How India Customs contributes to data-informed decision-making
15th June 2023By Brij Bhushan Gupta, Additional Director General, India
“In God we trust, all others must bring data”. Widely attributed to statistician and management theorist W. Edwards Deming, this quote epitomizes the importance of data in decision-making. With technologies for extracting, collecting, storing and processing data having progressed rapidly over the last couple of years, this statement is truer now than ever before.
This does not mean that knowledge and information acquired through human interaction, scenario building and experience are not important when analysing a situation. Rather, data enhances our cognitive abilities and at times provides a counterview for challenging or validating an initial judgement, as the case may be. The end result, all other things being equal, is a better outcome.
India Customs has taken various steps to leverage data, with some noticeable success. Being a large administration acting over a large territory, India Customs has adopted a two-level data analytics model. The first level consists of a team of highly-skilled officers who run the Advanced Analytics in Indirect Taxes (ADVAIT) programme at the Customs central office. The second level consists of teams located at regional offices, who bring specific knowledge and know-how on some issues. Below are some examples of how data is collected and analysed to inform decisions.
Conducting Time Release Studies to drive and assess changes
The WCO Time Release Study (TRS) is an internationally-accepted strategic tool for measuring the actual time taken for the release and/or clearance of goods – from the time of arrival until the physical release of the cargo – as well as the effectiveness and efficiency of border procedures relating to imports, exports and transit movements of goods. Moreover, TRS is considered to be a useful instrument when undertaking a comprehensive assessment of trade facilitation needs and priorities, as well as for the periodic monitoring and measurement of the outcomes of implementing specific measures and associated policies and programmes. A Time Release Study should be carried out regularly, and its results published in the public domain.
It is worth pointing out that TRS data and results are more reliable for assessing and making judgments on the performance of cargo clearance processes than other international performance indicators or indexes developed by institutions such as the Organization for Economic Co-operation and Development or the World Bank, where the data is gathered by means of a questionnaire. In a paper titled “Measuring trade facilitation: evidence from India”, Customs officers Vijay Singh Chauhan and Sruti Vijayakumar compared the various indicators available in the public domain; they concluded that data on cargo release time was more robust and meaningful than survey-based data for measuring the performance of border management agencies and practices.
India Customs uses statistics generated by TRS to improve processes, inform policy changes and drive discussions with trade stakeholders. For example, one TRS showed that traders were taking rather a long time to pay for the discharge of goods, thereby decreasing the efficiency of the ports. It was therefore decided to add interest after allowing a given amount of time for such payments, to nudge importers and brokers to accelerate the process. Similarly, changes made to promote the advance filing of bills of entry and AEO certification were backed by TRS data. The insights generated by the aforementioned TRS also led to the removal of certain processes which, over time, had become redundant or could be automated thanks to technology. The impact of all those policy interventions was in turn assessed in subsequent TRS.
The WCO strongly advocates that Customs administrations conduct a TRS in close collaboration with other relevant government agencies and private sector stakeholders. India Customs is gradually expanding its TRS by collecting data generated not only by Customs systems but also by logistics partners, port authorities and partner government agencies.
Mapping economic activity for targeted State intervention
Analysing import and export data enables governments to monitor levels of economic activity and identify the need for economic intervention or the development of infrastructures. In India, the mapping of export volumes according to districts has enabled policy makers to identify areas where measures to boost manufacturing and promote exports are needed. The mapping has also provided valuable inputs for optimizing transportation and warehousing costs.
Analysing item descriptions to identify Valuation anomalies
Products sharing the same HS code can have various Customs values, depending on their characteristics and specifications. One way to group them and uncover valuation anomalies is to analyse the description of each item provided in the Goods declaration. India Customs is using text mining algorithms (particularly Text Clustering) to do this. The first step was to create clusters according to historical data. A software then compares the text of the description to the text in each cluster and determines which one offers the closest match. The value of goods in the same cluster is then compared.
Providing data on the usage of preference treatments established by Free Trade Agreements
Trade negotiators need Customs data both when considering the signing of a Free Trade Agreement (FTA) and when evaluating the effectiveness of FTA commitments and the benefits they have brought for business. India Customs therefore analyses trends in imports for which preference treatment has been claimed, and looks at the factors which hinder the effective utilization of FTAs.
Classifying and analysing queries, to identify the areas where clarifications are more frequently sought, and the types of traders who are having difficulties in complying with the regulations
A Customs query is a request for information made by a Customs officer to a trade operator, during an import or export operation. Such queries are communicated by entering unstructured text in the Customs automated system, and the trader is expected to respond via the same system. It is sometimes necessary for a query to be tossed back and forth in order to solve an issue. A tool called “Query Classifier” has been developed to generate data on the types of queries issued by period, port of entry, frequency of interactions, type of clarification sought, type of trader involved, etc.
Actionable insights can be derived from this data. For example, having identified areas where clarifications are sought more frequently, Customs can offer targeted training to its staff and improve consistency in decision-making. It can also improve the compliance of trade stakeholders by providing training on processes and procedures which do not appear to be well understood. Also, an issue which crops up repeatedly may reveal the need for a policy intervention.
Customs has developed models for forecasting revenue from Customs and Goods and Services Tax (GST). There are three types of forecasting: monthly, quarterly and annual. The annual revenue forecast is updated regularly.
Predicting fraud by analysing applications for GST registration
GST is a value-added tax levied on goods and services sold for domestic consumption. It is paid by consumers, but is remitted to the government by the businesses that sell the goods and services. These businesses must register online. As those likely to commit fraud on GST are also likely to commit fraud when importing or exporting goods, an application has been developed to identify entities whose behaviour has been deemed suspicious when they apply for registration and on GST enforcement officer reports. The tool uses machine learning technology and is similar to those used to assess risk in connection with credit card applications.
Internal administrative processes also generate large amounts of data which can be mined. With a large workforce of more than 45 000 officers, India Customs has attempted to use data to guide human resources planning, manage vacancies, organize job rotations based on profiles and job requirements, grant promotions, and plan training and capacity building activities.
Building capacities across the administration, starting with its leaders
Often, Customs administrations have a handful of officers with a background in data science who form “islands of excellence” in data analysis. But this is not enough. Knowledge about data must be better shared across the organization, and a multi-layered approach needs to be adopted for this purpose. The starting point is to gain an understanding of officers’ behaviour towards data and analytics and provide tailored training that takes their abilities into account. India Customs has also organized bootcamps to train data masters and create a pool of trainers able to transmit their knowledge and skills to a larger set of people. It also organizes annual hackathons; these are events where teams (including developers, but sometimes designers and project managers too) collaborate on a computer project – usually the development of application software. Identified solutions are scaled up across the organization.
Most importantly, the drive towards data-driven decision-making in India Customs is being championed and promoted by its leadership, who demand data-based evidence when taking any managerial or policy decisions and lend their support to all sound, data-related initiatives.
 A district is an administrative division of a state or territory in India