One of the wonders of the modern, computerised world is the emergence of data science. Data science is, arguably, at the heart of most successful businesses today, and its favourite food is big data.
For Panjiva, data science and big data enables it to keep track of virtually all goods moving from one country to another, with a particular emphasis on the US market.
But while it might sound straightforward to some, the actual collation and preparation of the data, and then the presentation of it, is not a simple process.
Much of the data starts out in unstructured form – in technical terms, a mess. That mess then has to be given some structure, and put through various information processes to make it presentable to Panjiva’s customers, who can also access articles written about the organised information.
In an interview with EM360º, Chris Rogers, global trade analyst at Panjiva, explains more about the business.
“Panjiva is a global trade data platform,” says Rogers. “We bring together data on merchandise imports and exports from across the world, in particular for the US, China, Mexico, Brazil and a variety of Central and South American countries, as well as layering on high-quality data sources from other countries around the world.
“Our research business, which we started in May of 2016, is really designed to help our customers understand what they can use the data for. It’s designed to address questions that matter to different Industries and show how you could answer those questions using Panjiva systems and data.
“In that regard, we write articles about a lot of different industries. Obviously, logistics is a big area, and that varies from seaborne shipping through to things like handling.
“We actually write about industrial robotics as part of our electricals coverage – so capital goods more generally. Obviously a big part of industrial robotics, and international trade generally, is very much impacted by government policy, and that’s another big area, along with economics that we write research articles about.
“Our articles are shortform in nature and they’re really designed to raise a question, address how it can be answered, provide a high degree of transparency into our platform, as well as some clear and attractive graphics that our customers can either use in their own presentations or for their own business purposes.
“Some of our research is freely available, but most of it is only available to Panjiva customers, but most of it is only available to Panjiva customers through daily emails, and in summary form through our Panjiva podcast, which you can get on iTunes and Soundcloud. But the in-depth research is just available for Panjiva research customers on our subscription platform.”
So many words, so little time
The sheer volume of data out in the world can be overwhelming. If it’s an unstructured mess, it can be daunting. But once a company like Panjiva gets it all together, the results can be impressive.
Exploring the website, one feels like one is being given access to a very high-level view of global trade, which is what it more or less is.
The data points are often sea ports, where bills of lading – or transportation documents for goods – are required by law. There’s more to it than that, but even collecting data from those bills of lading is not a straightforward process.
The problem, as ever, is free will, or more specifically, what Rogers calls “free text” – which is basically the notes that a human person makes to describe the goods being shipped. Those humans with the problematic free will may be filling out highly structured forms, but what they write specifically isn’t always structured enough to be useful without some processing.
“In the case of some our highly-detailed US data, it’s actually delivered on a daily basis by physical delivery – in this case, on a CD-ROM.” Chris Rogers
“There is a lot of data available about global trade,” says Rogers. “But it’s in what we call in data science ‘semi-structured’ form. So, the typical unit of information in global trade is a shipping manifest – it’s a pretty standard form, very structured in terms of what different things are that are put into those forms. But what is written in those forms is generally free text.
“So, people say, for instance, a container full of parts for an industrial robot is being shipped by Fanuc in Japan to, let’s say, General Motors in the United States, and there’ll be some other characteristics around that.
“Now, in some countries – the ones that we have on our Panjiva platform – that data is actually publicly available. But you need things like freedom of information requests to get your hands on it, and when you get your hands on it, it’s not in a particularly user-friendly format.
“So, in the case of some our highly-detailed US data, it’s actually delivered on a daily basis by physical delivery – in this case, on a CD-ROM – and we then need to unpack that onto our database, and the we’ve got to interpret it.
“Our secret sauce, if you like, which isn’t so secret really because lots of people are into data science now, is really the machine learning we apply over time – and we’ve been going for nearly 10 years – to translate that free text into a properly structured data form and information that our customers can then search via our platform.
“And that involves things like tagging data to harmonise shipping codes, which may or may not be in the text of the shipping manifest, so our systems have to interpret a lot of that free text into those more structured, value-added data forms.
“So if I want to do a search for industrial robots, some companies will approach that directly, and some will put their own dimensions and wordings.”
“No data is ever perfect, not even government trade data, like the US Census, which restates their data several times during the year. The first version is normally accurate within two or three percentage points, but they have to keep refining that data over the course of the year.
“Invariably, data is big and messy, and we hope to do our part to make it cleaner and easier to understand.”
Let’s get down to specific business areas, like industrial robotics and automation
One of the key areas of interest for business and political leaders, especially in terms of who’s selling what to whom.
Shipping information would provide some insight into what international trade is going on, whereas the individual companies might not otherwise share that information.
Having all that shipping information at his disposal at Panjiva, Rogers is able to take a high-level view of the many markets and provide his perspective on trends he sees, and he seems to suggest that fears over the possible introduction of new, protectionist tax policies may be driving the market to some extent.
Rogers says: “Most of our research and insights are for our paying customers, but I think one of the interesting things that we found in industrial robotics is that companies are having to make difficult investment decisions in difficult or uncertain trade environments.
“So, one of the questions I ask myself now is, ‘To what extent are the imports we are seeing into the US is being driven by tax policies?”.
“Meaning, to what extent are the US-based dealers of companies like Fanuc, Kawasaki and others, and companies themselves, stockpiling ahead of a potential tax – a border adjustable tax, or destination adjustable cash flow tax as it’s often known, or any other restrictive trade policy – introduced by the Trump administration?
“The Trump administration has been pretty vocal about its concerns of currency manipulation by Japan and Germany, and you might see the possibilities within the trade policies and trade instruments that Trump has available to actually very quickly restrict imports from those two countries in particular, which for industrial robotics are very important.
“As you can see from research Panjiva has done and some charts I put together, more than half the total industrial robot imports into the US were from those two countries.
“And if you look at recent data, particularly after Trump and his administration started talking about these issues, there was a real jump in shipments from those two countries into the US.
“So that to me suggests that it isn’t just that the customers of these robotics and automation companies have started investing aggressively. It may simply be a matter of their supply chains within the US trying to avoid any new taxes that are brought in by the Trump administration.
“I’m sure Fanuc has done a great job, but I don’t imagine that it has doubled sales year-on-year in the US. Having said, I don’t cover the sales side of Fanuc and I’m not an investment analyst – you’d need to speak to them about that.
“Of course, ther’s a longer-term trend towards automation, and quite often you hear comments in the media that a lot of the jobs lost in manufacturing in the US haven’t been lost to China, they’ve been lost to automation.
“And while I have infinite respect for Bill Gates, I’m not sure that a tax on robots is necessarily the right answer. But it is quite clear that the world of automation is another country, if you like.
“Also, it should be mentioned that a lot of this equipment is sold on to other countries, through the US distributor network – into Mexico and Canada, for example.”
Fears of a huge trade war
Even before US President Donald Trump came to power, he was claiming that the US was being treated unfairly in global trade.
As soon as he got into the White House, one of the first things he did was to take the US out of the Trans-Pacific Partnership, a trade agreement between about a dozen nations around the Pacific, including:
- New Zealand
The US was one of the countries on that list until January 23, 2017 – just a few days after Trump was inaugurated as president.
Trump has also been talking about taking the US out of the North American Free Trade Agreement, which includes just three countries – the US, Canada and Mexico.
And while he may be looking to leave other trade agreements, such as the World Trade Organisation, and spark a global trade war, the measures he seems to want to introduce will likely alienate nations surrounding and close to the US.
As Rogers points out, if the US introduces punitive taxes on imports from Canada and Mexico, a number of industries within the US which export to those countries will almost certainly be subject to taxes imposed by Canada and Mexico on US imports into their countries.
“The US has a trade surplus when it comes to robotics and automotive – it sells more of those goods into Canada and Mexico than it buys from them,” says Rogers. “Those US companies and industries would be at risk of retaliation.
“And it’s not necessarily retaliation in the context of a trade war just about robotics and automation, it could be a trade war that spreads to and from other sectors.”
Rogers adds: “Free trade works for companies. It doesn’t always work for individual workers, especially if they are in companies that are displaced, but there’s lots of reasons why workers and companies are displaced, and it isn’t just free trade.”
Rogers makes the observation that Trump likes to get involved in business dealings on a micro level, directly talking to specific companies about their plans, trying, for example to dissuade automakers from investing in Mexico and attract them to the US instead.
But Rogers says it’s not as simple as that. “Trump’s preoccupation isn’t just with bringing back jobs, it’s also with the way the trade deficit actually works.
“If companies choose to move employment or move their operations back to the US, that helps the trade deficit to a certain extent, but if they use overseas equipment for that operation, that could actually worsen the deficit in the short term. I think the details of the implementation matter a lot.”
What might be even more complicated, says Rogers, is, for example, Trump’s trillion-dollar infrastructure investment plan, which includes requirements to “buy American, hire American”.
What if, asks Rogers, the materials and components of a complete system are all from outside the US, and it’s only the assembly and integration of that system that takes place in the US?
“It gets really complicated,” says Rogers.