January 28, 2022
The Changing of the Seasons - Tracking Trade Data Patterns
Tracking international trade statistics is integral to a successful supply chain strategy. One challenge is understanding how seasonality affects the data. This report outlines what seasonality is from both physical and economics perspectives. We show how supply chain decision-makers can best utilize seasonal data. There are analyses of ports, electronics manufacturing, consumer goods retailing and trade policy to illustrate the key points as well as some cautions regarding the technicalities of seasonal data.
International trade data is only a small part of the economist’s toolbox but is critical for supply chain management. The data can provide: a guide to the developing nature of supply and demand balances; market context for specific products and companies within their industries; insights into fluctuations during the year; and comparisons to prior years’ experience.
Data from national governments and supra-national agencies can be augmented with proprietary shipping metrics to gain deep insights, but a clear understanding of how metrics vary is needed to ensure the correct conclusions are drawn.
The Ups and Downs of Global Trade - Why Seasonal Adjustments Matter
US trade data for November 2021, sourced from the Bureau of Economic Analysis in Figure 1 above, showed that goods exports grew by 23.5% year over year in November to reach $156.4bn. At this point, a close observer might ask: “If exports grew, why do those green and black lines at the end of the graph tick down?”
Figure 1 looks at trade in goods. From October to November, exports of goods fell by $2.9 billion sequentially. The chart also shows there are more distinctions to be made, with the jagged line showing data which is not seasonally adjusted (NSA) while the smoother line is seasonally adjusted (SA).
Each serves a purpose.
If you want to know the value of goods passing through ports, you should use NSA. If, instead, you’re trying to detect a new trend in trade, you’ll be better served by seasonally adjusted numbers.
As perhaps the most striking visual example of the difference seasonal adjustment makes, look at the months of February each year. US goods imports in February 2019 fell by 9.1% with no seasonal adjustment, but were flat after adjustment as shown in Figure 2. This means that imports usually fall in February so, up to a point, this should not be taken as an important new trend.
We can quantify how much seasonal adjustment smoothes things. Over the time period shown, the average monthly change (in absolute value) was 5.6% for NSA goods imports and 6.2% for NSA goods exports. The corresponding SA numbers were 2.6% and 3.7%. This difference also serves as evidence of the importance of seasonality. It should be noted that “seasonality adjustments” made by economists vary over time, as discussed at the end of this report.
An additional implication of both the graph and the statistics above is that the monthly numbers jump around a lot, even with seasonal adjustment. Thus, to assess broader trends, it can also be helpful to compare longer stretches of time. An easy and obvious one is to look at totals for the year through November.
Comparing 2021 to 2020 is problematic, since the spring 2020 included the onset of COVID-19, which came with a dip in trade which distorts the comparison.
One option is to compare 2021 to 2019 and take the geometric average.
This shows that US goods exports grew at an annual average rate of 2.8% while goods imports grew by 6.1%. The pandemic era has been much harder on services trade: US exports fell at a 6.9% annual rate while imports fell at a 5.3% rate.
The bottom line is that US goods imports and exports have recovered from their COVID-19 plunge and are growing steadily, with imports outpacing exports. Whether that trend continues will depend a lot on major economies’ economic performance in the year to come, as discussed in Flexport’s 2022 Outlook report.
Businesses Need to Watch Details, Trends and Bends
So, what are the practical applications of examining seasonality?
As flagged already, the choice of NSA versus SA analysis is largely a function of the challenges supply chain participants are looking to address. In the case of seaport operations, non-seasonally adjusted data on a sequential, daily basis can provide a guide as to short-term moves resulting from policy shifts and whether prior years’ patterns may provide relief.
Figure 3 shows US imports of loaded containers to the west-coast US seaports through to December 2021 on a days-adjusted basis where January of each year is rebased to 100%. Note the significant dips in shipments into the west coast ports in February and March, which can be connected to the Lunar New Year in February and Golden Week holidays in October in Asia each year.
The two-week drift in the timing of lunar new year from year-to-year explains some of the shift in timing of the spring seasonal downturn. The “lunar new year effect” in 2017-2019 resulted in a sequential downturn in imports of 21% on a days-adjusted basis.
The surge in demand for consumer goods resulting from stay-at-home demand during the pandemic had inflated the volume of shipping through the west coast ports. There are signs that imports have firmly turned down, with shipments in December having fallen by 16.4% compared to their May peak. That’s not to say the system is in relief though - shipments were still 3.1% above December 2019 levels.
US imports of laptop computers also have a significant seasonal component as well as long-term trends.
US imports of laptop computers have two peaks during the year, as shown in Figure 4 above (HS 8471.30, Source US Census Bureau). The first peak in July can be linked to back-to-school purchases while the second in October and November is likely the result of new models and associated holiday gift buying.
Figure 4 also shows the data in volume rather than value terms, which can be affected by inflation, tariffs and industry consolidation. The volumes are important when planning the logistics elements of supply chains.
Finally it should also be noted that there’s been a further shift in imports from Mainland China, which accounted for 93% of imports in October and November 2021 compared to 89% in the same period of 2017. That came at the expense of imports from Vietnam and may reflect a growing importance of available capacity particularly during the peak season.
Trade War, Pandemic and Just-in-Case Act as Disruptors
Arguably there has not been a “normal” year for US import seasonality since at least 2018, with three major disruptive processes.
First has been the trade war between the US and China which reached its peak with the implementation of Section 301 duties on imports from China across a wide range of goods. In aggregate imports of the affected goods from China fell by 23.3% in the 12 months to Nov. 30, 2021 versus the 12 months to July 31, 2018 (when the first round of duties were applied) as shown in Figure 5 below.
Additionally the tariffs were applied in a series of waves and adjustments from July 2018 through February 2020 as widening lists of products and increasing tariff rates were applied. In some instances that resulted in a surge of imports ahead of both periods as shippers sought to beat tariff increases.
That can be seen for example in the “List 3” products which include consumer goods such as furniture and electronics where tariffs were launched at 10% in September 2018 and 25% in May 2019. The threat of an earlier increase in January 2019 arguably drove a 22.2% sequential increase in December 2018 versus November 2018.
Second, the COVID-19 pandemic has distorted consumer buying patterns versus prior history, as outlined in Flexport Research of January 21, 2022.
Figure 6 shows the seasonal pattern of US imports from China, deflated by import price inflation and indexed so that January of each year equals 100. The black line shows the pattern applying from 2010 to 2019.
The green line shows 2020 when the impact of factory closures in China at the earliest stage of the pandemic spilled over into a significant drop in deflated imports in March 2020. A subsequent surge in demand later in the year led to a higher-than-normal acceleration into peak season.
In that regard, in 2020 at least, the pandemic resulted in an exaggerated seasonality effect with the “normal” pattern yet to fully return in 2021 with March, September and November showing unusual patterns.
The third disruptor, which has yet to play out, are corporate strategies regarding inventories of both components and finished goods.
In the case of the furniture, electronics and appliances stores sector, shown in Figure 7 above, the seasonally adjusted value of sales in Oct. 2021 were 18.6% higher than October 2019 while inventories had increased by 8.2%.
As a result the inventory to sales ratio for the three sectors combined of 1.44x remained well below the pre-pandemic (2010-2019) average of 1.6x. For context the total retail industry’s ratio of 1.05x in October compared to 1.24x before the pandemic.
Corporations have two broad choices. One is to rebuild inventories to higher levels in order to ensure ongoing seasonal sales at the price of higher costs in the near-term and tied up cash flow. There’s some evidence of this already happening given October’s 1.44x level was already higher than the 1.31x trough of March 2021.
The second choice is to maintain inventories at lower levels on the presumption that seasonality will normalize, or at least become less extreme and unpredictable as it has been in the past two years.
Data Analysis of Seasonality in Consumer Goods
Bringing it all together, Figure 8 compares the two “flavors” of Flexport’s Trade Activity Forecast for the Consumer Goods sector – with and without seasonal adjustment.
The left hand chart shows imports on a non-seasonally adjusted Census Basis and provides a guide for physical supply chain planners as to how imports have and may vary month-by-month. It shows the usual downturn in Q1 before stability arrives near the end of the quarter.
The right hand chart compares the non-seasonally adjusted Census Basis data (green line) with the seasonally adjusted, balance of payments basis (red line) which is useful for long-term strategic and financial planning. The latter shows the trend towards higher consumer goods imports should continue during Q1 before stabilizing.
Two final important points to note are the degree of volatility in all the components involved. That should lead one to reserve judgment on short-term movements and focus on multi-month trends.
Figure 9 shows the percentage month on month changes in non-seasonally adjusted Census Basis data (the left hand chart of figure 8). The green bars show what happened in 2021, the red bars what happened in the previous 10 years on average and the dotted lines show the maximum and minimum over that time.
Two months to note:
In August 2021 imports rose by 9.8% sequentially versus July, the fastest growth in that month in over a decade. That may reflect the impact of retailers attempting to ship early to get ahead of earlier holiday shopping and avoid getting caught in logistics congestion.
In November 2021 imports rose by 0.6% versus October, the first time in over a decade they had increased rather than a more typical decrease. That may reflect the late arrival of products that had been intended to arrive in the normal peak month of October.
Finally, it’s worth noting that seasonality is not a constant. The patterns are real, but they shift. . The actual adjustment made by the U.S. government to get from non-seasonally adjusted to seasonally adjusted data varies over time, as shown in Figure 10 above.
There are many elements to the adjustment and getting them right can be vital for strategic and financial planners attempting to track the accuracy of their forecasts.
In October 2021 the NSA-to-SA adjustment was the smallest (negative) adjustment it had been in over a decade while in November 2021 it was the largest (negative) adjustment made in a decade.
Conclusion - Seasonality is Useful, but Tricky
This report has covered a wide range of seasonal patterns and statistics concepts. We provide these three simple conclusions:
Non-seasonally adjusted numbers provide the data that tell you how many containers are actually passing through ports. This is the flavor that is discussed least frequently in data releases but may be most useful to supply chain planners.
Seasonally-adjusted numbers are the ones that tell you whether the new data is in line with or deviates from regular patterns. This is the flavor one usually sees in news releases and used by finance sector participants.
Changes in seasonal adjustments, while seemingly arcane, tell us something about shifting patterns of business. Thus, tracking these changes is particularly useful for planning.
Disclaimer: The contents of this report are made available for informational purposes only and should not be relied upon for any legal, business, or financial decisions. Flexport does not guarantee, represent, or warrant any of the contents of this report because they are based on our current beliefs, expectations, and assumptions, about which there can be no assurance due to various anticipated and unanticipated events that may occur. This report has been prepared to the best of our knowledge and research; however, the information presented herein may not reflect the most current regulatory or industry developments. Neither Flexport nor its advisors or affiliates shall be liable for any losses that arise in any way due to the reliance on the contents contained in this report.