Revisiting the tax-spend nexus in Türkiye: evidence from time-varying Granger causality

Studies of the revenue-expenditure nexus support each of its four hypotheses, as does the Turkish evidence. Most assume a time-invariant causal link, which is implausible for an economy exposed to frequent macroeconomic and policy shocks. This study examines changes in the direction of causality between real tax revenues and non-interest expenditures, applying the lag-augmented recursive evolving Granger causality test to monthly Turkish general budget data over 2006M1 to 2026M4. The results reveal repeated shifts: institutional separation until 2015, interrupted by spend-and-tax in 2011 and 2012, tax-and-spend from 2015, fiscal synchronization in 2017 and 2018, spend-and-tax during the 2019 slowdown, separation during the pandemic, tax-and-spend from mid-2021 and fiscal synchronization from early 2022 onward. These episodes coincide with phases of fiscal policy, from the electoral cycles and tax amnesties of the mid-2010s to the post-earthquake consolidation. Causality from revenue to expenditure in the mid-2010s and the final synchronization survive conditioning on inflation, industrial production and the exchange rate, whereas controlling for activity largely removes the 2017–2018 synchronization. The question of which side of the budget leads therefore has no time-invariant answer in Türkiye, and the answerable question is when each side leads.