Now that telephone service is virtually ubiquitous, the new worry is that not all households are using these connections in a digital mode—that is, to connect to the Internet. Given the Internet’s recent vintage and, until the past year or two, the relatively high cost of personal computers, it should surprise no one that those who first embraced both were society’s more affluent members. Nevertheless, the Department of Commerce, the Federal Communications Commission, and outside lobbying groups such as Consumers Union are concerned that many households may be on the wrong side of a “digital divide.”
Incontrovertibly, poor people and minorities in the United States are less likely to own computers and be connected to the Internet than are middle- and upper-income citizens in general and whites in particular. What should we do? Policymakers have made several proposals. One—which died in the just-ended 106th Congress—would have extended the low-income earned income tax credit by up to $500 for 50 percent of the cost of “computer technology or equipment” used to connect to the Internet. A new subsidy program might prove politically attractive, especially to its recipients, but is there any real need for such a program? After all, if low-income households eschew personal computers and Internet connections, are they not telling us that they have more pressing needs?
Measuring the Digital Divide
For nearly two decades the Census Bureau has used its Current Population Survey to measure the extent to which households have telephone service. Telephone “penetration” has increased steadily across the population, but a 6-8 percentage point gap remains between telephone subscription in households with a white head and those headed by a black or Hispanic. Much of that gap may be attributed to income differences—all racial groups in households with annual incomes of $75,000 or more have essentially the same rate of telephone subscription. But low-income black and Hispanic households are significantly less likely (8-12 percent at annual incomes of less than $15,000) to have a telephone at home than are white households of similar income.
These disparities cannot be attributed to the high price of a residential connection to the telephone network. Regulators have kept rates for local residential service below its long-run incremental cost for decades, and direct subsidies are available to low-income households to reduce the cost further. For various reasons, including difficulty controlling long-distance spending, some poor and minority households find the benefits of a telephone in their residences—as opposed to access to a nearby pay phone or, now, a cellular phone—less compelling than do more affluent, white households. About 20 percent of very low income black and Hispanic households still feel that a wired telephone connection in their homes is simply not worth the roughly $13 a month charge, after federal and state subsidies, for unlimited local service.
The “analogue divide” in current voice telephone service is narrowing, however, largely because of the strong income growth of the 1990s. About 95 percent of black and Hispanic households with annual incomes of more than $15,000 now have a wired telephone in their homes. But greater affluence has made life more complicated. In addition to the telephones, refrigerators, television sets, washing machines, and stereo equipment that have become ubiquitous in American homes, comes now the personal computer. The PC is capable of performing many tasks, some that are very attractive to households. It can assist in financial management and planning, provide access to stored information, play recorded music, and connect to e-mail and other services over the Internet.
At first a luxury to which only a few aspired, the household personal computer is now much more widely affordable. Today, a rudimentary Pentium-based machine with a modem and a standard 15-inch monitor can be purchased for as little as $400-about the price of a large color television set. Internet connections can be had for $15 a month, or even for free if the subscriber does not mind being deluged by advertisements.
Still, this downward trend in prices does not guarantee the ubiquity of household PCs and household connections to the Internet. As of December 1998, the most recent period for which the Census Bureau has published data, “only” 42.1 percent of U.S. households had a personal computer, and “only” 22.2 percent accessed the Internet from home. A private market research firm, TNS, estimates that these numbers had grown to 61 percent (personal computers or similar devices) and 52 percent (Internet connections) by the second quarter of 2000. But the digital divide was clearly in evidence.
While 46.6 percent of white, non-Hispanic households had a computer, only 23.2 percent of black households and 25.5 percent of Hispanic households had a PC in 1998. Internet use was somewhat lower, but the digital divide exists here too. Among white households, 26.7 percent used the Internet at home, as against only 9.2 percent of black households and 8.7 percent of Hispanic households. Thus, the digital divide is greater for Internet use than for computer ownership even though the Internet is available to everyone for about $15 to $20 a month (see table 1).
What Causes the Divide?
Using Current Population Survey estimates of household computer ownership in 1994, 1997, and 1998 and of Internet use in 1997 and 1998, I have estimated the determinants of household demand for each. Clearly, computer ownership and Internet use depend on income, occupational status, residence (urban or not), household size, and whether a household is male-headed. But after correcting for these variables, I found that computer ownership is only slightly lower-and Internet use no lower-for Hispanic households, while both are significantly lower for black non-Hispanic households.
The Good News
While it is true that low-income households, black households, and to a lesser extent Hispanic households have less contact with the digital society at home, these groups are narrowing the gap over time. A good way to demonstrate that is to estimate the determinants of PC ownership based on the 1994 data from the Current Population Survey, and to predict household computer ownership in 1998 based on 1998 income and demographics. It turns out that the 1994 data underpredict computer ownership for low-income households by 99-224 percent, as compared with only 41 percent for households earning more than $75,000 a year. Similarly, the 1994 data underpredict computer ownership for white households by 82 percent, but by 104 percent for black households. In short, lower computer prices, income growth, and greater interest in computers by poor and nonwhite households have closed a good deal of the divide in just four years.
Should We Worry?
What if the digital divide were not being closed? Given systematic differences in consumption across demographic groups—the “music gap” between generations, the “pickup truck gap” among vehicle owners—should it be a matter of public concern if some groups in our society shun the $400 computer and a $15 a month (or less) connection to the Internet and use their income on something else?
It is possible that “consumption externalities” in computer and Internet use are not being exploited. Plainly speaking, you and I would gain if others connect to the Internet if these others are people that we wish to communicate with and have them reciprocate. But computer and Internet use by households is growing rapidly, suggesting that more and more of any such externalities are being exploited every day. Some households’ failure as yet to join the Internet world may simply reflect a lag in recognizing the benefits of its use and a lack of friends and associates who communicate through it.
It is also possible, even likely, that households on the wrong side of the digital divide would improve their educational, cultural, and professional status if they owned a PC and connected to the Internet. But such improvements are internal to these households and do not call for extraordinary market incentives for them to become connected. Households can choose for themselves.
The Imprudence of a New Subsidy Program
Would a new subsidy program directed at low-income households, minorities, or rural households have much effect? No one knows. We do not know how large a subsidy it would take to increase household computer and Internet use measurably. We do know that any program that made much of a difference could be very expensive. For instance, current low-income telephone subsidies cost about one-third of the local monthly telephone bill, yet their effectiveness is subject to considerable dispute. If we assume that the annual cost of a computer and Internet service is about $380 a year, a program similar to the telephone subsidy would cost about $127 per household opting for the subsidy. Even if only 5-10 million of the country’s approximately 40-50 million households not owning a computer or connecting to the Internet were to opt for the subsidies, the cost would be $635 million to $1.27 billion a year. And the subsidies would likely become a permanent part of the political landscape, whether needed or not, much as the low prices of telephone connections in rural areas have become untouchable.
Even more ill-advised is a nascent move toward trying to assure that all areas of the country have access to the new high-speed, “broadband” connections now being rolled out by telephone and cable companies. These new services require billions in investment to reconfigure existing networks and billions more to connect subscribers. A 1999 analysis by the FCC, Deployment of Advanced Telecommunications Capability: Second Report, laments that such services are not uniformly available throughout the country. Not surprisingly, the cable and telephone companies have initially targeted higher-income, urban areas to cover their capital costs. At this juncture, the deployment of these services is extremely risky because no evidence exists on the demand for them. With only about 4 percent of the nation’s households subscribing to such services, it is surely unwise for a government agency to pressure companies to roll out the services in high-cost and low-income areas before they are sure they have a winning product.
Government policymakers should relax and let the booming economy close the digital divide. To do otherwise risks committing a great deal of taxpayer money to technologies that could well be obsolete in a few years or that many households simply do not want.