Intention in Investing Digital Gold Through E-Commerce Platforms

. Gold can now be invested digitally through e-commerce platforms. However, Indonesian people generally invest in gold traditionally in physical form. This study aims to identify the intentions and behavior of investors in investing in gold on e-commerce platforms by proving the effect of Attitude, Subjective Norms, Perceived Behavioral Control, Perceived Usefulness, and Perceived Ease of Use variables on Intention to invest. The research uses quantitative methods with data collected through online surveys. The sample was selected using a purposive sampling method, with 261 respondents contained. The study's results prove that Attitude, Perceived Behavioral Control, and Perceived Usefulness are significantly and positively related to the Intention to Invest. Perceived Usefulness is also demonstrated to have a significant and positive relationship with Attitude. On the other hand, Subjective Norms and Perceived Ease of Use are not significant to Intention to Invest. Perceived Ease of Use also does not demonstrate a substantial Attitude. This study reveals that variables of attitude, perceptions of behavioral control, and perceptions of use value are proven to increase investor intentions in investing in gold digitally. This research can also add to the knowledge in the study of gold investment, especially those carried out digitally through e-commerce platforms.


Introduction
The increase in Covid-19 cases has caused Indonesia to experience an economic contraction.A very high spike in new cases occurred during 2021, forcing the government to implement the Implementation of Restricting Community Activities (PPKM).This PPKM has caused a cessation of economic activity in the non-essential sector, such as restaurants that only accept takeaway orders, closing public areas, and closing shopping centers or trade centers [1].
The existence of these conditions also has a direct impact on investment activities, especially gold investment.For the Indonesian people, gold investment activities are more commonly carried out traditionally because people tend to be more accustomed to direct gold transaction methods and physical gold investment [2].Thus, due to the implementation of PPKM during the pandemic, gold investors no longer have the flexibility to invest physically.
Suppose you look at the results of a survey from Katadata.id[3] regarding gold investment, which was carried out among 1207 Indonesian respondents, gold investment is the investment instrument that is most in demand by respondents in Indonesia.
Investor interest in gold remained high, even though they were constrained when they wanted to invest in gold due to restrictions on mobilization.This is also supported because gold is an investment instrument with high liquidity [4].Its value tends to be stable and is not affected by economic volatility.[5] Making it suitable for investing in difficult times such as a pandemic.
On the other hand, the pandemic and restrictions on people's mobility during PPKM led to the emergence of an alternative option for investing in gold, namely digital gold (e-gold), which can be supported online through a virtual savings account [6].The online platform manager is also starting to have digital gold investment facilities, and the growth is even higher.Some are Treasury, Lakuemas, Pluang, Sakumas [7], and Pawnshop Digital.The e-commerce platforms with digital gold investment facilities include Tokopedia, Shopee, and Bukalapak [8].
Not only are the platform providers experiencing growth, but digital gold investors have also started to grow during this pandemic.Based on the results of a report from DSInnovate, it can be seen that in 2021, the number of digital gold investors in Tokopedia has increased by 20 times over two years.In addition, the total number of transactions also increased by 20 times during the same period [9].
The existence of a new phenomenon in the form of digital gold investment requires the government to be present to protect investors legally.The government, through Bappepti, has regulated and recognized digital gold investment as legal and protected by law.The Commodity Futures Trading Regulatory Agency (Bappebti) started a new era in digital gold investment through online platforms and digital gold trading on futures exchanges.Gold investment is not limited to physical ownership but can also be invested online through digital gold.The Ministry of Trade guarantees that digital gold trading is easy, safe, and accountable (PRESS RELEASE Bureau of Public Relations, Ministry of Trade: A New Era for Physical Trading of Digital Gold on Futures Exchanges Begins).
The digital gold investment that is being loved by the public today is investing in pure 24-karat gold assets digitally at 99.9%.This digital gold can be funded through online gold investment applications, one of which is through an e-commerce platform.Transactions in buying, selling, or exchanging gold can be done online through gadgets, such as smartphones or laptops, as long as investors can access the e-commerce page.
Interestingly, investing in digital gold is considered more practical and easy because, on the one hand, the digital gold that has been purchased will be recorded digitally and will be stored in a virtual savings account, so investors will no longer obtain gold in physical form [10].
On the other hand, if necessary, the digital gold that has been invested can also be converted into cash which can be withdrawn to the investor's account by making selling transactions directly on the e-commerce platform.
In addition, digital gold investment also has a variety of flexibility and attractive features.Digital gold investment through e-commerce platforms can be started with very affordable capital.The initial capital required is less than 10 thousand rupiahs (depending on the policies of each forum).Thus, digital gold investment opportunities are open to all people from a comprehensive income and occupational group.In addition, payment transactions for digital gold investments on e-commerce platforms are also reasonably practical because investors can make payment transactions electronically through various digital wallet services available on the platform [10].
Based on the facts above, the researcher considers that investing in digital gold on e-commerce platforms has the potential to present new concepts and perspectives in investing in gold which is thought to influence investors' perceptions, attitudes, and intentions.This is interesting for further research because consumer expectations and perceptions will continue to change, and both will constantly develop along with the addition of new information and concepts [11].
Researchers assess the urgency of research to find out and re-examine the results of previous studies or find new facts related to digital gold investment on e-commerce platforms, which according to researchers, will have the potential to change investors' attitudes and behavior due to the availability of new ways of making investments.Gold through e-commerce platforms.

Problem statement
Due to restrictions imposed by the government, gold investors are faced with a dilemma where they can no longer come directly to gold sellers to make physical purchases of gold.This causes investors not to try to find new investment alternatives that can be done more flexibly, one of which is digital gold investment.
As a breakthrough, especially for the Indonesian people, digital gold investment through e-commerce platforms is not quite popular in Indonesia because the behavior of the Indonesian people in investing can be said to be traditional, where people tend to be more accustomed to direct gold transactions methods and physical gold investment.[2].Before investors can adapt to digital gold investment through e-commerce platforms, a push will likely be needed, both internal and external.
Then this pandemic condition also encourages people to save and reduce expenses [12].Based on a survey conducted by Pluang on 5500 millennial respondents, during the pandemic, the investment budget tends to decrease because millennials are more focused on saving.The enthusiasm for keeping in the millennial age group increased by 5-10% during the pandemic [13].
Aside from the reduced public investment budget due to the pandemic, there is also a problem where Indonesian investors, enthusiastic about investing, are still not ready to allocate capital for investing in large amounts.Quoted from the Jakpat survey conducted in 2017 among 2946 respondents in the millennial circle (20-35 years), it can be seen that most respondents only allocate 1 million to 5 million rupiahs per year for their investment activities.The allocation is considered very minimal for an investment.
When converted into daily periods, it can be seen that most respondents are only willing to allocate a maximum of 14 thousand rupiahs per day for their investment activities (assuming one year is 365 days).
To be able to invest in gold physically, investors need to come to a gold seller (e.g., Pegadaian or Antam Boutique) to make transactions and pay in cash.In addition, physical gold needs a unique safe place [14].However, investors who have invested in gold physically cannot find out changes in the price of the gold they have in real-time Based on the problems above, the research aims to investigate the factors influencing investors' intention to invest in digital gold on e-commerce platforms.Below, some research questions related to this problem are presented.
3 Literature review [15] stated that the decomposed Theory of Planned Behavior method offers a more comprehensive understanding of the phenomenon of intention and behavior.In this case, the TPB decomposed model has an advantage where the TPB decomposed model has properties like the TAM model, which can identify specific beliefs that are thought to influence the implementation of information technology.However, at the same time, other factors, such as social perception and perceived behavioral control, are also incorporated, which are not yet available in TAM but have proven to affect behavior significantly.
TPB (Theory of Planned Behavior) is a theory put forward by Ajzen (1991) [16], which reveals that human social behavior can be described by factors such as Intention, Attitude, Subjective Norms, and Perceived Behavioral Control.The combination of these factors is proven to explain the complexity of human behavior quite accurately.In addition, Ajzen also argues that behavior is a function of a variety of Intentions and Perceived Behavioral Control.In this case, both can explain the Behavior variable with a relatively high proportion of variance.
Ajzen (1991) [16], through the Theory of Planned Behavior, states that intention describes the motivational factors that influence behavior.In this theory, Ajzen also noted that the definition indicates how hard an individual or group of people is to try and how much effort they can make to carry out a behavior.
Then, subjective norms, according to Ajzen (1991) [16], refer to a social pressure that encourages individuals to do or not to do a habit.Ajzen (2012) [17] then again explained that subjective norms could also be interpreted as normative beliefs, which believe that individuals around us want us to perform certain behaviors.Ajzen also argues that the normative assumptions of various individuals in the outline will form social pressure.
According to Ajzen (1991) [16], behavior control refers to the ease or difficulty in carrying out behavior and can be assumed to reflect past experiences or anticipation of obstacles.Ajzen (2012) [17] then redefines perceived behavioral control as the extent to which an individual believes they can practice a specific behavior.On the other hand, Ajzen (2012) [17] also stated that behavior control could also be interpreted as an effort to control oneself in a behavior.
On the other hand, research also uses TAM as a reference, stating that attitudes towards a new technology implementation can be described through perceived ease of use and perceived value.Davis (1989) [18] defines perceived convenience as the extent to which individuals feel it is easy to use technology.Meanwhile, perceived value is determined by Davis (1989) [18] as the time to which an individual believes that the implementation of a new system or technology will improve their performance.

Research method
This research is quantitative, and by the research objectives, this research is explanatory.The research strategy used is survey-based because survey strategies can be used to provide information regarding the relationship between variables.Because this research was conducted in an actual environment, the setting was a noncontrived/field study.Nonetheless, the amount of disturbance caused by this research is minimal because the online survey conducted by the researcher will not cause too much trouble to the respondents and can be completed quickly.The unit of analysis that the researcher uses is individuals with a cross-sectional time horizon.Data will only be collected once [19].
The research population is Indonesian society in general.In this case, the research sample was selected through a non-probability sampling method, purposive sampling.Because this research aims to measure investors' intention to invest in digital gold, the appropriate respondent criteria for this research are respondents who have never invested in digital gold on ecommerce platforms.
Referring to the five-times rule method, the minimum sample size is five times the number of research indicators [20].In this study, the number of hands was 27.Therefore, the minimum sample in this study was 135 samples, which came from 5 (reference value) multiplied by 27 (number of research indicators).
Researchers will analyze the data that has been successfully collected and then use the PLS-SEM method.The device that the researcher will use to analyze the data is SmartPLS 3.0.There are two essential steps in the analysis process using the PLS-SEM method: the outer model test and the inner model test [21].
The outer model test is divided into two main stages: the validity and reliability tests.In carrying out the validity test, researchers will use two methods, namely convergent validity testing, and discriminant validity.Then, the researcher will also conduct a reliability test.A reliability test measures the questionnaire [22].In evaluating the reliability test results, researchers used two methods, namely Cronbach's Alpha and composite reliability [23].
Testing the structural model or inner model test is carried out to determine the strength of the causal relationship between the study's latent variables.Structural model testing can be carried out through several stages, namely by evaluating: the coefficient of determination (R-square), cross-validated redundancy (Qsquare), and effect size (F-square).
Next, the researcher will test the hypothesis using the path analysis method.In this case, the researcher will evaluate the hypothesis test results by looking at the significance value (p-value) and the path coefficient value.The p-value is used to determine whether a hypothesis is accepted or rejected.Meanwhile, the path coefficient value represents how strong the relationship between variables is, according to Hair et al. (2014) [21].
Then, the results of the significance of the hypothesis test can also be further analyzed using the Multi Group Analysis method.MGA can be used to explore the differences between two different data groups.In other words, MGA can be used to identify the nature of the heterogeneity of the two data groups [24].
In addition, in the SmartPLS device, indirect effect analysis can be done automatically by carrying out the bootstrapping process.The indirect effect can be evaluated by looking at the significance value (p-value).

Result and discussion
In this study, the authors collected 324 respondents in total (before screening).These respondents were successfully contained within three weeks.However, after the authors applied a screening filter (to get respondents who had never invested in digital gold on an e-commerce platform before), the number of respondents decreased by 63.Thus, the remaining research respondents who met the criteria were only 261.
The data characteristics of the 261 respondents will then be examined further using descriptive analysis to provide a general description of the respondent's profile based on the gender, age, income, and domicile of the respondent.Based on Table 1, we may have a clear understanding about respondents backgrounds involved in this research, and we may proceed to further step.
According to the data analysis method, the convergent validity test will be carried out in 2 ways: factor loading and the Average Variance Extracted (AVE) value.Based on the loading data obtained, it indicators are valid, where each hand meets the factor loading criteria> 0.7.Even so, one needle, PBC1, has a factor loading value of <0.7.However, according to the requirements of Hair et al. (2013), a loading value of > 0.4 is still acceptable.Therefore, the PBC1 indicator can still be valid because the loading value is more significant than 0.4, which is close to 0.7.
Apart from using the factor loading method, convergent validity tests can be evaluated by looking at each variable's AVE value.Based on the table above, the research variables are valid because the AVE value of each variable meets the convergent validity test criteria (> 0.5).These results also support the factor loading analysis, where the variable can still be declared valid even though there are indicators whose loading values are below 0.7.
Then, the discriminant validity test will be carried out using two methods: the Fornell-Larcker Criterion analysis and the cross-loading method.Based on the results of the Fornell-Larcker Criterion test, it can be seen that the square root value of AVE, when associated with similar constructs, has a value > 0.7.The correlation value is higher when compared to the correlation with other constructs.So, the variables used in the research model meet the criteria of a discriminant validity test.
Based on the results of cross loading, it can also be seen that the loading value for each indicator, when correlated with its construct, is > 0.7, which is higher when compared to the cross-loading value with other constructs.Even though, in this case, the PBC1 indicator only has a loading value of 0.676, the PBC1 gauge can still be valid because a loading value of >0.4 is still acceptable (Hair et al., 2013).
Furthermore, the reliability test will be analyzed using Cronbach's Alpha analysis and composite reliability values.Based on the test results, it can be seen that Cronbach's Alpha value and the combined reliability value obtained from each variable are> 0.7.Thus, the variables used in this study are reliable and meet the reliability test criteria.
Then, related to the structural model, there are three stages to evaluate it, namely the coefficient of determination (R-square), cross-validated redundancy (Qsquare), and effect size (F-square).The results of the Rsquare value for the Attitude variable of 0.658 symbolize that 65.8% of the variance of the Attitude variable can be explained by the Perceived Ease of Use and Perceived Usefulness variables.Meanwhile, the remaining 34.2% of the conflict is explained by other variables.Then, it can also be seen that the R-square value for the Intention variable is 0.598, which illustrates that 59.8% of the variance of the Intention variable can be explained by variables such as Subjective Norms, Perceived Behavioral Control, Attitude, Perceived Ease of Use and Perceived Usefulness.Then, the remaining 40.2% is explained by other variables not examined in this study.
Based on the Q-square value that has been obtained, it can be seen that the Q-square value of the Attitude variable is 0.456 or 45.6%.Meanwhile, the Q-square value for the Intention variable is 0.401 or 40.1%.The research has a good observation value because the Qsquare value is greater than 0.
The f-square value obtained shows that the effect relationship included at a significant level is Perceived Usefulness to Attitude.The effect size is essential because the F-square value is greater than the criteria (> 0.35).Then, a relationship between variables with a moderate effect size exists.In this case, the relationship is between Attitude and Intention, where the effect for this relationship is 0.142, which is considered close to a moderate level with the criterion F square = 0.15.Furthermore, the relationship between variables that have a significant effect on a weak story is the relationship between Perceived Behavioral Control on Intention and Perceived Usefulness on Intention, where each product has a value of 0.039 and 0.067 which are categorized as a weak effect because the value of the F-the impact is high, which is in the range 0.15 > F-square > 0.02.Based on Table 2, our hypothesis were not completely supported, as follow: H1, H4, and H6 were not supported.Then, for Table 3, we found out that statement perceived ease of use will affect intetion was not supported.
Based on the results of the hypothesis testing that has been carried out, the p-value in the relationship between Subjective Norms and Intention obtained is 0.071.The pvalue of 0.071 does not meet the criteria (p-So, in this case, the Subjective Norms variable is not significantly related to Intention.
Furthermore, it can be seen that the variable Perceived Behavioral Control has a p-value of 0.003.The p-value of 0.003 is still at the expected level (p-Therefore, it can be said that Perceived Behavior Control has a significant moderate positive effect on Intention.
Then, it can be seen that the p-value obtained from the relationship between Attitude and Intention is 0.000.
A p-value of 0.000 meets the desired criteria (p-0.05).Therefore, it can be said that Attitude has a significant relationship to Intention.The hypothesis test results prove that the p-value of the relationship between the variable's Perceived Ease of Use and Attitude is 0.208.This indicates that the relationship does not meet the expected criteria (p-value Perceived Ease of Use and Attitude is not significant.In other words, the various facilities available on ecommerce platforms, such as ease of buying and selling gold, ease of operation, and interaction with e-commerce platforms, will not necessarily lead to changes in investor attitudes.
Then, the results of hypothesis testing prove that the p-value of the relationship between the variables Perceived Ease of Use on Intention is 0.116.This proves that based on the results of hypothesis testing, it can be seen that the variable p-value obtained in the relationship between Perceived Usefulness and Intention is 0.000.The p-value of 0.000 still meets the expected criteria (p-value significant effect on Intention.There is no substantial relationship between the variable's Perceived Ease of Use and Intention because the p-value produced does not meet the expected p-value criteria (p- The results of hypothesis testing also indicate that the p-value obtained on the relationship between the variable's Perceived Usefulness and Attitude is 0.000.Because the p-value obtained meets the criteria (p-value tionship between the Perceived Usefulness variable and the Attitude variable is proven to be significant.

Conclusion and recommendation
Based on the research questions in Chapter 1 and the results of the discussion in Chapter 4, it can be concluded that of the seven hypotheses in this study, four hypotheses were declared acceptable.Invest, Attitude variable to Intention to Invest, Perceived Behavioral Control variable to Intention to Invest, Attitude variable to Intention to Invest, Perceived Usefulness variable to Attitude, and Perceived Usefulness variable to Intention.The three hypotheses that were rejected because they were not proven to have a relationship that had a significant effect were the Subjective Norms variable on Intention, the Perceived Ease of Use variable on Attitude, and the Perceived Ease of Use variable on Intention to Invest.
Regarding the analysis of mediation effects, the Attitude variable is only proven to be a mediator variable in the relationship between Perceived Usefulness and Intention, where the mediation relationship occurs partially.Meanwhile, the Attitude variable is not proven to mediate the relationship between Perceived Ease of Use and Intention to Invest.Thus the overall conclusion from the results of this study is that the variables of Attitude, perceived behavioral control, and perceived use value are variables that have a significant and positive influence and are proven to be able to increase investors' intentions to invest in digital gold online on the e-commerce platform.
Limitations based on the review and the results of the study of the research conducted by the author are in the process of collecting data, the information provided by respondents through questionnaires still has the possibility of not showing the actual opinions of respondents.This occurs due to differences in understanding and assumptions for each respondent, especially in the related knowledge of gold investment.Then the majority, the demographic characteristics of the research respondents are still not spread evenly, and the number of respondents is not proportional, so the data taken cannot represent the population of the research object as a whole, where female respondents are more dominant than male respondents.In addition, most respondents are still spread across the island of Java and DKI Jakarta, so respondents are not yet able to represent the wider community.
Then, the majority of respondents were selected research subjects because of ease of access, namely the ease in collecting data from respondents related to the environment around the researcher, so that the dominance of the sample is not evenly distributed, it is possible not to represent the entire population for this study.Therefore, a larger sample size is needed to produce more comparable and accurate demographic characteristics.
Based on this research, the researcher recommends that e-commerce platform developers be able to develop a gold investment platform by prioritizing benefits that can be felt and reflected directly by investors.In the short term, the developer can present new features that make it easier to use, the use of features that can monitor gold movements in real-time, anytime and anywhere, which is used to see gold price movements only through gadgets and at the same time investors can directly conduct gold investment transactions online with a low nominal or minimum investment value, p.This will undoubtedly be attractive to investors because of the additional benefits that investors feel as well as creating perceptions and encouraging their Intention to invest.
In the long term, the researcher suggests to platform developers take strategic steps aimed at educating consumers in various digital gold investment campaigns and promotions, which are expected to motivate and increase influence on attitudes, perceptions of behavior control, perceptions of use value, and perceptions of security that can influence consumer intention to invest in digital gold through e-commerce platforms.
Then, the researcher also suggests that further research can be carried out by considering a larger sample size to obtain more even demographic characteristics of respondents and applying more specific screening categories to target groups of respondents who already have an understanding of investing in gold on ecommerce platforms but have not yet never invest in the medium.In addition, the interview method is also expected to be implemented in the data collection process so that the sample data obtained is more comprehensive and accurate.

Table 1 .
Description of research respondents in general.