 12.925.Then it was also shown that attitudes were influenced by country and product image (β = 0.465, t = 7.278) and the Korean wave (β = 0.187, t = 2.741).In addition, Figure 2 shows the path coefficient values, and Figure 3 presents the T-statistical values between the paths connecting the country and product image and brand ambassador (β = 0.292, t = 4.437) and the Korean wave variables (β = 0.292, t = 2.741).In this paper, modeling is carried out to examine the influence of several unobservable latent variables.Table 6 shows the data processing results: the path coefficient values, T-statistic values, and P-values.It indicates that all paths' coefficient values are positive and more significant than zero, the T-statistic value is more than the critical value (2.00, 95% level of significance), and the p-value is less than 0.05.The positive value of the path coefficient means that the relationship between the two latent variables is unidirectional.For example, the coefficient on the attitude-behavior path is 0.597 (positive), meaning that if the attitude value increases by one unit, the behavior value will increase by 0.597.
The score of the respondent's answer cannot have quantitative meaning; if one causative variable changes one unit, then the value of the effect variable will vary as much as the path coefficient.Therefore, the value of the path coefficient in this structural model is only interpreted in terms of the direction and the alignment of relationships.
Meanwhile, the T-statistic value is more than the critical value (2.00), and the P-value is less than 0.05, indicating that the relationship between the two variables is significant and that the proposed hypotheses (H1-5) are confirmed.is the behavior variable.The indirect path with a single mediating variable (country and product image) or multiple mediations (country and product image and attitude) end in the behavior variable, indicating a significant path.This is evident from the value of the indirect path coefficient with a value greater than zero and supported by a T-statistic value higher than the critical T-value (2.00) and a significance lower than 0.05.

DISCUSSION
This study analyzes consumer attitudes toward Korean products.Attitude has an impact on behavior related to Korean cosmetic products.Behavior is reflected in the measuring indicators, which include tending to buy cosmetics made in foreign countries, specifically from Korea, or suggesting that other people also buy cosmetics made in Korea.Even more powerful is that consumers are proud of the cosmetics they use in front of people in their environment.
The relationship between the influence of attitude and behavior is the basic framework for the theory of planned behavior (TPB), widely applied in various research fields.The results of this study support that attitude is one explanation of behavior, even though the context of Korean cosmetic products is different from the contexts of previous research.
Nevertheless, the results are consistent.Previous research has shown that attitude is related to or influences buying behavior or purchase intention, for example, in the context of halal products (Garg & Joshi, 2018)  Attitude is a variable that can be formed via plural antecedents.In particular, the attitude created can be caused by the external environment or exposure to stimuli from outside oneself, as explained in the S-O-R framework, for example, in the context of or-ganic personal care products and natural products (L.T. H. Nguyen & Nguyen, 2022) and cosmetic product reviews (Aung, 2020).This means that specific stimuli can form attitudes toward something.
In this study, attitudes are measured by two indicators: feeling confident when using foreign-made cosmetics, especially from Korea (ATT1), and not sure when looking without foreign (Korean) cosmetics (ATT4).A consumer has such an attitude because a relevant stimulus can be a perception of an object, country, or situation.Cosmetic products made in Korea can be seen from two sides: the country of origin and the cosmetic products themselves, as well as the culture surrounding them.Therefore, Korean cosmetic products are covered by three aspects: the country, producers, and culture.The impression of Korea as a cosmetic-producing country and Korean cosmetic manufacturers as trustworthy can lead to a positive attitude toward such products.The modeling results show that attitudes are influenced by country and product image (β = 0.465, t = 7.278, and p = 0.000) and the Korean wave (β = 0.187, t = 2.741, and p = 0.000).This indicates that consumer attitudes are formed by an impression of the Korean image, the cosmetic products produced, and exposure to Korean culture.
The image of the country of origin and the product's image are the results of the influence of one or several variables.An image is the output or response of the stimulus that causes it.In the model, it is suspected that brand ambassadors and the Korean wave are two variables that can build an image of the country of origin.The modeling results show a significant favorable influence of brand ambassadors (β = 0.294, t = 4.437, and p = 0.000) and Korean wave (β = 0.469, t = 7.368, and p = 0.000).Indicators that measure brand ambassadors include: famous Korean celebrity, who is able to convey information on Korean products well, trustworthy, and compatible with the product, represents the brand and its products, and can influence and convey positive emotions.
The Korean wave includes: Korea as an idol and role model in cosmetics; Korean celebrities deserve to be emulated in the appearance; consumers choose Korean products used by such artists.The indicators can form an image regarding the country of origin and Korean cosmetic products.The ability and accuracy of ambassadors in presenting products from their homeland can create a positive impression about Korea and cosmetics produced in this country.Indonesian consumers believe that cosmetics made in Korea are unquestionably perfect; such products are superior compared to other cosmetics and suitable for supporting daily appearances formed because of the presence of qualified, relevant, and reliable ambassadors.In addition, this popularity is also caused by the massive exposure to Korean culture.
This study confirms the findings of earlier research.For example, Motsi and Park (2020) found a relationship between the perception and im-age of a product's country of origin.Vinata and Kusumawati (2021) showed that brand ambassadors affect appearance perception.The country's image has a direct positive effect on brand attitude, and through brand attitude, it has an indirect positive impact on purchase intention.The company's image and the brand's image have a positive impact on how people feel about the brand and a direct positive effect on their plans to buy particular products.Also, the country's image has a positive effect on corporate image, and through corporate image, the country's image has an indirect positive effect on the brand image.

CONCLUSION
This study aimed to analyze the determinants affecting the popularity of Korean beauty products among Indonesian consumers.The study chose such variables as consumer attitude, consumer behavior, country and product image, brand ambassadors, and the Korean wave.
The results show that consumer attitudes toward products made in Korea are formed due to the image built by consumers that Korea is a trusted country as an excellent cosmetics manufacturer.Consumers' positive impression of Korean products and the country as a superior, trusted cosmetics manufacturer is due to the massive exposure to the Korean wave and competent brand ambassadors who promote beauty products.Brand ambassadors and the Korean wave affect consumer attitudes and behaviors indirectly.Country and product image and attitude can mediate the relationship between brand ambassadors and the Korean wave toward behavior in products made by Korea.
The study offers several implications.First, brand ambassadors can be artists or well-known persons in the producing region.Second, products convey the impression of the country of origin; thus, the country of origin where the producer is located can help create a positive brand image.

Table 1 .
Variable measurements quantitative explanatory research design was used.The population in this study is buyers of Korean cosmetic products who are followers of @nacificcosmetics on Instagram, with a total of 118,916 (as of November 2021).Data were collected using a questionnaire designed with Google Forms, distributed to 250 sample subjects via Instagram, assisted by Twitter and WhatsApp, randomly selected from the follower list.The questionnaire distributed contains information concerns and demographic aspects of respondents.Valid and reliable variables measuring items, including the brand ambassador, Korean wave, country and product image, attitude, and behavior use a five-point Likert scale (Strongly Agree (5), Agree (4), Neutral (3), Disagree (2), and Strongly Disagree (1)).The items in the questionnaire were developed based on previous empirical studies.Details of measuring items for each variable are listed in Table1.The collected data were analyzed using Structural Equation Modeling (SEM), utilizing the Smart-PLS 4.0 software used in the model development.The analysis stages are:1.Compiling empirical and theoretical models

Table 6 .
Direct effect: Path statistics summary