Non-tariff measures and the regulatory shape of world trade

Once tariffs fell, the binding constraints on cross-border commerce shifted to technical rules, sanitary standards, licensing, price bands, and export restrictions. UNCTAD’s TRAINS database counts these measures at the HS6 line for 125 reporters. Which economies regulate most, what types dominate, and which product chapters bear the densest regulatory stack?

Who regulates most

Non-tariff measures (NTMs) are policy instruments other than ordinary tariffs that can affect traded quantities, prices, or both. The modern taxonomy comes from UNCTAD’s Multi-Agency Support Team (MAST) classification: sixteen chapters A–P, covering sanitary and phytosanitary (SPS, chapter A), technical barriers to trade (TBT, B), pre-shipment inspection (C), contingent trade protection (D), non-automatic licensing and quotas (E), price control (F), finance (G), subsidies and government procurement (H–M), intellectual property (N), rules of origin (O), and export-related measures (P). Counts below aggregate measures per HS6 product line across the latest year available per reporter.

What types of measures dominate

Not every NTM bites the same way. SPS and TBT are information-revealing: they force producers to prove compliance with food-safety or product-standard rules, and their trade cost typically runs five to ten percent ad valorem for affected sectors. Quantity controls (chapter E: licensing, quotas) and price measures (F) are more restrictive, and export-related measures (P: export taxes, bans, quotas) shift supply abroad. Murina and Nicita (2017, Journal of International Trade & Economic Development 26(7): 832–853, “Trading with conditions: the effect of sanitary and phytosanitary measures on the agricultural exports from low-income countries”) estimate ad-valorem equivalents (AVEs) that put SPS/TBT below licensing and price control on average, but with wide product variation.

Where on the tariff book NTMs concentrate

Kee, Nicita, and Olarreaga (2009, Economic Journal 119(534): 172–199, “Estimating Trade Restrictiveness Indices”) aggregate applied protection into country-level OTRIs; the underlying sectoral import-demand elasticities and AVE estimates come from Kee, Nicita, and Olarreaga (2008, Review of Economics and Statistics 90(4): 666–682, “Import Demand Elasticities and Trade Distortions”). Together they imply that agri-food and live-animal chapters carry the largest ad-valorem equivalents: SPS measures reach twenty-plus percent AVE in meat, dairy, and seafood chapters, dwarfing applied tariffs. The picture below uses simple average measure count per HS6 line as a pre-AVE proxy for regulatory density.

How heavily do regulators regulate

The original framing for this figure asked for NTM coverage by partner income group. UNCTAD TRAINS at HS6 is a reporter-only database: the measure applies to all imports of the HS6 line unless the legal text names a country of origin, and the bulk parquet does not carry partner codes. So this panel pivots to an adjacent question: how concentrated is the regulatory burden across reporters? The five-bucket histogram ranks the 125 reporters by measures-per-HS6. This relates to the Mattoo, Rocha, and Ruta (2020) Handbook on Deep Trade Agreements (World Bank) framing that deep regulatory integration co-exists with sharp asymmetry in regulatory depth.

Where regulation piles up: the densest reporter-sector cells

The spec asked for top bilateral NTM-dense pairs. Again, bilateral pairs are not encoded in TRAINS at HS6. The substitute is the analog on the reporter-sector plane: which reporter × HS chapter cells carry the highest total NTM count? This recovers the what-gets-regulated-wheresub-question. Count-based indices (share of lines subject to core NTMs, as catalogued by UNCTAD 2019 International Classification of Non-Tariff Measures, revised edition) are the raw material for AVE conversion and stringency comparisons.

Who regulates what: NTM stack by HS section

Chapter-level detail (Figure 3) is where most of the SPS tail lives. Rolling up to the WCO Harmonized System Section (I–XXI, HS 2022 nomenclature) collapses the 97 chapters to 21 economically meaningful aggregates and shows where the regulatory stack actually sits across the whole reporter panel. The join is HS6 → products.parquetsection id, then SUM(ntm_count) across every reporter-HS6 row in TRAINS.

NTM × tariff interaction by country income group

Tariffs and NTMs are the two canonical instruments of trade policy; the substitution and complementarity between them depends on development level. Kee, Nicita & Olarreaga (2009, Economic Journal 119(534): 172–199) document that richer economies often carry thin tariff books but dense NTM stacks (SPS, TBT, licensing), while poorer economies lean more on applied tariffs. The scatter pairs, for every reporter with both a UNCTAD TRAINS NTM snapshot and a WITS tariff summary, the simple-average MFN tariff against the average NTM count per HS6 product line, colour-coded by World Bank income group using NY.GDP.PCAP.CD and the 2024 GNI bands (low < 1,136; lower-middle < 4,466; upper-middle < 13,846; high ≥ 13,846 USD).

SPS + TBT share: food safety and technical regulation versus the rest

The NTM stack decomposes into two broad archetypes. SPS (MAST chapter A, sanitary and phytosanitary) and TBT (MAST chapter B, technical barriers to trade) are information-revealing: compliance documents the product against a standard. Licensing (E), price control (F), and export-related measures (P) are quantity- or price-restricting. Disdier, Fontagné & Mimouni (2008, American Journal of Agricultural Economics 90(2): 336–350) show that SPS/TBT coverage dominates the OECD regulatory stack, while developing-country regimes rely more on licensing and price controls. The ratio below ranks the top 25 reporters (total ntm_count ≥ 1,000) by the SPS+TBT share of their total measure-product lines.

The regulatory ratchet: NTM coverage by reporter snapshot year

TRAINS releases reflect when a national authority last submitted a complete legal-text coding mission to UNCTAD; the reporter pool therefore grows year by year as more economies enter the panel. This is not a continuous time-series of regulation in any single country (legal-text additions accumulate; UNCTAD records the snapshot as of the last coding mission), but it is the cleanest visible measure of how rapidly the observable universe of NTMs is expanding. Ederington & Ruta (2016, in Handbook of Commercial Policy Vol. 1B, ch. 16) frame this as the “regulatory ratchet”: once a measure is added it almost never gets withdrawn, so cumulative coverage drifts monotonically up. The bars show, for each year of the coverage panel, how many reporters last filed in that year and the median coverage ratio (share of HS6 lines carrying at least one NTM) across them.

Method note: from measure counts to ad-valorem equivalents

The counts above are a pre-tariff-equivalent proxy, not a welfare cost. To convert an NTM indicator into an ad-valorem equivalent (AVE) for a given importer-HS6 line, the Kee, Nicita & Olarreaga (2009, Economic Journal 119(534): 172–199) two-step works as follows. Step 1: estimate the import-demand elasticity ε_ij per product-country using a Kee, Nicita & Olarreaga (2008, Review of Economics and Statistics 90(4): 666–682) first-stage specification. Step 2: regress log imports on NTM-dummy, applied tariff, and fixed effects; divide the NTM coefficient by −ε to recover the price-equivalent. Cadot, Asprilla, Gourdon, Knebel & Peters (2015, Product-Policy Complaints and Non-Tariff Measures, World Bank / UNCTAD) note that AVEs vary orders of magnitude across products and are strongly positively skewed; median AVEs cluster at 5–10% for SPS/TBT, with long tails in agri-food and pharmaceuticals reaching 30%+.

Policy read

Three takeaways shape the policy debate. First, the regulatory shape of world trade is concentrated: a handful of reporters set the rules that the rest of the world complies with, and the densest chapters are agri-food (Figure 3). Second, SPS/TBT measures dominate the stack for most heavy regulators (Figure 2), so trade-facilitation wins come from mutual recognition, harmonisation, and third-party conformity assessment, not from negotiating the measures away. Third, because TRAINS at HS6 is reporter-indexed, a true cost-of-compliance map for exporters (especially LDCs) requires pairing these counts with AVE estimates in the Kee-Nicita-Olarreaga tradition plus firm-level compliance surveys (Cadot & Malouche eds. 2012, Non-Tariff Measures: A Fresh Look at Trade Policy’s New Frontier, World Bank).

Coding notes. Adapted figures: Figure 4 substitutes an intensity histogram for the specified partner-income split; Figure 5 substitutes a reporter-sector density ranking for the specified bilateral pair ranking. Both substitutions are forced by the TRAINS NTM-at-HS6 bulk schema, which is reporter-indexed and does not encode partner country at the HS6 level.