Is taxation a curse or a blessing? The case of Turkiye

The relationship between taxation and economic growth is a long-standing debate in economics. While neoclassical models often suggest a neutral impact of taxation on long-run growth (Solow, 1956; Harberger, 1964a, 1964b), endogenous growth models introduce the possibility of both positive and negative effects depending on how tax revenues are utilized (Jones and Manuelli, 1990; King and Rebelo, 1990; Capolupo, 2000). Some research posits a distortionary effect of taxation, advocating for minimal or zero taxation (King and Rebelo, 1990; Rebelo, 1991), while others highlight the potential for positive growth through productive public spending funded by taxes (Lucas, 1990). The impact of tax cuts on firm performance and innovation is also a significant area of inquiry, with studies finding stimulating effects from R&D incentives (Yue et al., 2023) but also potential negative impacts on innovation if tax increases interrupt the process (Atanassov and Liu, 2020; Chen et al., 2024; Lu and Cheng, 2024). A further complexity arises from the distinction between direct and indirect taxes, with some studies suggesting direct taxes negatively affect economic growth while indirect taxes might have a positive or smaller negative impact (Widmalm, 2001; Johansson et al., 2008; Arnold et al., 2011; Xing, 2012). This study addresses the lack of consensus in the literature by specifically analyzing the impact of direct and indirect taxes on Turkiye's economic growth, aiming to determine whether either type constitutes a "tax curse" and providing valuable insights for developing countries seeking to formulate effective fiscal policies. The study will explore these effects using quarterly data and robust econometric techniques.

The existing literature presents conflicting views on the relationship between taxation and economic growth. Studies examining the effects of tax revenues on economic growth yield mixed results, with some showing positive correlations while others indicate negative or neutral impacts. The impact of different types of tax structures on macroeconomic indicators has also been debated. Direct and indirect taxes, distinguished by their method of collection and impact on income distribution, are frequently analyzed, but the consensus regarding their effects remains elusive. Studies on the relationship between tax structure and economic growth in various countries show a range of findings, for instance, Stoilova and Todorov (2021) found that reducing direct taxes and increasing exports benefited Central and Eastern European countries; Pandey (2019) found corporate and indirect taxes positively affect India's economic growth; Sharabidze (2023) showed indirect taxes positively impact Georgia's economic growth but direct taxes are negative; Balasoiu et al. (2023) found negative correlations between corporate tax and economic growth in the EU; and Mamo (2023) showed that direct and indirect taxes show a mixed relationship to income growth. Research on the impact of tax revenue on economic growth in Turkey also presents conflicting evidence, with some studies finding positive effects (Özen et al., 2022; Saraç, 2015; Korkmaz et al., 2019) and others showing negative impacts (Saraç, 2015; Korkmaz et al., 2019; Özpençe, 2017; Mangir and Ertuğrul, 2012; Karayilmazlar and Göde, 2017; Organ and Ergen, 2017; Koç, 2019). This study aims to add to this complex body of research by focusing on Turkiye, separating the effects of direct and indirect taxes, and utilizing a robust econometric approach.

To test the tax curse hypothesis in the context of Turkiye, the study uses quarterly data from 1998/Q1 to 2023/Q2. The core model is a time-series regression equation (Eq. 1): GDPt = α + β1(IT)t + β2(DT)t + β3(EXP)t + εt, where GDP represents the gross domestic product growth rate, IT is the growth rate of indirect taxes, DT is the growth rate of direct taxes, EXP is the growth rate of export volume, and εt is the error term. The study first conducts unit root tests using the flexible Fourier form and Dickey-Fuller unit root test (Enders and Lee, 2012) to ensure stationarity of the time series data. Given the potential for autocorrelation, the Prais-Winsten (1954) method, an extension of the Cochrane-Orcutt (1949) method, is employed for regression analysis. This approach accounts for autocorrelation in the error terms (Dielman, 1985; Bimanto et al., 2023) and is considered superior to conventional OLS and ARMA models in specific contexts (Dielman and Rose, 1994; Sharma and Coleman, 2005; Bottomley et al., 2023). The Prais-Winsten method transforms the variables to remove the autocorrelation, improving the accuracy of parameter estimates. The Cochrane-Orcutt method is mentioned as a related technique, but Prais-Winsten is preferred as it doesn't exclude the first observation which can be valuable given the quarterly data of this study. The study uses equation 2 (εt = θ1εt-1 + ηt) and equation 3 (y0 = y0√(1 - θ1) and x0 = x/(1 - θ1)) to model the transformation process to account for first-order autoregressive process. To further investigate causality, the Cumulative Fourier-frequency Toda & Yamamoto Test (Nazlioglu et al., 2019) is applied, which accounts for structural changes in the data using equation 4 (yt = α + Σk=1p γt sin(2πt/T) + Σk=1p γt cos(2πt/T) + γt-1 +…+ γt-p+d yt-(p+d) + εt). Finally, Wavelet Transform Coherence (WTC) analysis (Torrence & Webster, 1999; Yilanci and Pata, 2022; Equation 5 & 6) is used to examine the long-run relationship between the variables at different frequencies over the study period. This multifaceted approach allows for a robust assessment of the relationship between different tax types and economic growth in Turkiye.

The unit root tests revealed that DT and GDP were stationary at level, EXP was stationary at first difference and IT was stationary at the second difference. After appropriate transformations to achieve first-difference stationarity for all variables, Prais-Winsten regression analysis was conducted. Results showed that only the IT and EXP variables were statistically significant (p<0.01), with positive coefficients indicating positive effects on GDP growth. The coefficient for DT was statistically insignificant (p>0.1) with a negative sign, implying a negative impact but lacking statistical robustness. The Cumulative Fourier-frequency Toda & Yamamoto causality test further supported these findings. A bidirectional causality was found between IT and GDP, confirming the positive impact of indirect taxes on economic growth. For DT and GDP, only a unidirectional causality from GDP to DT was observed, confirming that direct taxes do not significantly affect GDP growth and the effect is likely negative. The WTC analysis visualized these relationships across different frequencies over time. For IT and GDP, a strong positive relationship, especially at high frequencies, was observed, and this relationship became apparent at medium and low frequencies as well toward the end of the period. The relationship between DT and GDP was more complex and less consistent, showing a mainly negative relationship at medium frequencies in much of the period but turning positive at high frequencies towards the end. The relationship between GDP and EXP showed a consistently positive and significant association across frequencies, supporting the export-led growth hypothesis. The findings support the hypothesis that indirect taxes positively affect economic growth (H1), while rejecting the hypothesis that direct taxes positively affect economic growth (H2), supporting instead the alternative that direct taxes negatively affect GDP. The hypothesis that export growth has a positive effect on economic growth (H3) was also accepted. The negative impact of direct taxes on economic growth is statistically insignificant, but the negative sign is notable.

The study's findings provide strong evidence of a "tax curse" effect associated with direct taxes in Turkiye, while indirect taxes demonstrate no such negative impact. The negative effect of direct taxes, while not statistically significant in the main regression model, is consistently indicated by both the direction of the coefficient and the causality tests, and it is notable because it is visible in the WTC analysis as well. This is in contrast to various studies suggesting a neutral or positive relationship between taxation and growth. The temporary nature of the negative effect of direct taxes, becoming less evident towards the end of the analysis period, suggests a possible adaptation or mitigation mechanism over time. The positive effect of indirect taxes aligns with some studies suggesting positive impacts on growth. The significant and positive relationship between export growth and GDP growth is in line with the export-led growth hypothesis, which is frequently observed in developing economies. Policy implications are significant, especially for developing countries: Direct taxes must be managed carefully to avoid hindering economic growth. Governments should explore ways to optimize revenue generation through indirect taxes while minimizing the negative impacts of direct taxes, possibly by focusing on more progressive direct tax systems that are applied more carefully and with fewer loopholes that lead to tax evasion or avoidance.

This study contributes to the literature by providing strong evidence of a tax curse effect specifically related to direct taxes in Turkiye. This contrasts with some existing literature that suggests a neutral or positive effect of taxation on economic growth. The findings highlight the importance of considering the distinct impacts of direct and indirect taxes when formulating fiscal policies, particularly in developing economies. Future research could explore the specific mechanisms driving the observed relationships, potentially examining the impact of tax policy changes on investment and innovation.

The study focuses solely on Turkiye, limiting the generalizability of the findings to other countries with potentially different economic structures and policy contexts. The use of quarterly data may limit the precision in capturing certain short-term fluctuations and dynamic aspects of the relationship. While the econometric techniques employed account for autocorrelation and structural changes, the impact of omitted variables and potential endogeneity biases remains a consideration.