Affordable Connectivity Plan Enrollment and Digital Equity Planning

Benton Institute for Broadband & Society

Thursday, June 16, 2022

Digital Beat

Affordable Connectivity Plan Enrollment and Digital Equity Planning

John Horrigan
     Horrigan 

If the federal government’s investments in broadband connectivity are to be effective, different programmatic pieces must work together. Broadband infrastructure funds are necessary to ensuring universal access, but not sufficient to achieve full digital equity. Equitable broadband adoption depends on people having the financial means to maintain service, which the Affordable Connectivity Plan (ACP) facilitates, as well as access to wrap-around digital inclusion services (such as tech support and skills training). Effective coordination between infrastructure and digital equity investments can ensure that people subscribe to new networks that the Broadband Equity, Access, and Deployment (BEAD) program funds.

Examining ACP enrollment can help target resources to places with the most need

ACP enrollment data offers clues as to how well a community is positioned to take advantage of funds to promote digital equity. Abysmal ACP enrollment levels may indicate a capacity deficit; a community may have a dearth of institutions that can make people aware of ACP benefits and aid in enrollment. Strong ACP enrollment invites exploring why. Are particular places doing outreach that might explain high enrollment levels? If so, state policymakers would be wise to consult with digital inclusion advocates in these areas (as BEAD planning requires) and explore whether initiatives in high-enrollment areas might be replicated elsewhere. Understanding the geography of ACP adoption can therefore help states more effectively prioritize resources for digital equity. If, for instance, Digital Equity Act (DEA) funds will provide grants to entities providing digital inclusion services in cities and communities, wouldn’t it help to know which places have the greatest need? Patterns of ACP enrollment help answer that question.

There are three ways to think about ACP enrollment, which together can help policymakers and others understand digital inclusion needs in cities and communities.

  1. Enrollment rates
  2. Growth in enrollment
  3. Enrollment performance in particular places

Enrollment Rates for the Affordable Connectivity Program

The federal government’s subsidization program for home internet subscriptions is just over a year old – the amount of time the Emergency Broadband Benefit (EBB) and its successor program the Affordable Connectivity Plan (ACP) have been in operation. Since EBB’s inception in May 2021, some 11.3 million households have enrolled in either the EBB or ACP (through March of 2022).  ACP’s second year coincides with states beginning the planning process for using infrastructure and digital equity dollars that the National Telecommunications and Information Administration (NTIA) will distribute pursuant to Infrastructure Investment and Jobs Act funds.

Enrollment rates are a fairly straightforward calculation. ACP requires that applicants meet certain eligibility requirements, such as a household’s income being at or below 200 percent of the federal poverty level. Knowing how many households in a city meet that criterion and how many have enrolled means it is possible to calculate the percent of eligible households that have signed up for the ACP benefit. Using 2016-2020 combined American Community Survey data on the share of households living at or below the 200 percent poverty level (roughly 37 million households total in the U.S.) and end-of-March enrollment date (11.3 million households) means 30% of eligible households have enrolled in ACP.[1] As the table below shows, this figure varies looking across different cities. Much – but not all – of this variation has to do with how many people live at or below the 200% poverty line in those cities.

TABLE 1

CITY

March ACP Enrollment

Eligible households

Enrollment as a share of eligible households

Total number of households[2]

Detroit

98,460

184,949

53%

362,963

Cleveland

93,146

185,565

50%

514,066

Baltimore

70,770

147,178

48%

466,473

Philadelphia

113,757

256,635

44%

613,186

Los Angeles (city)

156,348

354,290

44%

905,303

Columbus

71,145

176,508

40%

550,892

Indianapolis

55,419

141,623

39%

381,317

Atlanta

50,729

130,314

39%

433,273

New York

416,128

1,080,294

39%

3,245,280

Washington DC

43,633

114,044

38%

611,533

Las Vegas

85,205

224,631

38%

680,331

San Antonio

81,799

218,027

38%

603,133

Charlotte

38,323

104,544

37%

363,598

Miami-Dade

89,821

256,506

35%

701,289

Phoenix

66,489

195,548

34%

534,412

San Diego

53,041

159,787

33%

624,476

Jacksonville

41,606

126,061

33%

387,850

Chicago

129,692

396,147

33%

1,164,657

Tucson

42,669

134,074

32%

366,557

San Francisco

23,399

74,401

31%

362,141

Dallas

62,168

210,378

30%

551,241

Fort Worth

38,290

136,532

28%

424,612

Portland

28,277

103,147

27%

422,778

Seattle

24,324

89,205

27%

448,151

Denver

52,127

200,188

26%

960,769

Boston

37,819

151,673

25%

565,187

Minneapolis

26,675

112,432

24%

450,804

Houston

95,708

449,632

21%

1,152,020

Austin

18,843

113,862

17%

450,246

Nashville-Davidson

16,189

98,116

16%

363,404

ALL

2,221,999

6,326,288

35%

19,661,942

Detroit’s high enrollment rate tracks with its large share (51%) of households living at or below the 200% poverty threshold. Cleveland has a 50% enrollment rate, but 36% of households there have incomes at or below the 200% poverty threshold. At the other end of the spectrum, Austin’s and Nashville’s low enrollment rates unfold in places where fewer households live at or below the 200% poverty line (25% and 27%, respectively).

Across these four cities, Detroit seems to be performing as expected, while Cleveland is a bit better-than-expected; Austin and Nashville are behind the curve. These figures tell us something about performance, but not a lot about what might be behind it or where in a given city it may be better or worse. Looking at performance permits a more precise understanding of why ACP enrollment rates vary in different places.

Performance

Understanding ACP enrollment performance is a more nuanced topic even if the question is clear: Do some places do a better job at signing up households to ACP than others? To answer that requires disentangling actual enrollment from predicted enrollment. That, in turn, requires a statistical model that predicts how many households enroll in ACP.[3] Such a model uses a number of a place’s characteristics (racial/ethnic make-up, the age of the population, levels of educational attainment, existing patterns of technology adoption, and income[4]) to predict how many households should enroll in ACP. That prediction may vary from the actual number of enrollees. If more households in a particular area have signed up for ACP than the model predicts, then that area is a high performer with respect to ACP enrollment.

In Baltimore, for instance, the model shows that the Broadway East and Johnston Square neighborhoods (in the 21202 and 21213 zip codes, respectively) have enrolled about 25% more households in ACP than predicted. Both zip codes have high rates of ACP enrollment, but nonetheless Broadway East has signed up some 25% more households than predicted. Why? In Baltimore, ReBUILD Metro is a local coalition that has aimed to improve the housing stock in Broadway East and integrate internet access into trying to improve the lives of residents. The initiative tries to leverage the presence of nearby anchor institutions – Johns Hopkins Hospital and a public library branch – in its work. This suggests that investments in social infrastructure may matter.

Another possible difference-maker is the public library. The statistical model shows a correlation between the presence of public libraries in a 5-digit zip code and ACP enrollment. The “library effect” is associated with 6% higher ACP enrollment in 5-digit zip codes with a public library compared to those without.

None of this demonstrates causation. Places with public libraries may have other characteristics that lend themselves to ACP uptake, although it is also true that many public libraries have sought to publicize the ACP program. From a planner’s perspective, however, causation may not be important. Knowing where ACP performance is strong (for whatever the reason) can spark a search for models that hold promise. If such models are found, these promising practices can be adapted to other places that may be underperforming. Findings from performance-based analysis of ACP enrollment can offer more targeted allocation of DEA funds, thereby helping to increase the chance that they pay off.

Comparing performance to enrollment rate shows several consistencies in outcomes across cities, but a number of cases where there is a disconnect between enrollment rate and performance. The high enrollment rates in Detroit and Cleveland track with high-performance indicators. Portland and Denver are two places where enrollment rates trail the average in cities sampled (and the nation at large) but whose performance is above expectations. Yet some cities—such as Jacksonville, Chicago, and Dallas—have ACP enrollment rates that do not differ greatly from the norm, but their performance is significantly below par. On the other side of the coin, there are cities whose enrollment rates are near the average (San Antonio, San Diego, Tucson, and Miami-Dade) but whose performance metrics are very good. Finally, there are a group of cities – Boston, Austin, Minneapolis, Houston, and Nashville – with below-average enrollment and poor performance.

TABLE 2

 

ACP Enrollment, March 2022

ACP Predicted Enrollment

Percentage Difference

Total Number of Households

San Antonio

81,655

61,322

25%

603,133

Cleveland

93,146

71,417

23%

514,066

Miami-Dade

89,821

69,545

23%

701,289

Tucson

42,669

34,495

19%

366,557

Los Angeles (city)

156,348

128,720

18%

905,303

Columbus

71,145

60,630

15%

550,892

San Diego

53,041

45,231

15%

624,476

Detroit

98,460

85,800

13%

362,963

Denver

52,127

45,736

12%

960,769

Portland

28,277

25,194

11%

422,778

Phoenix

66,489

59,567

10%

534,412

Indianapolis

55,419

51,412

7%

381,317

Baltimore

70,771

66,192

6%

466,473

Las Vegas

85,207

79,783

6%

680,331

Philadelphia

113,757

107,315

6%

613,186

Atlanta

50,732

49,904

2%

433,273

Washington

43,634

43,651

0%

611,533

New York

416,128

423,561

-2%

3,245,280

Seattle

24,324

24,766

-2%

448,151

Charlotte

38,323

40,619

-6%

363,598

Jacksonville

41,606

45,943

-10%

387,850

Chicago

129,275

145,035

-12%

1,164,657

Dallas

62,168

70,586

-14%

551,241

San Francisco

23,399

26,844

-15%

362,141

Fort Worth

38,290

45,325

-18%

424,612

Minneapolis

26,675

34,828

-31%

450,804

Boston

37,819

51,252

-36%

565,187

Austin

18,843

28,278

-50%

450,246

Houston

95,708

160,808

-68%

1,152,020

Nashville-Davidson

16,190

31,104

-92%

363,404

ALL

2,221,446

2,214,863

0%

19,661,942

What are the sources of performance variation? As noted, the presence of a public library has a positive, if modest, association with performance. But there are other drivers, each of which digital equity planners should note:

  • African American households are more likely to have enrolled in ACP (even when accounting for income, age, and educational attainment levels).
  • Hispanic and White households are significantly less likely to have enrolled in ACP.
  • Rural households are less likely to have signed up for ACP.[5]

A goal of Infrastructure Investment and Jobs Act broadband investments is to improve digital equity. The findings of inequity in these categories suggest which groups warrant special attention and, importantly, where they are. Note that San Antonio and Miami – areas with a high share of Hispanic households – fell into the category with average enrollment rates but high performance. Given the negative association between the share of Hispanics in an area and ACP uptake, it seems like something in the digital inclusion environment there has overcome the downward pressure the Hispanic variable has on ACP enrollment.

Growth

A final metric to consider is growth. Stakeholders in some cities who might be disappointed by their enrollment rates can potentially take heart if growth in ACP enrollment is strong. A simple comparison of EBB enrollment at the end of 2021 (when approximately 9 million households nationally had enrolled in the program) and March 2022 is helpful. During that time period, there was a 23% increase in enrollment in the federal government’s connectivity subsidy program. As table 2 shows, growth rates varied in the top 30 cities.

Looking at December to March growth does offer comfort to cities with low enrollment rates and poor performance metrics, such as Boston, Nashville, Austin, and Minneapolis. The strong growth rates for Miami, Los Angeles, and (to a lesser extent) San Antonio accompany strong enrollment and performance findings for those cities. In light of the fact that places with high shares of Hispanic households are generally less likely to enroll in ACP, the findings are striking. It is worth exploring if there were specific strategies to reach Hispanic households in those cities. Another takeaway from this table (in combination with Tables 1 and 2) is that the phrase “Houston, we’ve got a problem” clearly applies when it comes to ACP enrollment for that city.

TABLE 3

CITY

EBB, December 2021

ACP, March 2022

Difference

Percent growth

Total Households

Miami-Dade

62,912

89,821

26,909

43%

701,289

Boston

28,045

37,819

9,774

35%

565,187

Portland

21,365

28,277

6,912

32%

422,778

Washington DC

33,261

43,633

10,372

31%

611,533

Nashville-Davidson

12,491

16,189

3,698

30%

363,404

Austin

14,764

18,843

4,079

28%

450,246

Minneapolis

21,138

26,675

5,537

26%

450,804

Atlanta

40,208

50,729

10,521

26%

433,273

Fort Worth

30,444

38,290

7,846

26%

424,612

Los Angeles (city)

124,407

156,348

31,941

26%

905,303

Denver

41,591

52,127

10,536

25%

960,769

San Antonio

65,616

81,799

16,183

25%

603,133

New York

334,977

416,128

81,151

24%

3,245,280

Chicago

104,649

129,692

25,043

24%

1,164,657

San Francisco

18,994

23,399

4,405

23%

362,141

Dallas

50,928

62,168

11,240

22%

551,241

San Diego

43,529

53,041

9,512

22%

624,476

Philadelphia

93,805

113,757

19,952

21%

613,186

Jacksonville

34,327

41,606

7,279

21%

387,850

Columbus

59,428

71,145

11,717

20%

550,892

Phoenix

55,805

66,489

10,684

19%

534,412

Detroit

82,778

98,460

15,682

19%

362,963

Seattle

20,529

24,324

3,795

18%

448,151

Charlotte

32,396

38,323

5,927

18%

363,598

Baltimore

59,931

70,770

10,839

18%

466,473

Las Vegas

72,809

85,205

12,396

17%

680,331

Indianapolis

47,524

55,419

7,895

17%

381,317

Cleveland

80,157

93,146

12,989

16%

514,066

Houston

83,483

95,708

12,225

15%

1,152,020

Tucson

38,297

42,669

4,372

11%

366,557

ALL

1,810,588

2,221,999

411,411

23%

19,661,942

 

A final point about looking at growth rate is how cities that seemed to do well with EBB enrollment in its early months – such as Detroit, Baltimore, and Cleveland – experienced below-average ACP growth between December 2021 and March 2022.

What does this mean for planners?

Perhaps nothing could be a more frustrating outcome from BEAD network investments than a persisting gap between deployment and subscription. Understanding where there are ACP enrollment shortfalls can help address this “if you build it, will they log on?” concern. Looking at the performance of ACP uptake at a granular level can help avoid “networks to nowhere” for state broadband planners:

  1. Examining ACP enrollment in 5-digit zip codes can help target resources to places with the most need, e.g., places within cities where, for whatever reason, households are unaware of ACP.
  2. ACP enrollment can serve as a proxy for local capacity to foster digital inclusion. Understanding where enrollment is over-performing can launch productive inquiry into models that may be effective – and replicable.
  3. Findings on ACP enrollment can help structure community outreach initiatives that BEAD requires. Although focusing on promising practices for digital inclusion in areas with high ACP enrollment is an attractive avenue, community outreach in low-performing ACP areas has merit.

The National Telecommunications and Information Administration has emphasized that a key goal of BEAD investments is digital equity. State planners will need all the tools they can find to work toward that goal – and analysis of ACP performance is one such tool.


John B. Horrigan is a Benton Senior Fellow. He is a national expert on technology adoption, digital inclusion, and evaluating the outcomes and impacts of programs designed to promote communications technology adoption and use. Horrigan served at the Federal Communications Commission as a member of the leadership team for the development of the National Broadband Plan. Additionally, as an Associate Director for Research at the Pew Research Center, he focused on libraries and their impact on communities, as well as technology adoption patterns and open government data. 

Notes

[1] This is likely an underestimate of eligibility, since some households may have multiple eligible people (e.g., when more than one family shares a house or apartment).

[2] The number of households listed for cities comes from consulting a zip code database that identifies zip codes associated with a city. Those boundaries do not usually align with a city’s legal boundaries, meaning the figures for this column vary from totals found in the U.S. Census.

[3] The results on ACP performance are based on an ordinary least squares regression analysis that regress ACP enrollment per household at the 5-digit zip code unit of analysis on the variables noted above. The data comes from the American Community Survey (2016-2020 5-year data), the Public Library Survey from the Institute for Museum and Library Services, and the University of Michigan’s Population Studies Center which has developed a scale of “ruralness” at the 5 digit zip code level.

[4] The model does not include network availability or quality, as there is not reliable data on this at the 5-digit zip code level.

[5] The University of Michigan has devised a scale of whether a zip code is more or less rural, and that is used to derive this finding

 

The Benton Institute for Broadband & Society is a non-profit organization dedicated to ensuring that all people in the U.S. have access to competitive, High-Performance Broadband regardless of where they live or who they are. We believe communication policy - rooted in the values of access, equity, and diversity - has the power to deliver new opportunities and strengthen communities.


© Benton Institute for Broadband & Society 2022. Redistribution of this email publication - both internally and externally - is encouraged if it includes this copyright statement.


For subscribe/unsubscribe info, please email headlinesATbentonDOTorg

Kevin Taglang

Kevin Taglang
Executive Editor, Communications-related Headlines
Benton Institute
for Broadband & Society
1041 Ridge Rd, Unit 214
Wilmette, IL 60091
847-328-3040
headlines AT benton DOT org

Share this edition:

Benton Institute for Broadband & Society Benton Institute for Broadband & Society Benton Institute for Broadband & Society

Benton Institute for Broadband & Society

Broadband Delivers Opportunities and Strengthens Communities


By John Horrigan.