Transcript Nakamura_Soloveichik_slidesx

Valuing ‘Free’ Media Across
Countries in GDP
By Leonard Nakamura (FRB of Philadelphia)
and
Rachel Soloveichik (BEA)*
These represent our views and not those of the Philadelphia Federal Reserve, the
Federal Reserve System, or the Bureau of Economic Analysis.
www.bea.gov
Overview
▪ How to evaluate “free” media and its impact on personal
consumption expenditures (PCE)?
 What is the value of TV or Facebook or Google in GDP or PCE?
▪ Some researchers estimate that ‘free’ media in the US provided $2
trillion of consumer surplus (Brynjolfsson and Oh 2012)
 Their estimate is based on time use data for TV and Internet.
▪ For the same year, we estimate the ‘free’ media added only $76
billion to GDP in the United States.
 Our methodology is tied to payments to content providers
▪ Important for GDP to be tied closely to expenditures
 Our experimental methodology is in the tradition of valuing products at cost
even when the consumption is not purchased (e.g., government) or unpriced
(owner occupied housing or financial intermediation services)
 We calculate prices and real values by measuring input costs such as actor
salaries, software costs, server costs and consumer media costs like TV sets.
www.bea.gov
2
Global ‘Free’ Media vs. Global GDP (Nominal)
▪ From 1980 on, ‘free’ media hovered around 0.4% of global GDP
 Including ‘free’ media in GDP doesn’t change nominal GDP growth
 We will show that real GDP growth does rise slightly – but the change is very
small and sensitive to the price indexes used.
www.bea.gov
3
Current SNA Treatment of Free Media
and Our Experimental Treatment
▪ In the SNA and the US NIPA, free media is simply an
intermediate cost of the firms whose products are advertised
 A soap opera is a free byproduct of the sale of soap
 The utility gain to consumers is completely uncounted.
▪ Our Experimental Treatment:
 Measure the cost of the consumers’ desired content (soap opera) that is
subsidized by advertising (the sale of soap)
 The content is consumption, valued at the cost of producing the soap opera
 The advertiser and the consumer engage in a barter transaction in which the
consumer agrees to buy the TV content (computer, radio, newspaper) and
watch (listen to, read) the advertisement in exchange
 There is a balancing whereby the income paid to the consumer is exactly
equal to the consumption of the advertising (as in any barter transaction)

Thus the consumption of the soap opera doesn’t come out of nowhere
▪ We explore implications across nations for this methodology
www.bea.gov
4
Advantage of Experimental Treatment
▪ Under the current SNA and NIPA treatment, when content goes
from a paid format to “free” media, consumption declines
 For example, when TV came to the US in the 1950s, real consumption of
recreational services declined, although real incomes rose strongly
 When Jerry Seinfeld or Tina Fey write books, that contributes to PCE, but
their broadcast TV programs do not.
 Internet firms like Facebook do not show up in PCE
▪ With the experimental treatment
 Can make comparisons across countries, based on rates of adoption of media
 Public vs. Private TV broadcasts are now on equal footing
 We can begin to explore how to measure the technological progress in
content provision and in media
 We can study future technological progress in media platforms, distribution,
reception and content.
www.bea.gov
5
Historical Research on “Free” Media
▪ Borden (1935) was an early exploration of the proportion of
advertising devoted to subsidizing content provision
▪ Extensive discussion of measuring “free” media in national
accounts in the 1970s
 Ruggles and Ruggles (1970), Okun (1971), Jaszi (1971), Eisner (1978). Kendrick
(1979)
▪ Cremeans (1980) proposed a barter mechanism for measuring
free media similar to the one we propose and estimated it
▪ Vanoli (2000) discussed the issue in a review of the history of
national accounting
▪ Nakamura (2005) modeled the consumption gains from an
expenditure model
▪ Soloveichik (2014) revived this approach for US GDP
www.bea.gov
6
Empirical Analysis
▪ The World Advertising Research Council (WARC)
provided our main dataset
 WARC reports advertising expenditure by country, media
type and year from 1980 onwards.
 WARC’s data is better for recent years, larger countries and
wealthier countries.
 When calculating global totals, we impute missing country data.
▪ Other Datasets Used
 We took our data on public broadcasting from the European
Audiovisual Observatory Yearbooks and other sources.
 The World Bank provides background information on
national GDP, government quality, health, etc.
▪ This paper documents correlations, not causality.
 We focus on cross-country comparisons because most of the
variables studied remain very stable across time.
www.bea.gov
‘Free’ Media vs. GDP Per Capita in 2010
▪ ‘Free’ media accounts for a higher GDP share in wealthy countries
 Including ‘free’ media in GDP raises nominal inequality across countries.
www.bea.gov
8
Advertising-Supported Media by Language
▪ This correlation remains significant if we remove the US.
▪ The correlation isn’t explained by GDP per capita, government
spending, press freedom or anything else we could find.
www.bea.gov
9
Public Broadcasting Funding vs.
Advertising-Supported Television and Radio
▪ These results are consistent with the ‘crowd-out’ literature.
 However, we can’t determine the direction of causality.
www.bea.gov
10
Public Broadcasting Funding vs.
Advertising-Supported Print and Internet Media
▪ In countries with public broadcasting, advertisers appear to
substitute from broadcast to print or online.
 The net correlation between public broadcasting and advertising is small.
www.bea.gov
11
Advertising-Supported Media vs. Personal Consumption
▪ This correlation isn’t affected by including country-fixed effects
 Very few ads discuss government spending or investment goods, so it’s not
surprising that advertising is higher when consumer spending is higher
www.bea.gov
12
Measuring Media Prices Over Time
▪ Advertising-Supported Entertainment is hard to price.
 Consumer preferences differ across people and over time.
 The media experience depends on not only the media
program itself, but also consumer inputs like plasma TV’s.
▪ We use a two-step process to measure media prices.
 First, we estimate costs in the United States for each media
type from 1980 to 2013.
 In the United States, we combine input costs and output
prices for similar products to construct of price indexes
 Second, we estimate relative prices across countries.
 By construction, average global media prices track US prices.
www.bea.gov
US Prices for Online Media
▪ Software is the main input, so online media prices track it.
 Internet companies also require a few computers to run their software and a
few customer service people to deal with miscellaneous issues.
www.bea.gov
14
US Prices for Newspapers and Magazines
▪ We use book prices a proxy for newspaper writing costs.
 Newspapers (typically) require more outside research than books, so we
include telephone service and online media costs in our index.
www.bea.gov
15
US Prices for Television and Radio Broadcasts
▪ In a separate project on entertainment, BEA estimated the cost of
nonsports programs. We use that price index
 Consumer television prices also influence the watching experience.
www.bea.gov
16
Measuring Media Prices Across Countries
▪ We divide media into two categories: global and local
 Global media is equally valuable in every country and culture.
 Local media isn’t useful unless it’s customized.
▪ The Olympics is an example of global media
 The Olympics has a huge fixed cost to produce, but the
marginal cost of licensing rights is nearly zero.
 National content prices may not track local production costs.
▪ Newspapers are an example of local media
 Each city typically has its own newspaper, and almost nobody
reads global newspapers.
 Newspaper prices will depend on local production costs.
▪ Assumed local/global mix for each media type:
 Newspapers are 100% local; magazines are 95% local; radio is
95% local, television is 50% local and online is 25% local.
www.bea.gov
Olympic Fees vs. Total Broadcast Advertising
▪ We assume global media prices track total advertising spending.
▪ Small countries could increase consumer welfare with price
controls for imported media.
www.bea.gov
18
Newspaper Prices vs. Purchasing Power Parity
▪ Wealthy countries generally have higher costs on both variables.
▪ Controlling for wealth, there’s no relationship between
newspaper advertising prices and advertising market size.
www.bea.gov
19
Global Quantity Indexes for ‘Free’ Media
▪ We estimate that real ‘free’ media grew 6.7% per year, about 4%
faster than overall real GDP.
 Including ‘free’ media in GDP raises growth rates only 0.02% per year
www.bea.gov
20
Conclusion
▪ We present an experimental methodology for including “free”
media in personal consumption expenditures
▪ This would allow the Internet and TV, which consumers use a lot
as part of leisure, to enter personal consumption expenditures
▪ As media content transitions back and forth between pay and
free venues, this allows a more even-handed treatment
▪ Measuring media prices is challenging; we introduce a two-step
process that separates local from global media
▪ Having measured media, we can then explore its evolution across
countries and over time
 Public TV crowds advertising out of TV and into print in cross-section
 Internet gains as print declines over time
www.bea.gov
21