In discussions of changing computer and Internet use, a common question is whether inequality has been rising or declining. While previous chapters in this report show that inequality remains, this chapter shows that inequality has been declining by the standard measure of inequality used by economists. Just as income inequality declines when incomes grow faster among those with lower incomes, inequality in computer and Internet use declines when use rises faster among those with lower rates of use. Earlier chapters have noted that higher rates of growth in both computer and Internet use have been occurring among those groups with lower rates of use, such as those with lower income, with less education, from racial groups with low rates, or over 60 years of age.
Different rates of computer and Internet use result from such factors as income, education, use at school, and use at work in different occupations. Income still matters because computers and Internet subscriptions still cost a significant amount of money. On the other hand, income becomes less a factor as prices of computers and Internet subscriptions decline. For school-age children, we found substantial differences in home access to computers and the Internet according to household income. When school and library use are taken into account, however, differences in computer and Internet use among children were much smaller. Among adults, higher levels of education are associated both with greater income and with occupations that tend to use computers and the Internet at work. Once again, we found that computers and the Internet were becoming more common in occupations with lower rates of use.
How a Gini Coefficient for Computer and Internet Use is Computed
To analyze the distribution of computer and Internet use, we have adapted the standard methodology for assessing the distribution of income. In the case of income inequality, households are ranked according to their income and a Lorenz curve is drawn (starting with the lowest incomes) to indicate the cumulative income received by the cumulative population up to that point. For example, Figure 9-1 depicts the distribution of U.S. money income in 2000. Since the bottom 40 percent of the population received 12.5 percent of income, the Lorenz curve goes through (0.4, 0.125).
Figure 9-1: Lorenz Curve of
Household Money Income, 2000
The most widely used measure of inequality, the Gini coefficient, derives from the Lorenz curve. A Gini coefficient of 0 means that income is equally distributed among the population, while a value of 1 means essentially one person has all the income while everyone else has none. The Gini coefficient measures the area between the straight line connecting (0,0) and (1,1) and the Lorenz curve connecting those two points, as a proportion of the triangle formed by (0,0), (1,0), and (1,1). The Gini coefficient equals zero in the case of absolute equality because the Lorenz curve would lie along the straight line from (0,0) to (1,1). In a situation of absolute inequality (in which only one person had all income), the Lorenz curve would run from (0,0) along the X axis until virtually (1,0) and then abruptly rise to (1,1).
The standard approach to measuring income inequality differs from our approach to measuring inequality of computer and Internet use in one key aspect. In the case of income distribution, virtually every household has income that is reported at a very specific level. In our case, we divide the population into distinct groups (such as by income, education, or occupation) and compute the number of users within each group. Note also that we cannot measure intensity of use. In most of the calculations in this chapter, a person who occasionally uses the Internet at the library is counted the same as someone using a broadband connection for hours a day.
Figure 9-2 shows the distribution of household computer ownership by family income in 1984 and 2001. In 1984, the lowest 42 percent of households by income accounted for only 12 percent of computer users. By 2001, the lowest income 45 percent of households included about 27 percent of computer users. With lower income people accounting for a much higher share of computer users in 2001, the curve for 2001 “bows out” much less than the curve for 1984. The Lorenz curve for 1984 divides the lower right triangle almost in half, for a Gini efficient that year of .438. By 2001, the area between the curve and diagonal was less than a quarter of the triangle, for a Gini coefficient of .229 in 2001. (By comparison, the Gini coefficient for the distribution of money income among households indicated greater inequality, rising from .415 in 1984 to .460 for the most recent year, 2000.)
Figure 9-2: Lorenz Curve for Households with Computers vs. Income
Figure 9-3: Gini Coefficients for Households with Computers, Selected years
Figure 9-3 traces the descent of the Gini coefficient for household computer ownership beginning in 1984 and continuing through available data points to 2001. The figure shows that, even though significant disparities remain, the distribution of computers among households has moved continuously in the direction of less inequality. Most of the decline occurred in the second half of the period. (By contrast, the Gini coefficient for household money income dispersion went in the other direction, rising from .415 in 1984 to .460 in 2000)
Internet use figures have a shorter history. Even so, whether measured against income, education, family type, or race/Hispanic origin, the distribution of Internet use at home has moved in the direction of lower inequality. Figures 9-4 and 9-5 depict the reduction of inequality between 1998 and 2001 in Internet use at home based on income and education categories. In the case of income groups, the Gini coefficient declined from .361 in 1998 to .254 in 2001. For educational attainment groups, the measures of inequality for the two years were almost identical, falling from .364 to .262.
Figure 9-4: Lorenz Curve - Internet Use at home vs. Family income Figure 9-5: Lorenz Curve - Internet Use at home vs. Education
The Gini coefficient may also be used to calibrate the effects of work and school on inequality. Chapter 6 found that when someone in the household used the Internet at work, there were much smaller disparities in home Internet rates between high and low income households. That lower inequality is reflected by a Gini coefficient of only .083 among households with Internet users at work versus .298 among households in which no one uses the Internet at work.
Chapter 6 also noted that the rates of Internet use varied substantially by occupation, but were rising in some occupations that had been lower in 1998. The Gini measure confirms that, with a reduction of inequality by occupation falling from .374 in 1998 to .303 in 2001.
Similarly, Chapter 5 showed a substantial equalizing effect of school on both computer and Internet use compared to use at home. Among 10 to 17 year olds in 2001, the Gini coefficient for home computer use was .164 among income groups, but was only .026 for home and school computer use combined. In the case of Internet use, the disparity in home use was larger (.217) and the effect of schools was smaller (down to .126).
The measure of inequality for broadband fell from 2000 to 2001, but remains notably higher than the measure for household Internet subscriptions generally. The Gini coefficient for household broadband declined from .395 in August 2000 to .374 in September 2001. The Gini coefficient for household Internet subscriptions overall decreased from .309 to .270 over the same period. Because home broadband service costs substantially more than regular dial-up Internet service, it should come as no surprise that broadband is distributed more towards higher income groups than dial-up service.
These analyses show that substantial changes have occurred since the introduction of both home computers and the Internet when the initial user community tended to be dominated by those who had higher incomes or had them at work or both. The jobs involving computers and the Internet tended to require more education. As a result, inequality based on income and education was substantial. Over time, however, declining prices, increased availability in schools and libraries, and wider applications in many occupations have combined to reduce inequality in both computer and Internet use.
Table 9-1:
Selected Gini Coefficients
|
Population, Age 3 and Above: |
1998 |
2001 |
|
|
|
Use Internet At Home by Income |
.361 |
.254 |
|
|
Use Internet At Home by Householder Education |
.364 |
.262 |
|
|
|
|
|
|
Population, Age 10-17: |
1998 |
2001 |
|
|
|
Use Internet at Home by Family Income |
.367 |
.217 |
|
|
Use Internet Anywhere by Family Income |
.202 |
.115 |
|
|
|
|
|
|
|
At Home, 2001 |
At Home or School,
2001 |
|
|
|
Use Computer by Family Income |
.164 |
.026 |
|
|
Use Internet by Family Income |
.217 |
.126 |
|
|
|
|
|
|
Employed Persons, Age 25 and Over: |
1998 |
2001 |
|
|
|
Use Internet at Work by Occupation |
.374 |
.303 |
|
|
|
|
|
|
|
|
|
|
|
All Households: |
2000 |
2001 |
|
|
|
Internet Connection of Any Type by Family Income |
.309 |
.270 |
|
|
Broadband Connection by Family Income |
.395 |
.374 |
|
|
|
|
|
|
|
Someone in
Household Uses Internet at Work, 2001 |
No one in Household
Uses Internet at Work, 2001 |
|
|
|
Internet Connection by Family Income |
.083 |
.298 |
|
|
|
|
|