Some People Just Do Not Wish to Use the Internet

To solve a problem, one must have a theory about what causes the problem. Consider broadband non-adoption. Some households cannot subscribe because service is not available at the home. The solution there is to build and extend networks so the option to buy is available. 

For households that can purchase service, they do not for a variety of reasons, including the inability to pay the market price. Subsidies can help, in that case. Non-interest is a much-harder problem, and is not easy to solve, since it is hard to sell a product, no matter what the price, if the product solves no perceived problem. 

“Consistently, the U.S. Census Bureau surveys reveal that the primary cause for non-adoption is a lack of interest in what the Internet offers,” says George Ford, Phoenix Center for Advanced Legal and Economic Public Policy Studies chief economist. “A distant second reason is the expense of the service and/or the devices required to use it.” 

Surveys by the National Telecommunications and Information Administration clearly show that “do not need it” has become the main reason households do not buy fixed network internet access, by about a three to one margin over respondents who say it is  “too expensive.”

source: Phoenix Center

According to U.S. Census Bureau statistics, the main barrier to adoption now seems to be that potential buyers do not see the need. Very few say cost is the main barrier for not buying. 

source: Phoenix Center

The point is that internet access adoption cannot be “solved” completely by any single tactic. Extending networks works if one assumes those new locations want to buy and use internet access. Subsidizing usage might convince a small percentage of people to buy. But some consumers might simply refuse to buy, no matter what is available and at what price, because they do not wish to use the internet and do not find it useful.

Confounding Economic Data Might Mean "Not as Much Permanent Change" as We Think

For anyone who believes “everything has changed” since the advent of the Covid-19 pandemic, measurable economic data is confounding. What we can measure is equally confounding. The Institute for Supply Management non-manufacturing index already has shown a sharp “V” recovery, far more sudden than the recovery from the Great Recession of 2008. 

source: Wall Street Journal

The index is a monthly composite index based on surveys of 300 purchasing managers throughout the United States in 20 industries in the non-manufacturing area. The index is released on the first business day of the month and covers the previous month’s data, which makes it particularly timely. If the index is above 50, it indicates that the economy is expanding. Values below 50 indicate a contraction.

Within the overall index, there are nine subindices: new orders, production, supplier delivery times, backlogs, inventories, prices, employment, export orders, and import orders. The index is believed to reflect future movements in gross domestic product,

However haltingly, the index suggests the non-manufacturing part of the U.S. economy is rebounding sharply from its March plunge. That might indicate that “some things could change,” but likely not that “everything” will change. 

Similarly, about 80 percent of respondents say they are growing again after the Covid-19 economic shutdown. That is likely unprecedented in the modern era. 

source: ISM

Employment, as much of the retail, hospitality and transportation sectors of the economy remain closed or operating at a fraction of former capacity, has yet to return to “normal” levels. But job losses in the wake of the internet bubble burst of 2001 and the Great Recession of 2008 also suggest that normalization will happen. 

sources: Nordea, Macrobond

That is not to deny an acceleration of trends already underway prior to the pandemic striking. It is a caution that the amount of permanent change we see in daily life, schooling and work might not be anywhere near as great as many suggest.

Reducing Friction is Purpose of AI

If telecom companies truly operated with complete efficiency (maximum gain, minimum waste) and effectiveness (consistently doing the right things to operate and grow their businesses), we would never see failed acquisitions, mergers that failed to deliver intended benefits, marketing programs that failed to boost accounts or reduce churn, or technology initiatives that swiftly delivered the promised value. 

Telecom firms--like all others--suffer from “friction.” As a practical matter, when all business processes become more frictionless, it should lead to outcomes such as higher lead-to-customer conversion rates, lower churn and higher account retention, plus higher renewal rates, as well as enhanced productivity (the ability to produce and sell more while reducing the cost of doing so). 

To reduce friction, many agree that better insight is required of operations, customer experiences and expectations, supply chain and partner behavior. And almost everybody agrees at some level that artificial intelligence, machine learning or deep learning will underpin the efforts to glean much more insight. 

But it often is hard to imagine how artificial intelligence can be implemented, as a practical matter, since AI is a capability, not a product; a learning system, not a discrete set of attributes. 

One does not go “buy AI off the shelf,” so to speak. So it might be better to cast AI as a tool for reducing unwanted friction, where the theoretical scenario is:

• 100-percent efficiency and knowledge of buyer demands, preferences, tastes

• complete understanding, in real time, of the state of a firm’s supply chain

• as-good-as-can-be-expected employee productivity, based on knowledge of actual behavior

• full effectiveness of all information technology systems, devices and software

• real-time knowledge of any legal or regulatory compliance issues

• Robust feedback loops and intelligence gathering that aids in the development process for new products and features 

In practice, no organization operates that way, all the time. There are inefficiencies in all operations processes, capital allocation, employee alignment with organizational objectives, understanding of customer demand changes and supply chain processes. 

source: capgemini

Fragmented customer profiles, departmental silos, inefficient workflows, shadow IT, slow feature deployment and redundant processes are examples of friction. And friction matters because it gets in the way of customer and user experience, not to mention sales and profits. 

Friction also exists whenever a firm or organization has to work with other stakeholders outside the firm boundaries, including  customers, partners, suppliers or regulatory and political entities. 

Frictionless systems aim, at a high level, to deliver insight, allowing an organization to accomplish its desired outcomes in both an effective and efficient way. And it is hard to imagine effective friction-reducing knowledge systems not using artificial intelligence, deep learning or machine learning. 

So maybe we should speak less about AI as a technology and more about how AI enhances core business processes to remove friction.

When Better Broadband Might Not Help

The Wisconsin Economic Development Corporation, in detailing the impact of the Covid-19 pandemic on various sectors of the Wisconsin economy, recommends three priorities: 

• Get Everyone Back to Work

• Fix Broadband:

• Support Innovation

It is what one might expect, but also shows the difficulty of generating economic growth where it might not be happening organically. An isolated region heavily dependent on tourism might prefer to diversify, but there are good reasons why all sorts of firms and industries do not locate themselves in remote areas. 

A region with low population might similarly wish to spur economic growth by focusing on one or two new potential growth areas, but with no natural advantages in human resources, distribution networks or other underpinnings of an industry. 

That is not a criticism of a report by the WEDC, simply an observation that economic growth is not easy when low population, remote location, lack of skills or knowledge or other attributes that attract people and firms are lacking. 

Nor is it easy to suggest how the situation might be changed, by any set of feasible government policies other than reopening the economy.

There is a direct and measurable causal link between “going back to work” and worker income and tax revenue. The causal link between broadband and economic growth or innovation is much harder to demonstrate. 

In fact, there might be correlation without causation, in the same way that better broadband tends to exist where people are wealthier, live in higher-density areas, have higher education attainment and are younger. 

People generally believe there is such a causal relationship, but the reverse might generally be the case. Economic success leads to demand for better broadband. 

Virtually everyone “believes” (or at least acts as though they believed) that advanced technology (faster broadband, artificial intelligence, IoT, 5G) leads to an increase in productivity. People, organizations, firms and countries that have and use more of such assets are presumed to make faster productivity gains, and generate more economic growth.

The problem, aside from inability to measure precisely, seems to be that the evidence is suspect. It still does not appear that better, faster, more extensive broadband adoption actually is related to productivity gains.

source: Bureau of Labor Statistics

Broadband and innovation are good things. It simply is not clear that better broadband leads to economic growth in any direct way.

How Much Business Value from Influencing 5G Standards?

Historically, there has been business value when a firm is able to influence the setting of technology standards in ways that play to the firm’s strengths. 

Some argue 3G, 4G and 5G provide examples.  With the exception of “market-driven” standards such as Windows or IoS, it is less clear to me that vendor influence actually delivers so much value in setting telecom standards. 

“Standard setting does not always create market power for the standardized technology,” say economists at NERA Economic Research. 

“The gains from formal standard setting can be defined as the difference between the royalty that the technology owner can charge after being selected formally as the standard and the royalty that she could charge if no formal standard were set,” they say. 

source: Statista

Some will point to quantitative measures to suggest Chinese firms have a clear lead in 5G patents. But more patents do not necessarily mean better patents, some will note. 

By some accounts, patent portfolios are rather more distributed than often appears.

source: GreyB

In settings where compatibility requirements are high, standards competition may be very important as the choice of a standard may virtually eliminate, not merely disadvantage, competing technologies. 

One hears it argued that Europe benefited from its “leadership” of 3G, while the United States benefited from its “leadership” in 4G. It is hard to parse such statements. One might note that European firms Ericsson and Nokia benefited hugely in the 3G era as suppliers of infrastructure. 

The same argument is heard about U.S. firms in the 4G era, but the argument is nuanced. Firms such as Qualcomm lead in chipsets, but no U.S. firms lead in radio infrastructure. To the extent there was “leadership,” it was in applications and service development, not infrastructure or necessarily “economic benefits.” 

Also, the growth of open source and virtualization tends to degrade any potential value standards influence might have, as, by definition, open source means buyers have choices beyond the usual cast of supplier characters.

One way of putting it is that what mattered was the economic benefit derived from 4G, not the “leadership” in standards or infrastructure supply, and that mostly because it was application developers, platforms and content suppliers that drove the value of what could be done with 4G, not the network infrastructure or connectivity supply. 

The same “value” often tends to be believed about patent portfolios, and there is evidence that the causal link is quite weak: patent portfolios do not often lead to market dominance. 

It is not the infrastructure; it is the value a nation or a firm can wring from it that matters.

Ultra-Low Latency Use Cases is Where Most New 5G Apps Will Develop

Though capacity matters, the big use case upside for 5G is expected to come in the area of ultra-low latency applications or perhaps ultra-reliable and ultra-low-latency use cases. If prior history proves a useful guide, many of the new use cases will develop only after five or more yeras. Many could  flourish only in the 6G era, even if essentially prototyped on 5G networks. 

source: Nokia

It takes time to envision and then retool entire business processes to take advantage of the new network’s performance attributes. And many additional parts of the ecosystem, such as edge computing, also must develop in tandem.

Public Policy is Devilishly Hard Stuff

Public policy success always is harder than you might think, if only because the causal relationships between a policy and an intended outcome are tough to confirm, and often because unintended consequences are common as well. At other times policy failures are hard to diagnose or predict. But policy failure rates arguably are fairly high, despite good intentions. 

An examination of a recent program to increase broadband internet access adoption and quality might provide an example. “we find no positive effect on home broadband adoption from programs funded by the Broadband Technology Opportunity Program (‘BTOP’),” say economists George Ford,  T. Randolph Beard and Michael Stern. 

BTOP, which added about $4.7 billion to ongoing efforts to supply access to the estimated six percent of U.S. households without terrestrial network access to internet access at a minimum of 25 Mbps, seems to have had no effect. 

“We find no effect of the BTOP programs on home broadband adoption,” the economists say. “The evaluation of BTOP and similar programs has been done before. As for BTOP generally, econometric analysis by Hauge and Prieger (2015) found that the effect of the BTOP “stimulus spending on broadband adoption may well be zero.”

Another study found the “program had no significant impact on broadband adoption rates.” 

That is not to say success, measured as service adoption, is impossible. Comcast’s program for low-income customers seems to have added as many as eight million households. Generally speaking, lower prices tend to spur higher buying rates for any desired product. The issue is that many “non-buyers” of internet access have reasons for not buying that are not strongly affected by price reductions.

“Though using an admittedly crude calculation, Rosston and Wallsten (2019) estimate the own-price elasticity of demand for broadband (for low-income households) to be only about 0.10 to 0.13. A 10 percent reduction in price only increases subscriptions by about one percent,” the authors say. Other studies suggest similar low price elasticity. 

“It does not appear that price is the primary reason for non-adoption,” they conclude. According to U.S. Census Department surveys, most people who do not buy internet access report they do so because they “do not need it.” Perhaps only half a percent of non-buyers say “price” is the reason they do not buy. 

source: Phoenix Center

One might well draw the conclusion that the drive to make such access available is fully consistent with our presumed general positions on equity: everyone should be able to avail themselves of internet access. That is perhaps an obviously different goal from “everybody uses the internet,” which is a choice individuals make. 

It is worth noting that individuals also vary in the means they use to satisfy that need. A significant percentage of consumers consciously choose to buy mobile access rather than fixed access, as their sole form of access. 

And some people, especially some older people, might never decide they “need” the internet. That is a generational issue that fixed itself with time. There was a time when many older people believed they did not need linear video services. Many more might believe they do not need streaming video, either. Those demand profiles change over time. 

The point is that public policy is a devilishly complicated endeavor. Even well-meaning policy goals may miss the mark. Ensuring that everyone has quality fixed network internet access seems a well-established and rational policy aim. 

We might still find that some people do not wish to avail themselves of such services, even when price is not a barrier, especially when substitutes (mobile access) are available.