2.1 Currency substitution theory

The currency substitution (CS) theory was established by Chetty [20] and expanded by Ramirez-Rojas [21] and Calvo and Gramont [22]. Currency substitution means when a country’s citizens lose trust in their domestic currency or if the return on that currency is insufficient, they can substitute it wholly or partially with a foreign currency satisfying the functions [23]. Developing countries rely more on international currencies, especially the US dollar, which shows an asymmetrical currency substitution situation, while developed countries are the opposite, showing similar currency substitution. There are two types of currency substitution: indirect substitution and direct substitution. If investors desire liquidity and hope to change their currencies into other currencies, it is called indirect currency substitution. If there is a competitive dynamic between different currencies regarding their role as means of payment, it can be regarded as direct substitution [24].

2.2 Purchasing power parity theory

The purchasing power parity (PPP) theory was proposed by Cassel [25]. According to purchasing power parity, the same goods and services should cost the same in different countries without transaction costs and other considerations. To precisely determine the relative purchasing power of different currencies, the purchasing power parity is used to compare their values. This concept of purchasing power parity is widely used to compare the economic development of various countries [26], solve global poverty problems [27], and analyze foreign exchange rates [28]. Purchasing power, often linked to currency appreciation, reflects a currency’s increased ability to acquire more domestic goods and services as well as its stronger position in the foreign exchange market. Conversely, currency depreciation reduces purchasing power, making domestic goods more expensive and weakening the currency’s value in international markets, which limits the ability to purchase foreign goods and services [29].

2.3 Currency internationalization

International currency refers to a currency utilized beyond its issuing country for trade, investment, and reservation. It meets all the monetary functions of domestic currency and can be considered a special domestic currency. In academic research examining the factors that contribute to a currency’s widespread acceptance as an international currency and the potential emergence of new international currencies, the degree of currency internationalization is commonly analyzed as a dependent variable. Based on CS theory, Chinn and Frankel constructed a research framework for currency internationalization [9], including output and trade, money confidence, financial market, and network externalities. More international trade and macroeconomic stability contributed to the improved international status of currency [2]. Some studies have used the euro as an example, arguing that economic integration and greater economic ties will lead to using the same currency [30]. Liu et al. [31] examined the Chinese RMB internationalization process and identified economic size, trade networks, and financial market development as significant factors influencing the RMB’s international use. Cui et al. [32] found economic size, money confidence, financial market, and network externalities can influence currency internationalization and international competitiveness can be regarded as another influencing factor. The relationship between local currency debt and currency internationalization dynamics were examined, highlighting the importance of financial market development and technological innovation in currency internationalization [33].

2.4 Economic development

Economic development is always defined as the sustained improvement in a region’s or typically a nation’s general living conditions and economic well-being [34]. It plays a critical role in sharing international currency dynamics, influencing exchange rates, trade balances, and global financial stability. The countries with international currencies are always the countries with great power, suggesting that currency internationalization is associated with economic development [35]. Positive trade balances, increased investor confidence, and good economic policies all contribute to a country’s currency strength as its economy expands. A nation’s development level is strongly correlated with currency performance, showing a close relationship between the macroeconomic situation and the exchange rate mechanism [36]. Economic development is a key driver of currency internationalization. Previous studies found that there was a positive relationship between economic size and international use of its currency [32]. This indicates that as economic size grows, the process of currency internationalization will increase. The euroization and RMB internationalization are inseparable from their larger economies and the higher economic growth of China has also boosted its currency usage. The slowdown in economic development will also decrease the process of currency internationalization process [37]. GDP, FDI, and GDP growth rate are commonly used measures to capture economic development. GDP and its growth contribute to short-term currency appreciation, reinforcing its utility in international markets [38]. FDI has a significant influence on offshore currency usage, as vehicle currency use grows by 3.54% with every 1% rise in bilateral FDI [31]. The country’s economic development plays an important part in currency internationalization. Hence, the hypothesis is proposed:

2.5 Money confidence

Money confidence is the second important driver of currency internationalization. Stable currencies are preferred by people and nations because they can mitigate financial risks and maintain value over time [37]. Money confidence is only generated when a currency is deemed to possess a stable value within a specific timeframe. Reducing currency volatility means controlling transaction costs [39]. Maintaining the fundamental stability of currency value at a fair and sustainable level is the central banks’ major objective, benefiting both the nation and the global economy together. Recent studies examine the relationship between money confidence and currency internationalization. By using an experimental method, Cardozo et al. [40] introduced how the emergence of international currencies relies on confidence in the issuing country’s monetary stability and low inflation rates. Furthermore, empirical findings reveal that confidence in currency stability directly impacts trade efficiency and global currency circulation. Gold reserve and foreign exchange reserve are two important components of money confidence, maintaining a country’s economic stability and enhancing the trustworthiness of its currency. Gold reserve is considered high-quality hedging assets due to its exceptional security features and constitutes a significant portion of national asset allocations [6]. Previous papers have investigated that U.S. dollar internationalization was strongly linked to its gold reserve while euro global usage had a cointegration relationship with gold reserves in European countries [41]. In recent years, emerging economies, such as China, Russia, and India, have increased their gold reserves to reduce their dependence on the U.S. dollar and mitigate the impact of potential sanctions, thereby promoting the international use of their currencies [6]. Foreign exchange reserve is also held by central banks to protect against economic fluctuations, maintain price and financial stability, and improve export competitiveness [42]. The supply and demand of a currency are influenced by changes in its foreign exchange reserves. Abundant foreign exchange reserves give governments a benchmark and a foundation for modifying monetary policies. The diversification of foreign exchange reserves enables the country to reduce the reliance on a dominant currency, like the U.S. dollar, and enhance the global usage of other alternative currencies [43]. Accordingly, the second hypothesis is proposed:

2.6 Financial market

The third key driver to promote currency internationalization is an open, deep, and well-developed financial market [32]. A prosperous mature financial market enhances liquidity, reduces transaction costs, and increases currency exchange. A study on the internationalization of emerging market currencies illustrated by reducing “original sin”—the incapacity of nations to borrow internationally in their currency, financial market growth helps to promote currency internationalization [39]. Moreover, research on the Chinese RMB internationalization demonstrates that the RMB usage throughout the world has benefited from developments in China’s financial market [31]. Four indicators are commonly used to capture the development of the financial market, which are private credit to GDP, stock market capitalization to GDP, Chinn-Ito index, and M2 over GDP. A widely used indicator to evaluate the debt market is private credit to GDP. A regression model was utilized to analyze the influence of the credit market and exchange rate on currency internationalization, revealing that they did not significantly influence the process [44]. The percentage of private credit to GDP, measuring financial development, was negatively related to the invoicing share of an international currency [45]. However, it was also evaluated to have a statistically positive relationship with currency usage in trade [31]. Stock market capitalization to GDP is an indicator used to examine the valuation of a country’s stock market relative to its economic output, capturing the depth of the financial market. It was proven to have a significantly positive relationship with the reserve currency share and the use of international currency [46]. Chinn-Ito index (also known as KAOPEN) is an indicator to evaluate the openness of the financial market. An unrestricted financial market fosters currency internationalization, with a strong relationship between KAOPEN and currency internationalization [45]. M2 over GDP is an indicator for countries to evaluate the depth of the financial market. It showed a positive relationship with global currency use [47]. Accordingly, the third hypothesis is given:

2.7 Purchasing power

The purchasing power of a currency is generally regarded as the amount of goods and services that can be purchased by a unit of currency [10]. It is seen as a measure of the value of money. For different currencies, purchasing power explains why various currencies differ in value and why some currencies are more welcomed than others to become international currencies. The goal of holding a foreign currency is to acquire the purchasing power of that money, which people can utilize to afford products and services. If the purchasing power of a currency increases, which means the appreciation of the currency, people will be more willing to hold this currency, which will lead to its currency internationalization [11]. The complex relationships between currency value fluctuations, external debt, and economic growth were found in currency devaluation dynamics. This study shows the significance of currency value fluctuations can influence a country’s economic performance and its currency’s international position [33]. The innovation-driven economic growth leads to increased purchasing power through the creation and redistribution of financial resources, enabling higher demand for goods and services [48]. The welfare benefit gained from being an international currency depends on the inflation rate, usually measured by Consumer Price Index (CPI), and the Producer Price Index (PPI). Residents prefer the predominant international currency to have the lowest inflation rate, suggesting it has more purchasing power [49]. The Big Mac Index (BMI), which compares the price of hamburgers across countries, is also widely used to compare the purchasing power among currencies [50]. Hence, the fourth hypothesis is given:

2.8 Mediation effect

The purchasing power of money includes the inner value of money and the outer value of money. The inner value of money refers to the purchasing power in a certain country and the outer value of money refers to the exchange value of a country’s currency concerning other currencies, always known as the exchange rate. The internal value of a currency is inversely related to the inflation rate, and changes in the inner value can influence its outer value. The inner value of money is determined by money itself while the outer value is determined by the purchasing power relative to all other items [10]. Many scholars proved that the outer value of money is influenced by many factors such as economic development [12], currency stability [13], future expectations [14], and currency demand [15]. Economic size and macroeconomic stabilities were verified to promote the status of a currency and lead to currency appreciation. Higher purchasing power contributes to the currency internationalization process. Furthermore, money confidence is often reflected by a country’s precious metal reserves and foreign exchange reserves. Analyzing the prices of precious metals and seven local currencies reveals negative correlations between these two. Precious metals should not be utilized to hedge against local currency risk since they are unreliable tools for maintaining a local currency’s external purchasing power [6]. Therefore, purchasing power serves as a meaningful pathway through which economic development and money confidence influence currency internationalization. Four related hypotheses are listed below:

3.1 Research design

This study employs a quantitative research approach and utilizes the PLS-SEM method to examine the direct and indirect relationships between key drivers and currency internationalization. The PLS-SEM method is chosen because it is effective in handling complex relationships, small sample sizes, and non-normally distributed data, which makes it suitable for analyzing the dynamics between macroeconomic performance and international currency. The research framework, as illustrated in Fig 1, evaluates three exogenous variables (economic development, money confidence, and financial market) as direct drivers of the currency internationalization process. Additionally, it identifies purchasing power as a potential mediator via which currency internationalization is influenced by economic development and money confidence. All the variables involved are abstract and difficult to capture, so using only one or two proxies to evaluate is far from enough. In this paper, each variable is well segmented, and more indicators are covered in each variable to improve the accuracy of the research.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The research framework of this study.

https://doi.org/10.1371/journal.pone.0317765.g001

3.2 Data collection

This study relies on yearly secondary data consisting of macroeconomic and financial indicators at the country level from 2000 to 2020. All the data are hard data, which are collected from reliable and accessible databases, including the CEIC database, the World Bank website, the Chinn-Ito website, and the KOF Swiss Economic Institute website. This study includes three exogenous variables, which are economic development (ED), money confidence (MC), and financial market (FM), a mediator variable purchasing power (PP), and an endogenous variable currency internationalization (INT). Each construct is measured by at least 2 indicators to reduce potential measurement error and promote reliability and predictive validity. The indicators are chosen because they shape and form the basic idea of the constructs in past papers. Data standardization was performed to eliminate the scale differences and ensure consistency across variables, normalizing the data to a range from 0 to 1. This allows for a more accurate comparison and analysis of the relationships between these constructs. The data span from 2000 to 2020, chosen for several reasons. First, some indicators were only published on an annual basis without corresponding monthly or quarterly data. Second, some indicators could only be collected from the year 2000 while some indicators for Hong Kong were not publicly available. Keeping the same time interval for all nine issuers made it easy to collect and compare data. Yearly data presents a holistic evaluation of a country’s economic performance over a fixed period, making it easier to analyze long-term trends and fluctuations. Given the significant transformations experienced by many countries in the past two decades, especially China, this time frame was deemed appropriate for the study. Table 1 presents the definitions and data sources of all constructs and indicators.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. The summary of constructs, indicators, definitions, and data sources.

https://doi.org/10.1371/journal.pone.0317765.t001

3.3 Sampling

This study concentrates on 9 of the top 10 most-used currencies in the global financial market during the period from 2000 to 2020. The selected currencies are the Australian Dollar, Canadian Dollar, Chinese RMB, U.S. Dollar, U. K Pound Sterling, Japanese Yen, Swiss Franc, Sweden Krona, and Hong Kong Dollar. Euro is excluded because some indicators are published at the country level, and relevant data for the European Union during the required period is unavailable. Except for China, all the other issuers of the selected currencies are developed countries. The inclusion of these currencies is justified for several reasons. First, they are all among the top 10 most widely used currencies in the world financial market and are extensively used for cross-border trade, investment, and settlement, providing insights into optimal foreign exchange reserve allocation and risk mitigation. Second, the issuing countries of these currencies are major economies, playing an indispensable role in the operation of the world economy and significantly influencing international financial stability. Third, prior literature usually analyzes the main reserve currencies in the SDR basket, potentially limiting the accuracy of results. This study increases the sample size by including safe-haven currencies (like the Japanese Yen and Swiss Franc), commodity-linked currencies (like the Australian Dollar and Canadian Dollar), emerging market currencies (like the Chinese Renminbi), and traditional international currencies (e.g., U.S. Dollar, British Pound). Expanding the sample provides a more comprehensive understanding of the drivers influencing currency internationalization and better reflects the complexities of global currency dynamics.

3.4 Data analysis

The data will be analyzed by using the PLS-SEM method to analyze the complex relationships between the exogenous variables (economic development, money confidence, financial market), the mediator (purchasing power), and the endogenous variable (currency internationalization). The PLS-SEM method normally involves two parts. The first part is to evaluate the formative measurement model (the outer model), which examines the relationships between formative constructs and their respective indicators. Unlike reflective models, the indicators in the formative measurement model are not interchangeable and the relationships between these indicators can be positive, negative, or even not correlated. In this paper, since latent constructs, ED, MC, FM, PP, and INT, are impossible to measure directly, they should be represented by their corresponding indicators.

The formative measurement model assessment consists of examining collinearity and evaluating outer weight and significance to ensure the validity of the constructs. The PLS-SEM algorithm calculates the Variance Inflation Factor (VIF) values for multicollinearity testing while the outer weight and significance are determined by using a one-tail bootstrapping test. To enhance the precision of this study, the bootstrapping process is conducted by using 10,000 subsamples instead of 5,000 subsamples. Indicator collinearity represents the correlations between indicators within constructs, where higher correlations can result in multicollinearity issues. If VIF >  10, it means a high risk of a multicollinearity problem. Some papers select an acceptable threshold of 5 while others encourage to use a stricter criterion with a VIF value of 3.3 or less to better prevent multicollinearity [51]. The weight of an indicator indicates how much contribution it makes to its construct. This implies that indicators with higher outer weights have greater influences on identifying their constructs. The statistical significance of the indicator weights is always determined by T statistics and p values.

The second part is to test the structural model (the inner model), evaluating the relationships between different constructs and testing the hypotheses proposed in Part 2. This includes five steps, which are collinearity (VIF), path coefficient (β), coefficient of determination (), coefficient of effect size (), and predictive relevance (). The criteria for VIF and path coefficient significance in the structural model are the same as those in the formative measurement model. The path coefficient identifies the strength and direction of the direct relationship between latent variables. The sign of the path coefficient indicates the positive or negative relationship between an exogenous variable and an endogenous variable. The coefficient of determination () is a measure to examine how much variance in the dependent variable can be explained by other independent variables in the structural model. It is employed to evaluate the predictive accuracy, indicating how well the model fits the observed data. values range from 0 to 1, where 0 denotes no predictive accuracy and 1 means total predictive accuracy. A higher means a better fit of the structural modal and a higher level of predictive accuracy [17]. Chin proposed three criteria [52], which are 0.67, 0.33, and 0.19 respectively, to determine substantial, moderate, and weak levels of predictive accuracy, while Hair set the criteria at 0.75, 0.5, and 0.25 instead [17]. The coefficient of effect size is examining the relative effect of an exogenous variable on an endogenous variable. It is classified into four categories: no impact, small effect, medium effect, and large effect with thresholds of 0.02, 0.15, and 0.35 respectively [53]. Blindfolding is used to determine predictive relevance (). reflects the structural model’s power to predict the endogenous variable from the exogenous variables. >  0 indicates that exogenous variables have predictive importance for the endogenous variable. The values 0.02, 0.15, and 0.35 serve as standards for dividing predictive power into weak, moderate, and strong.

Additionally, the mediation effects of purchasing power on the relationships between economic development and currency internationalization and money confidence and currency internationalization are also tested by using a two-tail bootstrapping test with the 10,000 subsamples. When utilizing the bootstrapping technique to examine the mediation effect, it is also needed to consider confidence intervals. The 95% bootstrap confidences must not include the value 0 within the limits of their lower and upper bounds [54]. The formative measurement model, structural model, and mediating effects should all meet the criteria. Table 2 lists the models, assessment, and acceptable values for data analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. The models, assessment, and acceptable values for data analysis.

https://doi.org/10.1371/journal.pone.0317765.t002

3.5 Methodology justification

This paper adopts the PLS-SEM method to examine the relationships between latent variables and conduct relevant mediation analysis. This approach is chosen for the following reasons. First, PLS-SEM has fewer limits, particularly regarding data distribution and sample size. It is suitable to handle small sample data and non-normally distributed data, but the outcomes are just as trustworthy as those produced by using covariance-based structural equation modeling (CB-SEM) [17]. Second, the main aim of this paper is consistent with the research purpose of the PLS-SEM method. The PLS-SEM method aims to identify key drivers and extend the existing theory while the CB-SEM method is to test a theory or compare a theory with another theory [18]. Thirdly, using PLS-SEM, it is useful to assess a single indicator in a construct, solve complicated research models, and handle both formative and reflective measurement models. The relationships between indicators with secondary data and constructs are commonly formative measurement models, while CB-SEM can only deal with reflective measurement models. The relationships among the indicators in the formative measurement model can be positive, negative, or even not correlated while indicators in the reflective measurement model are highly correlated [51]. PLS-SEM is flexible with its assumptions, consistent with the research objectives, and adaptable to various research contexts.

4.1 Descriptive statistics

Table 3 shows the means, medians, and standard deviations for all indicators. It presents the basic results of 189 observations from 9 currency issuers between 2000 and 2020. Due to data standardization, all the data are in a specific range from 0 to 1. KAOPEN has the highest mean value of 0.833, indicating a greater average degree of openness in these places. The GR and SMC have extremely low averages of 0.109 and 0.104 respectively, indicating a lower amount of gold reserves, and the stock market is modestly undervalued in these places. Except for GDPS, GDP, and FXR, the variances between means and medians are minimal, smaller than 0.1, implying that the data are evenly distributed around the central point. KAOPEN has the largest standard deviation at 0.308, suggesting notable diversity in openness among the places examined. In contrast, with the lowest standard deviation at 0.129, the numbers for GGR tend to cluster closely around the mean.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. The descriptive statistics for all the indicators in this paper.

https://doi.org/10.1371/journal.pone.0317765.t003

4.2 Formative measurement model assessment

Fig 2 shows the results of the formative measurement model analysis and Table 4 depicts the results of the outer weights and significance and VIF values of the formative measurement model. The VIF values for all indicators in different constructs are all smaller than the threshold of 10 and all other indicators meet the stricter VIF criterion of being below 5, except for GDPS and GDP, which have VIF values of 7.209 and 8.455, respectively. This phenomenon means a lower risk of multicollinearity problems in each construct, which suggests that the indicators are not excessively correlated with one another, the model is quite stable, and the outcomes are trustworthy.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. The outer weights, significance, and VIF values of the formative measurement model.

https://doi.org/10.1371/journal.pone.0317765.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The results of formative measurement model analysis.

https://doi.org/10.1371/journal.pone.0317765.g002

Within the ED construct, GDPS and GGR are statistically significant with p-values of 0.005 and 0 and T-statistics of 2.558 and 10.638 respectively at a 1% significance level. Both GDPS and GGR have positive weights (0.853 and 0.863), indicating that an increase in GDPS and GGR correlates with an increase in economic development. Conversely, FDI and GDP do not satisfy statistical significance, showing p-values of 0.360 and 0.156 respectively. Given p values of 0, the two indicators in the MC construct, GR and FXR, are of statistical significance. The positive weight for FXR shows that the growth in foreign currency reserves is correlated with the growth in money confidence, whereas the negative weight for GR indicates greater gold reserve leads to lower money confidence. For the FM construct, three indicators, SMC, KAOPEN, and M2, are statistically significant at a 1% significance level while the t-statistics range from 4.395 to 15.012. M2 has a negative weight, indicating that a decrease in M2 corresponds to an increase in the financial market, while SMC and KAOPEN have positive weights, showing opposite relationships. In the mediation construct PP, BMI, and PPI are significant at 1% and 10% respectively and both show positive relationships with purchasing power. In the endogenous variable INT, the indicator FI is only significant at 10% and TI has a significant relationship with INT at 1%. Both FI and TI have positive relationships with the construct INT.

4.3 Structural model assessment

Fig 3 shows the results of structural model analysis and Table 5 shows the results of the hypotheses, path coefficients, VIF values, effect size (), coefficient of determination (), predictive relevance () and mediation effect. The first step is to assess the collinearity by comparing VIFs. The VIF values of constructs are all below 5 and even below 3.3, implying a very low potential for multicollinearity issues between these constructs [50]. The second step is to analyze path coefficients and significance. When the mediator variable PP is included, all hypotheses from H1 to H6 are statistically significant at the significance level of 1%. The path coefficients show that ED and MC have negative relationships with INT, while FM has a positive relationship with INT. PP also has a significant relationship with INT, with a p-value of 0.008, indicating that the higher purchasing power of a currency supports its currency internationalization process. What’s more, ED and MC negatively affect PP, while PP positively influences INT, suggesting that while economic development and market confidence may lower currency purchasing power, greater purchasing power contributes positively to international currency expansion. The third step is to analyze the coefficient of determination (). The value for the PP construct is 0.28, signifying that about 28% of the variance in PP could be explained by ED and MC. This model demonstrates a weak effect on predictive accuracy [51], suggesting that other unexamined variables may contribute to PP. In contrast, the value INT is significantly higher at 0.74, meaning 74% of the variance in currency internationalization can be explained by this model. This demonstrates a strong predictive accuracy for international currency, meaning that the model successfully captures the key factors driving internationalization. The fourth step is to evaluate effect size . The results show that economic development and money confidence have small effects on currency internationalization, with effect sizes of 0.072 and 0.039, suggesting their influence on internationalization is limited. The financial market has a large effect size with the value of 0.535, surpassing the threshold of 0.35 dramatically, which means a sound financial market plays a significant role in driving currency internationalization. Purchasing power has a small effect on INT with an effect size of 0.029, indicating a minimal direct influence. In terms of the relationships between ED on PP and MC on PP, both have small effect sizes with 0.121 and 0.113 respectively, implying that these two factors have a relatively weak impact on purchasing power. The main finding is the significant role of the financial market in promoting currency internationalization, while the effects of economic development, money confidence, and purchasing power are comparatively smaller. The fifth step is to examine predictive relevance (). The value for PP is 0.116, bigger than 0.02 and smaller than 0.15, suggesting that if purchasing power is regarded as the dependent variable, the two independent variables, economic development and money confidence have weak predictive relevance. This implies although these two exogenous variables can explain some variance in purchasing power, their predictive power is relatively limited. In contrast, the value for INT is 0.639, greater than 0.35, indicating a stronger power of predictive relevance. This means the whole model, including purchasing power as a mediator, provides a good fit for predicting the currency internationalization process.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. The results of path coefficients, VIF values, effect size coefficient of determination predictive relevance and mediation effect.

https://doi.org/10.1371/journal.pone.0317765.t005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The results of structural model analysis.

https://doi.org/10.1371/journal.pone.0317765.g003

4.4 Mediation effect assessment

As shown in Table 5, the results reveal that purchasing power plays a significant mediating role in the relationships between economic development and currency internationalization and money confidence and currency internationalization. PP positively influences INT (β =  0.103, p =  0.008), while both ED and MC have direct negative impacts on INT (β =  −0.163, p =  0.002; β =  −0.153, p =  0.004 respectively) and on PP (β =  −0.322, p =  0.000; β =  −0.311, p =  0.000). The mediation effects of PP in both relationships are statistically significant at the significance level of 5% (ED- > PP- > INT: β =  −0.033, p =  0.036; MC- > PP- > INT: β =  −0.032, p =  0.042). This indicates that the mediation effect of PP on ED and INT and on MC and INT are both partial mediations rather than full mediation. This means that a portion of the influence of economic development and money confidence on currency internationalization is transmitted indirectly by purchasing power. The negative products of the path coefficients for the relationships ED- > PP, PP- > INT, and ED- > INT (−0.322 *  0.103 *  −0.163) and for MC- > PP, PP- > INT, and MC- > INT (−0.311 *  0.103 *  −0.153) indicate the complementary mediation effects in both cases. Therefore, hypotheses H7 and H8 are both supported and purchasing power mediates the relationships between ED- > INT and MC- > INT. This finding suggests that economic development and money confidence may negatively influence purchasing power, thereby weakening the currency’s ability to act as an international currency. It highlights the complex dynamics between economic performance factors and currency internationalization.

6.1 Conclusion

Currency internationalization has attracted great interest due to the tendency of dedollarization and deglobalization. This study tries to examine the key drivers of currency internationalization and the complex mechanisms behind international currency by using the PLS-SEM method. Based on the data for 9 of the 10 most used currency-issuing countries between 2000 to 2020, this study reveals that economic development, money confidence, and financial market have significant direct relationships with currency internationalization. The financial market has a large effect size among the three exogenous variables, while economic development and money confidence have a small effect size. The findings of the study reveal that economic development and money confidence in advanced countries have a negative and significant effect on purchasing power and currency internationalization. However, the financial market has a positive and significant effect on the international currency, suggesting the development of the financial market greatly promotes the currency internationalization process. What’s more, purchasing power has a positive relationship with currency internationalization and it acts as a mediator in relationships between economic development and currency internationalization and money confidence and currency internationalization. Both mediation effects are complementary mediation effects, implying that purchasing power partially explains how economic development and money confidence influence currency internationalization, but there are other direct or indirect pathways influencing currency internationalization besides PP.

6.2 Research contribution

This study enhances the literature on currency internationalization by analyzing data from nine international currency issuers. Previous papers focused more on the reserve currencies in the SDR basket, while this study expands the scope by considering additional currencies, including those with the potential to become reserve currencies. It establishes a model of currency internationalization based on CS theory and PPP theory and explores the connections between economic development, money confidence, financial market, and currency internationalization. It also examines how purchasing power mediates the relationships between economic development and currency internationalization and money confidence and currency internationalization.

Furthermore, this study contributes to the literature on macroeconomic performance by highlighting the significance of economic development, money confidence, and the financial market in the process of currency internationalization. It suggests that for advanced economies, the increase in economic development and money confidence may not promote the process of currency internationalization and can even impair the process due to economic regulation. This insight provides valuable guidance for some emerging economies, such as China and India, to promote the process of currency internationalization.

The method effectively analyzes the complex relationships between exogenous variables (economic development, money confidence, financial markets), the mediator (purchasing power), and endogenous variable (currency internationalization). The complex interactions of these latent variables, as well as how they affect the currency internationalization process have never been studied by using this method before.

6.3 Policy recommendations

According to the main findings of the study, some practical policies are proposed for the promotion of currency internationalization. The results indicate that an open, deep, and well-developed financial market contributes to the process of currency internationalization. Since SMC, KAOPEN, and M2, are statistically significant in financial market construct. Countries should give priority to enhancing market depth by developing financial infrastructure, supporting the growth of local stock exchanges, and improving regulatory frameworks. Meanwhile, encouraging financial market openness will increase transparency, accessibility, and integration with the global financial market. Countries, like China, that have not yet fully opened up capital accounts need to reduce capital controls, attract foreign investments, and enhance the free flow of capital across their borders. Furthermore, central banks should notice the significance of monetary aggregates and sustain a balanced stable monetary policy. In addition, economic development and money confidence negatively influence currency internationalization for the developed economies involved. This suggests economic development and increased reserve assets can sometimes lead to currency depreciation. Governments had better maintain stable currency value by controlling inflation through monetary and fiscal policies, diversifying reserve asset management to avoid over-reliance on foreign exchange and gold reserves and promote a flexible exchange rate system to manage the outer value of currency.

6.4 Limitations and suggestions for future studies

Three limitations are covered in this study, which provides some enlightenment for future study. First, as one of the most essential international currencies, the euro is not covered in this study due to unavailable data. Many indicators, like gold reserve and KAOPEN, do not include the data of the whole European Union. Second, the data for all indicators are hard data and no soft data are included. The concept of money confidence is always related to people’s trust and feelings, which is better captured by soft data. Furthermore, owing to research capability, the sample size is small only consisting of 9 currencies for 21 years. Long-term observations will be more convincing. Third, by using the PLS-SEM method, more constructs can be considered and added to the conceptual model. The three independent variables in this study are all economic factors, and there are many other factors affecting currency internationalization, including politics, culture, and military, which also should be taken into the conceptual framework.

To summarize, in terms of currency internationalization, economic development, money confidence, financial market, and purchasing power are all essential factors. This paper addressed the gap in empirical research by examining the key drivers, the mediator purchasing power, and currency internationalization in a conceptual framework and extended the application scope of the PLS-SEM method, especially in dealing with secondary data in international finance and economics. Therefore, this study made great contributions to economic performance and international currency.