Article 1 of the Trade Facilitation Agreement (TFA) requires that each World Trade Organization member promptly publish tariff and regulatory trade information in a “non-discriminatory and easily accessible manner”.

While many member countries do, in fact, fulfill the letter of Article 1, few have realized its broader aim, which is to simplify the procedures associated with trade compliance. Most governments publish tariff schedule and regulatory trade information in a non-discriminatory form, but rarely make it “easily accessible”.

How easy is it for traders to access the trade compliance information they need?  Not very.

The starting point for finding relevant tariff and regulatory trade information is Harmonized System (HS) commodity classification. The problem is, most traders struggle with determining their product’s HS code. A recent KPMG-Thomson Reuters Global Trade Management Survey reported that 91% of respondents indicated that commodity classification is a challenge.

Accordingly, the degree to which governments offer HS classification assistance determines whether or not relevant trade information is easily accessible.

To gauge how easy it is to find relevant tariff and regulatory trade data on a typical customs website, we submitted 100 randomly-selected commercial goods descriptions (obtained from a national customs authority) to the ASYCUDAWorld Online Tariff search engine ( Developed by the United Nations Centre for Trade and Development and deployed in 90 countries, ASYCUDA is the world’s most widely used customs management system.

For the sake of clarity, only records with sufficient detail for HS-6 level classification were submitted. Also, it was necessary to limit the length of the commercial goods descriptions submitted to fewer than 30 characters because of constraints in the ASYCUDA search engine.

To assess the usefulness of the ASYCUDA tool, we measured its “precision”, which was determined by calculating the ratio between all results returned (“potential matches”) and correct results returned (“correct matches”). Ideally, each search should have resulted in a single, correct match.

Since ASYCUDA offers “contains all” and “contains any” searches, the data was run once using each parameter. Generally, we found that “Contains All” searches rarely deliver any results at all. Whereas “Contains Any” searches inundate users with endless amounts of irrelevant and erroneous information.

Here are some highlights from our ASYCUDA experiment:

  • Only 14 (out of 100) ASYCUDA “Contains All” searches returned any result. These included 22 “potential matches”, out of which 11 (50%) included the correct HS code.
  • Only 1 (out of 100) of the “Contains All” searches resulted in the optimal 1:1 potential-to-correct match.
  • 88 (out of 100) ASYCUDA “Contains Any” searches returned a result. However, these positive results included 5,448 “potential matches”, of which 65 (1.19%) included the correct HS code.
  • Only 3 (out of 100) of the “Contains Any” searches resulted in the optimal 1:1 potential-to-correct match.

Two additional points about ASYCUDAWorld’s Online Tariff Search results are worth noting.

First, the ASYCUDA search engine returns a maximum of 200 potential matches. Thus, the true ratio of potential-to-correct is lower than stated above. Ten (of the 100) declarations reached ASYCUDA’s 200-potential match limit.

Second, ASYCUDA’s results include potential matches at the Section, Chapter (HS2), Heading (HS4), and Subheading (HS6) levels. For 21 of ASYCUDA’s 65 correct results (32.31%), the user is required to navigate through the nomenclature to reach the final HS-6 code. Had the results been weighted to account for the HS-level returned, ASYCUDA’s precision score would have been lower than stated above.

Although the analysis was performed in ASYCUDA, we believe our results exemplify the lack of utility of most government publication efforts. Without useful classification assistance tools, published tariff and regulatory information is not really accessible at all.

You can request a copy of the full results of our study from