In recent years, the internet has grown steadily in the number of its users and the quantity of its content. Online television and videos have become increasingly prevalent, and non-television internet leisure activities, such as games and shopping, are abundant and popular. Many believe that internet-based entertainment options are becoming substitutes for standard television and will have an effect on the market for standard cable. This article is an examination of the validity of such beliefs.
Arguments could be made in support of several different hypotheses regarding the effects of the internet's growth on television consumption. One possibility is that on-demand services, such as Netflix (NASDAQ:NFLX) and Amazon (NASDAQ: AMZN) Prime, are preferable to consumers, and that viewers may soon turn away from standard television and towards online services that allow them to access specific content more readily. Another plausible hypothesis is that internet access increases awareness of specific television programs, perhaps through advertisements or through accessibility on Netflix, resulting in an overall increase in time spent watching television (through all outlets). It could also be the case that online shopping, games, or other forms of non-television internet leisure take away from time spent watching television, reducing total television consumption. All of these hypotheses revolve around questions of whether the internet is a complement for, a substitute for, or has no effect on standard television.
This investigation relies on two main data sources: The US Census Bureau's Data on Computer and Internet Use, and the American Time Use Survey. The former provides data on internet penetration, which I will define as the percentage of citizens who have high-speed internet access at home. The American Time Use Survey provides estimates of the amount of time Americans spend each day on various activities (including television watching), with data broken down into geographic and demographic categories.
A notable shortcoming of the Time Use Survey data is that television consumption is given as a combined total of time spent watching television through all outlets - through a standard television set, on a computer using services like Netflix or Amazon Prime, and all others. This detail is taken into account in this study.
The data I use in this study are made up of 357 observations - I obtained data for all 50 states and Washington, D.C. for each of seven years (2003, 2007, and 2009-2013). For each observation, I have data that includes average daily television consumption in that state, proportion of the population with high-speed internet access at home, unemployment rate for the year, and various other variables that serve as controls.
The strategy of this investigation is fixed effects regression. I regress television consumption on internet penetration while controlling for the state, year, and other variables. The regression model and variables are defined as follows:
tvit = β0 + β1internet_percentageit + β2Si + β3Yt + β4Xit + uit
tvit = average daily television consumption in minutes for an individual in state i during year t
internet_percentageit = proportion (on a scale of 0 to 100) of individuals in state i who had high-speed internet access at home in year t.
Si = state fixed effects
Yt = year fixed effects
Xit = a vector of control variables
The table below displays the results of the regression with various amounts of control variables. The control variables in the table represent the unemployment rate, average income, average age, and average household members for a given state and year.
In no form of the regression is there a statistically significant estimate of an effect of increased internet penetration on television consumption.
Interpreting the Results
The results of the regression seem to imply that increased internet penetration has little to no effect on television consumption. Put differently, there is no evidence that giving a person access to the internet has an effect on the total amount of television that he consumes.
Though the data show no effect of increased internet penetration on total television content consumed, there may be some implications regarding the amount of content consumed through a standard television set. Let us consider the following simple model:
Total TV Consumption = Standard TV Consumption + Internet TV Consumption
The model relies on the assumption that television content is consumed in one of two ways - through a standard television set (cable) or through the internet. From the model, it follows that:
Δ(Total TV Consumption) = Δ(Standard TV Consumption) + Δ(Internet TV Consumption)
Δ(Standard TV Consumption) = Δ(Total TV Consumption) - Δ(Internet TV Consumption)
The results of the regressions seem to imply that when internet penetration increases, Δ(Total TV Consumption) = 0. However, it is reasonable to assume that Δ(Internet TV Consumption) > 0 when internet penetration increases, since when more people have the internet, we would expect internet television consumption to be greater at least slightly. Using the above equations, this would imply that when internet penetration increases, Δ(Standard TV Consumption) < 0. In other words, an increase in internet penetration leads to a decrease in standard television consumption.
Limitations of This Analysis
Certain shortcomings of the data sources may pose threats to the validity of this analysis. It is important to consider that the data gathered from the American Time Use Survey are based on volunteers' records of their activities throughout the period of one day. Because the amount of time spent on each activity is self-reported, there may be some inaccuracies in the data. Another shortcoming is that time spent watching television is reported as a single number of minutes representing television watching through all outlets combined. It would have been ideal to have data for time spent watching television strictly through a standard TV set. Because such data was not available, the methods of this analysis were somewhat roundabout. With deeper investigation, it may be possible to find more optimal data and to experiment with different functional forms of the regression function.
Conclusions and Takeaways
There are two important takeaways from this study. The first is that the growth of the internet has not so far had a direct effect on the total amount of television content consumed by Americans. Increased internet penetration is not directly responsible for the increases in television consumption over recent years. The second, more interesting takeaway is that the growth of the internet has had a negative effect on the amount of time Americans spend watching television through their TV sets. This could be good news for internet television companies, such as Netflix, who seem to have been successful in wooing consumers away from standard television and toward online streaming services.
This research is representative of the state of television and the internet today, but the effects we observe in the future may be different. With the internet and online television content rapidly growing, we may be in the middle of a revolution in media delivery. It is impossible to predict the effects the internet will have in the future, but a revision of this study in several years' time may well yield more interesting and meaningful results.
Disclosure: The author has no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.