Is cost stopping people from getting online?

Congress has spent at least \$17.34 billion to get people online in the past two years. In late 2020, they funded the Emergency Broadband Benefit Program (EBB) by \$3.14 billion. Then, on December 31, 2021, the Affordable Connectivity Program (ACP) took over. This program is funded to the tune of \$14.2 billion. With all of the money going out, I’ve been wondering if there is any way we can understand how impactful the programs have been on uptake.

The Census Computer Use Supplement, the most reliable data, won’t be released for some time. So, instead I’ve pulled the data for the last couple of years from Leichtman Research Group’s Research Notes. The data is presented below alongside a the CPI for All Urban Consumers (CPI-U) that tracks Internet services and electronic information providers in U.S. city average, all urban consumers, seasonally adjusted. Find more on series CUSR0000SEEE03 at BLS.

The Leichtman Research Group was right to point out in their Q4 2021 Research Note that their “tracking of the top broadband providers in the U.S. finds that the industry had about 2.5 million net subscriber additions in 2018, and 2.5 million again in 2019.” This jumped massively with COVID, such that, “In 2020, the top broadband providers added nearly 4.9 million subscribers, the most in any year since 2008.”

What is most interesting, however, is the table of differences, located below. By comparing the number of new broadband subscribers from one quarter to the number of quarter adds from the year previous, as well as the first year of full data, some trends in subscriber growth seem apparent. In particular, there seems to be a real slow down beginning in Q1 2021 and stretching into Q3 2021, which Leichtman also cites to. Compared to the year previous, Q1 2021 saw a series decline. However, when compared against Q3 2019, for example, the number of new broadband subscribers in Q3 2021 was only off 77,146 subscribers. In other words, trends have flattened.

COVID acted as a massive demand shock. Kids and adults were sent home to study and work, so they had to use the Internet. But the days of just being online for class or work are over. A new equilibrium has emerged, a hybrid system. In reaction, the demand shock has subsided and growth rates dropped to pre-pandemic levels.

Leichtman again, “Given the increasing level of broadband penetration in the U.S. (further addressed in these Research Notes), it should not be surprising that broadband net adds would begin to return to pre-pandemic levels.”

Interestingly enough, the EBB first got up and running in May 2021, just as the trends crested. In October of last year, I dove into the EBB data at CGO and found that the biggest number of new adds to the EBB program came in the first couple of weeks. By June 2021, as seen below, the trend has flattened.

One of the common presumptions in broadband policy is that cost is stopping people from getting online. While cost is a headwind for some, and a solvable headwind at that, I think we should be careful about how much more of this lever can be pulled. Lowering the cost of broadband even further probably isn’t going to budge the needle all that much in getting new people connected.

I’ve got some evidence suggesting that the EBB and now the ACP are mostly acting as low income supplements or vouchers. To be clear, I think a properly funded long term vouched should be supported through the general operating fund. Congress needs to put in place a permanent system.

But a low income supplement is distinct from a cash incentive to get someone new online. As a cash incentive to get new people online, it is probably the case that these programs are getting less effective over time. My regressions aren’t definitive, and I do need to redo parts of the analysis, but they do suggest that the EBB program mostly targets income and not adoption.

Among the unconnected, cost is the second most cited reason that they’re offline. For nearly two decades now, Census Computer and Internet Use survey has tracked why households remain unconnected. Those trends are detailed below and give the most clear picture at the time.

However, the Census data is time lagged, so instead it is often helpful to look through Pew’s surveys on this topic. Much like this previous surveys, Pew said in April 2021 that the unconnected largely don’t want the service, “Fully 71% of non-broadband users say they are not interested in having such a connection at home.” But this is down from 80 percent in the previous survey. In a change of course, some 25 percent could be persuaded to get broadband in 2021, the highest level in years. The persuadable 25 percent.

Empirical research on this topic exits, but is sparse.

In 2012, the FCC launched a broadband pilot program to collect data on what policies might overcome the key broadband adoption barriers. The results show just how difficult it is to target non-adopters. In the end, 8,634 new Internet users got connected because of the project, yet, the agency estimated 74,000 households would opt in. One of the participating companies, XChange, expected 5,000 people to subscribe, but only found 214 takers. Frontier expected 1,500 people to sign up for the offer but got 118 instead.

Results from surveys conducted before and after the programs reveal the difficulty in getting people connected, as economist Scott Walstein first pointed out. Cost was the most cited reason for not being on the Internet before the program, and yet, participants later said the reduction in price wasn’t a driving factor in getting them online.

By and large, in advance of the pilot program, cost was the most cited reason for not being on the Internet. Below that reason is broken down by household income.

And yet, the post-program questionnaire flips this reasoning on its head. When asked afterwards, what is the reason for getting onto the Internet, the reduction in price doesn’t top the list. Rather, keeping in touch with family and friends, and getting medical information are more commonly cited. While there are some problems with the wording of the question that could have reduced response rates, each of the income levels surveyed show marked decreases.

If cost were truly were the barrier, this number should have gone up not down. The difference between the pre-program and post-program surveys seem to illuminate a problem that has seen intense study in recent decades under the banner of behavioral economics. How would you value a good or service that you haven’t consumed yet?

Price isn’t the same as value, but oftentimes it is treated as such. So when surveys ask those who haven’t gotten onto the Internet the reason for their decision, respondents gravitate towards price because it is the closest proxy to value. Yet, once online, they tend to cite items of value as reasons for being on the Internet, like connections with family and friends, and medical information. Reducing the price might induce some to consume broadband, but it doesn’t fundamentally alter the value proposition like sharing photos and getting medical information does. This should be an important lesson to policymakers. Value, more than just cost, is the reason people get onto the Internet.

Changes in the response to Pew surveys over time hint at this underlying trend. In a 2010 survey, the group found that 31 percent of non-Internet users didn’t subscribe because they weren’t interested in the service, while 10 percent cited cost as the reason for non-adoption. A similar report from Pew in 2009 found that 22 percent of those without Internet weren’t interested in getting online, while 16 percent agreed that cost was the barrier. In 2015, another version of this survey was conducted, but didn’t include an option for non-relevance. This time around, 33 percent of non-broadband users said the expense of the service was the primary cause. When non-relevance isn’t an option, respondents tended to gravitate towards cost.

Comcast’s Internet Essentials (IE) program stands as the longest running private version of a supplemental Internet program and, as such, provides some understanding of how these programs work. Though it is important to remember that IE isn’t like the EBB though because IE subscribers have to pay \$9.95 a month plus tax.

The best work on this program comes from Rosston and Wallsten (2020) which found that that roughly 292,000 households subscribed to broadband service between 2011 and 2015 because of Comcast’s Internet Essentials, out of 500,000 subscribers in the program altogether. As the authors note, “about 66 percent of IE subscribers represent true increases in low-income adoption as a result of the program, with the remaining subscribers being households that switched from a competitor and households that would have subscribed as part of a general upward trend in adoption.”

Continuing they explain that “IE makes computers available for \$150, but we found no difference in the change in low-income computer ownership across cable territories. As a result, it would be hard to conclude that subsidized computers made a difference in broadband subscription despite the visceral appeal of such programs.”

They concluded, “we generally do not know how best to entice the remaining broadband non-adopters to connect to broadband.”

Just to reiterate, the size of the unconnected, those do not access the Internet at all, can vary considerably depending on the source. Pew’s most recent survey from February 2021 suggests about 7 percent of adults in the US aren’t using the Internet. Since there are roughly 255 million adults, by their count, some 18 million people aged 18 and above aren’t using the Internet. Taking out from this group the persuadable 25, Pew data suggests that 4.5 million people could be.

The Census, on the other hand, estimates that 49.9 million people weren’t connected to the Internet by any means in November 2019. Their estimate clocks in at 18.9 percent of the population aged 15 and above, or about about 16 percent of all households.

Looking at this trend over time suggests two takeaways. For one, it is clear that broadband has had a strong adoption curve in a little over twenty years. But the growth rate of new people adopting broadband has slowed considerably. It is basically flat.

All of the available evidence suggests that a slowdown in broadband is underway. So in turn, it means that using lower prices to get people online is going to be much more difficult.