Introduction

States around the world have difficulties in raising the required level of tax revenue to finance planned activities (Mu et al., 2022). The potential problem of revenue collection especially occurs in moments of crisis, when economic activity slows down and leads to negative tendencies in the movement of fundamental macroeconomic indicators. The issue of the efficiency of revenue collection and collection is important for all fiscal authorities, where it is necessary to determine the main generators of tax revenue in state budgets. In order to provide public services and cover public needs, the creators of fiscal authorities must create an appropriate tax system that will generate the required level of income on a tax, but also non-tax basis. Adequate management of the revenue side of the budget is a prerequisite for optimal planning and execution of projected expenditures.

Tax collection represents an important instrument in ensuring economic stability and development (Majerová, 2016), as well as the most important item of state budget revenues (Kubjatkova et al., 2021). Tax funds can be used to direct the economy towards the realization of specific goals (Kowal, Przekota, 2021). Accordingly, tax revenues should provide funds for appropriate infrastructure, health, education, culture, employment, social income distribution and public safety. (Dobrović et al., 2021). Similarly, Barbu et al. (2022) identifies tax revenues as the main source of financing state budgets for investments in public infrastructure. In order to improve the collection of tax revenues, in addition to the appropriate institutional and legislative framework, it is necessary to improve the capacities of the tax administration, as well as the efficiency of their activities (Ramírez-Álvarez, Carrillo-Maldonado, 2020). Also, the issue of tax collection is essential for fiscal authorities and it is conditioned by the structure of the tax system of a particular country. Namely, the structure of the tax systems of developing and underdeveloped countries is primarily focused on indirect taxes, ie taxes on consumption. On the other hand, the tax systems of developed countries have a much higher share of direct taxes due to a significantly larger income base that is subject to taxation. In addition to serious borrowing constraints and obstacles to increasing resources through raising fiscal pressure, the huge growth in the volume of public spending in most developed countries indicates that economic policy makers must improve the efficiency of the public sector (Cordero et al., 2021). This issue is significant for the management of tax revenues and their efficient allocation for productive purposes.

Taxation of consumption in most countries around the world is based on value added tax and sales tax (Sokolovska, Sokolovskyi, 2015). For this reason, value added tax is gaining more and more importance in empirical measurements and analyzes of effects on certain macroeconomic indicators. Tóth et al. (2021) states that the role of consumption tax is one of the most important issues in the debates about the optimal tax structure from the perspective of efficiency and fairness. Acosta-Ormaechea et al. (2019) points out that economic policy makers often favor consumption taxes to stimulate economic growth over direct taxes on the example of 70 economies.

The structure of the work includes four segments. After the introductory discussion, there follows an overview of theoretical and empirical findings related to value added tax from the point of view of collection and efficiency of collected revenues. The third segment of the work includes a methodological framework that includes the sample, selection of observed variables, definition of hypotheses, application of panel tests in order to select the appropriate empirical model. The fourth segment of the work involves measuring and evaluating the effects of selected variables on VAT efficiency in the Benelux countries. The last segment of the work involves concluding considerations and recommendations for economic policy makers of the Benelux countries.

Theoretical framework of research

Facing the challenges of raising funds and covering the growing needs, many countries are introducing various tax forms, including value added tax (Aizenman, Jinjarak, 2008). Value added tax has become the most common consumption tax worldwide (Giesecke, Tran, 2012), with Keen (2013) pointing out that this tax form has been implemented in more than 150 countries worldwide. Value added tax, as a replacement for sales tax, stands out as an effective tax form, that is, an instrument that generates and increases tax revenues (Hosseini, Briand, 2020). According to OECD records, at the end of 2018, that number increased to 168 countries that implemented VAT in their tax systems (OECD, 2018). The theory predicts value added tax as an efficient tax form, which is one of the primary reasons for the rapid implementation of this tax form worldwide (Adhikari, 2020). The importance of value added tax from the aspect of revenue generation is discussed by Feltenstein et al. (2022) and Morrow et al. (2022), who emphasize that value added tax represents the main source of tax revenues. Also, Lin (2008) points out that this form of taxation generates significantly more revenue compared to other forms of taxation, that is, it represents the most stable and productive form of all components of the tax system (Abshari et al., 2021). Accordingly, value added tax should be a tax form that achieves greater economic efficiency compared to alternative indirect taxes (Mascagni et al., 2021). At the same time, this tax form is not suitable as an instrument of income distribution (Shiraishi, 2022). In countries without major administrative restrictions, this tax form should primarily serve for efficient and predictable revenue collection (Cnossen, 2012).

An efficient tax system is often characterized by a high level of tax collection with acceptable collection costs (Babici et al., 2019). Looking at the structure of the VAT system, Kalyva et al. (2016) emphasizes a simple taxation system with limited use of reduced rates. The introduction of value added tax implies an appropriate economic capacity in terms of absorbing potential price fluctuations. Benkovskis and Fadejeva (2014) state that value added tax rates are important for consumption, but also the price level, as well as that the movement or change of VAT rates can have essential implications for the inflation rate. Accordingly, Milenković et al. (2020) confirm the statistically significant influence of value added tax on the inflation rate in the Republic of Serbia. Campos-Vázquez and Medina-Cortina (2019) state that a lower degree of price dispersion is present in markets with pronounced competition. In addition to the possible price oscillation at the general level, the value added tax is a grateful tax form from the point of view of revenue collection and strengthening of tax discipline. Chan et al. (2017) states that the value added tax system has positive effects on the discipline of the executive in terms of tax revenue collection and management. Strengthening tax discipline is an essential task for fiscal authorities worldwide, given the existence of the problem of tax evasion (Martinez et al., 2022). In the research of Baum et al. (2017) confirmed a statistically significant relationship between revenue collection and tax administration capacity. According to the authors Capasso et al. (2021) it is important to ensure strong fiscal institutions which are essential to ensure a positive public attitude towards paying taxes. Also, authors Lin and Jia (2019), as well as Ndubula and Matiku (2021) indicate in their studies that tax education and tax rates are closely related to the performance of collected revenues.

There is a certain number of theoretical and empirical research that focused on the effectiveness of value added tax (Zídková, 2014; Tagkalakis, 2014; Baum et al., 2017; Bostan et al., 2017; Hodžić, Celebi, 2017; Đorđević et al., 2019; Hodroyiannis and Papaoikonomou, 2020; Kowal, Przekota, 2021). Authors Đorđević et al. (2021) point out that the highest efficiency of VAT collection is possible if fiscal policy measures are correlated with economic and structural policies, achieving a strong synergistic effect. Zídková (2014) confirms a positive relationship between final consumption and the total VAT gap in twenty-four EU countries, while Tagkalakis (2014) points out that improving economic conditions improves VAT efficiency. On the other hand, the empirical analysis of Bostan et al. (2017) shows surprising results confirming that there is no statistically significant difference between the VAT rate and VAT revenue, as well as the negative impact of the VAT rate on the fiscal efficiency of Romania for the time period 2006-2014. years. Acosta-Ormaechea and Morozumi (2021) emphasize that an increase in VAT has a positive effect on long-term growth, only if the collection of collected revenues is improved, but not the standard VAT rate. Also, Hodroyiannis and Papaoikonomou (2020) found that the increase in VAT revenues and their efficiency can be realized through greater use of card payments in countries within the Eurozone for the time period 2006-2016. years.

Since the goal of the research is aimed at determining the VAT efficiency in the Benelux countries, the obtained empirical results can be lucrative for the fiscal authorities of the observed countries when creating and profiling tax policies. The contribution of this research implies that the creators of the economic authorities can appreciate and apply the empirical findings as a guide when formulating the policy and the value added tax system. In this way, the optimal empirical model of VAT efficiency will contribute to maximum revenue collection with minimal harmful implications for the economy.

Methodological research framework

Assessing tax efficiency has always been the focus of public finance research, both at the national and international level (Mukherjee, 2020). The VAT efficiency testing methodology requires the collection of data on VAT revenues in relation to gross domestic product (GDP), standard VAT rate, final consumption and consumer price index (CPI). In recent years, research on VAT efficiency in the European Union has been linked to the model developed by Keen (2013). It determines efficiency by relating actual VAT revenues to theoretical VAT revenues. Based on the concept of Tanzi and Davoodi (2000), PDF efficiency can be calculated through a traditional measurement approach:

VAT efficiency = (1)

At the same time, the VAT efficiency coefficient, which is often denoted in the literature with the symbol C, can also be calculated according to the following mathematical formula:

C-efficiency= (2)

PV = purpose x (FC – V) (3)

V - realized VAT revenues, PV - theoretical (assumed) VAT revenues, SVAR - standard VAT rate, FC - final consumption.

The indicator C-efficiency of VAT shows what percentage of financial spending is collected for each percentage point of the standard rate of value added tax (Cnossen, 2015).

The optimal VAT efficiency coefficient of 100 percent should be considered totally efficient in the case where the tax base is covered by a proportional tax rate. In the case of a reduced VAT rate on certain goods and services, the value of this indicator will be below 100 percent (Hodžić, Celebi, 2017). After calculating these indicators/coefficients, it is possible to determine the VAT gap, as the difference between the achievable VAT income (theoretical) and the real VAT income in order to identify its increase or decrease over time, through the VAT gap rate. Based on the methodological research framework based on the research concepts of Keen (2013) and Popa and Botos (2021), the following hypotheses were developed:

H 0: Macroeconomic factors significantly affect VAT efficiency in the Benelux countries.

H 1: Economic growth rate has a positive effect on VAT efficiency in the Benelux countries.

H 2: The rate of final consumption has a positive effect on VAT efficiency in the Benelux countries.

H 3: The unemployment rate has a negative effect on VAT efficiency in the Benelux countries.

H 4: Inflation rate negatively affects VAT efficiency in the Benelux countries.

H 5: Government expenditures have a positive effect on VAT efficiency in the Benelux countries.

H 6: Higher standard VAT rates lead to less VAT efficiency in the Benelux countries.

H 7: Higher VAT revenues contribute to higher VAT efficiency in the Benelux countries.

The empirical framework for measuring and evaluating VAT efficiency in the Benelux countries is based on the research concept of the authors Keen (2013) and Popa and Botos (2021), which observes VAT revenues, the standard VAT rate, as well as final consumption as a key indicator when measuring the effect of indirect taxation., that is consumption tax. The paper presents a regression model that includes the dependent variable EF and the influence of independent variables such as GDP, FP, NEZ, INF, DR, VAT rate and VAT revenues (Table 1).

Table 1. Selection of explanatory variables

Source: Illustration by the author

Based on the panel construction of Brooks (2008), the regression model is presented as:

Y it = α + βx it + µ it (4)

Y it – dependent variable (EF) α - constant

βx it - coefficients of independent variables (GDP, FP, NEZ, INF, DR, PDVs, PDVp)

and – 3 countries (Belgium, Netherlands and Luxembourg) t – 2011-2020

µ it - residual

EF – VAT efficiency; GDP – gross domestic product growth rate; FP – final consumption; NEZ - unemployment rate; INF – inflation rate; DR – state expenditures; VATs – standard VAT rate; VATp - VAT revenues.

Empirical results and discussion

Within this segment of the work, the calculation of the VAT efficiency indicator in the Benelux countries was performed for the period 2011 - 2020. The obtained values of the observed indicator are a prerequisite for the panel evaluation of the effects of the explanatory variables.

Table 2. Comparative overview of VAT efficiency in the Benelux countries

Source: Author's calculation

Based on the results from Table 2, it can be noted that the average VAT efficiency is the highest in Luxembourg (81.73%) compared to Belgium (46.85%) and the Netherlands (51.04%), where the values of the observed indicator are significantly lower. Analyzing by year, VAT efficiency increased the most in 2012 in the Netherlands (+6.32%), then in 2020 in Luxembourg (+3.94%), and in 2016 in Belgium (+2.26%). If the relative decline of this indicator is analyzed, it can be seen that the VAT efficiency decreased the most in the Netherlands in 2012 (- 8.58%), followed by Luxembourg (-6.41%), as well as in Belgium in 2020 (- 4.98%).

In order to measure the impact of macroeconomic factors on the elasticity and efficiency of the value added tax in the Benelux countries, as part of this research, data from secondary data sources related to macroeconomic indicators (gross social product - GDP, unemployment, final consumption, inflation, government expenditures). All data were collected from the official pages of leading international institutions, so most of the data in the research was taken from the pages of the International Monetary Fund and the World Bank. The research uses data for the period from 2011 to 2020, and refers to the Benelux countries. Before examining and evaluating the impact of macroeconomic factors on VAT elasticity and VAT efficiency, it is necessary to determine potential differences in the level of observed indicators between the analyzed Benelux countries (Belgium, the Netherlands and Luxembourg) for the time period 2011-2020. years.

Table 3. Assessing differences in the level of VAT efficiency

Source: Author's calculation

Table 3 shows the results of the multivariate analysis for the Benelux countries regarding the assessment of the level of differences in the values of VAT elasticity and VAT efficiency for the time period 2011 - 2020. Based on the obtained values of Pillai's Trace = 0.000 for the selected variable VAT efficiency, it can be concluded that there is a statistically significant difference in the level of VAT efficiency in the mentioned countries for the analyzed time period.

After determining the differences in the level of VAT efficiency in the Benelux countries, the examination and assessment of the influence of selected macroeconomic factors on the VAT efficiency panel level for the time period 2011-2020 follows. years. The panel estimation was performed based on the random effects model and the fixed effects model. At the same time, panel causality, i.e. causality between macroeconomic factors and VAT efficiency, is presented. As a condition for an appropriate regression model, it is necessary to conduct stationarity testing, as well as to determine whether there is a problem of multicollinearity, that is, of pronounced correlation between the selected independent variables.

Table 4 shows different panel stationarity tests (LLC test, BTtest and HT test) on the example of three panels (Belgium, Netherlands and Luxembourg) which were applied to examine and identify the stationarity trend of the included variables. The results indicate first-order stationarity for most variables at the 0.05 significance level. The following is an examination of the mutual correlation of the explanatory variables using the VIF test.

Table 4. Stationarity tests

Source: Author's calculation

The results of the VIF test (Table 5) show an average value of 2.39, which leads to the conclusion that there is no problem of multicollinearity between the selected explanatory variables (value less than 4). Bearing in mind that all prerequisites for creating a suitable panel regression model have been met, the following is a comparative presentation of the random effects model and the fixed effects model.

Table 5. VIF test - multicollinearity test

Source: Author's calculation

Based on the results of the random effects model (Table 6), a statistically significant impact of the explanatory variables, i.e. GDP, final consumption, unemployment, inflation, government expenditures, VAT rate and VAT revenue, on VAT efficiency in the Benelux countries for the time period can be observed. 2011 - 2020. Analyzing the character and intensity of the impact, it is noticeable that the variables GDP, FP, DR and VAT revenues have a positive and statistically significant effect on VAT efficiency in the observed countries, which implies that their growth contributes to increasing the efficiency of VAT collection. On the other hand, the variables NEZ, INF and VAT rate have a negative and statistically significant impact on VAT efficiency, which implies that their increase results in a decrease in VAT collection efficiency. The high value of the coefficient of determination R-squared (0.876) indicates that the model is adequately designed, as well as Prob F (0.000) on the validity of the established empirical model.

Table 6. Random effects model

Source: Author's calculation

Based on the results of the fixed effects model (Table 7), it can be concluded that the explanatory variables, namely GDP, final consumption, unemployment, inflation, government expenditures, VAT rate and VAT revenue, have a statistically significant influence on VAT efficiency in the Benelux countries for the time period 2011 - 2020. Analyzing the character and intensity of the impact, it is noticeable that the variables GDP, FP, DR and VAT revenues have a positive and statistically significant effect on VAT efficiency in the observed countries, which implies that their growth contributes to increasing the efficiency of VAT collection. On the other hand, the variables NEZ, INF and VAT rate have a negative and statistically significant impact on VAT efficiency, which implies that their increase results in a decrease in VAT collection efficiency. The high value of the coefficient of determination R-squared (0.924) indicates that the model is set appropriately, as well as the Prob F (0.000) on the validity of the defined empirical model.

Table 7. Fixed effects model

Source: Author's calculation

The results of the Hausman test (Table 8) show that the fixed effects model is adequate when assessing the impact of macroeconomic factors on VAT efficiency in the Benelux countries. This means that the selected model includes explanatory variables that explain 92.4% of the VAT efficiency variations in the observed countries, which indicates high reliability and credibility of the obtained empirical findings.

Table 8. Model selection – Hausman test

Source: Author's calculation

In order to determine the potential causality between macroeconomic factors and VAT efficiency indicators, causality measurement was conducted between selected variables for the time period 2011-2020. years. The results of the causality test (Table 9) indicate bidirectional causality between gross domestic product (GDP), final consumption (FP), government expenditure (DR), VAT rate, VAT revenue and VAT efficiency (EF). On the other hand, unidirectional causality of unemployment (NEZ), inflation (INF) and PDF efficiency EF) was confirmed. In the case of identified unidirectional causality, it should be emphasized that a change on the side of macroeconomic factors such as unemployment and inflation leads to a change in VAT efficiency in the Benelux countries.

Table 9. Measuring the causality of macroeconomic factors and VAT efficiency

Source: Author's calculation

Conclusion

The paper includes an empirical analysis of measurement and assessment of the impact of selected macroeconomic factors on VAT efficiency in the Benelux countries (Belgium, the Netherlands and Luxembourg) for the period 2011-2020. years. Value-added tax rates in the Benelux countries ranged from 17% in Belgium to 21%, which was the value-added tax in the Netherlands and Luxembourg. The average VAT efficiency of the observed countries was 59.84%, with the highest average value of the indicator recorded in Luxembourg (81.73%). On the other hand, Belgium and the Netherlands achieved average values of 46.85% and 51.04% during the analyzed time period. The results of the multivariate analysis indicate statistically significant differences in the level of VAT efficiency in the Benelux countries during the observed time period. Panel regression analysis was performed based on the random effects model and the fixed effects model, with the results of the Hausman test confirming that the fixed effects model is adequate. Empirical findings indicate a statistically significant influence of all variables, which leads to the conclusion that the hypothesis H 0 can be accepted. Bearing in mind the positive impact of gross domestic product, final consumption, state expenditures and VAT revenue on VAT efficiency, it can be stated that hypotheses H 1, H 2, H 5 and H 7 are accepted. Specifically, the growth of GDP, FP, DR and VAT revenue by 1% contributes to the increase of VAT efficiency by 0.20%, 0.96%, 0.16% and 0.59%. On the other hand, unemployment, inflation and VAT rate have a negative impact on VAT efficiency in the observed countries, which allows the acceptance of hypotheses H 3, H 4 and H 6. Namely, the growth of NEZ, INF and VAT rates by 1% results in a decrease in VAT efficiency by 0.22%, 0.15% and 0.60%. The contribution of this paper is reflected in the expansion of the existing theoretical framework aimed at tax efficiency, as well as the provision of new empirical knowledge related to the efficiency of value added tax revenue collection in the Benelux countries. The obtained empirical results indicate that the Benelux countries must focus on higher rates of GDP growth and final consumption with a higher participation of state expenditures, in order to achieve positive implications for VAT efficiency. At the same time, a lower inflation rate and unemployment rate will have positive effects on VAT efficiency. Finally, a reduction in the standard VAT rate can generate more VAT revenue, which will undoubtedly have a positive impact on VAT efficiency in the observed countries. Causality results indicate bidirectional causality between gross domestic product (GDP), final consumption (FP), government expenditure (DR), VAT rate, VAT revenue and VAT efficiency (EF). On the other hand, unidirectional causality of unemployment (NEZ), inflation (INF) and PDF efficiency EF) was confirmed. The contribution of this paper is reflected in the provision of empirical guidelines to economic policy makers when profiling the VAT policy in the context of adjusting the macroeconomic framework of the Benelux countries. At the same time, the obtained empirical findings can be of help to countries around the world, especially countries where VAT has a significant share in collected tax revenues.