Global Marketing Monitor: Weekly Market Trends (Sept 19, 2020)

Key takeaways from this week’s note:

• The OECD’s global GDP forecast revisions or 2020 and 2021 are improved, especially for the U.S. and China, but continue to show a historic economic slowdown.

• In August 2020, U.S. and China retail sales were stable year-over-year, a deceleration for the U.S. and acceleration for China.

• A noisy, newsy week for each of Facebook and TikTok, with new concerns raised for the former company and uncertainty continuing for the latter one.

• Did cord-cutting in the U.S. actually improve in the second quarter, stay the same or worsen? It’s harder to know than we thought, as pandemic policies have temporarily altered Nielsen’s panel recruitment and turnover efforts.

We monitor earnings results from public companies and economic releases from countries around the world with a focus on consumer trends, advertising and digital transformation to gain a view on key trends impacting the world’s largest marketers. In this note, we summarize insights observed over the past week.

OECD economic growth forecasts improve, but still historically negative. Changing economic forecasts are always important to the advertising world, as related trends will generally be mirrored in marketers’ spending trends. The Organization for Economic Co-operation and Development (OECD) released its latest global economic forecasts this past week, updating prior forecasts published in June. In real (inflation-adjusted) terms, they now expect the world to see a 4.5% decline for this year versus their prior expectations for a 6.1% decline in June. While these figures are an improvement of sorts, they would still represent one of the worst economic outcomes on record.

The changes in the OECD’s outlooks for the United States and China were responsible for most of their revision. In the U.S., the OECD now expects a 3.8% decline for 2020 and 4.0% growth in 2021 versus the previously published 7.3% decline this year and 4.1% growth next year. For China, the OECD now forecasts 1.8% growth versus a 2.6% decline previously for this year along with 8.0% growth in 2021 versus 6.8% previously.

Expectations for the Euro area are now for a fall of 7.9% in 2020 and a gain of 5.1% in 2021 versus prior expectations of a decline of 9.1% this year and a gain of 6.5% for next year. In the U.K., expectations also improved, now at a 10.1% decline for this year versus an 11.5% decline previously, while they anticipate growth next year of 7.6% versus 9.0% previously. Importantly, the OECD states that its forecast presumes that there will be an EU-U.K. basic free trade agreement in effect at the beginning of next year.

Among major markets, India’s current year data saw the most significant negative revision in the OECD’s new data, with a 10.2% decline expected now versus 3.7% previously. A more significant rebound is now expected next year off this smaller base, as expectations are now for 10.7% versus 8.9% previously.

Assumptions underpinning the data are important to consider, including how Brexit plays out, the continuing trajectory of the pandemic, the nature of new lockdowns and government economic support for people and businesses. Among the important assumptions made beyond the Brexit resolution, “projections assume that sporadic local outbreaks will continue, with these being addressed by targeted local interventions rather than national lockdowns; a vaccination is assumed not to become widely available until late in 2021.”

Highlighting the uncertainty embedded in these numbers, they note “a stronger resurgence of the virus, or more stringent containment measures, could cut 2-3 percentage points from global growth in 2021.” They also note that the premature withdrawal of fiscal support in 2021 from governments around the world would stifle growth.

Separately, new data on retail sales for the month of August in the world’s two biggest markets, the U.S. and China, provides updated data on a part of the economy that more directly impacts marketers. Preliminary U.S. retail sales were released on Wednesday for August 2020. This activity decelerated to post year-over-year growth of 0.1%, below the revised 3.5% growth rate for July, which was similar to June and was essentially in-line with the “normal” growth of 2019’s full year levels after significant declines in March through May.

In assessing the U.S. data, it is important to consider the role of stimulus payments made to consumers that ended in July. U.S. data now looks more similar to Chinese data than it has in recent months, where China was gradually recovering but the U.S. was somewhat stronger.

To that point, in China, where the economy ended its relatively-more comprehensive lockdown at the end of February, new government data was released on Tuesday and indicated that retail sales there fell by 0.5%, better than the 1.1% decline observed in July and 1.8% decline in June.

​​​​​​​Within the United States, many categories of retail activity were up year-over-year. Motor vehicle and dealer sales grew by 1.6%, decelerating meaningfully from the 7% pace of July and 10% growth rate posted for June. Presumably the end of stimulus payments at the end of July made some difference, but some of the spending earlier in the summer likely related to pent-up demand as well.

Building material and garden equipment stores continue to be very healthy. For August they rose by 12% following on gains of 16% in July and 25% in June. Food and beverage stores are also still faring well, with an 8% rate of growth in August after growing by 13% rate in July and 11% in June.

By contrast, clothing and clothing accessory stores continue to be incredibly weak, falling by 24%. July and June saw a 21% and 26% decline, respectively, following on a near-evaporation of the sector in April and May as consumers may not be replacing as much clothing as they used to. They also presumably changed social and professional habits, thereby limiting the need for new purchases as well.

Food service (i.e.: restaurants) and drinking establishments (i.e.: bars) continue to be weak with another 17% decline to follow on July’s 18% fall, although these rates are still better than significantly worse declines from prior months during the pandemic. Electronics and appliances stores fell by only 3.4% in August after falling by a downwardly revised 4.7%, presumably aided by consumers continuing to buy products for work and school-from-home activities.

Department stores worsened to a 17% decline in August after losing 13% of sales in July following on a 14% decline in June, both much better than May’s 26% decline or April’s 45% fall. Other forms of general merchandisers (i.e.: warehouse clubs and supercenters) managed to continue growing, rising by 3.6% in August to following on July’s 7% growth level.

China is still seeing modest declines in retail sales. China’s relatively stable numbers were dragged down by a 7% decline in catering services (equivalent to restaurants and related food services activities). By contrast, automotive sales were relatively healthy, once again rising by 12%, just as they did in July. Year-to-date, automotive sales are still down by 9% versus the same period in 2019.

Within this data, online retail sales expanded by 17%, only slightly slower than July’s 19%, June’s 25% growth rate and May’s 23% pace. Its scale remains relatively high: in total these activities accounted for 23% of total retail sales during August 2020, up from 19% in August 2019. Excluding catering and petroleum sales, e-commerce accounted for 27% of retail sales in August 2020 versus 23% in August 2019.

On the media front, it was a very noisy week for TikTok. The U.S. and China, whose economies have been negatively impacted by the pandemic as well as a trade war, continued to play out a geopolitical drama around the media and technology industries this week.

News emerged that now Oracle, Walmart and ByteDance’s existing U.S.-based venture capital shareholders are proposing to the U.S. government that they would become minority investors in a new U.S. business housing TikTok, ostensibly to alleviate national security concerns. The U.S. government would have approval rights on membership of the board of the U.S. entity. The company would also aim to become a U.S.-listed publicly traded stock, which would add additional shareholders to the company.

Whether Bytedance would actually have control, how much cash Oracle and Walmart would invest, who would run it – Instagram founder Kevin Systrom was reported as a potential candidate for CEO of the entity – and, most critically, whether the U.S. or China would actually sign off on the agreement were among the many uncertainties heading into the weekend.

Another uncertainty would be the nature of the working relationship between TikTok and Oracle, which, at an earlier stage, was described as the entity that would be the acquiror of parts of the business. At least one element does seem relatively clear: Oracle would provide cloud services that would ensure the company would generate hundreds of millions of dollars in annual revenue for Oracle at minimum. For reference, Snap, whose scale is generally similar to TikTok’s, has commitments to Amazon and Google for similar cloud services that obliged them to spend at least $600 million with those two companies last year. This would add some heft to Oracle in a cloud services sector where they are well behind leaders Amazon, Microsoft and Google.

Whatever the structure, we think that most marketers would probably prefer that TikTok maintain as much operational independence from the newly acquiring entities as possible because they want to see new successful scaled alternatives in the media industry that can durably provide incremental competition to the world’s dominant digital media owners.

Toward these ends, we note the track record of companies primarily focused on enterprise software in sectors adjacent to ad-supported media has become so mixed, at best, in recent years, despite their capacity to invest resources and apply their data management capabilities into these properties.

To illustrate, this past week we saw new earnings results from Adobe that indicated while their core Digital Media (i.e.: Creative Cloud products) business continued to expand rapidly, the smaller Digital Experience business (ad tech and marketing tech products), outside of its recent acquisitions of Marketo and Magento, actually declined in the most recent quarter. While there are always company-specific factors at play, marketing tech, ad tech and digital advertising businesses probably fare best when operating independently from other software-related businesses.

Among scaled media owners, it was also a noisy, newsy week for Facebook. With Facebook this week we had:

1. Reporting in the Wall Street Journal that the U.S. Federal Trade Commission investigation was shifting toward a lawsuit;
2. A one-day boycott of Facebook and Instagram by Kim Kardashian and other celebrities;
3. Buzzfeed’s new reporting on the company’s recent tolerance of government misinformation around the world;
4. Buzzfeed’s new reporting on board member Peter Thiel’s efforts to build relationships with white nationalists in 2016 and 2017; and,
5. Bloomberg Businessweek’s reporting on Facebook’s extensive efforts to deepen its relationships within the Republican Party following on years when it was perceived as favoring the Democratic Party.

Any one of these issues has the potential to impact Facebook’s business, but they are unlikely to have much in the near-term.

The FTC investigation will, undoubtedly, take years to play out while the other items relate to issues that have evidently not resonated sufficiently with consumers to deter their use of the platform. The vast majority of the company’s advertiser base generally continues to mirror those activities in their budget-setting decisions. Sustained spending levels will occur in many instances because Facebook’s efforts to address problems are viewed by some of its stakeholders as sufficient.

Alternatively, for others, the benefits of using Facebook’s properties continue to outweigh societal and commercial consequences at the present time. Whether or not they should be is, of course, another matter altogether.

With respect to how they impact our advertising forecasts, these issues are unlikely to impact industry-wide numbers in any observable way this year, not least because of the nature of the economic recovery from the worst of the pandemic will be difficult to separate from any consequences that occurs because of non-economic factors.

And, speaking of scaled media, while television still dominates for most large media owners, it is scaled to a more uncertain degree than we thought previously. We have a clarification to make after we wrote two weeks ago of recent data from Nielsen that indicated that, within the United States, there was a moderation of cord-cutting trends, or a slowing in the rate of decline, as our analysis of Nielsen’s data indicated as much; however, as we were reminded, that data from publicly traded cable operators and owners of satellite services actually indicated a faster pace of decline over the past two quarters.

This prompted a review of the differences between data tracked by Nielsen in its cable universe estimates, the data provided by public companies and other gaps between those figures and actual underlying trends in order to better assess the actual consumer trends that are playing out during the pandemic.

Publicly traded owners of MVPDs in the United States, including Comcast, AT&T, Charter, DISH, Verizon and Altice all provide data on the number of video subscribers to their services every quarter; however, comparing numbers between MVPDs, let alone with Nielsen, is not as straightforward as it should be.

Let’s start with some of the elements that underpin the MVPDs’ reported data and compare that data to Nielsen’s approach for calculating pay TV subscribers.

• MDUs (Multiple Dwelling Units). Different MVPDs might have different policies to count subscribers from apartment buildings and different households might count differently within MDUs. A household within one MDU might be counted as a single video subscriber or as a percentage of a single video subscriber if there are bulk discounts for the MDU. Other MDUs might only count as single subscribers regardless of the number of households if residents are unable to receive multiple services. For Nielsen, a household is a household regardless of the MVPD’s billing arrangement with them.
• Business services subscribers. MVPDs sell their services to bars, restaurants, offices, schools and other institutions. Many of them also pay for video services. Nielsen only includes households and excludes other environments in its universe estimates.
• Double counting. A given household might have multiple subscriptions to different services within the same household – for example, paying for AT&T’s DirecTV to access NFL Sunday Ticket and concurrently paying for a cable operator’s video services because the video was bundled with a data service. A household might have multiple residences and multiple subscriptions as well. Each of these actions would lead to incremental subscriptions in public company filings. By contrast, Nielsen only tracks a household once.
• Exclusion of delinquent customers. Cable and satellite operators may no longer include in their figures subscribers who have stopped paying for services but who continue to receive those services. Nielsen makes no distinction as to whether or not a household is still paying for its services.
• Piracy/subscriber leakage. While the rise of digital distribution of TV services helped reduce historical forms of pay TV piracy, there are undoubtedly still many households who receive pay TV services who do not pay for those services, although they won’t typically come to light until they are shut down. In late 2018, Fast Company reported on a service pirating more than 180,000 customers from DISH Network before it was forcibly shut down. Such households would be counted as TV households in Nielsen’s data.

Beyond these factors, we were previously aware of differences that might result from Nielsen’s indirect reliance on U.S. Census Bureau data, which can be subject to its own methodological issues.

For example, analysts at MoffettNathanson have noted a recent announcement by the Bureau implying that, during the pandemic, they are ignoring newly vacated homes, leading to a spike in reported households, and the lagged application of this data into the Universe Estimates.

In essence, Nielsen uses projections of U.S. households made by Claritas based upon Census data made prior to the beginning of a given broadcast year projected forward to January of the following year, and then holds that number of households constant for the entire TV season. There may also be some vagaries in the data if some other factor made Nielsen’s panel of households not representative of the total population.

Pandemic policies have temporarily altered Nielsen’s panel recruitment and turnover efforts, which may help measure current period activity, but make assessing year-over-year trends more uncertain. However, in reviewing this data further, we learned that, with the arrival of the pandemic, Nielsen temporarily altered the process by which it rolls over its sample in order to maximize the quality of the current period data while it implements remote approaches to recruit and install new homes.

More specifically, households that become members of Nielsen’s panel are recruited, onboarded and maintained in the sample for a period of around 24 months. This means that, in normal times, around 4% of the panel turns over.

But, because of the pandemic, Nielsen recognized that recruiting new households would likely be challenging and so decided to extend the duration that households would stay on the panel. As Nielsen has noted that newly recruited households tend to be more likely to be Broadband-Only Homes (i.e.: no subscription to traditional pay TV), this change likely reduces the pace at which we calculate that traditional pay TV subscribers are cutting the cord, at least when we look at the data on a year-over-year basis. It also probably understates the growth rate at which we might calculate that consumers are adopting vMVPDs for their video services.

It’s certainly possible that the changes to U.S. society that followed the beginning of the pandemic may have skewed the reported growth trends from cable and satellite operator; however, it’s probably very likely that Nielsen’s changes had a more significant effect.

Whether the true number is closer to the 3-to-4% rate of decline observed in the September to March time frame or the 1% rate of decline observed more recently is difficult to know, although, in accounting for the change in the process Nielsen has applied, a number approaching 3-4% appears more likely than 1%.  As with so much of the rest of the world, however, uncertainty remains an operative word.