Pushing Boundaries: jumping hurdles in the blue flow

For a tighter grip on goods from unknown and lesser-known companies, there’s one thing above all: customs supervision has to become even more data driven.

The Pushing Boundaries vision focuses a lot of attention on goods from unknown and lesser-known market parties. After all, we’re talking here about the biggest volumes, which also carry the highest fiscal and security risks. Dutch Customs is gradually getting a better grasp on this so-called blue flow, in all kind of ways. To get a really tight grip, it’s important that the agency becomes even more data-driven in its supervision. The envisioned transition demands a great deal of time and effort, according to enforcement expert René Doolhoff.

People sometimes think that blue flow goods only come from non-compliant companies. But that’s not the case. “For many of these actors, we haven’t – or haven’t yet – found out whether they are meeting their customs obligations. At the moment, we just don’t have a broad enough view of their way of operating,” Doolhoff says. “It happens occasionally that an expeditor with AEO status files a declaration on behalf of a new client he hardly knows, and who is also unknown to us. That particular shipment of goods is automatically put into the blue flow – and not into the green flow of market parties with proven reliability.”

Making progress
In recent years there’s been progress in the monitoring of the blue flow, according to Doolhoff. “One positive step is the placement of permanent scanners in all the container terminals on both of the Maasvlakte areas, and the realisation of the State Inspection Terminal in Rotterdam and the Joint Inspection Center at Schiphol Airport. Scanning goods is the frontline detection procedure, but it’s also very effective and efficient because at these locations it’s completely embedded in the logistical process. There’s also camera surveillance, which has a tremendous value for our organisation as it’s shaped up in the port of Rotterdam and at the national airport. Thanks to excellent public-private partnership in this area, we can keep a sharp eye on large areas of these logistical hubs. In addition, we have an extra device in hand to strengthen our monitoring on the spot, and we receive useful input for our risk analysis at the Customs National Tactical Centre. Moreover, all those cameras hanging up everywhere have a preventive effect.”

And there’s more, Doolhoff says: “We implemented a work distribution tool, that helps us to schedule the control of goods by automation, which is more efficient. Furthermore, we have equipped our staff with handheld scanners for physical controls, and there are also things like an app for detecting prohibited Chinese medicines. Applications for sorting out flora and fauna and other types of goods are currently being developed. Briefly put, customs officers are constantly able to do their work faster and more efficiently.”

Insight into declaration behaviour
Another positive development is that Dutch Customs has more methods in-house for monitoring and influencing declaration behaviour. Doolhoff: “In recent years we ran the project Quality of Declarations, which has since become part of the standing operations of our organisation. We analyse which fields in declarations import companies have difficulty with, and try to make improvements. We can also apply our business rules to gain insight into which companies are submitting a lot of declarations that are initially rejected by our AGS system. These validity rules recognise certain invalid combinations of filled-in fields – e.g. a document code that doesn’t belong with a goods code. Such a declaration will not be processed, but is returned immediately to the person who sent it to be corrected. If you think for a moment that this happens with 1.5 million import or export declarations each year – 8% of the total volume – you realise there’s lots of ground that can be won. When we have a good picture of what is going wrong, we can instruct companies what to watch out for next time. This saves them administrative costs and prevents wasted time. And we ourselves can collect useful information about market parties. Do they have qualified people making declarations? Are they doing things wrong consciously or unconsciously? Is it worth giving them some extra attention?”

Auto detection of data
If Customs seriously wants to make leaps and bounds in monitoring the blue flow, then the service will have to think outside the box and be more innovative, Doolhoff stresses. “In order to keep a grip on the growing volume of goods, it’s essential to make heavy investments in the auto detection of goods and data. We have to develop automated systems that can independently recognise deviations in normal patterns of goods and data streams with the help of Artificial Intelligence and machine learning. In fact we’ve been working on this for a while now. For example, we’ve got a project working with risk filters, which is dedicated to reducing the relatively high numbers of false positives that turn up in our traditional risk selection. Our data experts have built a few test filters based on historic data collected where the output gets crunched a second time. This tells us whether a suspected risk is actually a real risk, and whether it’s worth taking the trouble to check a shipment of goods. We can supervise things more sharply with this: we’re less quick to send out officers for inspections, and the chance of a hit when we do check is greatly increased. That’s also good news for the outside world, because it means fewer unnecessary delays.”

“Right now the filter project is momentarily on hold”, Doolhoff continues. “It happens fairly often in our service that such innovative processes are delayed because other IT issues are given priority. When the already limited capacity for automation is overstretched, people look first for trouble-spots in operations right now. Logical, naturally, but I’d like to see the organisation give innovative processes more priority – ultimately, that’s what’s going to determine the future for our service. What’s more, you have to create internal support for promising new working methods: get staff members enthusiastic and make sure they stay that way. When you interrupt the continuity of such a development, there’s a chance that the growing spirit of excitement will ebb away.”

Auto detection of goods
A second track that, as already mentioned, Customs has great expectations of, is the co-creation of scanning equipment that doesn’t only make images, but can also identify irregularities. Self-thinking systems that can signal when, for example, there’s a mismatch between the declaration submitted and the goods physically present. Or that can identify things in a container, suitcase or parcel that are prohibited or require an authorisation. Doolhoff: “We’re already working on prediction models based on algorithms for our scan arsenal. But it’s not so simple. If you want to do forecasting then you need an archive with a huge number of existing images to base it on. The problem is that we didn’t collect these in the past; at the time we thought it wasn’t necessary. We made images, checked them, and then they were immediately erased. What we have to do now is store scanned images on a really big scale for reuse and analysis. But there are all kinds of formal and legal aspects attached, too, particularly related to privacy regulations: what are we allowed to save and share? And how should we do it so it’s secure? These are problems that are new for us and time-demanding.”

“And then there’s something else that makes things stagnate,” Doolhoff continues. “To get a context for what you’re looking at in a scanned image, you have to label every type of image – and do it very carefully. There’s no reliable algorithm without solid data preparation, and that’s an especially labour-intensive process. But to get a good output we have to allow ourselves such a lengthy preparation period. We don’t want to find out later that we’ve got a programme that can’t tell the difference between a pistol and a hairdryer, just to mention one example.”

Steering towards transformation
Doolhoff underscores how crucial it is that Customs move with confidence in the direction of an enforcement process that is, in large part, data driven. “This is a fundamental transition, and the responsibility of the entire organisation. Every manager in the service has to mark this point on the horizon and steer a direct course for it. Together we all have to record and save data – completely, correctly, and unambiguously: the scanner operator his X-ray images, the control officer her findings from inspections, and so on. We have to standardise our methods of registration and archiving more, and make them uniform, so that we have access to more reliable data. And we have to optimise our systems to that they can process all this information fast. Wouldn’t it be great if we could adjust our risk profiles to our supervision results in near real-time, for example. That’s the goal we want to reach in the long-term.”

E-commerce, a story apart
E-commerce stands out as a phenomenon that has given Customs major headaches in the blue flow. Doolhoff: “Online shopping is fragmenting the whole cross-border traffic in goods. All those packages sent by mail maybe don’t count for much in terms of weight and value, but in terms of articles and in terms of declarations they reveal an enormous growth in volume. Apart from the massiveness, it’s evident that the quality of declarations in this sector leaves a great deal to be desired. Taken all together, it’s clear our traditional methods of profiling and physical controls don’t work for e-commerce. So we’ve started a special project to come up with a separate way of dealing with it. We’re looking for a total approach: intensive collaboration with the Tax Administration and the Fiscal Information and Investigation Service at home, and with other customs authorities abroad. Together we’ll be searching for signs of fraud, compiling dossiers on certain market parties and start criminal prosecution at an earlier stage. That’s how we try to motivate companies to demonstrate compliant behaviour.”

Share this post