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

Modern governments use tax policy to raise revenue for public activities and to promote social cohesion by addressing income inequality. The literature debates how taxes affect economic growth in the short and long run. While neoclassical growth models (e.g., Solow 1956; Harberger 1964a,b) argue taxation does not affect long-run growth, endogenous growth models (e.g., Jones and Manuelli 1990; King and Rebelo 1990; Capolupo 2000) show tax policy can influence long-run growth positively or negatively, depending on how revenues are used (e.g., human capital, public goods). Some studies find taxation distorts long-run growth and recommend low or zero taxes, while others note potential benefits when taxes finance productive expenditures. Research also highlights channels through which taxes affect firm innovation and income inequality, with mixed evidence on whether tax cuts or increases stimulate growth. Importantly, different tax types may have different growth effects: several studies find direct taxes more harmful to growth than indirect taxes, but there is no consensus. Against this background, the study investigates whether tax is a blessing or a curse in Turkiye by distinguishing direct and indirect taxes and assessing their respective impacts on economic growth. The study proceeds with a literature review, data, model, and methodology, followed by empirical results and policy recommendations.

The review documents mixed empirical findings on the growth effects of taxation and underscores the importance of distinguishing tax types. Definitions differentiate direct taxes (levied on income/wealth and borne by the intended taxpayer) from indirect taxes (levied on transactions and potentially shifted to consumers). Indirect taxes can be regressive and may worsen income inequality. Cross-country evidence is mixed: some studies suggest reducing the share of direct taxes in GDP supports growth (e.g., Stoilova & Todorov 2021), and that corporate and indirect taxes may foster growth while personal income taxes may hinder it (Pandey 2019). Other works find direct taxes negatively affect growth and disposable income, whereas indirect taxes may have positive or smaller negative effects; some find negative effects of indirect taxes on growth and inequality. Studies combining tax revenues, public expenditures, and growth report varied causality patterns across countries (e.g., G7, Indonesia). For Turkiye, evidence is mixed: some find positive effects of both tax types on growth, others find positive effects for indirect but negative for direct taxes; the tax burden’s effect is also debated. Inspired by the resource curse literature, the authors propose a “tax curse” hypothesis, examining whether certain types of taxes impede growth. The study posits three hypotheses: H1: Indirect taxes (IT) positively affect economic growth (no tax curse from indirect taxes). H2: Direct taxes (DT) positively affect economic growth (no tax curse from direct taxes). H3: Export volume growth (EXP) positively affects economic growth.

Data and model: Quarterly data for Turkiye from 1998/Q1 to 2023/Q2 are used. The model relates economic growth (GDP growth rate) to growth in indirect taxes (IT), direct taxes (DT), and export volume (EXP): GDP_t = α + β1 IT_t + β2 DT_t + β3 EXP_t + ε_t. GDP is the seasonally adjusted GDP growth rate (TUIK). IT and DT are tax revenue growth rates (Turkiye Public Finance Statistics; World Bank WDI and TUIK). EXP is export volume growth. All variables are measured as quarter-over-quarter growth rates. Descriptives and correlation: Variables are non-normally distributed (Jarque-Bera rejects normality at 1% for all). Mean values: GDP 0.00305; IT 0.00666; DT 0.00735; EXP 0.00348. Correlations are positive across pairs, with notably high correlations between GDP and IT (0.587) and GDP and EXP (0.559), all significant at 1%. Unit root testing and transformations: Stationarity is assessed using the flexible Fourier-form Dickey-Fuller unit root test (Enders & Lee, 2012) that allows smooth structural breaks. DT and GDP are stationary in levels; EXP becomes stationary at first difference; IT becomes stationary at second difference. To harmonize integration orders, the analysis proceeds using first differences where needed (IT treated accordingly) so that variables are stationary. Estimation approach: To address autocorrelation in time series residuals, the Prais-Winsten (1954) adjusted OLS regression with AR(1) errors is employed, which performs well relative to standard OLS and ARMA approaches and does not require normally distributed residuals. The estimated AR(1) coefficient (rho) accounts for serial correlation. Causality analysis: To examine directional relationships under potential structural shifts, the cumulative Fourier-frequency Toda & Yamamoto causality test (Nazlioglu et al., 2019) is applied. Wald tests evaluate Granger non-causality with bootstrap p-values (1,000 runs), optimal lags via AIC, and Fourier frequencies capturing smooth shifts. Long-run comovement: Wavelet Transform Coherence (WTC) using the Morlet wavelet assesses time–frequency coherence between GDP and each regressor over the sample. Statistical significance is evaluated by Monte Carlo simulations; coherence near one indicates strong association across scales. Model diagnostics and presentation: The Prais-Winsten regression reports coefficients, standard errors, t-statistics, p-values, R-squared, and Durbin-Watson statistics (original and transformed). The final equations are: GDP_t = −0.00808·DT_t + 0.02564·IT_t + 0.18055·EXP_t + 0.00250 + ε_t; ε_t = 0.4614·ε_{t−1} + η_t.

• Unit roots: DT and GDP are stationary in levels; EXP is I(1); IT is I(2). Variables are analyzed after appropriate differencing to ensure stationarity for inference.
• Prais-Winsten AR(1) regression (dependent variable: GDP):
  • IT: β = 0.02564, SE = 0.01086, t = 2.36, p = 0.020 (positive and statistically significant at 5%).
  • DT: β = −0.00808, SE = 0.00982, t = −0.82, p = 0.413 (negative, statistically insignificant).
  • EXP: β = 0.18055, SE = 0.04638, t = 3.89, p = 0.000 (positive and highly significant).
  • Constant: 0.00250 (p = 0.000). R-squared = 0.2083; Adj. R-squared = 0.184; rho = 0.4614; Durbin-Watson (transformed) = 2.2567 (indicates effective correction of autocorrelation).
• Causality (Cumulative Fourier-frequency Toda & Yamamoto):
  • IT ↔ GDP: Bidirectional causality. IT → GDP: F = 16.099, p = 0.048 (lags 7, freq 3). GDP → IT: F = 21.964, p = 0.004.
  • DT → GDP: No evidence (F = 6.392, p = 0.382). GDP → DT: Unidirectional causality (F = 15.919, p = 0.028; lags 6, freq 3).
  • EXP → GDP: Unidirectional causality (F = 5.278, p = 0.040; lags 1, freq 3). No causality from GDP to EXP (F = 0.97, p = 0.303).
• Wavelet Transform Coherence (WTC):
  • IT–GDP: Strong positive coherence throughout, especially at high frequencies; toward the end of the period, coherence extends to medium and low frequencies.
  • DT–GDP: Limited and complex; largely negative relationship at medium frequencies over much of the sample; in the last three years, relationship turns positive at high frequencies.
  • EXP–GDP: Positive and significant coherence across much of the period, mainly at medium frequencies.
• Hypothesis evaluation: H1 (indirect taxes positively affect growth; no tax curse for indirect taxes) accepted. H2 (direct taxes positively affect growth; no tax curse for direct taxes) rejected—results indicate a tax curse for direct taxes. H3 (exports positively affect growth) accepted.

The study set out to determine whether taxation is a blessing or a curse for economic growth in Turkiye when distinguishing between direct and indirect taxes. Empirical evidence shows a robust, positive effect of indirect taxes on growth and strong bidirectional causality with GDP, indicating that increases in indirect tax revenues are closely linked with economic activity. Conversely, direct taxes exhibit a negative (though statistically insignificant) coefficient and no causal influence on GDP, while GDP influences direct taxes—suggesting direct taxes respond to, rather than drive, growth. Time–frequency analysis corroborates these findings: indirect taxes co-move positively with growth across scales, whereas direct taxes are associated with negative coherence over much of the sample, turning positive only near the end. These outcomes support a tax curse for direct taxes in the Turkish context during much of the period, particularly for an economy in a high-growth phase, while indicating no tax curse for indirect taxes. The results align with strands of the literature that find direct taxes more detrimental to growth and suggest that tax structure matters for growth outcomes. Policy-wise, reducing the burden of income-based direct taxes on firms and allocating tax revenues toward productive investments (industrialization and exports) could enhance growth performance.

Taxes are essential for financing government functions but can influence economic growth in different ways depending on their structure and use. For Turkiye (1998/Q1–2023/Q2), indirect taxes are associated with higher growth and exhibit strong two-way linkages with GDP, indicating no tax curse for indirect taxes. Direct taxes, in contrast, show a negative association with growth and no causal impact on GDP—evidence consistent with a tax curse for direct taxes during much of the period, though this adverse effect diminishes toward the end of the sample. Policy implications include: moderating the direct tax burden (especially on productive enterprises), closing loopholes and preventing evasion, reviewing exemptions (e.g., high-income sectors) to improve efficiency and equity, and channeling tax revenues toward productivity-enhancing uses (industrialization, export capacity). These steps can mitigate the tax curse associated with direct taxation and leverage taxation as a blessing for sustainable growth. Future research might extend the analysis with alternative identification strategies, richer tax disaggregation, and cross-country comparisons to generalize findings.