Can U.S. Shipping Transactions Data Predict Sales of Non-U.S. Companies?

With many companies currently facing supply chain issues, inventory and sales forecasting can be trickier than usual. For any investor, it is therefore important to identify the extent to which the companies held in a portfolio could potentially be affected by supply chain disruptions. A leading indicator that gauges the effect of these disruptions on company revenues could prove extremely valuable.

One potential indicator is maritime shipping data. Thousands of ships arrive at U.S. ports every day and the registration of each’s ship cargo is done almost in real time. When the cargo is unloaded, the details for each transaction are registered with the local customs officers. These details are listed in the bill of lading, which includes information such as the shipper, consignee, transaction number, number of containers, weight, manifest description, and other details about the specific shipment.

The FactSet Shipping data set provides a structured view of any maritime shipment that has been registered by U.S. Customs and Border Protection (CBP) as imported goods. The CBP data is received daily and is systematically linked with FactSet’s entity structure to properly identify shippers and consignees at the subsidiary, parent, and ultimate parent level.

While most shipments arriving at U.S. ports are destined for U.S. companies, shipping data can also provide insight on foreign companies’ maritime exports to the U.S. Can CBP maritime shipments data be used as a leading indicator of sales for non-U.S. companies?

Defining the Universe of Companies

The first step in our analysis was to define our universe of companies. We wanted to focus on non-U.S. companies with a large share of their revenue generated from customers based in the U.S. We started with the over 6000 constituents of a major global index excluding U.S. companies and then used FactSet Geographic Revenue Exposure (GeoRev) data to identify which of these companies have more than 20% revenue exposure to the U.S. According to GeoRev, about 670 companies satisfy this criterion with high certainty.

Next, we looked at company sales data. To conduct a bulk analysis of the companies’ shipment vs. sales, we needed to ensure that data was available for both concepts for the same period and frequency. To simplify the analysis, we wanted to use a calendar quarterly frequency. Therefore, we further limited our investment universe to only those companies that report earnings quarterly. This left us with 485 companies and of these, about 450 report on a calendar quarterly basis.

At the same time, we filtered out companies in the Cargo Transportation and Logistics industries as for this study, we are mainly interested in revenues generated by the manufacturers rather than the shipping companies themselves. This gave us a list of approximately 440 target companies.

Capturing Shipments by Subsidiaries

Given that the total revenue of a target company includes the sales of its subsidiaries, we needed to ensure that the shipping data used in our analysis reflected shipments by company subsidiaries. We used FactSet’s corporate hierarchy data to identify the subsidiary lines of our 440 target companies. This brought our company universe to approximately 3600.

For this analysis, we wanted to examine the quarterly sales data for the past five years (back to 2017) for these target companies and their subsidiaries. According to FactSet data, 2270 entities linked to 252 of our target companies have shipped products to the U.S. since 2017.

Compiling Quarterly Shipment Data

Because the raw CBP data is reported daily, we summed the daily shipments for each target and its subsidiaries listed among the 2270 shippers to create calendar quarterly shipment statistics for three metrics: number of unique transactions, containers, and weight (in metric tons). We then compared the changes in these metrics with the change in quarterly sales for each target company.

When we examined the average yearly transactions by the target companies (including transactions by their respective subsidiaries), we found that not all companies have completed enough transactions to yield significant results when performing a correlation analysis. Therefore, we filtered out those target companies with fewer than 100 distinct transactions per year on average.

Correlation Analysis

Once we had compiled our quarterly shipping statistics for our target companies, we calculated the correlations with quarterly sales figures to establish whether U.S. shipments can be used as a leading indicator for the total revenue of a company. In our correlation analysis, we also introduced a lag factor to see if the change in sales lags the change in shipments by zero, one, or two quarters.

To avoid forward-looking bias, we selected the period January 2017-June 2020 as our test frame. We then compared the change patterns in the different shipping indicators with sales figures to see if we could have predicted the direction of sales for the period September 2020-June 2021.

The Results

Limiting to companies that have a correlation above 0.5 (an indication of strong correlation) for any of the lagged periods, we see that most of the companies meeting this threshold belong to either the Industrial or Technology sector. This result wasn’t a surprise given the importance of these two sectors to the U.S. economy.

The correlation heatmaps shown below indicate that for many of our target companies, there is a strong enough correlation between one or more shipping indicators e.g., total number of unique transactions, containers, items, or the total weight of the shipments, and quarterly sales to justify further analysis.

Correlation Analysis for Specific Companies and Industries

• Spin Master Corp., Bayer AG, and Gränges AB show strong correlations between each of the selected shipping indicators and their overall quarterly sales with no lags
• Kawasaki Heavy Industries and Daikin Industries demonstrate relatively strong correlations between the shipping indicators and their respective quarterly sales with a one-quarter lag
• Trelleborg Ab and SSAB AB exhibit strong correlations between just one shipping indicator and quarterly sales: number of transactions for Trelleborg AB and weight of shipments for SSAB. For these two companies, the correlation is strongest when we introduce a two-quarter lag.
• For technology companies such as Sony, the correlation is strong across the shipping indicators with zero lag
• For LG Electronics and Konica Minolta, only the number of quarterly transactions and number of quarterly containers show strong correlations, with one- and two-quarter lags, respectively

To better illustrate these relationships, the charts below highlight a few examples of strong correlations between the quarterly change in the various aggregated shipping indicators (transactions, containers, and weight) versus the change in quarterly sales.

Visualization Analysis for Specific Companies and Industries

• For Kawasaki Industries, the two lines seem to follow the same pattern closely for most of its U.S. shipments indicators
• LG Electronics shows a one-quarter lagged signal between the change in shipments and sales; the strong correlation is most evident for containers vs. sales and transactions vs. sales
• Konica Minolta shows the strongest correlation in the change in quarterly U.S. transactions versus the change in its quarterly sales, notably with a two-quarter lag

U.S. Revenue Exposure

Spin Master, Bayer, and Gränges have between 30-50% U.S. revenue exposure according to FactSet GeoRev, and the strong correlations shown in the heatmaps above for these companies should therefore have been expected. However, Kawasaki Industries, LG, and Konica Minolta all have significantly lower exposure to U.S. markets, generating 20-30% of their total revenue from sales in the U.S., as shown in the table below. Still, it seems the information captured through some of their U.S. shipment indicators can serve as valuable signals when estimating changes in their total revenue.

U.S. Share of Total Company Revenues (%)

Source: FactSet GeoRev

Conclusion and Further Studies

As indicated above, the correlations presented here were calculated for the period between January 2017 and June 2020. Based on our analysis, shipping data could have been used to forecast changes in sales for the period September 2020-June 2021 for many of our target companies with relatively high confidence. The purpose of this study was simply to verify the existence of a potential signal within the shipping data. A more comprehensive and detailed study would be required to justify any investment decisions.

Here we examined correlations based on the quarter-over-quarter changes in several non-U.S. companies’ U.S.-bound cargo shipments and their respective overall sales. Pivoting the analysis to year-over-year quarterly changes could potentially identify other companies with strong links between their U.S. shipping indicators and changes in their overall sales.

Given the ongoing global supply chain issues and shipping delays, it would be interesting to see which companies have been most impacted by these disruptions and whether these delays have had any significant impact on the degree of correlation between their U.S. shipments and overall sales.

Disclaimer: The information contained in this article is not investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.