Macroeconomic Consequences of Tariffs

The paper investigates the macroeconomic effects of changes in tariffs, motivated by renewed use of trade policy seemingly for macroeconomic objectives despite longstanding economic consensus favoring free trade. While conventional wisdom argues tariffs are welfare-reducing, much prior evidence is theoretical, micro-level, or dated. Using recent cross-country aggregate data, the study asks how tariff changes affect key macro and distributional outcomes—output, productivity, unemployment, inequality, the real exchange rate, and the trade balance—while also exploring heterogeneity across business cycle phases, development levels, and the direction of tariff changes. The goal is to provide transparent, empirically grounded benchmark evidence on tariffs’ macroeconomic consequences.

Theoretical work indicates no definitive presumption about tariffs’ effects on output or trade balance; impacts depend on factors such as expected duration, wage and exchange rate behavior, elasticities, and institutions (Ostry and Rose, 1992; Dornbusch, 1974; Edwards, 1989). Empirical research has examined trade liberalization and openness in relation to currency movements and external balances (Santos-Paulino and Thirlwall, 2004; UNCTAD, 1999; Ju, Wu, and Zeng, 2010; Li, 2004) and to productivity/output (Feyrer, 2009; Alcalá and Ciccone, 2004). Distributional implications have been debated, with studies parsing the roles of trade versus technology (Helpman, 2016) and firm-level evidence on wage inequality (Artuç and McLaren, 2015; Klein, Moser, and Urban, 2010). More recent work links protectionism to macro fluctuations using high-frequency temporary trade barrier data (Barattieri, Cacciatore, and Ghironi, 2018). Compared to prior studies focusing on liberalization or openness, this paper isolates tariffs and provides broader coverage across countries and outcomes, arguing that model-based gains/losses are often too small and that a reduced-form approach with wide data coverage is valuable for policy.

The study employs a reduced-form panel approach with local projections (Jordà, 2005) to estimate impulse response functions (IRFs) of macro variables to tariff changes. Country-level analysis: For an unbalanced panel of 151 countries over 1964–2014 (annual data), the baseline specification includes country and time fixed effects; the regressor is the annual change in the (import-weighted) tariff rate aggregated from product-level tariffs (mainly WITS and WDI; also WTO, GATT, BTN). Controls include two lags of changes in the dependent variable, tariff changes, log output, log real exchange rate, and the trade balance-to-GDP ratio. Outcomes analyzed: GDP (log), labor productivity (GDP per worker), unemployment rate, inequality (Gini from SWIID), real effective exchange rate, and trade balance/GDP (WEO, WDI, INS). IRFs are presented up to five years after the shock; Driscoll–Kraay standard errors are used. Heterogeneity and asymmetries: The authors extend the baseline to allow for asymmetric effects between tariff increases and decreases, split effects by income level (advanced economies per IMF vs others), and allow business-cycle state dependence via a smooth-transition function F(Z) (Auerbach and Gorodnichenko, 2012), using GDP growth or unemployment changes as Z. Robustness strategies: (1) Alternative tariff transformations (percentage change; lagged tariffs). (2) Estimation approaches addressing endogeneity: a VAR with Cholesky ordering (Δlog output or productivity, Δtariff, Δlog REER, Δtrade balance/GDP); specifications controlling for contemporaneous changes in REER and trade balance; specifications including expected future growth (IMF WEO forecasts). (3) Instrumental variables: instrument for domestic tariff changes using the weighted average of tariff changes in the top five trading partners (weights are bilateral trade shares from DOTS/WDI). First-stage strength is documented (Kleibergen–Paap F-statistics above Stock–Yogo critical values); exclusion is supported by tests showing partner tariffs affect domestic outcomes only via domestic tariffs and by controlling for contemporaneous REER changes. (4) Sample perturbations: dropping series with gaps or short runs (<20 consecutive years), high-inflation episodes (>100%), small countries (<1 million population), outliers (top/bottom 1st percentile regression residuals), early years (pre-1980), very high tariffs (>66%, 99th percentile), and regional exclusions (Americas; Asia and Sub-Saharan Africa). Industry-level analysis: A sectoral panel (16 manufacturing sectors in 39 countries, 1991–2014) combines UNIDO sectoral outcomes (output, value added, employment, productivity), WITS tariffs, and OECD input–output tables to construct sectoral output tariffs (2-digit) and input tariffs (input-weighted averages across supplying sectors). The specification includes country–industry fixed effects, country–year fixed effects (controlling for macro shocks/reforms common to all sectors), and industry–time fixed effects (common shocks across countries to a given industry), with lags of sectoral outcomes and tariffs. This allows decomposing impacts via output vs input tariff channels.

Aggregate results: A one-standard-deviation tariff increase (≈3.6 percentage points) reduces real GDP by about 0.4% after five years and lowers labor productivity by about 0.9% after five years; unemployment rises modestly (statistically marginal), and inequality (Gini) increases, becoming statistically significant from year 2 onward. The real exchange rate appreciates (short-run significance), and the trade balance response is small and statistically insignificant. Consumption declines by roughly 0.4% after four years. Asymmetries and heterogeneity: Tariff increases have larger negative effects than the (smaller) positive effects from tariff reductions; output and productivity losses from tariff increases exceed baseline responses. Advanced economies experience larger adverse effects: output falls by about 1% four years after a tariff increase (vs ≈0.4% baseline), with stronger productivity declines than in emerging/developing economies. Business-cycle dependence: Negative impacts are amplified in expansions; a standard-deviation tariff increase during good times lowers medium-term output by about 1 percentage point more than baseline and similarly magnifies productivity losses. During recessions, effects on output/productivity may be mildly positive but are not statistically significant. Inflation rises about two years after tariff increases, with stronger effects in expansions, suggesting that monetary tightening in response can magnify contractionary effects. Robustness: Results are robust to alternative tariff measures, VAR identification, inclusion of contemporaneous REER and trade balance shocks, inclusion of growth forecasts, IV estimation (which yields even larger medium-term output declines), and numerous sample exclusions. Industry-level results: Input tariff increases depress sectoral performance significantly—sectoral output falls by about 6.4% and productivity by about 3.9% five years after the shock. Output tariff increases raise sectoral output by about 3.1% in five years, with a positive but statistically insignificant productivity effect. These patterns indicate that increases in tariffs on imported inputs are the primary channel driving aggregate losses.

The findings provide macro-level empirical evidence that tariff hikes reduce output and productivity, increase unemployment and inequality, appreciate real exchange rates, and leave trade balances largely unchanged—consistent with the view that protectionism introduces distortions and deadweight losses without improving external balances absent changes in saving–investment fundamentals. The amplification during expansions and in advanced economies underscores that protectionism is particularly damaging when economies are operating above potential or when supply chains and input use are sophisticated and globally integrated. Asymmetric responses—tariff hikes hurting more than liberalizations help—are consistent with intertemporal demand dynamics and adjustment frictions. The inflationary response and potential monetary tightening suggest an additional policy interaction that can exacerbate real losses. Sectoral evidence pinpoints input tariffs as especially harmful, aligning with theory that raising costs of intermediates propagates across industries and undermines productivity, thereby explaining the aggregate results.

Using local projections on a large cross-country panel and complementary sectoral data, the paper establishes that tariff increases have economically and statistically significant adverse macroeconomic and distributional effects: they lower output and productivity, raise unemployment and inequality, appreciate the real exchange rate, and have negligible effects on the trade balance. Losses are larger in advanced economies, during expansions, and for tariff increases relative to decreases. Sectoral analysis shows input tariffs drive much of the negative impact. The results support the case for free trade on macroeconomic grounds and provide contemporary empirical benchmarks relevant to current policy debates. Future research could further address endogeneity with alternative instruments, explore longer horizons and dynamic general equilibrium effects, incorporate spillovers to trading partners, and analyze interactions with monetary and structural policies in more detail.

The dataset, while extensive, is constrained by limited information on concomitant structural policies and non-tariff barriers, raising concerns about omitted variables despite extensive controls and robustness checks. Endogeneity cannot be entirely ruled out, though addressed via VAR specifications, controls for contemporaneous shocks and expectations, and an IV strategy using trade-partner tariff changes. The postwar period sample predominantly features liberalization, which may limit inference about prolonged protectionist regimes. The analysis focuses on domestic effects, abstracting from international spillovers. Results are truncated to a five-year horizon, potentially understating longer-run impacts. Measurement issues in aggregate tariff rates, inequality indices, and cross-country comparability may also affect estimates.