#研究分享#【量化研究:互联网对消费者的价值】

#研究分享#【量化研究:互联网对消费者的价值】在线消费时间已成为一项越来越重要的休闲活动,很难衡量休闲在线产生的消费者剩余。文章的目的是量化互联网对消费者的价值,通过运用Goolsbee和Klenow提出的方法,使用来自Nielsen Clickstream的客观数据,其中包含5个国家/地区超过12,000名互联网用户的整个点击流以及有关其人口统计特征的信息。估算了欧盟五大经济体中花费在线时间的收益程度,通过花费时间的方式来估算互联网产品价值,估值对于确保互联网的基础设施的提供和改进政策非常重要,最终文章表明,一般互联网用户的在线休闲价值每年在524至785欧元之间,每个国家的消费者剩余总额在180亿欧元(意大利)和440亿欧元(德国)之间。http://www.looooker.com/?p=58993

 

The Value of the Internet as Entertainment in Five European Countries

Pages 16-30 | Published online: 22 Feb 2016

We estimate the value of leisure online by applying Goolsbee and Klenow (2006) method to Nielsen Clickstream dataset, which covers the clickstream of more than 12,000 internet users in France, Germany, Italy, Spain and United Kingdom, in 2011. We find that the equivalent variation of welfare from leisure online for the average internet user was between 524 and 785 euros per year. At country level, it amounted to between 18 and 44 billion euros.

The objective of this article is to quantify the value of internet for consumers. Estimating this value is important for evidence-based policies related to the provision and improvement of infrastructure that ensures access to internet. Improving internet access and connectivity is at the heart of the Digital Agenda for Europe, a flagship Europe 2020 initiative, and the recently adopted Digital Single Market. An important policy area within these initiatives is ensuring universal broadband coverage across the European Union (EU) and increasing the penetration of high speed broadband. The main motivation behind these policies is the expectation that broadband and the use of internet stimulate economic and productivity growth, decrease prices and reduce the isolation or exclusion of individuals in isolated areas or in specific socioeconomic groups. However, it is increasingly recognised that internet access may create other benefits for consumers, which have not been previously fully acknowledged, such as the value created by access to free online services. Their importance can be inferred from the increasing time spent online using these services. Not taking into account such benefits underestimates the impact of such policies on consumers.

In this article, we show that benefits from online leisure for consumers are considerable, thus indicating that access to internet and to free online services, leads to important welfare gains.

Several studies have attempted to estimate the value of the internet (Greenwood and Kopecky, 2007Greenwood, J., & Kopecky, K. A. (2007). Measuring the welfare gain from personal computersEconomic Inquiry, 51(1), 336347.[Crossref][Web of Science ®][Google Scholar]; Kalapesi et al., 2010Kalapesi, C.Willersdorf, S., & Zwillenberg, P. (2010). How the internet is transforming the U.K. economyBoston, MABoston Consulting Group. [Google Scholar]; Greenstein & McDevitt, 2011Greenstein, S., & McDevitt, R.(2011). The broadband bonus: estimating broadband internet’s economic valueTelecommunications Policy, 35(7), 617632.[Crossref][Web of Science ®][Google Scholar]2012Greenstein, S., & McDevitt, R.(2012). Measuring the broadband bonus in thirty OECD countries (OECD Digital Economy Papers, No. 197). Paris, FranceOECD.[Crossref][Google Scholar]; Cooper, 2012Cooper, R. (2012). Measuring the impact of innovations in public IT infrastructure on the standard of living in OECD economies (OECD Digital Economy Working Paper). Paris, FranceOECD.[Crossref][Google Scholar]; OECD, 2012OECD. (2012). Internet economy outlook 2012Paris, FranceAuthor. [Google Scholar], 2013) using the market value of telecom hardware and services that give access to the internet. This approach is problematic because it does not take into account consumers’ time expenditure online, which can be substantial. Another approach (Bughin, 2011Bughin, J. (2011). The Web’s 100 billion surplusMcKinsey Quarterly, January. Retrieved fromhttp://www.mckinseyquarterly.com/The_Webs_100_billion_euro_surplus_2724 [Google Scholar]) is to estimate this value based on survey data on internet users’ stated preferences about the internet. However, the stated preferences may differ from users’ actual preferences.

In this article, we estimate how much consumers benefit from spending time online in the five largest economies in the EU: France, Germany, Italy, Spain, and the United Kingdom. We apply a method proposed by Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]), which is based on the revealed preferences of the internet users and information on their time spent online and their opportunity cost of time. It was previously used to estimate the value of internet for in the United States by Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]), Bayrak (2012Bayrak, E. (2012). Valuing time intensive goods an application to wireless and wired internet. In S. AllegrezzaA. Dubrocaut (Eds.), Internet economics (pp. 130-141)Basingstoke, UKPalgarve Macmillan.[Crossref][Google Scholar]), and Brynjolfsson and Oh (2012Brynjolfsson, E., & Oh, J. H.(2012). The attention economy: Measuring the value of free goods on & the internet. Paper presented at the International Conference on Information Systems (ICIS'12), Orlando, FL. [Google Scholar]) and the value of other time intensive goods by Loomis (2011Loomis, J. (2011). New approach to value urban recreation using visitors’ time allocations. Urban Public Economics Review [online]. [Google Scholar]). We use objective data from Nielsen Clickstream, which contains the entire clickstream of more than 12,000 internet users in five countries studied and information on their demographic characteristics.

There are several reasons why the consumer surplus from leisure online could differ between the United States and Europe. First, it has been shown that individuals in Europe and the United States differ in their allocations of time between leisure and work and that these differences are largely due to Europeans’ cultural preference for leisure (Alesina, Glaeser and Sacerdote, 2005Alesina, A.Glaeser, E., & Sacerdote, B. (2005), Work and leisure in the U. S. and Europe: why so different?Harvard Institute of Economic Research Working Papers 2068 Cambridge, MAHarvard Institute of Economic Research.[Crossref][Google Scholar]). Second, the EU has been lagging behind the United States in ICT investment and ICT use (OECD, 2013). The only other estimate of consumer surplus from internet use in the EU countries is provided by the McKinsey study, whose results are reported by IAB Europe (2010IAB Europe. (2010). Consumers driving the digital uptake: the economic value of online advertising-based services for consumers(White paper). Brussels, Belgium: Author. [Google Scholar]) and Bughin (2011Bughin, J. (2011). The Web’s 100 billion surplusMcKinsey Quarterly, January. Retrieved fromhttp://www.mckinseyquarterly.com/The_Webs_100_billion_euro_surplus_2724 [Google Scholar]), and it is based on survey data and conjoint analysis and extrapolation methods.

We find that, according to our most conservative measure, leisure time spent on the internet generated a consumer surplus of between between 524 and 785 euros per year. The total consumer surplus for each country amounted to between 18 billion euros (Italy) and 44 billion euros (Germany) and total for the five countries to 147 billon euros. As a percentage of gross domestic product (GDP), it represented between 1.1% in Italy and 2.2% in the United Kingdom. These estimates suggest a much larger consumer surplus than the one obtained by the McKinsey study, who found a total consumer surplus for the United States and 22 European countries of 100 billion euros (IAB Europe, 2010IAB Europe. (2010). Consumers driving the digital uptake: the economic value of online advertising-based services for consumers(White paper). Brussels, Belgium: Author. [Google Scholar]; Bughin, 2011Bughin, J. (2011). The Web’s 100 billion surplusMcKinsey Quarterly, January. Retrieved fromhttp://www.mckinseyquarterly.com/The_Webs_100_billion_euro_surplus_2724 [Google Scholar]). The differences in the estimates can be explained by the differences in methodology, the use of objective data as opposed to subjective survey data, and in the definition of internet users (this study considers all users with access to internet at home, while McKinsey study focused only on those with access to broadband).

These large values indicate important welfare gains for individuals in France, Germany, Italy, Spain, and the United Kingdom from access to internet. They confirm the importance of existing policies of improving access to internet and broadband, availability of high speed internet and access to digital content. It is important to notice that these types of welfare gains were not the main objectives of Digital Agenda for Europe and the Digital Single Market. Therefore, they indicate additional benefits of the existing policies. The magnitude of these benefits suggests that it is important to consider them in future policy initiatives. This is particularly relevant for countries that still have low share of households with access to internet and low broadband penetration rates.

Related literature

Estimating the price elasticity of demand for internet services is difficult for several reasons. First, the monetary cost of internet use is usually a fixed monthly subscription fee independent of the volume of use. Second, there is little variation in the monetary price paid by consumers. Finally, internet is a time-intensive good and for such goods the monetary costs are small compared to total costs, which include money and time spent.

A standard method to overcome these problems is to estimate the difference between how much a consumer would be willing to pay for a service and the market price he actually pays. This approach was adopted by McKinsey study (IAB Europe, 2010IAB Europe. (2010). Consumers driving the digital uptake: the economic value of online advertising-based services for consumers(White paper). Brussels, Belgium: Author. [Google Scholar]; Bughin, 2011Bughin, J. (2011). The Web’s 100 billion surplusMcKinsey Quarterly, January. Retrieved fromhttp://www.mckinseyquarterly.com/The_Webs_100_billion_euro_surplus_2724 [Google Scholar]), who applied conjoint analysis and extrapolation to data from a survey of a total of 4,500 web users in six European countries and estimated a total net consumer surplus for users with access to broadband in the United States and 22 European economies11 Austria, Belgium, Bulgaria, Croatia, Denmark, France, Finland, Germany, Greece, Hungary, Italy, Netherlands, Norway, Poland, Romania, Russia, Slovakia, Spain, Sweden, Switzerland and Turkey, the United Kingdom, and the United States.View all notes of 100 billion euros. The main limitation of this method is that it is based on survey data that reflects stated preferences, which may differ from actual choices of consumers.

Several studies used a revealed preferences approach to estimate the consumer surplus from goods related to internet using data on prices and money expenditure on these goods. Greenwood and Kopecky (2007Greenwood, J., & Kopecky, K. A. (2007). Measuring the welfare gain from personal computersEconomic Inquiry, 51(1), 336347.[Crossref][Web of Science ®][Google Scholar]) estimated consumer surplus from personal computers using changes over time in the price and expenditure on personal computers. They estimated a consumer surplus from using personal computers of 2.4% of consumption expenditure when using a measure derived from their model and a consumer surplus of 0.16% when using measure proposed by Hausman (1999Hausman, J. (1999). Cellular telephone, new products and the CPIJournal of Business and Economic Statistics, 17(2), 188194.[Taylor & Francis Online][Web of Science ®][Google Scholar]). Greenstein and McDevitt (2011Greenstein, S., & McDevitt, R.(2011). The broadband bonus: estimating broadband internet’s economic valueTelecommunications Policy, 35(7), 617632.[Crossref][Web of Science ®][Google Scholar]) estimated changes consumer surplus due to switching from dial-up access to internet to broadband access using a survey of dial-up and broadband users in United States between 1999 and 2006. They found that this switch generated a consumer surplus equivalent to a decline in the price of internet access of between 1.6% and 2.2% per year. Greenstein and McDevitt (2012Greenstein, S., & McDevitt, R.(2012). Measuring the broadband bonus in thirty OECD countries (OECD Digital Economy Papers, No. 197). Paris, FranceOECD.[Crossref][Google Scholar]) extended these estimations to thirty OECD nations to the period 2006 to 2010. They used changes in prices and quality of internet access and found large differences across countries in consumer surplus generated by this switch. Cooper (2012Cooper, R. (2012). Measuring the impact of innovations in public IT infrastructure on the standard of living in OECD economies (OECD Digital Economy Working Paper). Paris, FranceOECD.[Crossref][Google Scholar]) estimated changes in consumer welfare due to improvements in ICT infrastructure in thirty OECD countries using data on government investment in ICT infrastructure and found that the effects on consumer welfare differ considerably across countries. The main limitation of the empirical approaches used in these studies is that they consider only the money expenditure on internet access or personal computers and not the time expenditure. For time intensive goods, such as internet, time expenditure is likely to represent a large part of the consumers’ total expenditure.

A revealed preference approach that takes into consideration both money and time expenditure of consumers on internet was proposed by Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]). Their method uses differences in time spent online and in the opportunity cost of time, to measure consumer surplus from access to internet. They applied this method to a cross section of internet users in United States in 2005 and estimated a consumer surplus equal to 2.3% of full income. This method was also used by Bayrak (2012Bayrak, E. (2012). Valuing time intensive goods an application to wireless and wired internet. In S. AllegrezzaA. Dubrocaut (Eds.), Internet economics (pp. 130-141)Basingstoke, UKPalgarve Macmillan.[Crossref][Google Scholar]) to study consumer surplus from internet use in the United States and by Loomis (2011Loomis, J. (2011). New approach to value urban recreation using visitors’ time allocations. Urban Public Economics Review [online]. [Google Scholar]) to study consumer surplus from urban recreation in Wyoming. Brynjolfsson and Oh (2012Brynjolfsson, E., & Oh, J. H.(2012). The attention economy: Measuring the value of free goods on & the internet. Paper presented at the International Conference on Information Systems (ICIS'12), Orlando, FL. [Google Scholar]) used a variant of this method to measure the value of internet for consumers and found a consumer surplus of 3.3% of full income in the United States in 2007.

Overall, the existing literature on estimating consumer surplus from internet or related goods is very heterogeneous in terms of methods used and estimates obtained. We follow Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]), which relies on consumers’ revealed preferences and takes into consideration both money and time expenditure. In addition, this approach is suitable for our study because it has a similar purpose (estimating consumer surplus from leisure online) and uses similar types of data (time use data and the demographic information on internet users).

Estimation strategy

This section explains the Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) model and how it is applied to estimate consumer surplus from leisure online. In this model, consumers obtain utility from consuming two goods: an internet good and a composite good. The internet good represents online leisure activities. The composite good represents all other goods and services consumed, including offline leisure activities. Consuming these two goods requires both time and money. The amount of time that can be spent on the two goods is limited by the fixed time in a day and the amount of money by the total money available, which is determined by hourly wages and number of hours worked. Consumers choose the time and money to spend on the two goods subject to combined time and money budget constraints. Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) assumed that the utility function takes the following form:(1)

The budget constraint is:(2)

U represents utility for a consumer. θ is the contribution of an internet good to utility and takes values between 0 and 1. LI is the fraction of non-sleeping time spent on internet per week. LO is the fraction of non-sleeping time spent on the composite good per week. PO and CO are the price and quantity consumed of the composite good. PI and CI are the price and quantity consumed of the internet good. FI is the fixed fee for subscribing to internet. 1 - LI – LO is the fraction of non-sleeping time spent working per week. W is the wage consumers earn in labour markets. αI and αO are the money intensities of the internet and they are defined as time and money expenditure shares of total (time + money) expenditure spent on the internet and composite goods:(3)

Following most of the time allocation studies since Becker (1965) and Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) assumed that the opportunity cost of time is given by the wage.22 The assumption that opportunity cost of time spent on leisure is given by income that could be earned in labour markets is common in studies related to leisure. It is important to notice that is has been shown that income is an appropriate measure for the value of time even for people working fixed hours (Larson, 1993Larson, D. M. (1993). Joint recreation choices and implied values of timeLand Economics, 69(3), 270286.[Crossref][Web of Science ®][Google Scholar]). Another measure used for value of time is 1/3 of income (Loomis, 2011Loomis, J. (2011). New approach to value urban recreation using visitors’ time allocations. Urban Public Economics Review [online]. [Google Scholar]). Because of functional form chosen, the estimates of elasticity of substitution and of consumer surplus would not be affected if the opportunity cost of time represented a fixed proportion of income.View all notes The time intensities of the internet and composite goods are (1-αI) and (1-αO), respectively. The internet and composite goods have different money and time intensities. The internet good is time intensive: Time expenditure represents a large part of total expenditure and money expenditure a small part. The composite good is relatively less time intensive and more money intensive.33 It is important to notice that the composite good may also include time intensive offline leisure activities. However, overall, the composite good is relatively less time intensive and more money intensive than internet good, which includes only online leisure activities.View all notes σ is the elasticity of substitution between the internet and the composite good. By solving the utility maximisation problem subject to the budget constraint, Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) obtain the optimal consumption of the internet and composite goods.

Consumer surplus is measured as equivalent variation (EV). It indicates how much the income of a consumer without access to internet would need to increase so that the consumer obtains the same utility as when he or she consumes the internet good. Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) derived the following expression for the EV (hereafter referred to as the GK measure):(4)

This measure may overestimate the consumer surplus from leisure online due to the functional form of the utility function from which it is derived. It implies that when consumption of leisure online approaches zero, the marginal utility from leisure online becomes infinite. This would imply that all consumers would spend at least some infinitesimal time online independently of the price of leisure online. This is unrealistic, as there are individuals who do not spend any time online. Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) acknowledge this problem and suggest that an adaptation of the measure proposed by Hausman (1999Hausman, J. (1999). Cellular telephone, new products and the CPIJournal of Business and Economic Statistics, 17(2), 188194.[Taylor & Francis Online][Web of Science ®][Google Scholar]) provides a better measure of consumer surplus:(5)

Both measures depend on σ (the elasticity of substitution between internet and composite goods), which has to be estimated. Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) derived from the equations of the optimal consumption quantities of internet and composite goods the following expression:(6)

The left-hand side variable is the logarithm of the ratio of the share of nonsleeping time spent offline to the share of nonsleeping time spent online. The right hand variables are a constant term,44 In Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]), A is a function of the price of internet good and time intensity of internet good and the price and time intensity of the composite good. This ratio is constant across individuals because it is assumed that all internet users are price takers and face the same prices.View all notes the opportunity cost of time, and a term that represents individual differences in the preference for the internet good relative to the composite good, which depend on the users’ demographic characteristics (x):

Then, to obtain the parameters of interest, we estimate:(7)

i represents an internet user. From the estimated coefficients we can recover the parameters of interest, using:(8)

We estimate equation (7) using OLS and heteroskedasticity robust standard errors. One concern is that high income and low income consumers may differ systematically in their preferences for the internet good, which would lead to biased estimates of σ. There is some empirical evidence that this might be the case (Goldfarb & Prince, 2008Goldfarb, A., & Prince, J.(2008). Internet adoption and usage patterns are different: implications for the digital divideInformation Economics and Policy, 20(1), 215.[Crossref][Web of Science ®][Google Scholar]; Pantea & Martens, 2014Pantea, S., & Martens, B.(2014). Has the digital divide been reversed?: Evidence from five EU countriesElectronic International Journal of Time Use Research, 11(1), 1342.[Crossref][Google Scholar]). We address this potential problem by allowing the internet users’ preferences for the internet good relative to the composite good to depend on their occupation and education. We control for differences in the quality and speed of internet access in different regions by including regional dummies.

Data sources and variable measurement

Internet usage in France, Germany, Italy, Spain, and the United Kingdom

The countries covered in this analysis represent the five largest economies in the EU: France, Germany, Italy, Spain and the United Kingdom. In addition to their economic size, these countries are interesting to study because they differ in the development of ICT infrastructure and internet use. Table 1presents descriptive statistics on these indicators.

Table 1. Internet Use in 2011.

Theses descriptive statistics show that access to internet varied considerably across the countries studied in 2011. In France, Germany, and the United Kingdom, close or more than 80% of households had access to internet, whereas in Italy and Spain only around 60% had access to internet in 2011. Among the households who had access to internet, most had broadband connection (above 90% in Germany, Spain, France, and the United Kingdom and above 80% in Italy). The shares of individuals that used internet for leisure related purposes tend to be higher in Germany and the United Kingdom, than in the other countries and these shares were lowest for all purposes in Italy. Spain registered relatively low shares for certain types of internet uses (sending/receiving e-mails and downloading games, images, films or music), but high relative use of internet for social media sites, listening to radio/watching TV and for reading newspapers and news online. Overall, there is considerable variation in access to internet and in internet usage of leisure purposes across the countries studied.

Data description

The main dataset used was collected by Nielsen NetRatings. It contains the clickstream of 25,000 internet users in France, Germany, Italy, Spain, and the United Kingdom, during the year 2011.

Data is collected through a piece of software that internet users voluntarily install on their personal computers (PC). It is collected for each internet user in a household.55 This is done by prompting the users in households where there are more than one user to log in.View all notes The data collection corrects for errors in measurement of the time spent on websites due to periods of inactivity or tabbed browsing66 This is done by using information about the website in focus and keyboard input.View all notes and it is transmitted to Nielsen NetRatings in real time.

Acording to Nielsen, the sample of internet users is representative of the online population in these countries in terms of gender and age.77 Nielsen provides incentives to participate and to remain in the panel in the form of vouchers and points which can be redeemed from their reward website or used in online games and sweepstakes(prize drawing), which might bias our sample towards people who are more likely to value these activities. As a robustness check we repeated the estimations excluding time spent on online games and gambling websites to make sure that our results are not driven by time spent on these websites. These estimations are not reported here.View all notes This data is used extensively by private companies, especially media companies for measuring their audience, which suggest that it is widely regarded as reliable and of high quality.

The dataset is in the form of a clickstream and for each click it contains information on the website visited and time spent on it. It also contains information on internet users’ demographic characteristics (age, gender, marital/cohabitating status, presence of children in the household, size of household, household income range, highest educational level attained, and occupation). This data is collected through a questionnaire at the time of the installation of the Nielsen software.

In the empirical analysis, we use only the clickstream of employed internet users, who were between 16 and 74 years old and provided complete information on the relevant demographic characteristics. We focus on employed internet users because in our model the opportunity cost of spending time online is given by the wage that the internet user could earn on the labour market. We exclude internet users in the highest and the lowest 1% of the distribution of weekly time spent online, as their records might be affected by recording problems or by problems related to internet addiction, which are outside the scope of this empirical analysis. There are 12,311 individuals in the sample used in the empirical analysis. The number of observations in each country is reported in Table 2.

Table 2. Summary Statistics for Time Allocation Variables.

Time allocation variables

The time allocation variables used are calculated as shares of total non-sleeping time. In the absence of data on sleeping hours, we assume that all internet users spend 8 hours sleeping, which amounts to a total of 112 nonsleeping hours per week.88 Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) made the same assumption. Given the lack of data on other possible non-discretionary time use activities, we assume that the 8 hours include all such activities. Lack of data on sleep hours also prevents us from taking into account the possible endogeneity of the sleeping patterns, despite evidence that they might depend on employment opportunities (Brochu et al., 2012Brochu, P.Armstrong, C. D., & Morin, L. P. (2012). The “trendiness” of sleep: An empirical investigation into the cyclical nature of sleep timeEmpirical Economics, 43(2), 891913.[Crossref][Web of Science ®][Google Scholar]).View all notes

Leisure time online is defined as the average number of hours spent on websites related to leisure activities on their personal computers per week.99 Internet users may also have access to internet from portable devices, which are not covered by the Nielsen meter. However, OECD (2012OECD. (2012). Internet economy outlook 2012Paris, FranceAuthor. [Google Scholar]) suggests that internet traffic from such devices accounted for 6.8% of internet traffic in the UK, 4% in Spain and less than 3% in France of all internet traffic in these countries in August 2011. To the extent that these figures are representative of other months of the year and of other two countries, they suggest that the internet traffic on these devices represented only a small share of total internet activity.View all notesWe consider leisure websites to be of the following types (based on the Nielsen classification):1010 Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) assumed that all time spent online is leisure. However, activities, such as ecommerce or online banking are not considered leisure by most people. Time spent on these activities contributes to the utility as an input in the production of the composite good.View all notes entertainment (adult, digital arts and graphics, books and magazines, broadcast media, events, gambling and sweepstakes, humour, comics and novelties, multicategory entertainment, music, online games, sports and videos and movies), family and lifestyle (family resources, genealogy, kids, games and toys, multicategory family lifestyle, personals, pets and animal care, religion and spirituality), news and information, social networks (member communities and targeted member communities), and telecom and internet services (email, instant messaging, long distance and local telephone carriers and internet tools, which include popular illegal downloading websites).1111 This definition is chosen because these categories of websites are related to leisure, but they may include several activities not leisure related. We have tried two other definitions: all time spent online, as in Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) and Nielsen entertainment category. Both these definitions are problematic. Classifying all time spent online as leisure, would result in including websites clearly no related to leisure such as e-banking, e-government, job search, education resources online. The Nielsen entertainment category does not include social networks websites, such as Facebook, which is the most popular leisure website (in terms of time spent on it). For comparison, with Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]), we report the results using all time spent online.View all notes This definition has been used by Pantea and Martens (2014Pantea, S., & Martens, B.(2014). Has the digital divide been reversed?: Evidence from five EU countriesElectronic International Journal of Time Use Research, 11(1), 1342.[Crossref][Google Scholar]) and it is similar to the definition used in the McKinsey study (IAB Europe, 2010IAB Europe. (2010). Consumers driving the digital uptake: the economic value of online advertising-based services for consumers(White paper). Brussels, Belgium: Author. [Google Scholar]; Bughin, 2011Bughin, J. (2011). The Web’s 100 billion surplusMcKinsey Quarterly, January. Retrieved fromhttp://www.mckinseyquarterly.com/The_Webs_100_billion_euro_surplus_2724 [Google Scholar]).

Because of the lack of data on working hours for individual internet users, we assume that they work the average number of hours of employees in their countries, which are available from Eurostat Labour Force Survey. The time spent on the composite good is defined as non-sleeping time that is spent neither on online leisure nor working.

The descriptive statistics on these variables are in Table 2. Time spent on leisure online varies from 2.3 hr per week in France to 3.8 hr in Germany. In all countries, the time spent on leisure online is lower than the 7.7 hr per week reported by Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) for the United States. This happens because internet users in the countries studied spent less time online (on all websites) than users in the United States and because we focus on the time spent on leisure websites. Time spent leisure online accounts for between 2% and 3% of the nonsleeping time in the countries studied. Time spent on composite good ranges between 65 and 68 hr, which represents between 59% and 61% of nonsleeping time in the countries studied.

Time and money intensities of the two goods

Following Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]), we assume that the marginal monetary cost of using internet is 0, which means PI =0.1212 This assumption is not unreasonable. Van Dijk (2012Van Dijk Management Consultants. (2012). Broadband internet access costs, European Commission, Directorate General for Communications NetworksContent and Technology Report. [Google Scholar]) found that more 80% of the internet access offers were unmetered offers.View all notes By substituting this in (3), if follows that αI =0 and that αO is:

LW is the share of nonsleeping time spent working and its calculation is explained in the section on time allocation variables. WLW is the total money consumption expenditure. As there are only two goods in this model, the share of consumption expenditure on the composite good is equal to one minus the share of consumption expenditure on the internet good, which is zero. The values of αO obtained for the five countries are reported in Table 3. These values are between 0.37 and 0.39, similar to 0.38 obtained by Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) for the United States.1313 We considered several alternative measures of αo,including we calculated it assuming that the share of consumption expenditure on internet is equal to share of consumption expenditure with all telecommunication services and using different measures of average working hours. Overall, the estimates were very similar to the baseline estimates.View all notes

Table 3. Time and Money Intensities.

Income and other demographic characteristics

The dataset does not provide information on individual income, only on the household income range. We calculate income as the average of the lowest and highest income in the income range. Admittedly, this is a rough proxy for wage. However, the dataset does not provide information on work related income or on the number of employed persons in the household. In the empirical analysis we address these issues by controlling for household size and by focusing on the employed internet users.

The descriptive statistics in Table 4 show that the average income in the sample is between 32,439.81 euros per year in Spain and 42,896.53 euros per year in the United Kingdom. In France and in the United Kingdom, the samples contain larger shares of internet users in higher income ranges than in the other countries. This is due to larger shares of internet users with managerial/executive and professional occupations in these countries. In the empirical analysis we control for these differences by including controls for internet users’ occupations and other demographic characteristics.

Table 4. Summary Statistics for Income.

In line with previous studies (Goldfarb & Prince, 2008Goldfarb, A., & Prince, J.(2008). Internet adoption and usage patterns are different: implications for the digital divideInformation Economics and Policy, 20(1), 215.[Crossref][Web of Science ®][Google Scholar]; Pantea & Martens, 2014Pantea, S., & Martens, B.(2014). Has the digital divide been reversed?: Evidence from five EU countriesElectronic International Journal of Time Use Research, 11(1), 1342.[Crossref][Google Scholar]), we assume that preferences for the internet good relative to the composite good depend on users’ gender, age, marital status, household size,1414 The dataset provides information on whether the household is composed of 1–2 persons, 3–4 persons or more than 5 persons.View all notes education, occupation, and region of residence. Table 5 presents the definition of these variables and their descriptive statistics. The descriptive statistics show that the sample used in the empirical analysis includes a large variety of internet users in terms of gender, age, education, occupation, and family composition. Several demographic characteristics, in particular household income, having tertiary education and occupation dummies, such as executive/managerial occupation, are positively and significantly correlated. However, the magnitude of these correlations is between 0.20 and 0.41, which suggest that multi-collinearity is unlikely to be an important concern.

Table 5. Summary Statistics for Demographic Characteristics.

Estimation results

The results of the estimation of Equation (7) and the implied parameters, computed using Equation (8), are reported in Table 6. The coefficient of income is always positive and statistically significant. Its values vary between 0.24 in Spain to 0.47 in the United Kingdom. These results indicate that internet users with higher incomes spend less time online for entertainment, in line with previous evidence (Goolsbee & Klenow, 2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]; Goldfarb & Prince, 2008Goldfarb, A., & Prince, J.(2008). Internet adoption and usage patterns are different: implications for the digital divideInformation Economics and Policy, 20(1), 215.[Crossref][Web of Science ®][Google Scholar]; Pantea & Martens, 2014Pantea, S., & Martens, B.(2014). Has the digital divide been reversed?: Evidence from five EU countriesElectronic International Journal of Time Use Research, 11(1), 1342.[Crossref][Google Scholar]). The estimated elasticity of substitution between the two goods varies between 1.64 in Spain and 2.18 in the United Kingdom. These values are higher than the value obtained by Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) for the United States in 2005 (1.62).

Table 6. Estimation Results.

The relationship between preferences for an internet good relative to the composite good and demographic characteristics is given by equation (8), which can be rewritten as:

Thus, a positive and significant γx implies that internet contributes less to utility for internet users with the characteristic x, and a negative coefficient implies that internet contributes more to utility for internet users with the characteristic x.

The results reported in Table 6 suggest that, in most countries, education has a mixed and often insignificant effect on the relative preference for leisure online. In most countries, leisure online contributes less to the utility of women than of men and less to the utility of older consumers (over 50) than younger ones. It also contributes more to utility of single people than to the utility of married/cohabitating ones. A large household size has a negative effect on the relative preference for leisure online. Generally, these results are plausible and in line with previous findings on the use of internet of different demographic groups.

Next, we compute consumer surplus from leisure online as a percentage of full income, as defined by Becker (1965): The income that could be achieved if a person dedicated all his nonsleeping time to earning income. We also report the annual consumer surplus per internet user. Using data on population and access to internet from Eurostat, we also calculated the total economic value of time spent online for internet users in each country. We assume that the value of time spent online for internet users who were not employed is the same as for those employed, which is the best estimate we have for these internet users. Table 7 reports these estimates.

Table 7. Estimates of the Welfare Gain From Leisure Time Spent Online.

For the average internet user, the estimates based on Goolsbee Klenow measure suggest welfare gains from leisure online between 2.2% and 5.3% of full income, while those based on Hausman measure indicate welfare gains between 0.6% and 1.0 % of full income. In monetary terms, the average internet user would need to receive between 2,082 euros in France and 4,125 euros in Spain to achieve the same utility that they achieve when being able to spend time online, according to Goolsbee Klenow measure, or between 524 euros and 785 euros according to the Hausman measure. At country level, Goolsbee Klenow estimates indicate economic values of leisure online between 74 and 165 billion euros, whereas those based on Hausman measure suggest more conservative, but still large values, between 18 and 44 billion euros.

How do these estimates compare with previous estimates? They are lower than those obtained by Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) for United States: 12.6 % when using the measure derived from their model and 2.3% when using the Hausman measure, which they consider more realistic. These differences can be explained by the less time spent online by internet users in the EU countries1515 In Table 9, in the annexes, we report the estimates obtained when considering all time spent online as leisure online, as in Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]). As expected, these estimates are larger and closer to those obtained by Goolsbee and Klenow for the United States. However, we do not emphasise this definition of leisure online, as it includes many non-leisure activities, such as e-banking and job search.View all notes and by the higher elasticity of substitution between online leisure and the composite good in these countries than in the United States. However, these estimates are much larger than those obtained for the same countries by McKinsey study, which were, in billion euros, 7.5 in France, 6.9 in Germany, 5.6 in Italy, 4.9 in Spain, and 7.8 in the United Kingdom (IAB Europe, 2010IAB Europe. (2010). Consumers driving the digital uptake: the economic value of online advertising-based services for consumers(White paper). Brussels, Belgium: Author. [Google Scholar]). The definition of leisure online is very similar in the two studies. The differences in the estimates are mainly due to differences in estimation methods used, the use of objective data as opposed to subjective survey data, and in the definition of internet users (this study considers all users with access to internet at home, while McKinsey study focused only on those with access to broadband).

To gain a better insight into the welfare gain from time spent online we also compare these estimates with relevant economic indicators. These comparisons are reported in Table 8.

Table 8. How Large Is the Welfare Gain From Time Spent Online?.

Table 9. Estimates of the Welfare Gain From All Time Spent Online.

First, we compare the equivalent variation estimate of welfare from an hour spent online with the average hourly wages, which can be interpreted as an average opportunity cost of spending time online. Estimates based on Goolsbee Klenow measure are close to or higher than the average hourly earnings in most countries. In Spain, they are more than double the average hourly earnings. The Hausman estimates are smaller (around 4 Euro per hour) and amount to between 23% and 41% of average hourly earnings. These estimates suggest that leisure online is a low value activity for consumers, consistent with the results of Goldfarb and Prince (2008Goldfarb, A., & Prince, J.(2008). Internet adoption and usage patterns are different: implications for the digital divideInformation Economics and Policy, 20(1), 215.[Crossref][Web of Science ®][Google Scholar]) and Pantea and Martens (2014Pantea, S., & Martens, B.(2014). Has the digital divide been reversed?: Evidence from five EU countriesElectronic International Journal of Time Use Research, 11(1), 1342.[Crossref][Google Scholar]).

Second, we compare our estimates of total welfare gains per country with the GDP, reported in Table 7. The total welfare gains suggest that despite the low value of an hour spent online, the total welfare gains are large because internet is consumed in large quantities and the monetary cost of using it is essentially zero (for internet subscribers). As a percentage of GDP, the estimates based on Goolsbee Klenow measure suggest that the economic value of time spent online represents between 4.7% and 10.7%. The estimates based on Hausman measure suggest that it amounts to between 1.1% and 2.2%., similar to the estimates obtained by Brynjolfsson and Oh (2012Brynjolfsson, E., & Oh, J. H.(2012). The attention economy: Measuring the value of free goods on & the internet. Paper presented at the International Conference on Information Systems (ICIS'12), Orlando, FL. [Google Scholar]) for the United States for 2010. Overall, the Goolsbee Klenow and Hausman measures provide quite different estimates. The comparisons with previous studies, with opportunity cost of time and GDP suggest that those based on the Hausman measure are more plausible.

Another informative comparison is with the value of other leisure activities. To the best of our knowledge the only comparable estimate for the countries studied is the estimated value of BBC, for United Kingdom. BBC and Human Capital (2004BBC. (2004). Measuring the value of the BBC: A report by the BBC and human capitalLondon, UKBritish Broadcasting Corporation & Human Capital. [Google Scholar]) estimated that the annual value of BBC for a user was 282 pounds per year. This amounted to 1.1% of the GDP per capita in 2004. Our estimates of the value of internet for a user in the United Kingdom in 2011 (768.84 Euros per year and 2.6% of GDP per capita in 2011, using the most conservative measure) compare favorably to these estimates of the value of BBC. This comparison is suggestive of the important value provided by free online leisure services for consumers in the United Kingdom.

Overall, the estimates of the value of online leisure presented in this study, in particular those based on Hausman method, compare in plausible ways to economic indicators and to estimates of the value of comparable leisure activities. They suggest that leisure online provides considerable welfare gains for internet users.

It is important to notice several limitations of the estimates presented. First, the estimation strategy assumes that the opportunity cost of spending time online is given by the wage. In reality, people may value their leisure time less than their wage. Nevertheless, if people value their leisure time at a constant fraction of their wage,1616 Loomis (2011Loomis, J. (2011). New approach to value urban recreation using visitors’ time allocations. Urban Public Economics Review [online]. [Google Scholar]) suggested that a suitable estimate of price of leisure is one-third of the wage.View all notes the estimates of elasticity of substitution and the estimates of equivalent variation of welfare would not be affected. Secondly, in the model used, there is only one possible substitute for leisure online: a composite good. In reality, there might be closer substitutes to spending time online, such as other types of leisure. A model in which the consumer chooses between leisure online and other types of leisure activities and a composite good would be more suitable. We are not able to pursue this extension due to lack of data on internet users’ time allocation on other leisure activities. Thirdly, this model does not take into consideration a possible effect on leisure online on productivity and wages, but the use of internet for leisure purposes could enhance users’ e-skills and, thus, improve their employment opportunities. Finally, we assume that internet users benefit from all the time spent online and that the principle of revealed preferences applies. However, the evidence on this topic is mixed. There are studies, such as Penard et al. (2011Penard, T.Poussing, N., & Suire, R. (2011). Does the internet make people happier? (CEPS/INSTEAD Working Paper Series 2011-41). Luxembourg: CEPS/INSTEAD. [Google Scholar]), which found that using internet has a positive effect on well-being. However, Nie and Hillygus (2002Nie, N. D., & Hillygus, D. S.(2002). The impact of internet use on sociability: time-diary findingsIT & Society, 1(1), 120. [Google Scholar]), Ward (2012Ward, M. R. (2012). Does time spent playing video games crowd out time spent studying? 23rd European Regional ITS Conference, Vienna 2012 60374, International Telecommunications Society, ViennaAustria. [Google Scholar]), and Wallsten (2013Wallsten, S. (2013). What are we not doing when we’re online (NBER Working Paper No. 19549). Cambridge, MA: National Bureau of Economic Research, Inc.[Crossref][Google Scholar]) found that spending time online may negatively affect activities such as socialization, other leisure activities, and studying, which are generally associated with higher wellbeing. We partially address this problem by focusing on economically active internet users and by excluding those who spend extremely high amounts of time online (in the highest 1% of the time spent online distribution), who are most likely to be affected by a possible negative effect from spending too much time online.

Conclusions

Spending time online has become an increasingly important leisure activity. It is difficult to measure the consumer surplus generated by leisure online because money expenditure represents only a small part of the total expenditure on it and variation in price of access to internet is limited.

In this article, we quantify the consumer surplus from online leisure in the five largest economies in the EU using an innovative method proposed by Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) and a unique dataset that covers the clickstream of a large number or internet users in these countries. This method uses differences in time spent online and in the opportunity cost of spending time online to obtain a measure of economic value of internet as entertainment. We use Nielsen Clickstream dataset, which contains the entire clickstream of more than 12,000 internet users in France, Germany, Italy, Spain, and the United Kingdom, during 2011, and information on their demographic characteristics.

We find that, using the most conservative estimates, the value of leisure online for the average internet user was between 524 and 785 euros per year. The total consumer surplus for each country amounted to between 18 billion euros (Italy) and 44 billion euros (Germany).

The value of leisure online for consumers could be somewhat overestimated due to assumptions made regarding the opportunity cost of time, possible substitutes of leisure online and the assumption that time spent online always increases utility for internet users. However, the estimates are in line with estimates of previous studies on this topic and compare in a plausible way with a variety of related economic indicators. Most importantly, they constitute, to the best of our knowledge, the first such estimates for EU economies based on objective data.

These estimates suggest that there are considerable welfare gains for individuals in the EU from accessing and using free online services. This evidence highlights the importance of existing policies of improving access to internet for households, broadband coverage, and availability of high speed internet. Although these types of welfare gains were not the main focus of the existing policies in this area, the magnitude of these benefits suggests that it is important to consider them in future policy initiatives. This is particularly relevant for countries that still have low share of households with access to internet and low broadband penetration rates. These findings are also informative for policies that aim to increase access of consumers to online content, such as the completion of Digital Single Market. The magnitude of the estimates to the value of online leisure presented in this paper suggests that such policies could lead to important welfare gains.

Acknowledgments

The views and opinions expressed in this article are the authors’ and do not necessarily reflect those of the JRC or the European Commission. The authors thank Marc Bogdanowicz, Russel Cooper, Ibrahim Kholilul Rohman, Piotr Stryszowski and anonymous referees for comments and suggestions. Errors and omissions remain the responsibilities of the authors.

 

Notes

1 Austria, Belgium, Bulgaria, Croatia, Denmark, France, Finland, Germany, Greece, Hungary, Italy, Netherlands, Norway, Poland, Romania, Russia, Slovakia, Spain, Sweden, Switzerland and Turkey, the United Kingdom, and the United States.2 The assumption that opportunity cost of time spent on leisure is given by income that could be earned in labour markets is common in studies related to leisure. It is important to notice that is has been shown that income is an appropriate measure for the value of time even for people working fixed hours (Larson, 1993Larson, D. M. (1993). Joint recreation choices and implied values of timeLand Economics, 69(3), 270286.[Crossref][Web of Science ®][Google Scholar]). Another measure used for value of time is 1/3 of income (Loomis, 2011Loomis, J. (2011). New approach to value urban recreation using visitors’ time allocations. Urban Public Economics Review [online]. [Google Scholar]). Because of functional form chosen, the estimates of elasticity of substitution and of consumer surplus would not be affected if the opportunity cost of time represented a fixed proportion of income.3 It is important to notice that the composite good may also include time intensive offline leisure activities. However, overall, the composite good is relatively less time intensive and more money intensive than internet good, which includes only online leisure activities.

4 In Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]), A is a function of the price of internet good and time intensity of internet good and the price and time intensity of the composite good. This ratio is constant across individuals because it is assumed that all internet users are price takers and face the same prices.

5 This is done by prompting the users in households where there are more than one user to log in.

6 This is done by using information about the website in focus and keyboard input.

7 Nielsen provides incentives to participate and to remain in the panel in the form of vouchers and points which can be redeemed from their reward website or used in online games and sweepstakes(prize drawing), which might bias our sample towards people who are more likely to value these activities. As a robustness check we repeated the estimations excluding time spent on online games and gambling websites to make sure that our results are not driven by time spent on these websites. These estimations are not reported here.

8 Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) made the same assumption. Given the lack of data on other possible non-discretionary time use activities, we assume that the 8 hours include all such activities. Lack of data on sleep hours also prevents us from taking into account the possible endogeneity of the sleeping patterns, despite evidence that they might depend on employment opportunities (Brochu et al., 2012Brochu, P.Armstrong, C. D., & Morin, L. P. (2012). The “trendiness” of sleep: An empirical investigation into the cyclical nature of sleep timeEmpirical Economics, 43(2), 891913.[Crossref][Web of Science ®][Google Scholar]).

9 Internet users may also have access to internet from portable devices, which are not covered by the Nielsen meter. However, OECD (2012OECD. (2012). Internet economy outlook 2012Paris, FranceAuthor. [Google Scholar]) suggests that internet traffic from such devices accounted for 6.8% of internet traffic in the UK, 4% in Spain and less than 3% in France of all internet traffic in these countries in August 2011. To the extent that these figures are representative of other months of the year and of other two countries, they suggest that the internet traffic on these devices represented only a small share of total internet activity.

10 Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) assumed that all time spent online is leisure. However, activities, such as ecommerce or online banking are not considered leisure by most people. Time spent on these activities contributes to the utility as an input in the production of the composite good.

11 This definition is chosen because these categories of websites are related to leisure, but they may include several activities not leisure related. We have tried two other definitions: all time spent online, as in Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]) and Nielsen entertainment category. Both these definitions are problematic. Classifying all time spent online as leisure, would result in including websites clearly no related to leisure such as e-banking, e-government, job search, education resources online. The Nielsen entertainment category does not include social networks websites, such as Facebook, which is the most popular leisure website (in terms of time spent on it). For comparison, with Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]), we report the results using all time spent online.

12 This assumption is not unreasonable. Van Dijk (2012Van Dijk Management Consultants. (2012). Broadband internet access costs, European Commission, Directorate General for Communications NetworksContent and Technology Report. [Google Scholar]) found that more 80% of the internet access offers were unmetered offers.

13 We considered several alternative measures of αo, including we calculated it assuming that the share of consumption expenditure on internet is equal to share of consumption expenditure with all telecommunication services and using different measures of average working hours. Overall, the estimates were very similar to the baseline estimates.

14 The dataset provides information on whether the household is composed of 1–2 persons, 3–4 persons or more than 5 persons.

15 In Table 9, in the annexes, we report the estimates obtained when considering all time spent online as leisure online, as in Goolsbee and Klenow (2006Goolsbee, G., & Klenow, P. J.(2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.[Crossref][Web of Science ®][Google Scholar]). As expected, these estimates are larger and closer to those obtained by Goolsbee and Klenow for the United States. However, we do not emphasise this definition of leisure online, as it includes many non-leisure activities, such as e-banking and job search.

16 Loomis (2011Loomis, J. (2011). New approach to value urban recreation using visitors’ time allocations. Urban Public Economics Review [online]. [Google Scholar]) suggested that a suitable estimate of price of leisure is one-third of the wage.

References

  • Alesina, A.Glaeser, E., & Sacerdote, B. (2005), Work and leisure in the U. S. and Europe: why so different? Harvard Institute of Economic Research Working Papers 2068 Cambridge, MAHarvard Institute of Economic Research.

    ,

  • Bayrak, E. (2012). Valuing time intensive goods an application to wireless and wired internet. In S. Allegrezza & A. Dubrocaut (Eds.), Internet economics (pp. 130-141)Basingstoke, UKPalgarve Macmillan.

    ,

  • BBC. (2004). Measuring the value of the BBC: A report by the BBC and human capitalLondon, UKBritish Broadcasting Corporation & Human Capital.
  • Brochu, P.Armstrong, C. D., & Morin, L. P. (2012). The “trendiness” of sleep: An empirical investigation into the cyclical nature of sleep timeEmpirical Economics, 43(2), 891913.

    ,

  • Brynjolfsson, E., & Oh, J. H. (2012). The attention economy: Measuring the value of free goods on & the internet. Paper presented at the International Conference on Information Systems (ICIS'12), Orlando, FL.
  • Bughin, J. (2011). The Web’s 100 billion surplusMcKinsey Quarterly, January. Retrieved from http://www.mckinseyquarterly.com/The_Webs_100_billion_euro_surplus_2724
  • Cooper, R. (2012). Measuring the impact of innovations in public IT infrastructure on the standard of living in OECD economies (OECD Digital Economy Working Paper). Paris, FranceOECD.

    ,

  • Eurostat. (2012a). Eurostat Information Society Indicators Database. Retrieved from http://ec.europa.eu/eurostat/web/information-society/data/database
  • Eurostat. (2012b). Eurostat Labour Force Survey Indicators Database. Retrieved from http://ec.europa.eu/eurostat/web/lfs/data/database
  • Eurostat. (2012c). National Accounts Indicators Database. Retrieved from http://ec.europa.eu/eurostat/web/national-accounts/data/database
  • Goldfarb, A., & Prince, J. (2008). Internet adoption and usage patterns are different: implications for the digital divideInformation Economics and Policy, 20(1), 215.

    ,

  • Goolsbee, G., & Klenow, P. J. (2006). Valuing consumer products by the time spent using them: an application to the internetAmerican Economic Review, 96(2), 108113.

    ,

  • Greenwood, J., & Kopecky, K. A. (2007). Measuring the welfare gain from personal computersEconomic Inquiry, 51(1), 336347.

    ,

  • Greenstein, S., & McDevitt, R. (2011). The broadband bonus: estimating broadband internet’s economic valueTelecommunications Policy, 35(7), 617632.

    ,

  • Greenstein, S., & McDevitt, R. (2012). Measuring the broadband bonus in thirty OECD countries (OECD Digital Economy Papers, No. 197). Paris, FranceOECD.

    ,

  • Hausman, J. (1999). Cellular telephone, new products and the CPIJournal of Business and Economic Statistics, 17(2), 188194.

    ,

  • IAB Europe. (2010). Consumers driving the digital uptake: the economic value of online advertising-based services for consumers (White paper). Brussels, Belgium: Author.
  • Kalapesi, C.Willersdorf, S., & Zwillenberg, P. (2010). How the internet is transforming the U.K. economyBoston, MABoston Consulting Group.
  • Larson, D. M. (1993). Joint recreation choices and implied values of timeLand Economics, 69(3), 270286.

    ,

  • Loomis, J. (2011). New approach to value urban recreation using visitors’ time allocations. Urban Public Economics Review [online].
  • Nie, N. D., & Hillygus, D. S. (2002). The impact of internet use on sociability: time-diary findingsIT & Society, 1(1), 120.
  • OECD. (2012). Internet economy outlook 2012Paris, FranceAuthor.
  • OECD. (2013a). Science, Technology and Industry Scoreboard 2013Paris, FranceAuthor.

    ,

  • OECD. (2013b). Measuring the internet economy: a contribution to the research agenda (OECD Digital Economy Paper, No. 226). Paris, FranceAuthor.

    ,

  • Pantea, S., & Martens, B. (2014). Has the digital divide been reversed?: Evidence from five EU countriesElectronic International Journal of Time Use Research, 11(1), 1342.

    ,

  • Penard, T.Poussing, N., & Suire, R. (2011). Does the internet make people happier?(CEPS/INSTEAD Working Paper Series 2011-41). Luxembourg: CEPS/INSTEAD.
  • Van Dijk Management Consultants. (2012). Broadband internet access costs, European Commission, Directorate General for Communications NetworksContent and Technology Report.
  • Wallsten, S. (2013). What are we not doing when we’re online (NBER Working Paper No. 19549). Cambridge, MA: National Bureau of Economic Research, Inc.

    ,

  • Ward, M. R. (2012). Does time spent playing video games crowd out time spent studying? 23rd European Regional ITS Conference, Vienna 2012 60374, International Telecommunications Society, ViennaAustria.

原文来源:Journal of Media Economics

原文作者:Smaranda Pantea & Bertin Martens

原文网址:https://www.tandfonline.com/doi/full/10.1080/08997764.2015.1131701


Comments are closed.



无觅相关文章插件