Introduction

Modern governments use tax policy decisions to raise capital for government activities and to ensure social peace by eliminating inequality in income distribution. The literature has long discussed that governments can affect economic growth through tax policies. These debates are mainly centered on the long-run and short-term effects. Although there is considerable literature on the effects of fiscal policies on economic growth (e.g., Schultz, 1961; Barro, 1990; Blanchard & Perotti, 2002; Mountford & Uhlig, 2009; Alm & Rogers, 2011), within the framework of the neo-classical growth model, Solow (1956) argues that taxation does not affect long-run growth. Harberger (1964a, 1964b) contends that taxation does not significantly impact a country’s long-term economic growth rate. His research indicates that alterations in the composition of direct and indirect taxes exert minimal influence on the growth of labor supply and the share of income attributed to labor. The marginal impact on labor’s share of income implies corresponding marginal effects on saving and investment rates, ultimately resulting in an inconsequential alteration in output growth (Harberger 1964a). However, endogenous growth models considered in some later studies have shown that governments can affect (positively or negatively) long-run growth rates through tax policies (Jones and Manuelli 1990; King and Rebelo 1990; Capolupo 2000). These studies have mainly provided new evidence that taxation can lead to faster growth when it is used for productive purposes (by including human capital or by specifying a particular production function). Even Lucas (1990), who found little or no effect of taxation, found small negative effects when examining the welfare effects of growth and taxes in a model with human capital, but noted that if taxation is used to provide education or public health, the traditional result may be reversed. Another group of studies argues that taxation has a distorting effect on long-term growth. Those who hold this view argue that taxation should be zero or close to zero (King and Rebelo 1990; Rebelo 1991). In a study by Shin (2012), which has important findings, the indirect effect of tax on economic growth is analyzed based on the effect of tax on income inequality. Although keeping tax rates low in the early stage of economic growth does not have a corrective effect on income inequality, it paves the way for correcting income inequality through economic growth. This policy achieves the goals of rapid economic growth and correction of income inequality. The same policy (low tax policy) does not affect economic growth and income inequality when the economy is close to a steady state. A high-tax policy has a small corrective effect on income inequality in the early stages of economic growth but a large corrective effect on income inequality when the economy is steady. However, correcting income inequality through economic growth is impossible in both cases. Another important perspective is the view that tax cuts will support economic growth through their impact on firm performance. The study by Yue et al. (2023) revealed that tax cuts have a stimulating effect on firms’ research and development (R&D) innovation. Accordingly, tax incentives play an important role in stimulating corporate technological innovation and support the economic growth process. The results of many similar studies on the subject have revealed that tax increases interrupt the innovation process, while tax cuts support the innovation process (Atanassov and Liu 2020; Chen et al. 2024; Lu and Cheng 2024).

Evaluating all these and similar studies, there is no consensus in the literature. While some studies argue that taxation has a negative impact on the growth rate, others argue that the negative impact can be offset by positive growth or even lead to positive growth, given the benefits generated by tax revenues. An important point to note here is that direct and indirect taxes may affect economic growth differently. For example, Widmalm (2001) finds a strong negative relationship between direct taxes and economic growth, while he finds a positive relationship between indirect taxes and economic growth. Johansson et al. (2008) found a similarly strong negative relationship between direct taxes and economic growth, while indirect taxes have a relatively smaller but still detrimental effect on economic growth. Arnold et al. (2011) rank taxes in terms of their impact on per capita income eventually and argue that indirect taxes come first and direct taxes second. Xing (2012) argues that there is no clear evidence that different types of taxes are more or less harmful.

As can be seen, there is no consensus in the literature on whether tax is a blessing or a curse. From another perspective, is there a difference between tax types in terms of being a blessing or a curse? There is also no consensus in the literature on the answer to this question. This study aims to analyze the impact of direct and indirect taxes on economic growth by separating direct and indirect taxes from a correct perspective within the confusion in the literature. The results of this analysis will provide important information on whether tax is a blessing or a curse.

The study consists of six chapters. In the next stage, the literature is discussed in detail, and empirical analyses are carried out by presenting the data, models, and methodology. In the final stage, the analysis results are evaluated, and policy recommendations are made.

Literature review

As mentioned earlier, there is no consensus in the literature on whether taxes are a blessing or a curse. In the case of the distinction between direct and indirect taxes, the issue becomes much more important and has not been sufficiently addressed in the literature. This is the most important originality of the study. Although limited in the literature, it has been mentioned that different tax structures affect macroeconomic indicators (Ilaboya and Ohonba 2013). Taxes, which are also an indicator of the level of development of countries, vary from country to country. However, they are basically categorized into two types: direct and indirect taxes. The main criterion for determining which taxes are considered indirect and which are deemed direct is that indirect taxes are imposed on all individuals, regardless of their wealth, primarily during the transfer of goods within distribution channels. In contrast, direct taxes are levied in direct proportion to an individual’s income, earnings, or wealth, or are directly related to the ownership of assets. According to Buchanan (1970); “Direct taxation is defined as taxation imposed on the person who is intended to be the ultimate bearer of the burden of payment. Indirect taxation, on the other hand, is taxation imposed on persons other than the person intended to bear the ultimate burden”. Another relevant definition is provided by Shoup, who argues that “the most useful distinction between direct and indirect taxes is that indirect taxes can be ‘personalized’ or tailored to the particular economic and social characteristics of the taxed household”. Indirect taxes, on the other hand, are not so adaptable (Chase 1964). Another distinguishing factor is whether a tax office intervenes in tax payments (Erdem et al. 2023). If taxes are collected directly from the taxpayer through tax offices, they are direct and direct taxes; if they are collected through intermediaries (e.g., VAT), they are indirect taxes. In indirect taxes, the burden on the taxpayer is gradually transferred to the next level until it reaches the final consumer. The most important problem with indirect taxes is that since they are levied on everyone, regardless of rich and poor, the share of taxes paid by low-income people in their total income is very high and increases inequality in income distribution (Barnard 2010).

In this section, studies on taxes and economic growth are discussed more broadly. The number of researchers analyzing the relationship between tax revenues and economic growth is quite high. This relationship has generally been analyzed within the framework of empirical studies. The results of empirical studies have yielded contradictory findings. Fiscal instruments are an important factor in achieving macroeconomic objectives. Researchers are curious about the direction and intensity of the relationship between these fiscal instruments and economic growth. Therefore, the existing literature in the relevant field plays a crucial role in guiding policymakers and researchers.

Fiscal policies implemented by states in the context of their taxation powers reveal their tax structures. These policies are of vital importance. Mistakes in tax and debt policies have the potential to put states in difficult situations that they cannot get out of (Hassan et al. 2024). Tax structures that have changed from the past to the present may differ according to countries’ fiscal, social, and economic characteristics. In this respect, the extent to which countries’ tax structures can achieve macroeconomic objectives has become a subject of interest for researchers. In this respect, studies analyzing the impact of tax structures on economic growth across countries have produced mixed results. For example, Stoilova and Todorov (2021) analyzed the impact of three fiscal instruments on economic growth in some Central and Eastern European countries for the period 2007–2019. The study favored indirect tax revenue, direct tax revenue, and public expenditure as the three fiscal instruments. In addition, ten new member states of the European Union from Central and Eastern Europe (Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia) were included in the analysis. The researchers drew some conclusions based on the analysis results to ensure economic growth. According to these conclusions, economic growth in Central and Eastern European countries can be achieved by reducing the share of direct taxes in GDP. In addition, increasing the share of exports and investment in GDP will benefit economic growth. Pandey (2019) analyzed the impact of tax structure on economic growth in India based on data from 1973–74 to 2018–19. This study examined the impact of personal income tax, corporate income tax, and indirect taxes on economic growth. According to the results, corporate income tax and indirect taxes positively impact economic growth, whereas personal income tax has a negative impact. In terms of policy recommendations, the study suggests a gradual increase in corporate and indirect taxes, coupled with a corresponding gradual reduction in individual income tax, to promote sustainable economic growth.

The results of studies examining the impact of tax revenue on economic growth show variability depending on the specific category of tax under consideration. Within certain studies, the impact of both direct and indirect taxes on economic growth is positive and negative. Sharabidze (2023) tested the effect of tax revenues on economic growth in Georgia. The autoregressive distributed lag (ARDL) model was used for the test. The analysis results show indirect taxes’ effect on economic growth is positive, whereas direct taxes are negative. The researcher argues that taxes on income reduce disposable income, thereby reducing investment opportunities and negatively impacting economic growth. Balasoiu et al. (2023) investigated the impact of direct taxation on economic growth in 27 EU countries based on data covering the period 2008–2020. In this study, the clustering method is used, and EU countries are divided into two groups in terms of high and limited fiscal efficiency. The findings show that corporate tax significantly negatively impacts economic growth in both groups. Moreover, individual income tax is associated with lower economic growth for countries with limited fiscal efficiency. According to the researchers, lowering direct taxes would increase disposable income and consumption, supporting economic growth. Moreover, direct taxes will become an incentive tool, thus creating employment opportunities. Mamo (2023) analyzed the estimation of the relationship between tax structures and per capita income growth in the state. The study found a positive relationship between indirect and direct tax rates and income growth. The analysis revealed a consistently negative relationship between indirect taxes, property taxes, and growth. Corporate income tax, individual income tax, and sales tax have no relationship with growth. In this regard, the researcher also emphasized the negative impact of indirect taxes on economic growth and unequal income distribution. Abd Hakim et al. (2022) examined the impact of direct and indirect taxes on economic development in 47 developed and 90 developing countries. The results showed a significant negative association between direct and indirect taxes and economic development in developing countries. Consequently, the analysis suggests that the prevailing tax structure does not contribute to increased economic growth in developing countries. In contrast, a positive relationship was found between direct taxes and economic development in developed countries. In another study by Mgammal et al. (2023), the impact of changes in VAT rates—one of the most significant components of the indirect tax category—on corporate financial performance was analyzed. Using the ARIMA model, the results of the study show that a sharp increase in VAT has negative effects on company financial statements and may even increase the probability of bankruptcy. These results suggest that a sharp increase in VAT may lead to a decline in tax revenues in the long run by affecting unemployment and may ultimately have a negative impact on economic growth.

Some studies measure the impact of public expenditures on economic growth. In such studies, both the effect of tax revenues and the impact of public expenditures on economic growth have been analyzed. Furthermore, the causal relationships among tax revenues, public expenditures, and economic growth have been investigated. For example, Gurdal et al. (2021) examined the causality relationship between tax revenues, public expenditures, and economic growth in G7 countries, namely “Canada, France, Germany, Italy, Japan, the United Kingdom, and the United States”, using annual data for the period 1980–2016. According to the findings, a bidirectional causality exists between economic growth and public expenditures, and a unidirectional causality exists between tax revenues and public expenditures. Furthermore, the study did not find a causal relationship between “economic growth and tax revenues”. This study emphasizes the importance of developing tax policies in line with the economic situation of the G7 countries as a key fiscal instrument for achieving economic objectives. A similar study by Maulid et al. (2022) analyzed the causality between tax revenues, government expenditures, inflation, and economic growth in Indonesia for the period 1973–2019. According to the findings, a positive bidirectional causality relationship was found between tax revenues and government expenditures and between tax revenues and economic growth. While government expenditures cause economic growth, the reverse is not valid. The effect of inflation on tax revenues, government expenditures, and economic growth is negative. Generally, the government is claimed to generate tax revenues that can finance its expenditures successfully. In addition, an increase in revenues enhances the government’s ability to spend, making public expenditures an important tool for ensuring economic growth. Another study that examines the relationship between public finance and economic growth in positive and negative directions is Romero-Ávila and Strauch 2008. This study focuses on the relationship between “public finance and economic growth” in the EU. The results of the analysis show that public expenditures and transfers have a negative impact on economic growth. On the other hand, the findings show that public investment positively impacts growth. Direct taxation has a strong negative impact on capital accumulation.

Given the specificity of our study to Turkiye, a review of the relevant literature examines the impact of tax revenue on economic growth in the Turkish context. Within this framework, the review includes evidence from studies examining the impact of direct and indirect taxes on Turkiye’s economic growth. In this context, a positive relationship was found in a study that concluded that both indirect and direct tax revenues positively affect economic growth (Özen et al. 2022). There is also evidence that indirect taxes positively impact economic growth in Turkiye (Saraç 2015; Korkmaz et al. 2019). In such studies, it is stated that indirect taxes have a positive impact on economic growth. Increasing indirect taxes will increase government revenues. Therefore, although increasing this type of tax is important in terms of economic growth and government revenues, it is also deficient in tax justice. There are also results in the literature that direct taxes have a negative impact on economic growth (Saraç 2015; Korkmaz et al. 2019). In this case, an increase in direct taxes has a negative impact on economic growth. A heavy tax burden on income can significantly affect taxpayers’ willingness to work. In particular, such taxes on income tend to reduce individuals’ disposable income. Therefore, an increase in income taxes may have a negative impact on the savings and investments of individuals and institutions. The literature also analyzes the relationship between tax revenues and economic growth in Turkiye. Some analyses have found positive effects between tax revenues and economic growth (Polat 2019), whereas others have found negative effects (Ozpence, 2017). Moreover, the relationship between tax burden and economic growth in Turkiye has also been analyzed. Although the findings of the analyses generally suggest that the tax burden has a negative impact on growth (Mangir and Ertuğrul 2012; Karayilmazlar and Göde 2017; Organ and Ergen 2017), some studies conclude that an increase in tax burden contributes to economic growth (Koç 2019).

In this study, the concept of “tax curse” is discussed, inspired by the “resource curse” hypothesis in the literature. Therefore, briefly mentioning the “resource curse” hypothesis would be appropriate.

The resource curse hypothesis in literature is used in many research areas. Studies examining the relationship between “natural resource wealth and economic growth” in countries occupy an important place in the literature. Atkinson and Hamilton (2003) found a significant negative relationship between natural resource wealth and economic growth. In the analysis findings of this study, it is stated that resource revenues should serve as a financial source for public expenditures, and thus, increased public expenditures will support public investments. Moreover, it has been argued that increasing public investment in resource-rich countries can avoid the resource curse. Crivelli and Gupta (2014) studied the impact of increased resource revenues on domestic (non-resource) incomes in a cohort of 35 resource-rich countries. The results of this study show a statistically significant negative correlation between resource revenues and domestic (non-resource) incomes. Shahbaz et al. (2019) examined the correlation between oil prices and economic growth, intending to provide evidence for the resource curse hypothesis. The empirical results indicate the existence of a long-term relationship between the variables. Moreover, the observed negative relationship between natural resource wealth and economic growth supports the natural resource hypothesis. This study reveals the existence of unidirectional causality from natural resource wealth to economic growth. Henri (2019) aimed to identify the damages caused by natural resource rents to African institutions and economic indicators. In the findings obtained from the analysis, the institutional problems caused by natural resource rents were identified as corruption, lack of accountability, and low efficiency in public services. In terms of economic indicators, the phenomenon caused fluctuations in GDP per capita, resulting in lower levels of physical and human capital. The researcher concluded that African countries should promote good institutional governance and undertake economic diversification initiatives. Marques and Pires (2019) tested whether there is a relationship between “natural gas and economic growth” in the context of the resource curse hypothesis. Three approaches were used to test the resource curse hypothesis by comparing production, reserves, and rents. The analysis found evidence of the resource curse phenomenon only in the production approach. According to the analysis results, natural gas production eventually hampers economic growth.

To the best of our knowledge, the “tax curse” has not been addressed in the literature from this perspective. Only Sun et al. (2022) partially addresses this issue. In this study, the authors examine the impact of resource taxation on the resource curse in China. In addition, the study revealed findings on resource tax policy. According to the findings, it is emphasized that resource taxes of resource-rich states should be regulated to avoid the resource curse. This study makes an important contribution to the literature by approaching the issue from the “tax curse” hypothesis framework and analyzing taxes in terms of direct and indirect taxes.

Hence, the hypotheses of the study are as follows:

H1: Indirect taxes (IT) positively affect economic growth. There is no tax curse in terms of indirect taxes.

H2: Direct taxes (DT) positively affect economic growth. There is no tax curse in terms of direct taxes.

In addition, our hypothesis for exports added to the model as a control variable is as follows:

H3: Export volume growth (EXP) positively affects economic growth.

Model and data

To determine whether the tax curse hypothesis is valid (whether tax types have a positive or negative impact on economic growth) in the case of Turkiye, a time series model with all variables defined is constructed as in Eq. (1).

$${{GDP}}_{t}=\alpha +{\beta }_{1}{({IT})}_{t}+{\beta }_{2}{({DT})}_{t}+{\beta }_{3}{({EXP})}_{t}+{\varepsilon }_{t}$$
(1)

In the model, t = 1998/Q1, …, 2023/Q2, GDP is the growth rate of gross domestic product (economic growth), IT is the growth rate of indirect taxes, DT is the growth rate of direct taxes, EXP is the growth rate of export volume, εit is the error term, αi is the country-specific fixed effect. β1, β2, and β3 correspond to the long-run elasticities of GDP with respect to indirect taxes (IT), direct taxes (DT), and export volume (EXP), respectively.

Table 1 shows the definition and scope of the variables and the sources from which they were obtained.

Table 1 Description of the variables.
Full size table

Compared with the previous quarter, growth rates are considered for all variables included in the analysis. A summary of the descriptive statistics of the variables is presented in Table 2.

Table 2 Descriptive Statistics.
Full size table

As shown in Table 2, the series does not conform to a normal distribution, as indicated by the robust rejection of the Jarque–Bera statistic at the 1% significance level for the four variables. The series is left-skewed, characterized by negative values and skewed variables. The smallest value belongs to the DT variable (−0.04374), and the largest value belongs to the DT variable (0.10222). The smallest mean value belongs to the GDP variable (0.00305), and the largest mean value belongs to the DT variable (0.00735).

The correlation relationship between the variables is as follows:

As shown in Table 3, there is a high correlation between the variables.

Table 3 Correlation matrix.
Full size table

Methodology

We use the Prais-Winsten (1954) adjusted ordinary least squares (OLS) regression model to test the tax curse based on the correlation between the variables in the model. The main reason for using the Prais-Winsten (1954) adjusted regression model is that it is a recommended method to overcome autocorrelation (Dielman 1985; Bimanto et al. 2023) and performs better than conventional OLS and ARMA models (Dielman and Rose 1994; Sharma and Coleman 2005; Bottomley et al. 2023). Like OLS, the Prais–Winsten method does not require the residuals to be normally distributed. However, inferences about the coefficients may be more reliable when the residuals are approximately normally distributed. The variables considered in this study are not normally distributed, which makes the model appropriate for use.

First, the unit root test is applied in the study. For this purpose, the flexible Fourier form and Dickey-Fuller unit root test proposed by Enders and Lee (2012) are used. Because the related unit root test formulations are widely used in the literature, they are not detailed here. Then, an adjusted ordinary least squares (OLS) regression model with the Prais–Winsten method is applied.

The Prais–Winsten method is an extension of the Cochrane–Orcutt (1949) method and provides an approach to overcome the problem of autocorrelation in time series data (Park and Mitchell 1980). “The method involves estimating the correlation [corr (\({\varepsilon }_{t},\,{\varepsilon }_{t-1}\))] between the error at ‘t’ and at ‘t-1’. This estimate is then used to transform the outcome and forecast variables such that the correlation is subtracted from the error when a linear regression model is fitted to the transformed data. The basic assumption behind the method is that the error follows a first-order autoregressive process, so autocorrelation is only completely removed if the error follows this model” (Bottomley et al. 2023). The Cochrane–Orcutt method, which is based on the Prais–Winsten method, “involves removing the first observation in the data and therefore the sample must be large enough to follow this method” (Culas and Timsina 2019). Because the study includes quarter-based data, this sample is large enough in this sense. However, the Prais–Winsten transformation preferred in this study already eliminates this disadvantage of the Cochrane–Orcutt method (Greene 2017).

Accordingly, in the Cochrane–Orcutt method, the error term in the regression equation includes the following:

$${\varepsilon }_{t}={{\rm{\theta }}}_{1}{\varepsilon }_{t-1}+{n}_{t}$$
(2)

Here, \(\left|{{\rm{\theta }}}_{1}\right| < 1\) and \({n}_{t}\) are independent disturbances with zero mean and \({\sigma }^{2}\) variance. The restriction on the autoregressive (AR) parameter \({{\rm{\theta }}}_{1}\) ensures that the process is stationary, i.e. that \(\mathrm{cov}({\varepsilon }_{t},{\varepsilon }_{t+h})\) is independent of “t” and hence the variance remains constant over time (Bottomley et al. 2023).

The Prais–Winsten method includes the elements indicated by \({\widetilde{y}}_{0}\) ve \({\widetilde{x}}_{0}\) in the analysis and thus increases the precision of parameter estimates compared with the Cochrane–Orcutt method. In this method, \({\widetilde{y}}_{0}\) ve \({\widetilde{x}}_{0}\) are described as follows (Bottomley et al. 2023):

$${\widetilde{y}}_{0}={y}_{0}\sqrt{(1-{\hat{{\rm{\theta }}}}_{1}^{2})}\,{and}\,{\widetilde{x}}_{0}={x}_{0}\sqrt{(1-{\hat{{\rm{\theta }}}}_{1}^{2})}$$
(3)

In addition, the Toda-Yamamoto causality test developed by Nazlioglu et al. (2019), which takes into account structural changes, including gradual/smooth changes, is used to support the Prais-Winsten (1954) Regression Model. Nazlioglu et al. (2019) defined the relevant equation as in Eq. (4).

$$\begin{array}{l}{y}_{t}={{\rm{\alpha }}}_{0}+\mathop{\sum }\limits_{k=1}^{n}{\gamma }_{1}\sin \left(\frac{2\pi {kt}}{T}\right)+\mathop{\sum }\limits_{k=1}^{n}{\gamma }_{2}\cos \left(\frac{2\pi {kt}}{T}\right)+{\vartheta }_{1}{y}_{t-1}\\\qquad+\,\ldots {\vartheta }_{p+d}{y}_{t-\left(p+d\right)}+{\varepsilon }_{t}\end{array}$$
(4)

In the Toda-Yamamoto framework, the null hypothesis for Granger causality is based on the zero restriction on the first “p” parameters of the “K”th component of the “\({y}_{t}\)” series (\({{H}_{0}:\vartheta }_{1=\ldots =}{\vartheta }_{p}=0\)). The Wald statistic used to test this hypothesis has an asymptotically χ² distribution with “p” degrees of freedom.

Finally, Wavelet Transform Coherence (WTC) analysis was conducted to analyze the long-run performance of the relationship between the variables used in the model. Torrence & Webster (1999) defined the equation of the adjusted wavelet coherence coefficient as follows (Eq. 5):

(5)

In Eq. (3), M is represented as a smoothing operator. The range of the square wavelet coherence coefficient is “\(0\le {W}_{n}^{2}(m)\le 1\)”. A value close to zero reflects a weak correlation, while a value close to one reflects a strong correlation.

For the Morlet wavelet (MW), M can be represented as time (τ) and frequency (⨑), respectively, as in Equation (Yilanci and Pata 2022) (Eq. (6)):

(6)

Here \({\varkappa }_{1}\) and \({\varkappa }_{2}\) are the normalization constants and Π is the rectangle function. The coefficient 0.6 in the equation is the scale-averaging factor empirically determined by Torrence & Compo (1998). The significance of the estimated wavelet coherence was determined using Monte Carlo simulations.

Empirical results

First, the unit root test is performed, and the results are presented in Table 4.

Table 4 Augmented Dickey–Fuller Unit Root Tests with Flexible Fourier-Shaped Structural Breaks.
Full size table

As can be seen in the table of unit root test results for the variables (Table 4), the DT and GDP variables are stationary at the level. While the EXP variable becomes stationary at the first difference, the IT variable becomes stationary at second difference. To make all variables stationary at the first difference, the first difference value of the IT variable was taken as the level value, and all variables were made stationary at the first difference.

At this stage, the regression model adjusted by the Prais–Winsten (1954) method was estimated Table 5.

Table 5 Prais-Winsten (1954) Regression Model.
Full size table

Among the coefficients of the variables in the model, only p > 0.1 (statistically insignificant) for the DT variable and p < 0.01 (statistically significant) for all other variables. The fact that the DT variable is statistically insignificant and has a negative sign indicates that direct taxes have no effect on economic growth. However, it is understood that direct taxes are inefficient and ineffective in Turkey, both in terms of taxation being the source of public expenditures and in terms of fulfilling the function of taxation to encourage production. Since the statistical value of the DT variable is p > 0.1, the causality relationship between the DT and GDP variables is further analyzed. For this purpose, the Toda-Yamamoto causality test developed by Nazlioglu et al. (2019), which takes into account structural changes, including gradual/smooth changes, was used. The causality test results are presented in Table 6.

Table 6 Cumulative Fourier-frequency Toda & Yamamoto Test Results.
Full size table

The causality test results in Table 6 confirm the significance of the coefficient of the IT variable in the Prais-Winsten (1954) Regression Model. Indirect Taxes are effective on GDP, and there is a bidirectional causality relationship between them. In terms of the DT variable, Direct Taxes have no effect on GDP (there is only a unidirectional causality relationship from GDP to DT). These results support the conclusion of Prais-Winsten (1954) Regression Model that DT does not have a positive coefficient. These results of the empirical analysis confirm the hypothesis “H1: Indirect taxes have a positive effect on economic growth. There is no tax curse in terms of indirect taxes”. Accordingly, hypothesis H1 is accepted. Secondly, the empirical results confirm the hypothesis “H2: Direct taxes have a positive effect on economic growth. There is no tax curse in terms of direct taxes” but the alternative hypothesis is valid. Accordingly, “There is a tax curse in terms of direct taxes” and H2 hypothesis is rejected. Our last hypothesis “H3: Export volume growth has a positive effect on economic growth” is also confirmed by empirical analysis. Table 7 summarizes the results of the hypothesis tests performed by Prais-Winsten and Cumulative Fourier-frequency Toda & Yamamoto Test.

Table 7 Summary of the hypothesis test results.
Full size table

For all these reasons, then, the Prais-Winsten compatible model can be defined as follows:

$$\begin{array}{l}{{GDP}}_{t}=-0.00808\,{({DT})}_{t}+0.02564\,{({IT})}_{t}+0.18055\,{({EXP})}_{t}\\\qquad\qquad+\,0.00250+{\varepsilon }_{t}\end{array}$$
(7)

and

$${\varepsilon }_{t}=0.4614{\varepsilon }_{t-1}+{n}_{t}$$
(8)

Finally, the long-run performance of the relationship between the variables used in the model is analyzed. For this purpose, the long-term relationships between the variables are evaluated by Wavelet Transform Coherence (WTC) analysis to reflect the changes throughout the analysis period. WTC analysis examines the relationship between variables at different frequencies throughout the analysis period. In the WTC analysis, while the red color indicates a high probability relationship, it is understood that the probability value decreases as the color turns blue. In addition, black lines indicate a significant relationship during the analysis period. The arrows between the black lines indicate a positive or negative relationship. The figures of the variables obtained with the WTC analysis are shown below. At this point, the relationship between IT and GDP variables is first analyzed (Fig. 1):

Fig. 1: The Relationship Between IT and GDP Variables.
figure 1

The graphical presentation of the wavelet transform consistency analysis of the relationship between IT and GDP variables is given in this figure. In Figs. 1–3, the horizontal axis represents time, while the vertical axis represents frequency density. In the graph, the cone of influence showing the region remaining at 5% significance level is separated by a thin white line. In the figures, red color represents high coherence, while yellow, green and blue represent low coherence. The shapes formed by the black lines show the areas of influence and the arrows inside them provide information about the direction of the influence. Arrows pointing up and to the right are interpreted as positive relationships, while arrows pointing left and down are interpreted as negative relationships. The x column gives date information, while the y column gives frequency density information.

Full size image

When the relationship between IT and GDP shown in Fig. 1 is analyzed, it is observed that there is a strong and positive relationship, especially at high frequencies throughout the analysis period. Towards the end of the analysis period, it is observed that the relationship existing at high frequencies is also valid at medium and low frequencies. Second, the relationship between DT and GDP variables is analyzed (Fig. 2):

Fig. 2: The Relationship Between GDP and DT Variables.
figure 2

The graphical presentation of the wavelet transform consistency analysis of the relationship between DT and GDP variables is given in this figure.

Full size image

Figure 2 shows the relationship between GDP and DT. The relationship between the two variables was limited and complex throughout the analysis period. In the early years of the analysis period, there was no relationship, while later on and for a long period of time, a negative relationship was observed at medium frequency. These results are confirmed by the Prais-Winsten (1954) Regression Model and the Cumulative Fourier-frequency Toda & Yamamoto Test. Towards the end of the analysis period (in the last three years), there was a change and the relationship turned into a positive relationship at high frequencies. Finally, the relationship between EXP and GDP variables is analyzed (Fig. 3):

Fig. 3: The Relationship Between GDP and EXP Variables.
figure 3

The graphical presentation of the wavelet transform consistency analysis of the relationship between EXP and GDP variables is given in this figure.

Full size image

Figure 3 analyzes the relationship between GDP and EXP variables. It is observed that the relationship between GDP and EXP is positive and significant throughout the analysis period, and generally, significant relationships occur at medium frequencies. The positive effect of EXP, which is the control variable, on GDP supports the export-led growth hypothesis, which has an important place in the literature and whose effect is clearly seen especially in developing countries (Fosu 1990; Rivera-Batiz and Romer 1991; Buffie 1992; Thornton 1996; Trlaković et al. 2018; Waheed et al. 2020; Nam and Ryu 2024).

This study analyzes the impact of direct and indirect taxes on GDP using data for the period 1998/01–2023/Q2 for Turkiye. The correlation matrix between the variables evaluated in the study shows a strong correlation between the variables. Then, the unit root test and Prais–Winsten (1954) analysis were applied after stationarity corrections. The results of the analysis showed that direct taxes have a negative impact on economic growth and that the tax curse hypothesis is valid in this respect, whereas the tax curse cannot be mentioned in relation to indirect taxes. The causality analysis also supports this, and the WTC analysis applied to the variables. The results have important implications for government policymakers. In light of these results, governments should reduce the income-based direct tax burden on firms. As many studies in literature and this study have shown, direct taxes have a negative impact on economic growth (Easterly et al. 1994; Widmalm 2001; Stoilova and Todorov 2021; Sharabidze 2023; Balasoiu et al. 2023). Although it is not possible to eliminate direct taxes, as some studies in the literature suggest, a proportional reduction will contribute positively to the economic growth process (Chamley 1986; King and Rebelo 1990; Rebelo 1991; Shin 2012; Stoilova and Todorov 2021). Furthermore, if governments allocate the resources raised through direct and indirect taxes to productive areas (industrialization and exports) that support the economic growth process, this will have a positive impact on the economic growth process (Jones and Manuelli 1990; Lucas 1990; Capolupo 2000; Romero-Ávila and Strauch 2008; Tian et al. 2020; Hasan et al. 2021). Allen (2011) argues that economic growth and success are “the result of secure property rights, low taxes, and minimal government”. Whether state power in the economy is effective or not is another matter, but as Allen (2011) says, economic growth cannot be realized if “arbitrary government policies” and “high tax policies” open the door to rent-seeking. Therefore, governments should limit direct taxes and use them in productive areas by determining an optimal equilibrium point.

When the regression results are evaluated, it can be said that direct taxes negatively impact economic growth; in this respect, the tax curse hypothesis is valid. In terms of indirect taxes, there is no tax curse. The results show that direct taxes have a destructive effect in economies such as Turkiye, which is in the middle of the economic growth process and has a high growth rate. As seen in the Wavelet Transform Coherence (WTC) analysis, this effect only disappeared in the last years of the analysis period. The results make an important contribution to the literature in addressing the impact of direct and indirect taxes separately on economic growth and revealing the factor called the tax curse. In terms of its results, the study differs from studies such as Solow (1956), Harberger (1964a), Harberger (1964b), and Gurdal et al. (2021), etc., which state that taxes do not affect economic growth. Again, the study does not support studies that say that indirect taxes have a negative effect on economic growth (Abd Hakim et al. 2022; Mamo 2023), while it supports studies that say that indirect taxes have a positive effect on economic growth (Widmalm 2001; Pandey 2019; Korkmaz et al. 2019; Sharabidze 2023). When the situation is evaluated in terms of direct taxes, the study does not support studies that say that direct taxes have a positive effect on economic growth (Özen et al. 2022), while it supports studies that say that direct taxes have a negative effect on economic growth (Widmalm 2001; Stoilova and Todorov 2021; Sharabidze 2023; Balasoiu et al. 2023). In light of these results, the possible ways to remove the tax curse can be by keeping tax rates at reasonable levels (King and Rebelo 1990; Rebelo 1991; Stoilova and Todorov 2021) and transferring the taxes collected by the public authority to productive and industrialization-supporting elements (Jones and Manuelli 1990; Lucas 1990; Capolupo 2000; Romero-Ávila and Strauch 2008).

Conclusion

For governments in today’s modern economic systems, taxes are the only resource they should perform their promised policies. Taxes, on the one hand, ensure the implementation of government policies and, on the other hand, have an impact (positive and/or negative) on economic growth and development, which is one of the main objectives of governments. A comprehensive evaluation of the literature shows that many conflicting studies indicate that taxes have a positive or negative effect on economic growth, and at times may have no effect at all. Some of these studies have tried to show that there are differences between types of tax in terms of their effect on economic growth, some studies have tried to show that there are differences between tax rates, and some studies have tried to show that country conditions affect the result. On the other hand, this study analyses direct and indirect taxes as separate variables in the case of Turkiye and whether they are a blessing or a curse in terms of their impact on economic growth. The main motivation behind this study is the authors’ belief that taxes are indispensable for the governments of countries, but they can have different results in terms of how they are implemented. In this respect, the results can provide guidance to governments.

Direct taxes are expected to have more significant and direct effects on economic growth. To achieve this, some policies should be implemented, or existing policies should be reviewed. The existing practices for direct taxes, which are the ideal source of public revenue for taxing the rich or those who earn more, should be reviewed. For example, the exemption from taxation of revenues from sports branches, especially football, which is applied in Turkey, causes serious losses in public revenues. The practice of tax-free income for the soccer community needs to end. On the other hand, there should be more audits in the taxation processes of high earners and tax evasion should be prevented. Loopholes in tax laws and regulations should be eliminated and tax avoidance should be prevented. In Turkey, some institutions and organizations are entitled to tax exemption. It is necessary for the efficiency of the tax system to continuously audit these tax-exempt institutions and organizations and to make decisions to subject them to re-taxation if necessary.