United States Department of Labor

and the New Workplace of the 21st Century

ROBERT KOMINSKI, Ph.D., Assistant Division Chief, Population Division, U.S. Census Bureau

The ability for people to work from their home has been facilitated in large part during the last two decades by an enormous rise in the amount and kind of computing technology available to people and businesses. The transformation of the computer from a large, industrial object to a small, relatively inexpensive appliance, has allowed a variety of transformations in social life. Included among these is the capability to use computers in the home as an extension of the workplace. Nevertheless, despite the rapid and widespread growth of home computing technology, sizable differentials still exist in home computing, and, therefore, the ability to utilize this tool for homebased work.

This paper discusses data collected in a variety of Current Population Survey supplements that has allowed the Census Bureau to track home computer diffusion as well as the uses for which computers are employed. These data show that the use of home computers for accomplishing work-related tasks have risen about as quickly as has the general diffusion of the tools in the population itself. However, as with the basic distribution of home computing technology, there do exist differentials in who uses computers to work at home.

The Current Population Survey (CPS) is a large, monthly nationally-representative household-based survey of the noninstitutional population of the United States. Conducted monthly since the early 1940s, the CPS is best known as the official source of the national monthly unemployment rate for the United States and individual states. The design uses a multistage stratified cluster design, to both minimize data collection costs and maximize efficiency in variance. Interviews are conducted with all members of the household, or a designated proxy. The collected sample data is then weighted using independent estimates of the population, disaggregated along some of the same general aspects of stratification (e.g., race, age and geography) used in the sampling design. In this way, the CPS is able to provide weighted population estimates of a wide variety of social, demographic and economic phenomena for the nation and some subnational geography, as well as for some selected demographic subpopulations.

Over the years, the CPS has come to be known as much for the supplemental data collections that it conducts, as for the monthly labor force data collection. Each March, for example, sees the fielding of the Annual Demographic Supplement, which focuses on the earnings, income, assets and selected income transfer program activities of individuals in the sampled households. It is from these data that the yearly national estimates on the poverty population are prepared.

Other months are used to sponsor different supplemental data collections. Since the late 1940s the month of October has routinely been used by the Department of Education and, in recent times, the National Center for Education Statistics (NCES), to field a supplement on the school enrollment of the population. This supplement, jointly sponsored with the Census Bureau, collects a series of data items that allow us to identify all persons enrolled in school, their level of enrollment, and a variety of other school-related questions. Some of these latter items change from year to year, depending on the specific data needs of the education community.

In the early 1980s, NCES approached the Census Bureau about the feasibility of collecting information on computers and especially, home computers, as a way of measuring one of the skill capacities of the population. At that time, home - or personal - computers were just beginning to emerge as a viable product. The Apple computer had first appeared in the late 1970s, and by the early 1980s a variety of companies had started producing small computers designed for home use. We designed a CPS supplement that was focused on the concepts of access and use of computers in general, in three venues - work, home and school. This initial supplement was fielded as part of the October 1984 CPS. Subsequently, we have fielded additional supplements with the same fundamental focus, as parts of the October CPS in 1989, 1993 and 1997.

It should be noted here that the focus of these supplements has been on the general levels of computer access and use in the three venues, with limited attention paid to the kinds and levels of use. The surveys have not been designed to answer questions specific to the issue of telecommuting and home-based work, but are useful in illuminating some of the general trends and underlying patterns in the growth and spread of home computing infrastructure and the ways it is being put to use.

An important strength of the CPS is that it relies on a sample design of all household addresses in the United States. In the context of studying home computers this is not a trivial consideration. While the vast majority of United States households do have telephones, a small but important proportion - about 6% do not. Households such as these are seldom accounted for in conventional telephone-based surveys conducted by marketing organizations, and overlook a segment of the population that is highly likely not to be a part of the home computing population.

One might reasonably argue that the growth in the use of computers at home has been facilitated, at least in part, by the clear fact that computers have in general become a part of everyday life, especially in the workplace. Certainly, in the era prior to the introduction of the personal computer, many workers were employed in positions that required them to use a large (in physical terms) computer, often referred to as a "mainframe." These computers originally used a system of "batch processing," whereby user requests were processed, one at a time. Entry of information or task requests was often accomplished by use of punched cards or instructions coded onto magnetic tape. Later versions of these machines began to focus on multiuser processing, whereby users at different computer "terminals" could simultaneously use the central processing power of the central computer, directly typing in instructions, using a keyboard with visual echoing on a computer screen. While today we take much of the ease of use of our interface with computers for granted, the reality is that until about 1980, most computer interactions were accomplished in fairly tedious and involved ways with machines requiring large amounts of physical space and special heating and cooling needs. The move, first to multiprocessing computers, and then to physically small-scale computers, allowed the workplace to extend computing capacity to many workers.

Data from the four CPS supplements demonstrate that by the time of even our earliest survey, many people reported that they did, in some way, use a computer at work. Table 1, which provides a variety of summary measures for the child and adult population, shows that even as early as the first supplement in 1984, a quarter of the employed population was using a computer at work. By 1997 this had risen to one half of all employed persons. Disaggregating these users by a variety of social and demographic factors, as in Table 2, shows that variation has existed at all points in time. Some of these, such as the racial and ethnic differences, are most likely accounted for by the differential distribution of persons with these characteristics across occupation and educational categories, which are themselves, highly correlated.

Examination of Table 2 shows that the highest levels of computer use in the workplace are by persons who have the highest reported educational attainment as well as those working in managerial and professional occupations. Not coincidentally, the highest level of use across family income categories is also demonstrated in the highest income category. Figure I arrays the levels of use at work for a variety of categories in the various sociodemographic variables examined in detail in Table 2.

Obviously, computer use at work is most prevalent among persons in occupations that we routinely associate with computers - managers and high-skilled technical and professional jobs, generally requiring high levels of education and, on average, paying higher wages. But by 1997 there were also relatively high levels of computer use at work for other groups as well. Over a third of persons with just a high school degree used a computer at work, as did about 70% of those employed in technical and sales jobs. Low levels of computer use at work are in fact reported by only small segments of the population - those working in farming, forestry and fishing occupations, or those who have less than a high school degree. Clearly, by 1997, the computer became a fundamental tool for many workers.

Throughout the 1980s and 1990s, the personal computer took up residence in a large number of homes across the United States. As Table 1 shows, the proportion of households with a computer rose from just 8% in 1984, to 37% by 1997. In population terms, 40% of adults, and 50% of children, had access to a home computer by 1997. A number of factors fueled this growth - dramatic reductions in costs and dramatic increases in usability and applications being two of the primary forces.

Table 3 shows that the distribution of home computers, as with those in the workplace, was not even across all segments of the population. The single strongest dimension of distribution, even in 1997, is the family income or economic resources of the household. While the dollar cost of computers, relative to their capability, has declined massively during the period of study here, the fact remains that the actual dollar cost of a "basic" home computing ensemble is still in the range of $1,000-$2,000. For many households a $ 1,000 investment is not a trivial cost. Even in 1997, only about 1 -in- 10 households with family income of $15,000 or less a year, had a home computer. This compares to households with family income of $75,000 or more a year, three-quarters of which had a home computer.

Figure 2 focuses on extreme variations in home ownership across the categories of a number of sociodemographic factors. Computers were more likely to be found in homes where the householder had higher levels of education, and where the householder was employed. Married-family household with school-age children were also more likely than average have a home computer - an encouraging sign that the presence of a computer is mediated to some extent by the desire of parents to expose their children to this important tool.

Lower levels of home computer ownership are seen in Black and Hispanic households, among householders with low educational attainment levels, the unemployed, and as noted before, households with low family income. Some geographic variation is also evident. While 43% of the households in the West had a home computer, just 33% of those in the South did.

Thus, while home computers have become a part of many households in the United States, there are still a sizable proportion which do not have a computer. (An updated estimate produced from the December 1998 CPS showed a prevalence level of 42.8% based on this trend it is not unreasonable to expect that about half of all U.S. households now have a home computer of some type.) More importantly, it is clear that there are serious sociodemographic differentials in the distribution of home computers that are likely to continue for at least the near future.

Just because a computer is present in the home does not automatically mean that every household member will use it, use it in the same way, or use it at the same level. As we have seen, home computers have a particularly high prevalence in homes with school-age children and where there is disposable income. In the early years of CPS data collection on this topic, we were also able to show that home usage rates were much higher for children in these homes than for adults (75% compared to 53%). By our most recent collection in 1997, the difference in usage rates had considerably decreased. Among children 3-17 who had a computer at home, 82% were reported to use it in some way. For adult with a computer at home, 71% were using it. The overall gap between children and adults is clearly diminishing, even with usage rates for children in schools also on the increase.

Tables 4(A) through 4(D) show the levels of home computer use, given that one is available, for the collection of socio-demographic subgroups we have discussed up to this point, tabulated across the four data collection points. These tables help detail some of the changes in the use of home computers of adults over time, as well as demonstrate some of the patterns that have remained relatively constant. As in the earlier figures, Figure 3 draws attention to categories of widest variation in levels of use of home computers.

While there are a number of factors considered in these tables, only several which are of most significant interest to the broader topic of home work are discussed here. The first pattern of interest is the sizable increase in the proportion of older persons who report they use a home computer. In 1984, about 36% of persons ages 55 or above who had a computer at home reported that they used the computer. By 1997, this level had risen to 57%. In numeric terms we went from about 500,000 older adults using home computers in 1984, to 7.5 million in 1997, a fifteen-fold increase.

The next dimension of interest is gender. In 1984 for both adults and children alike, there was a sizable gender difference in the use of computers, especially in the home. In 1984, 63% of adult males reported using their home computer, but just 43% of women did the same. By 1997, this difference had nearly disappeared - 72% of men and 70% of women reported that they used their home computer. In fact, reported computer usage at work by employed persons has always been higher for women since these data were first collected in 1984 (29% of women and 21% of men in 1984; 57% of women and 44% of men in 1997). Clearly, as the home computer has evolved to a more multipurpose tool, the variety of applications has made it a more useful tool for everyone.

The final dimension of interest in home usage levels concerns the patterns with regard to family income. As we have seen, family income is a strong correlation of the ability to own a home computer. Since income is strongly associated with other factors such as education and occupation, there is reinforced selectivity in who uses a home computer. However, as the data in Table 4 show, once a computer is in the home, there is very little effect of income level on whether or not the computer is used. Usage levels at each point in time are highly similar, although the overall level of use among all groups has been rising over time. In short, once the barrier of the ownership cost is passed, factors such as income have little effect on actual use of the tool.

In summary, the CPS data show what is now commonly accepted knowledge - that the use of home computers is on the upswing for virtually all sectors of the population. As computing machines become more commonplace in the home, people of all ages and groups are likely to find ways to put them to use. Of course, there are many different things that computers can be used for.

One of the strong values of the CPS data on computer ownership and use is high level of consistency maintained in the questions asked in each survey administration. By doing this, we hope to get answers and data which will behave in a consistent fashion. The goal is to avoid changes over time that are simply an artifact of the survey administration procedure, or in this case specifically, the questions asked of respondents. In this regard, the baseline measures provided by the CPS supplements are quite strong. The fundamental questions asked in the supplement access to and use of computers in the domains of home, work and school were posed to respondents in virtually the same way in each of the four administrations.

It is in the finer detail of the questions that variation starts to arise across the supplements. Part of the reason for this is that these more detailed issues have changed over time as computing technology and its uses have evolved. What was an "appropriate" response in 1989 ("bulletin boards") is replaced by something in 1997 that didn't even exist for most people a decade earlier ("Internet").

The CPS data available to measure computers used for "working at home" shows just this sort of variation. In 1984, respondents were offered seven choices of activities for which they used their home computer. One of these was the choice, "job or business related activities." In 1989, the question on home uses had been expanded to seventeen choices, among them of which were the choices, "connect to computer at work/work at home" and "home-based business." By 1993, 24 choices for home use were offered, including, "connect to computer at work"; "home-based business"; and " work at home." By 1997, enough self-control was regained to pare down the list to just 14 responses, including, "connect to computer at work or school" and "work at home." Obviously, the changes over time reflect both the growth in the number of uses, but also the importance of kinds of uses as well.

While the inconsistency in the kinds of response categories do not allow clean indications of the growth of home computers for work uses, the data do show marked increases in the number and percentage of people who are using computers in the home to do their work. The various panels of Tables 5(A) - 5(D) show that in 1984 about 2.9 million people said they used their home computer for job-related work. By 1989, just over 2 million people said they used their home computer to connect to or work at home, while 2 million people also said they used their computer for a home-based business. It is important to note that since respondents could choose as many reasons as they liked, there is undoubtedly some, but not complete, overlap in the 2 million people that chose these two answers.

In 1993, with three possible choices concerning work, the numbers increase dramatically. About 2.6 million people said they used the home computer to connect to work; 1.7 million said they used it for a home-based business; and - 8.6 million - said they used it to "work at home". Again, the likelihood is that some overlap exists among the people who made these choices. Nevertheless, the largest single number 8.6 million -,defines the SMALLEST possible overall number of "homeworkers" (that is, if all persons who chose either of the other two responses also all chose the response, "work at home.") Thus, by 1993 it is not unlikely that over 10 million persons were using a computer at home to do their job at least part of the time.

Finally in 1997, with only two relevant categories, 8.2 million people reported that they used a home computer to log in to school or work, and 19.4 million people said they used a computer to work at home. Again, using the logic of unduplicating these counts, it is not unreasonable that certainly over 20 million, and perhaps something as high as 25 million persons, were using a home computer for work purposes.

The subtables depicted here as Tables 5(A) through 5(D) show the percentage of all adult home computer users - not the total adult population - who said they used the computer at home for the category of examination. So, for example, in Table 5(A) for 1984, about 37% of all adults who used a computer at home said they used it for a job related purpose. Similarly, in Table 5(D), 34% of all persons who said they used a computer at home, said they used it to "work at home." As noted, these percentages are heavily mediated both by the overall numbers of users (which have risen considerably over the period) as well as the number and kinds of uses afforded to them to choose from. Focusing only on the data of the most recent collection, it is clear that using a computer to work at home is a routine part of many persons' lives. Figure 4 helps to illustrate some of the strong levels of variation among different sociodemographic factors. Race differences, while present, are not overwhelmingly large; the gender difference still favors males slightly. The largest levels of variation in the level of home use for work appear across educational levels and occupational categories. Related factors, such as family income, reflect this variation as well.

Much of the CPS data and discussion provided here focuses on a relatively small set of socio-demographic dimensions that help to identify users of computers in the general population, then in the home context, and lastly, in the home context specifically with regard to work. As noted at the beginning of the paper, the CPS computer supplements are not ideally designed to measure the level or complexity of home working or telecommuting in the United States. The data they do provide allows us to gauge some of the areas of inequality, and therefore, concern, to which focus might be brought. Other areas, however, are completely out of the scope of the CPS supplements, and as such, will require other data to be appropriately addressed.

The CPS data do show us that over time, increasingly larger numbers of people are using home computers to do some of their work. While there is not an absence of inequality in this behavior across race and gender groups, the levels are small compared to other dimensions. The factors where the larger variations in levels of home use occur are educational attainment and occupation type. Among persons with a bachelor's degree or more who had a home computer in 1997, 47% reported that they used the computer to work at home. By contrast, less than 15% of persons who had less than a high school degree, and who had a computer at home reported that they used the computer to work at home. Moreover, because we know that ownership of a home computer is itself related to educational attainment, these differences are even more disparate if calculated on the total adult population, as opposed to the computer-owning population, as has been shown here.

Across occupation categories, half of managers and professional workers who had a computer at home reported they used it to work at home. But far smaller proportions of persons in service and laborer occupations are able to report the use of a computer at home to do their Job. To some extent this is understandable. Many managers and technical professionals are doing work which requires perhaps nothing more than a computer, while obviously, persons working in an industrial production setting are probably going to have a difficult time running the factory machinery from their home - at least for the time being. Since the CPS supplements are focused so broadly, they cannot really begin to get at some of the finer social and cultural details that emerge once people are enabled to work in the home. There are numerous issues that quickly come into play: the use of (and boundaries between) work and non-work time;

• the ability to accommodate family and other activities;
  
• economic considerations;
  
• social relationships among workers; and
  
• management control.

Each of these, and probably many others, are topics worthy of study and investigation, and in order to do so, will require dedicated, focused data to inform the analyses. In some cases it may well be that small scale studies will be able to provide useful data, but for some topics large-scale data will be necessary. A thoughtful research agenda can help to determine where resources need to be spent, and where the best return for the investment can occur.

The discussion in this paper relies on a series of analyses of the CPS Computer Use Supplements conducted in October 1984, 1989, 1993 and 1997. Other supplements, with some related data, were conducted in November 1994 and December 1998, for the National Telecommunications and Information Administration (NTIA). All of the associated reports and analyses prepared from these data sources are accessible via the Census Bureau's internet site (www. census. gov). Upon accessing the site, one should use the alphabetic locator to identify the "Computer Use" topic area, then click on this item for full access to the reports and associated tabulations of these data by both Census and the NTIA. Public-use microdata samples for further analyses are available for all of these data sets. The Census Bureau currently is undertaking the development of a continuing yearly CPS supplement on computer and Internet use. This activity involves multiple federal agencies including Census, the National Center for Education Statistics (NCES), the Bureau of Labor Statistics (BLS) and the NTIA. The proposed questionnaire is currently in cognitive testing. The tentative goal is to field the first new yearly supplement in 2001.

Comments by Harriet West, Manager, Telework Resource Center, Metropolitan Washington Council of Governments

The author's paper examines the change in computer access throughout the United States in three areas home, work, and school - over a 13-year period, from 1984 to 1997. The data source is a nationwide telephone household survey and related supplements known as the Current Population Survey (CPS) supplements conducted by the Census Bureau. The survey does not reflect responses of the estimated 6% of U.S. households without telephone service.

The research finding from this paper address three primary areas: growth in home computers and differentials among user groups; the evolution of home computer use; and trends in using home computers for work. Key findings from each of these areas are summarized below.

Based on data from the CPS, the proportion of U.S. households with a computer rose from 8% in 1984 to 37% in 1997. By 1997, 40% of adults and 50% of children had access to a home computer. The two primary factors attributed to the surge in home computer access are the dramatic reductions in equipment costs and the dramatic increases in usability and applications. Although the availability of computers in the home has grown significantly, the distribution of home computers is not even across all segments of the population, with family income or economic resources of the household being the single most important determining factor. In 1997, 10% of households with income of $15,000 or less had a computer while 75% of households with income of $75,000 or more had a home computer. Computers were also more likely to be found in homes where the residents had higher levels of education and where occupants were employed. Lower levels of computer use were identified in Black and Hispanic households and some geographic differences were noted as well with 43% of households in the West having a computer while only 33% of households in the South did.

72% of men and 70% of women reported using their home computer compared to 63% of men and 43% of women who reported doing so in 1984. The data also show that once cost of ownership is overcome, usage is fairly consistent.

In 1984, 2.9 million people reported using their home computer for job-related work. This estimate had grown to between 20 and 25 million by 1997; however, the type of work-at-home use is not clearly identified or defined. In some instance, respondents may be teleworking (i.e., using information technology and telecommunications to replace work-related travel to a traditional work site). In other instance, respondents may be using their home computers to complete work after hours, or on weekends or holidays. Yet in other instances the respondents may be home-based business owners. Regardless of the nature of the work being completed on home computers, the growth rate has been significant. Education level was identified as the single most important indicator of work-related use of home computers. In 1997, 47% of people with a bachelor's degree reported using their home computer for work compared to just 15% with less than a high school degree.

Another important variable, which is fueling the growth of working at home, is the Internet. The latest National Telecommunications and Information Administration's report, Failing Through the Net: Defining the Digital Divide, shows significant gaps in access to computers and the Internet based on race, location, education, and income and that in some cases, these variations are expanding. Following are key findings from this report which is based on December 1998 U.S. Department of Commerce Census Bureau information.

• U.S. households in urban areas with incomes of $75,000 or more are 20 times more likely to have Internet access and 9 times more likely to have a computer than those at the lowest income levels in rural areas.
  
• Gaps for home Internet access between White and Hispanic households and White and Black households have grown 5% since 1997.
  
• Gaps based on income and education levels have grown 25%-29% in the past year.
  

Although this report shows significant and expanding gaps in some cases, the findings also show that as computer prices continue to decrease, more people are able to afford them at home.

The International Telework Association and Council (ITAC), a non-profit organization dedicated to promoting the economic, social and environmental benefits of telework, has been tracking the number of U.S. teleworkers since the early 1990s. ITAC's findings show the number of teleworkers has grown from 8.7 million in 1996 to nearly 23 million in 2000. ITAC projects that nearly 30 million U.S. workers will be teleworking by 2004.1

With nationwide unemployment in the U.S. at a 30-year record low, employers are seeking new ways to attract and retain skilled workers. Teleworking enables employers to significantly expand their geographic reach, either nationally or globally, to gain access to the skilled workers they need. Between 1996 and 1998, the Metropolitan Washington Council of Governments conducted a 21-month Telework Demonstration project with eight Washington-area employers to help them start or expand telework programs. During the course of the project, four of the eight employers retained employees who moved out of the Washington region or hired new employees who lived in distant parts of the country.

Rapid growth in the information technology field has created a glut of unfilled jobs. The Information Technology Association of America (ITAA) projects that nearly half of the 1.6 million new information technology jobs created in 2000 will go unfilled. This represents a shortfall of nearly 850,000 skilled information technology workers in 2000. Teleworking could provide the means to filling some of these positions.

The author's paper shows that there are serious socio-demographic differentials in the distribution of home computers which are likely to continue for the foreseeable future. These findings are supported by other recent research. The predominant factor limiting home computer ownership is household income. However, the use of home computers is on the upswing for nearly all sections of the population and as home computers become more commonplace, people of all ages and groups are finding ways to use them.

Increasingly large numbers of people are using their home computers to do some of their work. This trend is also expected to continue with an estimated 30 million U.S. workers regularly teleworking by the end of 2004. The largest variances in the use of home computers for work are based on education and occupation; whereas, variances based on race and gender are minor.

A strong economy with low unemployment and continued rapid, growth in the information technology field is creating a demand for skilled labor which may not be met by existing venues. This demand provides new employment opportunities for populations being left behind in the digital divide, namely those in inner cities, rural areas, and Indian reservations. With additional resources, these individuals could be provided with the training and access to technology necessary to perform work for employers in other areas through teleworking.

Public and private-sector grant programs provide funding for numerous technology-based programs, especially for under served populations. For example, the U.S. Department of Commerce's Technology Opportunities Program (TOP) has provided $150 million in funding for 456 projects and leveraged an additional $221 million in matching funds. Private-sector organizations like the Gates Foundations provide funding for technology-based initiatives including a new program that focuses solely on Indian reservations. Additional technology-related programs are available through other federal agencies like the U.S. Department of Education's Community Technology Centers Program to increase use of technology in urban and rural areas and economically distressed communities.

While these programs serve a worthy purpose by providing under served populations with increased access to technology, they fall short in actually providing access to new employment opportunities. In April 2000, Senator Paul Wellstone introduced legislation that would fund and establish National Centers for Distance Working which would provide assistance to individuals in rural communities and on Indian reservations with access to employment in distant areas through teleworking.²

The following measures are provided as suggestions for increasing access to employment for under served populations in the U.S. through the use of teleworking while, at the same time, providing employers access to a larger pool of skilled workers in a tight labor market:

• Continue and expand upon public and private sector initiatives that provide under served populations access to computers and technology through grant programs or other incentives.
  
• Offer the private sector with incentives to provide satellite technology for hard to access rural areas and Indian reservations.
  
• Create incentives and provide training for employers so they can and will hire people where unemployment rates are significantly above the national average (inner cities, rural areas, Indian reservations) rather than increasing reliance on offshore labor through HB- 1 visas. Fund and conduct pilot programs to demonstrate how teleworking can be used to provide new employment opportunities for under severed populations. For example, ITAC is collaborating with the Northern Arapaho Business Development Corporation in Wyoming to establish the Wind River Indian Reservation Technology Portal. This facility will provide Indians from the Reservation with training and access to employment with organizations in the Washington metropolitan region through the use of Information technology and telecommunications.
  
• Continue to share information and findings on efforts to bridge the telework divide through organizations like ITAC.
  

¹Based on Telework America 2000 research results issued by the International Telework Association and Council on October 24, 2000.

² Senator Wellstone's bill was originally introduced as S. 2447, Rural Telework Act of 2000.