APPENDIX

TRENDLINE STUDY ON ELECTRONIC ACCESS BY HOUSEHOLDS: 1984-1998

A. Introduction

The "trendline study" in this Appendix identifies household trends in telephone, computer, and Internet penetration rates from 1984 through 1998. This historic review relies on data collected by the U.S. Department of Commerce's Bureau of the Census during the watershed period of pro-competition and regulatory reform policies following the breakup of the former Bell System through court decree. Specifically, we have relied on the Census Bureau's Current Population Surveys (CPS) conducted in 1984, 1989, 1994, 1997, and 1998. In a special supplement to each CPS, the Census Bureau compiled the requisite data by surveying 48,000 or more households. The Census Bureau also cross-tabulated the information gathered according to specific variables, such as income, age, race, educational level, and geographic categories (such as urban, rural, and central city, as well as nationally and by region).

The following analysis highlights many of the significant trends in electronic access during the past fifteen years. For each topic, we have summarized some key findings, examined the changing profile of the "most" and "least" connected, and analyzed the impact of specific demographic variables. The accompanying charts provide a wealth of data that we have only begun to tap. These data and charts should provide invaluable information for policymakers and researchers continuing to explore this area.⁽¹⁾

Stable penetration rates. A review of the CPS time series data demonstrates that Americans have maintained consistently high telephone penetration rates between 1984 and 1998. On average, 91.6% of Americans were connected by telephone in 1984. That percentage rose to 94.2% by 1993 but has failed to surpass that level over the last 5 years. In 1998, U.S. penetration stood at 94.1%. During 1994-1998, rural, urban, and central city areas increased slightly, and by 1998 all areas approximated 94% except for central city (92.9%). (Chart A-1)

Narrowing subscribership gap. The most significant change during the 1984-1998 time period is the particularly high rate of growth in telephone subscribership among households that have traditionally been the least connected. Households that were far less likely to own telephones in 1984 (such as those earning less than $10,000 and unemployed households) still lag behind the national average, but are now far more likely to own telephones. (Chart A-2) Low-income households (those earning less than $5,000) experienced a particularly high rate of growth (nearly 10%) between 1984 and 1998. Id.

At the same time, households that were the most connected (e.g., those earning higher incomes, seniors, and employed households) have experienced a slight decline in telephone penetration rates in the last few years. The reasons for this decline are unclear. It may be due, in part, to the growing prevalence and substitution of wireless devices, which were not included in the survey results. At any rate, with the rapid growth in telephone penetration among the least connected and a slight decline among certain connected groups, the gap between those with and without telephones has decreased.

2. Profiles of the Most and Least Connected

Not surprisingly, households earning high incomes continue to be the most connected, and those with low incomes (particularly those earning less than $5,000) have remained the least connected. In 1989, low-income households living in central cities had the lowest telephone subscription rates (72.6%, compared to 73.8% in rural areas and 76.2% in urban areas); in 1998, by contrast, low-income families living in rural areas were the least connected (76.3%, compared to 78.7% in central cities and 79.2% in urban areas). Geographic location has played far less of a role with regard to penetration rates at higher incomes.

3. Variables in Telephone Penetration

• Income directly correlates with the rate of telephone ownership among households, although the gap between the higher and lower income brackets is decreasing. (Chart A-2) While those at the lowest income levels have continued to lag behind higher income households in telephone penetration, the lowest income bracket experienced a significant growth in penetration between 1984 and 1998 (from 71.8% in 1984 to 78.7% in 1998), considering that telephony is a mature technology. In the meantime, while penetration rates have remained consistently high across higher income households, there has been a slight decrease in penetration rates among all households earning more than $10,000 in the last several years. This has occurred no matter whether the household is located in a rural, urban, or central city area.
• Race/Origin. Telephone penetration also continues to vary by race/origin. Blacks and Hispanics lagged nearly ten percentage points behind Whites in 1989 (86.5% and 86.4%, compared to 95.9%). (Chart A-3) That pattern has changed little over time; in 1998, Blacks still lagged by more than seven percentage points and Hispanics by more than six points.
• Household type also plays a significant role, as single-parent households have continued to trail all other household types. (Chart A-5) In 1997, their telephone rates declined to their lowest recent point (87.1% for male-headed households, and 86.3% for female-headed households). Connectivity for these two groups has improved in the last year, although not to 1994 levels.
• Age has also traditionally been a factor in the rate of telephone penetration, with the 55+ households constituting the most connected group in the last fourteen years. (Chart A-6) Nevertheless, while those householders below 25 years of age remain the least connected, telephone penetration for this group has increased significantly (from 76.1% in 1984 to 87.6% in 1998). As a result, the 19.4 percentage point gap between the oldest and youngest age groups in 1984 narrowed to an 8 percentage point gap in 1998.

At the same time, age and employment status are beginning to become less significant variables, particularly as seniors and those "not in the labor force" buy computers with increasing frequency. In 1984, 2.5% of households 55 and older owned personal computers (PCs), compared to 15.5% of 35-44 year-olds. In 1998, one-quarter of seniors (25.0 %) owned PCs. Whether a family has children is also becoming a less significant determinant of whether the household owns a computer. (Chart A-13) In 1984, non-family households were particularly unlikely to own PCs (3.7%), followed by family households without children (5.1%). In 1998, these two categories of households are still less likely to own computers than households with children, but nevertheless now buy computers at a far higher frequency (27.5%, and 43.2%, respectively).

Widening Ownership Divide. The rate of growth has also had a more significant impact on some groups than others. Those that were most likely to own PCs in 1984, are now especially likely to own them in 1998, even though they may have experienced a lower rate of growth than other groups. For example, for the highest income group (those earning $75,000 and above), ownership has grown nearly fourfold (from 22.1% to 79.9%). While the growth rate for the lowest income group (those under $5,000) was nearly tenfold during the same period, only one of six households at this income bracket owned computers in 1998. (Chart A-8)

The trend of seeing the "computer-rich get richer" means that the digital divide among groups is widening over time. The twenty percentage point difference that existed between the highest and lowest income levels in 1984 has now expanded to a 64 percentage point difference. What was a fifteen percentage point gap in 1984 between those with a college degree and those with elementary education is now nearly a 61 percentage point gap. These trends will continue to occur until the relative growth rates among the least connected significantly surpass the growth rates for the more connected on a sustained basis.

2. Profiles of the Most and Least Connected

Certain households have continued to own PCs at higher rates: those earning higher incomes, those with a college degree or higher, households consisting of married couples with children, households located in the West region of the country, and those that are employed. One significant change has occurred based on race/ethnic origin. In 1984, White households accounted for the highest computer penetration rate; in 1989, they were surpassed by "other non-Hispanic" (e.g., Asians, American Indians, and Eskimos) households, which held an even more significant lead in 1998. (Chart A-11)

In examining the "least connected," households earning lower incomes, those with lower education levels, those located in the South, and those under the age of 25 have consistently had lower computer ownership rates. In particular, households earning low incomes and living in rural areas have repeatedly reported the lowest penetration rates. Rural black households have also remained the least likely group to own a PC (2.7% in 1984, 17.9% in 1998), followed by Hispanics living in central cities (3.1% in 1984, 21.4% in 1998).

3. Variables in Computer Penetration

• Income continues to strongly influence computer ownership rates. (Chart A-8) In 1984, households earning $75,000 and higher were far more likely -- by twenty percentage points -- to own a PC, than households earning less than $5,000. Since 1984, the gap in PC-ownership rates has continued to widen. Nearly 80% of households earning above $75,000 owned a PC in 1998 -- sixty-four percentage points above (or five times more than) those at the lowest income level. Low-income households in rural areas have remained the least likely to own a computer. (Chart A-9)
• Race/Origin also remains closely correlated with computer ownership. (Chart A-10) In 1984, White households owned nearly twice the number of PCs as Black and Hispanic households. "Other non Hispanic households" trailed White households, on the other hand, by only 0.4 percentage points. Between 1984 and 1998, White households' penetration rates increased approximately fivefold, and all other race/ethnic groups experienced approximately a sixfold increase. Because of their similar growth rates, White households continued to own computers at a rate roughly twice that of Black and Hispanic households in 1998. Beginning in 1989, however, "other non Hispanic" households began to exceed all groups in PC ownership. Location has also influenced the rate of PC ownership among race/ethnic groups. (Chart A-11) Black and "other non Hispanic" households in rural areas have remained far less likely than those in urban areas or center cities to own a computer, while Hispanic households in central cities have lagged behind those in rural and urban areas.
• Education remains closely correlated with computer ownership. (Chart A-12) Indeed, between 1984 and 1998, the gap between households with an elementary education and those with a college degree or higher has increased significantly. In 1984, the gap between these two groups was approximately fifteen percentage points; in 1998, the gap rose to approximately sixty-one percentage points.
• Household type also plays a significant role. Married couples w/children have remained the most connected group, particularly in urban areas. (Charts A-13) Non-family households have remained the least connected group, especially those in rural areas. One significant change is the rapid increase (more than eight-fold) in computer ownership among "family households without children." In 1984, that was the second least connected group; in 1998, it was the second most connected group. Also significant is the widening disparity between male and female-headed households. In 1984, there was little difference between the two (6.7% for female-headed versus 6.9% for male-headed). In 1989, however, PC penetration rates among male-headed households began to soar, creating a significant gap with female-headed households (31.7% for female-headed versus 35.0% for male-headed). This gap has only begun to narrow in recent years.
• Age is becoming less determinative of computer ownership. (Chart A-14) In 1984, the 35-44 year old group was significantly ahead of other age groups. In 1998, it was only marginally ahead of the 45-54 year old group and slightly ahead of the 25-34 year group. Most significantly, seniors appear to be catching up, due to a ten-fold increase in PC ownership between 1984 and 1998. In 1984, there were approximately six times as many with PCs in the 35-44 year category as in the 55+ category (15.5% compared to 2.5%); in 1998, that ratio dropped to a little more than double (54.9% versus 25.8%). Households under 25 are also gaining ground: in 1984, the 35-44 year group was three times as likely to own a PC as those under 25, but was about 1.7 times as likely in 1998. Those households under 25 living in rural areas are still the least likely, however, to own a PC.
• Region. The West has remained significantly ahead of other regions in computer ownership from 1984 to 1998. (Chart A-15) Significantly, the Northeast was the second highest connected region in 1984 (8.5%); but in 1997, the Midwest took second place (42.9%. Meantime, the South (particularly the rural areas) has lagged far behind other areas in PC penetration rates. (Chart A-16)

The data used in this discussion pertain to modem ownership and e-mail access among households. Until 1998, modem ownership was measured as a means of determining the level of Internet access. That practice ceased in 1998 because nearly all computers contain modems today and because modems, in practice, are not always used to connect to the Internet. Nevertheless, for historical purposes, modem ownership serves as the best proxy available for measuring Internet access.

1. Modems ⁽²⁾

     a. Highlights

Growing information access and exchange. As gauged by household ownership of modems, Americans have dramatically increased their ability to access the Internet. In 1989, only 3.3% of the nation's households possessed modems; by 1997, the figure had rocketed to 26.3%, an eightfold increase. Viewing Internet access through various demographic perspectives provides a similar picture of tremendous growth in modem ownership.

Widening digital divide. In general, underserved groups (such as low-income users) and rural areas have fallen further behind the modem ownership leaders in their respective categories in recent years.

     b. Profiles of the Most and Least Connected

The CPS data affords some illuminating profiles of the Americans who are the most and least connected.

• Those Americans enjoying the greatest connectivity today are typically high-income households. Holding income constant, other highly-connected groups include Whites or Asians, middle-aged, highly-educated, employed, and/or married couples with children, most often found in urban areas and the West. Conversely, the least connected generally are low-income, Black, Hispanic, or Native American,⁽³⁾ senior in age, not employed, single-parent (especially female-headed) households, those with little education, and those residing in central cities or especially rural areas. These profiles generally prevailed during 1989-97 albeit some changes occurred (e.g., the South fell into last place among regions).
• Income. From 1989 through 1997, modem ownership increased for all income levels. Penetration rose tenfold in income brackets below $20,000 and increased at a decreasing rate in the higher income brackets, registering growth of 4.2 times at the $75,000 and over bracket. Despite greater growth rates by the lower income households, the percentage point gap between lowest and highest penetration (for the $5,000-9,999 group versus the wealthiest households) grew from 13.4 percentage points in 1989 to 56.5 percentage points in 1997. During this period, rural areas have generally experienced greater growth than central city and especially urban areas, but generally still trail the other two.
• Race/origin. During 1989-97, household modem penetration rose in every category of race/origin. White non Hispanic, Black non Hispanic, "Other non Hispanic," and Hispanic each grew eightfold or more. Because White and "Other non Hispanic" households started from a higher proportion, the digital gap has widened considerably compared to Blacks and Hispanics. For example, the frontrunner "Other non Hispanic" group (e.g., American Indians, Aleuts, Eskimos, Asians, Pacific Islanders) outdistanced Blacks and Hispanics by more than 22 percentage points in 1997, compared to 2.0 and 2.22 in 1989. That pattern generally holds whether rural, urban, or central city, although White households have the highest penetration in rural areas (24.6%).
• Education. During the eight-year period (from 1989 to 1997), the digital gap mushroomed to more than a fivefold increase (from a 8.6 to a 46.3 percent point difference) between those households of the lowest and the highest educational levels. This result can be explained largely in terms of the very low penetration rates exhibited by the less-educated households in 1989. This pattern generally holds in rural, urban, or central city areas, with the largest disparity in rural environs.
• Household type. The ownership of modems by all types of households grew substantially during 1989-97, registering sevenfold gains or more. As in 1989, households comprised of a married couple with children eighteen years old or younger led all other categories (4.9% in 1989, 42.5% in 1997). Single-parent households with children lagged considerably; female households with children trailed all others throughout the period (1.0% in 1989, 15.4% in 1997) but grew fifteenfold over the span -- faster than any other category. The digital gap expanded from 3.9 percentage points to 27.1. Both types of single-family rural households with children registered only a 0.5% modem penetration in 1989, but rocketed more than thirtyfold by 1997.
• Age. Modem ownership in each age bracket grew approximately sevenfold or more. Middle-aged householders (35-54 years) led all other categories, equaling more than 35.0% in 1997. Senior citizens exhibited the lowest penetration throughout the period, registering 13.2% in 1997. However, the seniors' elevenfold growth rate in modem ownership exceeded all other brackets. Between 1989-1997, the digital gap in terms of percentage points increased by roughly sixfold between the two groups. Urban middle-aged householders possessed the highest ownership rate (36.0%+), while rural seniors had the lowest penetration (11.2%) but the greatest growth rate (almost 13.0% from 1989-97).
• Employment. Modem ownership rose more than seven times for both the employed and unemployed, and more than 14 times for the many-faceted not-in-labor-force category. In 1997, the highest penetration occurred among the urban employed (34.5%), while the lowest gauge belonged to the not-in-labor-force category in rural areas (9.0%). From 1989 to 1997, the digital gap increased from a 3.6 to a 21.7 percentage point differential. The greatest growth over the period 1989-97 occurred in rural America for the employed and unemployed, and in urban areas for the not-in-labor-force.
• Region. The West exhibited the highest modem penetration in both 1989 (4.5%) and 1997 (30.8%), and no region experienced less than a sevenfold increase. The digital gap grew modestly, from 1.9 percentage points to 6.4. Whether in areas that were rural, urban, or central city, the Midwest grew more than any other region, bumping the South into last place. Rural areas frequently experienced greater growth rates than their urban or central city counterparts but often fell further behind in percentage-point differentials in urban-rural comparisons.

2. E-mail Use

     a. Highlights

E-mail usage also grew at a tremendous rate during the shorter, more recent interval 1994-98. Usage by all races or ethnic origins grew at least 3.5 times during the span. Every income and educational level as well as labor force category exhibited growth of four times or more. All age groups have increased by at least 4.5 times, while usage by household types and regions rose some fivefold.

Widening digital divide. The digital gap also widened in all major categories with respect to e-mail usage during 1994-98. With the exception of regional use (where the West's lead more than doubled), such usage rose by fourfold (or more) regarding income, race/origin, age, educational level, household type, and labor force. Interestingly, the gap actually declined from the 1997-98 period with respect to income, but grew substantially over the longer period 1994-98.

     b. Profiles of the Most and Least Connected

The profiles for e-mail usage are similar to those of modem users. Some notable changes between 1994-1998 include surges by certain rural areas. In the Northeast, rural households led all others in the region and all other rural areas across regions. Additionally, female households with children in rural areas overtook those types of households in urban America or central cities.

• Income. E-mail usage increased at all income levels by fourfold or more during 1994-98. Although usage rose more than four times to 6.2% at the lowest-income level (under $5,000) by 1998, the digital gap grew from 8.8 percentage points to 37.5. (Chart A-18) Unexpectedly, both the lowest-income and the highest income levels experienced declines in usage from 1997 to 1998 (i.e., 7.2% to 6.2%, and 45.1% to 43.7%, respectively). At each income level, rural areas lagged behind urban and central city areas.
• Race/Origin. All groups registered growth of 3.5 times or more. Black and Hispanic household usage remained substantially behind Whites and non Hispanics in both 1994 and 1998. The digital gap more than quintupled during the period. (Chart A-19) The Whites' category usage of e-mail led in rural and urban areas, and also in central cities in 1998. (Chart A-20)
• Age. E-mail usage grew at least 4.5 times at each age level during the period. Senior citizens (55+ years) trailed all other age brackets in both 1994 and 1998. The digital gap more than quadrupled, from 3.5 percentage points to 14.6. (Chart A-23) Although exhibiting some of the highest growth, rural areas consistently lagged behind other areas in each age category.
• Education. E-mail use and level of education are also correlated. All levels experienced at least a fourfold growth rate in usage. The digital gap between elementary school-educated and college-educated households grew from 8.7 percentage points in 1994 to 37.5 in 1998. (Chart A-21) Those households in rural areas have consistently lagged behind urban areas but surpass central cities, except at the lowest and highest age brackets.
• Household type. Every type of household increased their e-mail use by at least a factor of five except for male householders with children (three times). Usage by households composed of married couples with children led all other categories throughout the period by a substantial margin and equaled 25.9% in 1998. The latter figure decreased from that registered in 1997, 26.2%. Least usage occurred with female householders with children over the same time, registering 10.1% in 1998. The digital gap widened during the span, increasing from 3.9 percentage points in 1994 to 15.8 in 1998. (Chart A-22) Surprisingly, the gap decreased from 1997 to 1998 by 4.2%. Throughout the 1994-98 span, rural area usage consistently trailed other areas for all household types except for the female householder with children category, where over time rural overtook both urban and central cities.
• Region. E-mail usage experienced fivefold growth or more during the period in every region. The West has consistently led all other regions during the four-year period. The digital gap grew somewhat, increasing from 2.3 percentage points to 5.9. (Chart A-24) The West has led in urban and central city areas, but the Northeast led in rural America. Except for the Northeast, rural areas generally trailed other areas.