Why do people not subscribe to cable television? A ... - CiteSeerX

7 downloads 0 Views 62KB Size Report
cable non-subscribers on broadcast television stations for their video .... Abstract of the United States prepared by the U.S. Bureau of the Census. ... 10 Interestingly, Mueller and Schement (1995) provide evidence that low income ..... Rubinovitz, R. N., 1993, Market power and price increases for basic cable service since.
Why do people not subscribe to cable television? A Review of the Evidence*

Robert Kieschnick Federal Communications Commission B. D. McCullough Federal Communications Commission

Current Draft: September, 1998

*The authors wish to thank Gary Biglaiser, Jerry Duvall, Judy Herman, Jeff Prisbrey, Mike Riordan, and Doug Webbink for comments on prior versions of the paper. The authors especially wish to thank Jim Eisner for preparing column 3 of Table 2. The opinions expressed are solely those of the authors and not of the FCC.

Why do people not subscribe to cable television? A Review of the Evidence

Abstract We review the literature on the demand for cable television and on the characteristics of cable subscribers and non-subscribers to address two issues. First, what is the influence of household income on household use of cable television. Second, how do cable subscribers and non-subscribers differ. On the first issue, the evidence examined suggests that while household income is an influence on a household’s decision to subscribe to cable television, it is not a significant influence. Rather, household income is a more important determinant of household purchases of premium cable video programming services. On the second issue, the evidence suggests that characteristics of cable subscribers and non-subscribers have changed significantly over time. Quite some time ago households subscribed to cable to obtain clearer broadcast signals. More recently, households are subscribing to cable to obtain video programming uniquely available on cable. Further, cable subscribers increasingly differ from non-subscribers in their TV viewing habits and preferences.

Why do people not subscribe to cable television? A Review of the Evidence

1. Introduction

The question of why people do not subscribe to cable television arises up in a variety of policy issues. The reasons offered are often used to justify special treatment of the terrestrial broadcast television industry. Three examples illustrate the varied nature of the justification. First, the Report & Order in MM Dkt. No. 93-48 (Policies and Rules Concerning Children’s Television Programming) argues that the Children’s Television Act focuses upon the provision of children’s programming on broadcast television stations because such “may be the only source of video programming for some families that cannot afford, or do not have access to, cable or other subscription services.”1 Second, the Further Notice of Proposed Ruling Making (MM Docket No. 91-221) raised the issue of whether or not cable television should be counted in the competition and diversity analyses of the Federal Communications Commission’s local television ownership rules. The expressed concern was many households who do not subscribe to cable, do so because they cannot afford cable, and so are dependent upon broadcast television for their video programming.2 Third, the Supreme Court in Turner Broadcasting System, Inc. versus The Federal Communication Commission states: “Must-carry is intended not to guarantee the financial health of all broadcasters, but to ensure a base number of broadcasters survive to provide service to noncable households.”3 Repeatedly the Supreme Court talks about the dependency of cable non-subscribers on broadcast television stations for their video programming. 1

Report & Order (FCC 96-335) at page 22.

2

Further Notice of Proposed Rulemaking (FCC 94-322) at pages 50-51.

3

Turner Broadcasting System, Inc. versus The Federal Communications Commission, 117 S. Ct. 1174 (1997) at page 1202.

1

We see in these different contexts two primary issues. First, do many households not subscribe to cable television because they can not afford the service? Second, what other reasons are there for households not to subscribe to cable television? The purpose of this paper is to review available evidence on these two issues. We organize our discussion as follows. First, we examine data on cable expenditures and cable subscription by household income level to see what these data suggest about the influence of household income on cable subscription and use. Next, we summarize the literature on the demand for cable television and on the differences between cable subscribers and non-subscribers. We do this because our first analysis ignores the effect of other relevant factors on cable use (e.g., price), and because we want to develop a more detailed characterization of the differences between cable subscribers and non-subscribers. Finally, we summarize the implications of the evidence for the income-based argument for special regulatory treatment of terrestrial broadcast television. Before we proceed, we need to define a few terms carefully. In the literature, the phrase "cable access" has been used two ways. It can refer to whether or not a cable system passes a given household or it can refer to whether or not a household subscribes to a cable system and therefore has access to some of its programming. In the telephony literature, access payments by consumers refer to a consumer's payment to become hooked up to a telephone system. Such payment is typically included in the price for basic cable service. However, unlike a telephone subscriber, a cable subscriber does not make additional payments for viewing video programming distributed under the basic service plan.4 Consequently, we will use the phrase cable availability to refer to whether or not a cable system passes a consumer's household. We will use the phrase cable subscription to refer to whether or not a consumer has subscribed to basic cable services, and we will use the phrase cable use or usage to refer a household's expenditures on whatever bundle of cable services they choose to purchase.

4

There are a number of different schemes for pricing a call that a consumer places through a telephone system. Some price each call, some price local calls above some limit, and so forth. Thus, bundling of access and a basic service level is not a uniform practice in telephony.

2

2. Univariate Statistical Evidence

The U.S. Department of Labor collects data on household expenditures on various products and services annually through its Integrated Consumer Expenditure Survey. In Table 1, we report some data that was derived from this survey for 1995 by the income bracket of the responding consumer unit or household. Generally these data represent average expenditures of consumers in the Survey sample with particular income characteristics. Because these are sample data, we report in the first row of Table 1 the number of responding households in each income group. These data reveal two interesting points. First, as household income increases, expenditures on community and cable TV services increase. However, the percentage change is rather small. Table 1 suggests than a 1% increase in household income is associated with a 0.16% increase in cable expenditures which suggests that cable expenditures are rather insensitive to changes in household income.5 Further, consumer expenditures on cable television appear less sensitive to income changes than a number of other consumer expenditures reported in the Survey. For example, similarly calculated income elasticities for food (0.23), housing (0.24), or clothing (0.32) suggest that consumer expenditures on cable television are less sensitive to changes in household income than many other goods or services that are traditionally thought of as insensitive to household income. Second, households typically spend substantially more on other forms of entertainment than they do on cable television. The national average monthly rate for basic cable service in 1994 was $21.62.6 Thus average expenditures for one year's worth of basic cable service would be approximately $259.44 nationally. So if households earning less than $5,000 spend on average $719 for entertainment, as Table 1 suggests, many of these households could spend $259.44 to

5

This elasticity measure was calculated as the percentage change in expenditures on cable tv divided by the percentage change in their income (mid-point of the $5,000 to $9,999 bracket to the mid-point of the $50,000 to $69,999 bracket). 6

See National Cable Television Association, Cable Television Developments: Spring, 1996, at page 3, citing Paul Kagan Associates, Inc., The Cable TV Financial Databook, 1995.

3

subscribe to basic cable television services. They could do so by simply re-arranging their entertainment expenditures (i.e., by spending less on other forms of entertainment).7 Further, data for these same households suggests that they made discretionary expenditures (e.g., entertainment, alcoholic beverages, etc.) that substantially exceed what they would have had to spend to buy cable television services. Thus, these data suggest that even for households in the lowest income bracket, the decision not to subscribe to cable television is more often the result of a preference than an inability to afford services. One problem with the above data is that cable operators charge one rate for basic services and additional amounts for additional services. Consequently we are unsure whether the average level of dollar expenditure on cable services indicates many people purchasing inexpensive cable service, few people purchasing expensive cable service, or some mix thereof. The figures in Table 1 probably represent consumer expenditures on a mix of basic and other cable television services (e.g, pay channels). Further, some consumers do not pay directly for access to cable services (e.g., renters are sometimes provided basic cable television service as part of their rent). Thus cable subscription is better reflected by cable penetration figures.8 Table 2 shows data on cable penetration for different categories of household income. Column 2 of Table 2 was prepared using data from the Telecommunications Industries Analysis Project (TIAP). TIAP used cable penetration data from Nielsen's People Meter Sample to determine cable penetration by household income groupings delineated in the 1996 Statistical Abstract of the United States prepared by the U.S. Bureau of the Census. Column 3 of Table 2 was prepared using data from the TLC MarketShare Monitor, a joint product of PNR and Associates Inc. and Market Facts Inc.9 TLC MarketShare Monitor represents a panel of 7

It is important to note that for the lowest income group in the Survey, expenditures exceed income. This is due in part to the fact that BLS imputes expenditures, but not income (e.g., food costs for food stamp recipients). Additionally, households using debt or savings to finance expenditures would also have expenditures greater than income. Finally, some of these households are college students and so may have different characteristics from other low income households. 8

Cable penetration represents the number of households subscribing to cable relative to the number of households that could subscribe to cable television because it is available to them. 9

The TLC MarketShare Monitor provides a significant amount of information about surveyed households and their use

4

telephone subscribing households and began in February, 1997. Either Panel A or Panel B of Table 2 suggests that while cable penetration does increase as household income increases, there are a substantial number of households in higher income brackets that do not subscribe to cable television services. The similarity of cable penetration estimates in Panels A and B, despite their use of different samples, gives us confidence in the reasonableness of this conclusion. Because of the different number of households in each of the income brackets it is not clear how many of those households that do not subscribe to cable television services are from each income group. To address this issue, we prepared Table 3 from Table 2, Panel A and data on the number of households in each income grouping presented in TIAP’s Calculations and Sources for Revving up the Communications Economic Engine: Household Services, Monthly Bills, and Barriers to Competition. Table 3 suggests that over 70% of the households not subscribing to cable television services are above the poverty line and Table 1 suggests that these households spend more than $900 on average per annum on entertainment. Thus, the majority of households that do not subscribe to cable television services do so for other reasons than their ability to afford such services. In fact, one can argue that these households are simply choosing a different mix of entertainment activities.10 Taken altogether, the above evidence on household expenditures on cable television, cable penetration by household income bracket, and the breakdown of non-subscribers by household income bracket strongly suggest two points. First, reported statistics on national cable penetration largely reflect the decisions of households who choose not to subscribe to cable television for reasons other than their ability to afford such services.11 Second, household income may be more a determinant of the bundle of cable television services that a household purchases of different communication services on a monthly, quarterly, and annual basis. 10

Interestingly, Mueller and Schement (1995) provide evidence that low income households are more likely to subscribe to cable television than to purchase telephone access. Apparently these households viewed subscription to cable television as an inexpensive source of entertainment that they value more highly than having a phone in the house. 11

Separately, it is interesting to note that a 1985 survey reported in LaRose and Atkins (1988a) suggests that up to 40% of the then cable non-subscribers were former cable subscribers. LaRose and Atkins= research suggests that satisfaction with the service provided by the cable operator is a more important determinant of continued subscribership than household income.

5

rather than whether or not the household subscribes to any cable television service.

3. Multivariate Statistical Evidence

One criticism of the above analyses is that they fail to account for the influence of other factors (e.g., price of cable services) on the use of cable television services and so may misestimate the influence of household income on household cable use. Further, the above analyses do not reveal much about whether there are any other significant differences in the characteristics of cable subscribers and non-subscribers. Consequently we turn to an examination of the published multivariate statistical studies of cable television subscribership. In conducting this examination, we reviewed research in two disparate literatures. Economists have focused upon the demand for cable television. As a consequence, they have focused upon the effect of price on cable subscribership and have tended to ignore certain attitudinal or demographic variables that are central to studies conducted by communication researchers. On the other hand, communication researchers have tended to ignore the role of price in a household's decision to subscribe to cable television. Thus the two literatures tend to tell different stories about why households subscribe to cable television. We provide a summary and critique of the more important papers in these literatures in the Appendix. Based upon this survey, we draw three important conclusions. First, the reasons that people subscribe to cable television have changed over time. Second, household income is a significant influence on the level of use of cable television but not on the subscription decision. Third, there are significant differences in the viewing habits or preferences of cable subscribers and non-subscribers.

3.1 The reasons that households subscribe to cable television have changed over time.

Early studies, such as Park (1972), suggested that households subscribe to cable television primarily to obtain clear reception of terrestrial broadcast television signals, and only secondarily

6

to obtain distant (i.e., outside of their geographic area) terrestrial broadcast television signals. However, as Sparkes and Kang (1986) point out, cable television has diffused throughout the American economy and people's motivation for subscribing to cable television has changed appreciably over time. This point is reinforced by the fact that apparently most current nonsubscribers are former subscribers (Umphrey (1989)), which suggests that consumer satisfaction with a cable operator's service and offerings plays an important role in determining cable subscribership. Recent evidence suggests that households now subscribe to cable television not simply to obtain clear broadcast signals, but also to obtain the unique video programming offerings supplied by cable operators (e.g., Umphrey (1989), Atkins (1992), and Albarran and Umphrey (1994)).

3.2 Household income is not a significant influence on a household's decision to subscribe to cable television.

Early studies, such as Park (1972), suggested that household income was a significant influence on a household's decision on whether to subscribe to cable television. However, as cable television became more widely available and its services and prices changed, the influence of household income on cable television usage has changed. Since Webbink's (1986) study, econometric evidence (e.g., Beil, Dazzio, Ekelund, and Jackson (1993) and Crandall and Furchtgott-Roth (1996)) has tended to suggest that household income is not a significant influence on a household's decision to subscribe to cable television, but it is a significant influence on a household's purchase of additional cable services (e.g., pay per channel video programming services). This is not to say that all studies since 1986 have found no differences in the income of cable subscribers and non-subscribers. Rather that these differences are not significant influences on a household decision on whether to subscribe to cable television once price and other relevant variables are accounted for in the analysis.12 Thus the effect of household income on the use of 12

Most of the research in the communications literature uses discriminant analysis of cable subscribers and nonsubscribers to infer differences in their characteristics. As Albarran and Umphrey (1994) point out, this literature has found income be a significantly negative influence, a significantly positive inference, and no influence on a household

7

cable television has changed since the 1970's. Today it primarly affects what additional cable television services a household chooses to purchase.

3.3 There are significant differences in the viewing preferences of cable subscribers and nonsubscribers.

As suggested above, as cable television services became more widely available and the video programming it offered became more diverse, differences between cable subscribers and non-subscribers have decreased in demographic terms and increased in attitudinal terms. For example, earlier studies (e.g., Park (1972)) suggested that larger households, with children and higher incomes, were the typical cable television subscribers. More recent studies, such as LaRose and Atkin (1988) or Albarran and Umphrey (1994), suggest that typical cable television subscribers are smaller households, with younger heads of the household, and much more ethnically diverse households (e.g., African-Americans constitute a larger proportion of cable subscribers than non-subscribers). Further, these households tend to watch more television and are particularly attracted to the kinds of programming uniquely offered by cable television operators (e.g., all-sports channels, etc.). Thus the data are consistent with the notion that the television viewing habits and preferences of cable subscribers and non-subscribers are different. Cable non-subscribers appear to find the video programming available over the air more than adequate for their viewing preferences.

decision to subscribe to cable television. These conflicting results are due in part to the limited samples studied, the lack of consistency in included variables, and the use of statistical methods that are not robust with respect to deviations from the presumed distributional characteristics of the sample. Thus we have focused upon econometric evidence which uses larger samples, includes variables that are found to be significant in prior studies, and uses statistical methods that are appropriate to the data.

8

4. Summary

The question of why people do not subscribe to cable television comes up in a variety of policy issues. We address this question by focusing on two issues. Do many people not subscribe to cable television because they can not afford such services? What other reasons are there for people not to subscribe to cable television? We find little evidence to support the argument that many of the people who do not subscribe to cable television services do so because they are unable to afford these services. We find that while there is a positive relationship between household income and household expenditures on cable television, this positive relationship is fairly small and largely reflects the purchase of additional cable services, and not the initial decision to subscribe to cable.13 Even households in the lowest income bracket spend substantially more on all entertainment combined than they spend on cable television. A number of these non-subscribing households could purchase basic cable service and still have money left over to spend on other entertainment. Consequently the majority of households that are not subscribing to cable television are doing so for reasons other than affordability. Further, it appears that the diversity of video programming available to non-subscribing households through terrestrial broadcast television is sufficient for their preferences. Such a conclusion is consistent with available evidence on the viewing preferences of cable subscribers and non-subscribers which suggests that cable subscribers tend to watch more television than nonsubscribers and watch a different mix of television programs than non-subscribers. So, at the margin, those households who value more video programming than available over the air are subscribing to cable television.

13

The relatively low income elasticity of cable subscribership is both consistent with and explains why consumers are so irate at cable price changes. Such price changes tend to represent real wealth losses for cable subscribers.

9

Table 1 Selected annual household expenditures by different income brackets

Less than $5,000

$5,000 to $9,999

$10,000 to $14,999

$15,000 to $19,999

$20,000 to $29,999

$30,000 to $39,999

$40,000 to $49,999

$50,000 to $69,999

$70,000 and more

Number of consumer units (in thousands)

4,693

9,827

8,744

7,729

12,658

10,652

8,200

10,3756

10,577

Average expenditures on community or cable tv

113

137

163

176

203

236

251

300

320

0.16

Average expenditures on food ($)

2519

2373

3073

3882

4064

4709

5499

6228

8030

0.23

Average expenditures on housing ($)

5687

5440

6745

7407

8417

10460

11242

14453

21211

0.24

Average expenditures on clothing ($)

679

746

919

1187

1470

1656

2079

2437

3810

0.32

Average expenditures on entertainment ($) Average discretionary expenditures ($)

719

664

889

903

1216

1764

1924

2509

3780

0.39

1032

962

1280

1358

1742

2330

2577

3306

4714

0.34

ei

Source: Bureau of Labor Statistics' 1995 Consumer Expenditure Survey Note (1) Discretionary expenditures represent expenditures on entertainment, alcoholic beverages, and tobacco products and supplies. Note (2): ei represents an estimate of the income elasticity of particular consumer expenditures (i.e. by what percentage would a consumer's expenditures on x increase if their income increased by 1%). It is calculated as the percentage change in expenditures on x divided by the percentage change in their income (mid-point of the $5,000 to $9,999 bracket to the mid-point of the $50,000 to $69,999 bracket). Note (3): One can not add up rows within a column to discern total expenditures by an income group on the implied collection of products or services as some of the rows are subsets of other rows.

10

Table 2 Cable Penetration Estimates for different income brackets

Household Income

TIAP Estimates of the percentage of households with cable TV service*

TLC MarketShare Monitor based estimates of households with cable TV service**

Under $10,000

47.8%

43.3%

$10,000 - $14,999

55.5%

49.6%

$15,000 - $24,999

60.2%

53.9%

$25,000 - $34,999

64.0%

56.0%

$35,000 - $49,999

68.0%

61.6%

$50,000 - $74,999

72.1%

65.4%

$75,000 and over

82.2%

65.5%

All Households

65.0%

57.9%

*Source: Figure 6, ACalculations and Sources for Revving up the Communications Economic Engine: Household Services, Monthly Bills, and Barriers to Competition@, Telecommunications Industries Analysis Project, July 22, 1997. **Source: TLC MarketShare Monitor, PNR and Associates Inc. and MarketFacts Inc. Based upon a sample of 24,368 households.

11

Table 3 The Number and Percentage of Cable non-subscribers by income brackets

Household Income

Number of Household without Cable TV Service (in thousands)

Percentage of Households without Cable TV Service which are in this income group

Under $10,000

6,396.59

18.3

$10,000 - $14,999

3,857.26

11.06

$15,000 - $24,999

6,304.72

18.08

$25,000 - $34,999

5,092.92

14.61

$35,000 - $49,999

5,387.84

15.45

$50,000 - $74,999

4,753.04

13.63

$75,000 and over

2,624.61

7.73

Total Households

34,869.45

100

Source: Column 2 is based upon data in Figures 2 and 6, ACalculations and Sources for Revving up the Communications Economic Engine: Household Services, Monthly Bills, and Barriers to Competition@, Telecommunications Industries Analysis Project, July 22, 1997.

12

References Albarran, A. B. and D. Umphrey, 1994, Marketing Cable and Pay Cable Services: Impact of Ethnicity, Viewing Motivations, and Program Types , Journal of Media Economics 7, pp. 47-58. Atkin, D., 1992, A Profile of Cable Subscribership: The Role of Audience Satisfaction Variables, Telematics and Informatics 9, pp. 53-60. Beil, R. O., P. T. Dazzio Jr., R. B. Ekelund Jr., J. D. Jackson, 1993, Competition and the Price of Municipal Cable Television Services: An Empirical Study, Journal of Regulatory Economics 6, pp. 401-415. Becker, L.B., S. Dunwoody, S. Rafaeli, 1983, Cable's Impact on Use of Other News Media, Journal of Broadcasting 27, pp. 127-140. Collins, J., J. Reagan, and J.D. Abel, 1983, Predicting Cable Subscribership: Local Factors, Journal of Broadcasting 27, pp. 177-183. Crandall, R.W. and H. Furchtgott-Roth, 1996, Cable TV: Regulation or Competition?, Washington, D.C.: The Brookings Institution. Department of Labor, Bureau of Labor Statistics, Consumer Expenditure Survey, 1994. Ducey, R. V., D. M. Krugman, and D. Eckrich, 1983, Predicting Market Segments in the Cable Industry: The Basic and Pay Subscribers, Journal of Broadcasting 27, pp. 155-161. Greenberg, B.S., C. Heeter, D. D'Alessio, and S. Sipes, 1988, "Cable and Noncable Viewing Style Comparisons", Chapter 16 in Cableviewing, edited by C. Heeter and B.S. Greenberg, Norwood, N.J.: Ablex Publishing Company. LaRose, R. and D. Atkin, 1988a, Satisfaction, demographic, and media environment predictors of cable subscription, Journal of Broadcasting and Electronic Media 32, pp. 403-414. LaRose, R. and D. Atkin, 1988b, Understanding Cable Subscribership as Telecommunications Behavior, Telematics and Informatics 5, pp. 377-388. Mayo, J. W. and Y. Otsuka, 1991, Demand, pricing, and regulation: evidence from the cable TV industry, RAND Journal of Economics 22, pp. 396-410. Mackey, C., 1988, "Free Cable Service as an Incentive", Chapter 19 in Cableviewing, edited by C. Heeter and B.S. Greenberg, Norwood, N.J.: Ablex Publishing Company. 13

Metzger, G. D., 1983, Cable Television Audiences, Journal of Advertising Research 23, pp. 4147. Mueller, Milton and Jorge R. Schement, 1995, Universal Service from the Bottom Up: A Profile of Telecommunications Access in Camden, New Jersey, Report, Rutgers University School of Communication, Information, and Library Studies. National Cable Television Association, 1996, Cable Television Developments: Spring, 1996. Nielsen Methodology Research, 1996, NPM Sample Cable Penetration, January 1, 1996. Park, R. E., 1972, Prospects for Cable in the 100 Largest Television Markets, Bell Journal of Economics and Management Science 3, pp. 130-150. Reagan, J., R. Ducey, and J. Bernstein, 1985, Local predictors of basic and pay cable subscribership, Journalism Quarterly 59, pp. 397-400. Roddy, D. J., 1994, Demand and Revenue Elasticities for Cable Television Service, unpublished working paper, Economics and Technology, Inc. Rothe, J.T., M.G. Harvey, and G. C. Michael, 1983, The Impact of Cable Television on Subscriber and Nonsubscriber Behavior, Journal of Advertising Research 23, pp. 15-23. Rubinovitz, R. N., 1993, Market power and price increases for basic cable service since deregulation, RAND Journal of Economics 24, pp. 1-18. Sparkes, V.M., 1983, Public Perception of and Reaction to Multi-Channel Cable Television Service, Journal of Broadcasting 27, pp. 163-175. Sparkes, V.M. and N. Kang, 1986, Public Reactions to Cable Television: Time in the Diffusion Process, Journal of Broadcasting and Electronic Media 30, pp. 213-229. Statistical Research, Inc., 1996, S.M.A.R.T Television Ownership Survey, 1996. Telecommunications Industries Analysis Project, 1997, Calculations and Sources for Revving up the Communications Economic Engine: Household Services, Monthly Bills, and Barriers to Competition, Public Utility Research Center at the University of Florida. Umphrey, D., 1989, A Comparison of Cable Disconnecters and Subscribers, Journalism Quarterly 66, pp. 628-31.

14

U.S. Bureau of the Census, Statistical Abstract of the United States, various years. U.S. Government Printing Office. Webb, G. Kent, 1983, The Economics of Cable Television, Lexington, MA: Lexington Books. Webbink, D., 1986, The Demand for Cable TV and for MDS, unpublished working paper, Federal Trade Commission. Webster, J. G., 1983, The Impact of Cable and Pay Cable Television on Local Station Audiences, Journal of Broadcasting 27, pp. 119-126.

15

Appendix A: Summary of Research on Factors Influencing the Demand for Cable Television Services

Title:Prospects for Cable in the 100 Largest Television Markets Authors:Rolla Edward Park Source:Bell Journal of Economics and Management Science, 1972, 3, 13-150 Research Design: The basic purpose of this study is to forecast the effect of the FCC's allowing cable operators to import distant terrestrial broadcast television station signals by examining the determinants of cable penetrations in markets with clear broadcast signal reception. Park examines 1971 data on 63 cable service areas in which broadcast signals were clear. Specifically, he examines the effect of the number of primary network stations, the number of duplicate network stations, the number of independent stations, the number of educational stations, the number of Canadian stations, the mix of UHF and VHF stations serving the community, average household income, cable service pricing, and the proportion of households with color receivers on cable penetration in these 63 communities. Research Results: Park's primary result was the quality and quantity of a household reception of terrestrial broadcast signals were the critical determinants of a household's use of cable television services. With respect to the influence of household income, Park finds the coefficient on his average household income variable to be significantly positive and he estimates the income elasticity of demand for cable services to be around 0.96. So for his sample, a 1% increase in household income would be associated with a 0.96% increase in cable penetration (% of homes passed that use cable). Research Limitations: Park=s study was a pioneering study, but has limited applicability to current market conditions. For one reason, cable operators now provide a diversity of cable programming that they did not provide in 1971 and so consumers have other reasons to subscribe to cable. Further, Park=s statistical analysis is suspect as he implicitly treats cable penetration as following a log-normal conditional distribution - which it does not.

16

Title:Predicting Cable Subscribership: Local Factors Authors:Janay Collins, Joey Reagan, and John D. Abel Source:Journal of Broadcasting, 1983, 27:2, 177-183 Research Design: This study examines the characteristics of households that subscribe to cable television in a single top 100 market because forecasts of cable penetration (e.g. Park (1972) ) in these markets have exceeded actual penetration. Specifically the authors focus upon Meridian Township, Michigan, a community within the Lansing (market rank number 91) with clear reception of several terrestrial broadcast television signals (all the existing networks and several independents). They conducted a phone survey of 730 households with the Meridan Cable service area to collect their data. They then examined the influence of households characteristics (e.g., television use, radio use, income sex, education, etc.) on the decision to subscribe to cable television services. Research Results: The authors uses discriminant analysis to examine their data and report five significant influences on a household's decision to subscribe to cable television services: television use, persons in household, children in household, home ownership, type of residence, and income. Interestingly, lower income was associated with a higher probability of subscribing to cable television services. The authors explain this contrasting inference to Park's result on the effect of household income as being due to the significantly lessened effect of clear reception of broadcast television signals to cable subscription. They point out that cable systems are increasingly providing additional nonbroadcast channels of video programming and so the determinants of the demand for cable television use are changing. Research Limitations: First, the study uses a sample of households from a single community and so it is difficult to generalize their results to other communities. Second, the study uses discriminant analysis to test the significance of the selected influences, but ignores the non-normality of the observed multivariate distribution under study. Consequently its statistical inferences are suspect.

17

Title:Predicting Market Segments in the Cable Industry: The Basic and Pay Subscribers Authors:Richard V. Ducey, Dean M. Krugman, and Donald Eckrich Source:Journal of Broadcasting, 1983, 27:2, 155-161

Research Design: The purpose of the study is to examine the differences between basic cable service subscribers and pay cable service subscribers in terms of their: demographics characteristics, their access to clear broadcast signals, their use of television, and other attributes (e.g., price of pay cable services). The authors conduct a mail survey of households in four cable system service areas operated by TeleCable Corporation and use data on 1,185 subscribers (662 using basic only, 523 using basic and pay services).

Research Results: Applying discriminant analysis to their samples, the authors report that pay cable television subscribers differ from basic cable television subscribers in being more interested in movies, being younger, having more children, and having greater income.

Research Limitations: First, the research does not directly address the effect of household income on basic cable subscribership. Second, the research does not account for the effect of price of whether a subscriber purchases more than basic service. Third, the research uses discriminant analysis to draw inference, but fails to account for the non-normality of the examined multivariate distribution.

18

Title:The Demand for Cable TV and for MDS Authors:Douglas W. Webbink Source:Federal Trade Commission working paper, April 28, 1986, presented at 14th Telecommunications Policy Research Conference

Research Design: Demand for cable is measured by the proportion of homes passed that subscribe to cable, with separate equations for basic and pay or premium services using 1983/4 data. Thus, the research focus is on market level, rather individual level, demand for cable services. Randomly sampled from 6400 cable systems, deleting systems without key data, resulting in a sample of 252 cable systems.

Research Results: Income is not statistically significant for basic cable, but is significant for pay or premium services.

Research Limitations: First, the income variable was a measurement of county income for 1981 and thus did not exactly measure average household income for a particular service area, in a contemporaneous period. Second, regression model assumed conditional normal distribution, which is inappropriate for proportional data. Consequently statistical inferences are suspect.

19

Title:Satisfaction, Demographic, and Media Environment Predictors of Cable Subscription Authors:Robert LaRose and David Atkin Source:Journal of Broadcasting & Electronic Media, 1988, 32(4), pp. 403-413.

Research Design: The purpose of this study is to examine the influence of satisfaction variables on a consumer's decision to continue subscribing to cable television. The authors argue that as cable television has matured as a technology, new subscribers are increasingly former subscribers who decide to subscribe or not depending upon the satisfaction they receive from using cable television. The authors collected data from a telephone survey of a random sample of cable subscribing households for a stratified sample of 100 cable systems. The dependent variable in their study was a consumer intention to disconnect cable in the next month (a 4 point scaled variable: 1=very unlikely,..., 4=very likely).

Research Results: The authors’ results are based upon a series of stepwise linear regressions on their intention variable. They find, as hypothesized, that a consumer's decision to continue subscribing to cable television is based upon the extent to which they are satisfied with the service. Of lesser importance to this decision were media market variables (e.g., the number of off-air signals available to the consumer) and demographic variables (e.g., household income). While household income and household size positively influenced a consumer's decision to continue their cable subscription, they were not the most important influences on this decision.

Research Limitations: First, the authors ignore that the regressand is an ordinal variable and so their statistical tests are suspect. Second, the authors ignore explanatory variables, such as price, as a determinant of a household’s intention to disconnect cable service.

20

Title: A Comparison of Cable Disconnecters and Subscribers Authors: D. Umphrey Source: Journalism Quarterly (1989) 66, pp. 628-31

Research Design: The purpose of this research is to examine if people who drop cable exhibit different attitudes towards television or different usage of television. A sample of 1,384 households was drawn from the telephone directory for Austin, Texas during April, 1986. From this sample, interviews with 389 subscribers to Austin Cablevision led to completed surveys. These same households were then called in April and May, 1987 - a year later - to discern if they still subscribed to Austin Cablevision, and if not, why. 134 still subscribed, 34 had disconnected, and the remainder did not respond. The authors then conducted a series of simple ANOVA and Chi-Square tests of the differences between these two groups according to different measures.

Research Results: 38.2% of disconnectors mentioned costs and/or the desire to save money and 23.5% of disconnector mentioned time as the primary reason for their decision to discontinue their cable service. Interestingly disconnectors had higher monthly bills than continued subscribers, which says that they purchased more cable services (watched more movies) than had those who continued cable service, and tended to be younger than continued subscribers. Further, over 44% of the disconnectors had been previous subscribers - so they had a past history of subscribing and unsubscribing to cable.

Research Limitations: As with most communication research studies, the data are for one community and so subject to questions about their generalizability. Also the data are not always distributed as assumed by the testing procedures, and some tests were statistically suspect.

21

Title: Household-Level Demand for Cable Television: A Probit Analysis Authors: M. O. Wirth and H. Bloch Source: Journal of Media Economics (1989), pp. 21-34.

Research Design: The purpose of this research was to determine what factors were most important in determining the likelihood that a household would subscribe to either basic or pay cable services. The authors drew a probability sample of 1,780 households from eight markets served by American Television and Communications Corporation during 1982. Of the 1,780 sample households, 1,380 subscribed to cable, 400 did not. Research Results: They find the likelihood that a household subscribes to basic cable is significantly increased by viewing preferences (number of televisions in the household and hours of television watched), annual household income. People who value television more subscribe to cable. People who make more income subscribe to cable. They also find that this same likelihood is significantly decreased by the presence of a rooftop antenna, the view that broadcast TV is better than cable TV, and the education attainments of the head of household. Interestingly, the effect of household income on the likelihood of a household subscribing to cable is substantially less than other factors.

Research Limitations: Unlike many prior studies, the authors use probit analysis of their data and so do not face the criticism we have made for studies using discriminant analysis. Unfortunately, however, the authors do not account for the fact that they use a choice based sample in their estimation procedures and so both their parameter and variance estimates are biased. Further, the data are dated (1982) and so changes in the video marketplace may change some of their results and their do not describe how they measured a number of variables; for example, household income sufficiently to know if there is significant measurement error.

22

Title:Demand, pricing, and regulation: evidence from the cable TV industry Authors:John W. Mayo and Yasuji Otsuka Source:RAND Journal of Economics, 1991, 22:3, 396-410

Research Design: The proportion of cable penetration is modeled via demand equation two ways: the proportion of households subscribing to basic and the proportion subscribing to premium channels. The variables in each equation are weighted by a constant proportionality factor to correct for heteroscedasticity, and 2SLS is applied to each equation. The sample size is 1355 for demand equations and 1276 for price equations, using 1982 data for cable systems from a variety of sources.

Research Results: Income (not specifically defined) is a statistically significant determinant of each equation. Demand elasticities are estimated, but without confidence intervals. Income elasticity of demand for basic estimated to be around 0.09, and for pay around 0.35.

Research Limitations: While the authors recognize that they face heteroscedacity in OLS estimation of a regression on cable penetration, they fail to recognize that cable penetration is not a normally distributed random variate and hence their equations are mis-specified. Thus it is unclear what confidence what can put in their results.

23

Title:Competition and the Price of Municipal Cable Television Services: An Empirical Study Authors:Richard O. Beil, Jr., P. Thomas Dazzio, R. B. Ekelund, and J. Jackson Source:Journal of Regulatory Economics, 1993, 6, 401-415.

Research Design: This paper builds onto the model developed in Mayo and Otsuka (1991) a component to explain a muncipality=s choice of competitive regime. Thus the paper is primarily intended to test Hazlett=s public choice model of municipal franchising. To conduct this test, they use 1989 data on a sample of 178 jurisdictions (48 with competing cable operators, 130 with monopoly operators). Demand equations are estimated for both basic cable and premium cable, using 2SLS.

Research Results: Their main result is that cable competition significantly lowers cable prices and that municipalities that use monopoly franchising conform to Hazlett=s public choice arguments. With respect to the effect of household income, they find that income (as measured by state per capita income) is a statistically insignificant determinant of basic cable use, but a statistically significant determinant of pay cable use.

Research Limitations: Because they use Mayo and Otsuka=s (1991) basic model and estimation procedures for their basic and pay penetration equations, they face the same criticism as above - their statistical model does not conform to the data generating process and so their inferences are suspect.

24

Title:Demand and revenue elasticities for Cable Television Service Authors:David J. Roddy Source:Unpublished working paper, June 20, 1994, Economics and Technology, Inc.,

Research Design: The basic purpose of this study is to model the demand for cable television services in order to estimate the revenue effects of the 1992 Cable Television Consumer Protection and Competition Act which imposed constraints on cable prices. Specifically the study breaks cable television service into three categories matching FCC's cable regulations (i.e., basic service tier, cable programming service tier, and à la carte services). The author models the demand for the different cable television services as a function of an unobservable index, which allows him to treat each cable demand function as a probit model. He then estimates what he calls a basic access model (i.e., a comparison of basic cable service users to non-subscribers) and a CPS access model (i.e., a comparison of cable programming service users to non-subscribers) using data on 273 households obtained from a national telephone survey conducted during January, 1994.

Research Results: The authors report results that suggest that cable basic tier prices and household income are significant determinants of a household's decision to use basic cable services (basic tier) and cable programming services (CPS tier). Interestingly, the author estimates the income elasticity of demand for cable services to be 0.20 which he then compares to estimates for other goods and services (1.44 for books, 1.02 for clothing, 0.83 for long distance telephone service, 0.43 for housing, and 0.15 for local telephone access). Thus the author implicitly provides evidence on how the use of cable television services is less sensitive to household income changes than many other goods or services that consumers purchase - which is consistent with our univariate comparisons. Research Limitations: As the author acknowledges, he fails to account for the fact that a consumer must purchase the basic cable services tier before purchasing the CPS tier in his estimation - a point addressed in both the Mayo and Otsuka (1991) study and the Beil, Dazzio, Ekelund, and Jackson (1993) study. Further, the author ignores factors (e.g., household TV viewing) and so it is unclear the extent to which his results are biased by unaccounted for factors in his estimations. This second concern is the more important for the purposes of discerning the effect of household income on household use of cable television services.

25

Title:Marketing Cable and Pay Cable Services: Impact of Ethnicity, Viewing Motivations, and Program Types Authors:Alan B. Albarran and Don Umphrey Source:Journal of Media Economics, 1994, 7:3, 47-58

Research Design: The primary purpose of this study is to examine the influence of ethnicity, motives for watching television, and television viewing habits on whether a household subscribes to basic cable, pay cable, or to neither of these options. To accomplish this purpose, the authors collected data on 1,199 households (425 non-subscribers, 196 cable only subscribers, and 299 pay subscribers) through a phone survey of households in Dallas, Texas during 1991. All households in the survey were served by the same cable operator, Tele-Communications, Incorporated.

Research Results: The authors find cable and pay cable subscribers tend to watch more movies and sports than nonsubscribers, possess VCRs, be younger, and were more frequently either African American or Hispanic. Put differently, non-subscribers tended to be White, older, and have smaller households. There was little difference in the household income of non-subscribers and cable only subscribers. However, pay cable subscribers were distinguished from non-subscribers by higher income.

Research Limitations: Like a number of prior studies of cable subscribership, this study is limited by the fact it uses data for a single system and uses discriminant analysis to analyze the data.

26

Title:Cable TV: Regulation or Competition, Appendix B Authors:Robert W. Crandall and Harold Furchtgott-Roth Source:Washington, D.C.: The Brookings Institution, 1996.

Research Design: The use of cable system data based on aggregated data has problems that can be alleviated by the use of disaggregated data, and so in this appendix the demand for cable is estimated based on household data. A 1992 survey of 750 households yielded 664 respondents -- no location for the survey is given. In addition to household data, data on the cable service providers were obtained. Household demand is considered to be a function of three broad categories of variables: geographic characteristics of the community in which the household is located (e.g. rural areas have fewer entertainment opportunities and so a greater demand), cable system characteristics (e.g. older systems have higher penetration than younger systems, the number of broadcast stations carried by the cable system), and household characteristics (e.g. length of residency, number of children).

Research Results: A multinomial logit model for household purchase of cable services is constructed with three alternatives: no cable, basic cable only, and basic plus premium. Income is not significant. A dichotomous logit model, no cable or cable (whether basic or basic plus premium) is estimated: income is significant. However, an indepedendence of irrelevant alternatives test suggested that basic and basic plus premium cannot be conflated, so the multinomial model is to be preferred. A subsection (p. 147) explicitly considers the effect of income on the consumer's decision. A 10% change in income results in only a 2% change in the demand for cable. The demand for premium services is slightly more sensitive. The demand for cable is found to be sensitive to household characteristics other than income.

Research Limitations: The authors ignore variables (e.g., household viewing preferences) that are shown in communication type studies to be important determinants of cable subscribership. Consequently there is the potential that significant variables were excluded and important statistical problems introduced.

27