Whats Fit to Print - The Effect of Ownership Concentration of Product Varieties in Daily Newspaper Markets - Lisa George

INFORMATION

Information Economics and Policy 19 (2007) 285–303

ECONOMICSAND POLICY

www.elsevier.com/locate/iep

What’s fit to print: The effect of ownershipconcentration on product variety in

daily newspaper markets

Lisa George *

Department of Economics, Hunter College and The Graduate Center, CUNY,

695 Park Avenue, New York, NY 10021, United States

Received 7 June 2006; received in revised form 5 December 2006; accepted 17 April 2007Available online 1 May 2007

Abstract

This paper examines the effect of ownership concentration on product position, product varietyand circulation in the US daily newspaper market. The effects of consolidation in differentiated prod-uct markets cannot be determined solely from theory. Because multi-product firms internalize busi-ness stealing, mergers may encourage firms to reposition products, leading to more, not less, variety.Using data on the assignment of reporters to topical areas at 706 newspapers in 1993, 1999 and 2004,results show that both differentiation and variety increase with ownership concentration. Moreover,greater concentration increases variety over a range of topics and does not reduce readership.� 2007 Elsevier B.V. All rights reserved.

JEL classifications: L1; L8

Keywords: Media; Newspapers; Ownership concentration

1. Introduction

Regulation of media markets in the US presumes that larger numbers of owners andproducts in a market lead to more variety. Limits on the number of radio stations in amarket owned by a single firm, protection of newspaper joint operating agreements, a

0167-6245/$ - see front matter � 2007 Elsevier B.V. All rights reserved.

doi:10.1016/j.infoecopol.2007.04.002

* Tel.: +1 212 772 5437.E-mail address: [email protected]

mailto:[email protected]

286 L. George / Information Economics and Policy 19 (2007) 285–303

prohibition against cross-ownership of broadcast and print media products in a market,and antitrust enforcement against newspaper mergers all attest to this presumption.

Yet, it is not obvious that more owners give rise to greater variety. Media are differen-tiated products produced with large fixed costs and advertiser finance. It is well knownthat such markets can deliver too much, or too little, variety. Excess variety in the formof duplicative coverage can arise if revenue from capturing only a fraction of one con-sumer type covers fixed costs (Steiner, 1952). At the same time, markets can fail to providespecialized coverage when revenue obtained from targeting particular groups is less thanthe cost of developing new content (Spence and Owen, 1977). Because multi-product firmsinternalize business-stealing externalities, mergers can lead owners to eliminate duplicativeproducts and change the content of others. Various production economies, as well ashigher revenues, brought about by consolidation can also make new content viable. Theeffect of ownership concentration on content variety is therefore an empirical question.

This paper examines the effect of ownership concentration on product position, productvariety, and readership in markets for daily newspapers. Newspapers provide a useful set-ting for studying the effects of concentration on variety for several reasons. First, policyinterest in newspapers suggests the importance of understanding factors that lead to greatervariety. Second, high fixed costs and other aspects of newspaper production limit to a hand-ful the number of dailies produced in any market, raising the consequences of positioningdecisions by individual firms. Newspaper owners also appear to have little scope for pricediscrimination that might otherwise allow for provision of content demanded by smallgroups.1 Finally, the 1990s saw a sharp increase in newspaper mergers and acquisitionswhich aids empirical investigation of the effect of concentration on variety.

This study uses newspaper-level information on the assignment of more than 25,000reporters and editors to different topics in 1993, 1999 and 2004 to characterize the separa-tion between products and amount of content variety available in US newspaper markets.Using a simple distance formula and the number of unique topics to measure the degree ofdifferentiation among papers in each market at each point in time, results show that adecrease in the number of owners in a market leads to an increase in separation betweenproducts. Moreover, the number of topics covered per market also increases with owner-ship concentration. Finally, there is evidence that the additional coverage brought aboutby consolidation does not reduce readership.

The paper proceeds as follows. Section 1 reviews relevant theory and literature. Section2 outlines the data and content measures used in the analyses. Section 3 describes theempirical strategy and Section 4 presents results. Section 5 considers welfare and policyimplications and concludes the paper.

2. Theory and literature

Several features of differentiated product markets complicate simple theoretical predic-tions about the effects ownership concentration. When products are somewhat substitut-able for one another, they tend to divert or ‘‘steal’’ business from each other. With

1 See Reddaway (1963) and Rosse (1970) for background on the newspaper industry. There is very littlevariation in newspaper prices across markets or over time, with approximately 80% of daily newspapers auditedby ABC charging $.50 from 2000 through 2004. Increased ownership concentration is also not associated withhigher cover prices. High fixed costs also seem to make the proliferation of tailored content through specializednewspapers very costly, since only the largest markets can support legal, industry, or business dailies.

L. George / Information Economics and Policy 19 (2007) 285–303 287

atomistic ownership and fixed costs, there can be a tendency for duplicative products(Spence, 1976a,b and Dixit and Stiglitz, 1977). Joint production of multiple products,sometimes brought about by mergers, internalizes this business-stealing externality. As aresult, a decrease in the number of owners in a market may increase, rather than decrease,available variety.

In newspaper markets, an owner who acquires a local competitor’s paper similar to hisown would likely not continue to operate both papers in their previous forms. Rather,the owner might differentiate them by altering their content, or close one of the papersaltogether. Differentiation might take the form of simply eliminating duplicative contentfrom one paper, replacing duplicative content with new material, or shifting emphasisamong reporting topics. In closing one of the papers, the owner might add content tothe remaining product to prevent competitor entry or maintain readers with particularpreferences.2 If the acquisition allowed the owner to increase advertising revenue perreader, the owner might also introduce content previously unavailable in the market(Anderson and Coate, 2005). Hence even mergers that reduce the number of newspaperswould not necessarily reduce variety. The effect of ownership concentration on variety isan empirical question.

Although there is an extensive theoretical literature on the effects of concentration onproduct position and product variety, there is relatively little empirical evidence on thequestion.3 Most closely related to this study is Berry and Waldfogel (2001), which exam-ines the effects of ownership concentration on programming variety in radio broadcasting.They find that consolidation triggered by the Telecommunications Act of 1996 reducedentry but increased the number of radio formats broadcast, both absolutely and relativeto the number of stations in a market. In newspaper markets, a large literature in sociologyand communications examines relationships between ownership characteristics and cover-age using data on the content of individual newspaper articles. Lacy (1991), one of themore extensive studies in this literature, finds that chain-owned papers contain shorterarticles and devote less space to news and editorial beats than independently ownedpapers, but that they also devote more staff resources for a given allocation of space. Lacy(1987) also found higher staff allocations in markets with competing dailies. Hicks andFeatherstone (1978) find evidence that newspapers under common ownership tend to pro-duce less overlapping content.4

A related literature examines the industrial organization of media markets generallyand the effects of concentration in particular. Early papers by Reddaway (1963) and Rosse(1970) establish the importance of fixed costs in newspaper production. Chaudhri (1988)and Blair and Romano (1993) examine the pricing decisions of a newspaper monopolistearning revenues from circulation and advertising. Both find that features of the newspa-per market can lead to lower consumer prices with monopoly ownership. Empirically,Bucklin et al. (1989) find no evidence of a relationship between concentration and circu-

2 The tendency for media firms to produce content appealing to groups with distinct preferences is considered inWaldfogel (2003) generally and George and Waldfogel (2003) for newspapers.

3 Hotelling (1929) provides a foundation for the literature on product positioning. Spence (1976a,b) and Dixitand Stiglitz (1977) form the basis of the theoretical literature on product variety.

4 These studies are the most closely related to this work due to their focus on the relationship betweenconcentration and aspects of variety. However, a large literature in communications considers the relationshipbetween market or owner characteristics and content more generally. See Compaine (1982, 1995) and Picard(1988) for useful reviews.


lation or advertising prices, although Reimer (1992) provides evidence from a small (35city) sample that concentration may lead to lower prices for advertisers. Dertouzos andTrautman (1990) examine the effect of concentration on firm costs, finding significant scaleeconomies in the production of news but no evidence of efficiencies from consolidation. Afinal line of research uses newspaper firms as an avenue for the study of technology adop-tion and diffusion (Dertouzos and Quinn, 1985; and Genesove, 1999).

This paper contributes to this literature in two key ways. First, a unique and compre-hensive set of data allows characterization of newspaper content across an entire marketover time, allowing a more general understanding of the effects of ownership concentrationon coverage than is possible using content analysis over a limited subset of papers in across-sectional framework. Second, this data allows straightforward measures of productdifferentiation in a market that can test theoretical predictions of the effects of ownershipconcentration.

3. Data sources and content measures

3.1. Data overview

The basic data used in this study is a market-level panel of product differentiation, vari-ety and circulation measures along with the number of newspapers and newspaper ownersin each market. The data covers 263 Metropolitan Statistical Areas (MSAs) in 1993, 1999and 2004.

The market-level panel is constructed from newspaper-level data on the reporting top-ics covered by individual reporters and editors. This data comes from the 1994, 2000 and2005 editions of Burrelle’s Media Directory, a directory of newspaper staff produced byBurrelle’s Information Services, a media monitoring organization. Each edition reportsthe job title of more than 20,000 reporters and editors at the 1400–1500 daily newspaperspublished across the US. This research considers papers published within MSAs, whichtotal 750–850 papers over the period of the study. The final sample, which excludes thesmallest 10% of papers (circulation below 5700) and 8 MSAs, considers 12,435 reportersat 706 papers in 1993, 14,298 reporters at 664 papers in 1999 and 17,845 reporters at 646papers in 2004.5 For this study, job titles are linked by keyword to 76 reporting topics.The distribution of reporters across topics is then used to construct market-level mea-sures of product differentiation and variety. Construction of these measures is discussedbelow.

Newspaper data are supplemented with zip code-level circulation data from the AuditBureau of Circulations (ABC) Circulation Data Bank and population data from the 1990census. ABC is a membership organization that sets standards for reporting newspapercirculation and audits publisher statements for use by advertisers. The data reflect paidnewspaper sales at newsstands, by home delivery, and by mail. Zip code-level circulationdata are available for two periods only, 1995 and 2000.

5 Reporter data are missing for many very small papers in several time periods, so the smallest 10% of papers(circulation less than 5700) are excluded from the sample. Results are not sensitive to the cutoff. Reporter data isalso missing for several large papers, necessitating exclusion of 7 markets (Atlanta, Detroit, Goldsboro,Honolulu, Lake Charles, Little Rock, Washington-Baltimore, Rochester, NY). The sensitivity of results to thesize of papers in the sample is considered at the end of Section 5.


3.2. Measuring variety

Burrelle’s Media Directory is essentially a staff directory of daily newspapers publishedacross the US which categorizes reporters and editors into about 100 topical areas such asbusiness, food, entertainment, and travel. For this study, job titles are linked by keywordto 76 reporting topics based on those delineated in Burrelle’s. For example, if a reporter’stitle includes the word ‘‘food’’, ‘‘dining’’, ’’cooking’’, ‘‘restaurant’’ or ‘‘wine’’, the reporteris assigned to the Food and Dining category. If the title contains ‘‘music’’, the reporter isassigned to the Music category. To maintain exclusive categories, some topics were con-solidated. For example, both restaurant reviewers and cooking writers are combined intothe Food and Dining category because titles such as ‘‘Food Writer’’ cannot be disaggre-gated. Burrelle’s directory includes administrative and support positions such as researchand classifieds, but these are excluded from the analysis.

The distribution of reporters across topics is then used to construct market-level mea-sures of product differentiation and variety. The basic measure of variety used in this paperis the number of different topics covered by reporters in a market. For example, if onepaper in a two-paper MSA assigns one reporter to local news and one to sports and theother paper assigns one reporter to local news and one to entertainment, the total numberof beats covered in the market would be three.

The primary advantage of this reporter-based measure over characterizations based oncontent analysis is that using reporters allows straightforward characterization of a com-prehensive sample of papers over time. In contrast, the complexity of measuring output byevaluating published articles typically limits researchers to a small fraction of all publishedpapers in a cross-sectional framework, introducing the potential for sample selection andunobserved heterogeneity to bias results.6 Using reporter assignments to characterize mar-kets also allows simple scalar measures of product differentiation. Finally, the notion thatmore variety in reporter assignments corresponds to more variety in coverage is alsohighly intuitive. Does a market have a travel editor or not? A music critic? A political ana-lyst? The presence or absence of a reporter in a particular area is directly related to choicesavailable to readers and hence constitutes a meaningful and interesting variable to study.

The primary limitation of using reporter assignments to measure variety is that cover-age areas are inputs rather than outputs and hence do not directly correspond to the num-ber and types of articles published across newspapers over time. Reporter assignments arethus likely to be a noisier measure of coverage than content-based measures, as reporterswith similar titles may cover slightly different areas at different papers. The standard errorson regression coefficients may thus in some cases be too large for valid inference.7 Overall,results are robust to alternative categorization, which reduces concerns about the accuracyof reporter assignments or categorization.

6 Content analysis, as frequently employed in the communications literature, involves categorizing publishedarticles, counting references to particular subjects, or measuring the space devoted to different topics. Even largestudies that use content analysis such as Lacy (1987, 1991) cover only about 100–200 papers.

7 Of greater concern is the potential for biased coefficient estimates, which would occur if the link betweenreporter assignments and the content of newspaper articles differs across markets in ways that are correlated withownership concentration. Since small markets tend to have higher ownership concentration than large markets,coefficient estimates would tend to bias results against finding a negative relationship between ownershipconcentration and variety.


It is worth noting at this point that the number of topics covered in a market is not theonly variety measure of interest to consumers and policy makers. While consumers mayprefer more topics to fewer, they may also value coverage of a single topic from differentperspectives. This paper does not address what is sometimes called ‘‘viewpoint diversity’’directly. However, many aspects of viewpoint diversity are captured with reporter data.For example, consider two markets each with two newspapers. If one market has two gen-eral business reporters and the other has one business reporter and one labor union repor-ter, the second would be recorded in this data as having more variety than the first. Thesecond market might also be expected to have greater viewpoint diversity as well. Simi-larly, a market with two popular music critics clearly offers less variety than a market withone popular and one classical music critic and may also be considered to have less view-point diversity. Overall, the extent to which reporter-based measures of variety also reflectsviewpoint diversity hinges on the level of detail available in reporting categories. However,despite this link, this paper does not attempt to evaluate the effects of ownership concen-tration on all aspects of content variety. Instead, the goal is to carefully measure the effectson a single but meaningful measure, the number of topics covered in a market. Conclu-sions about welfare and policy that might be drawn, and should not be drawn, from thisanalysis are discussed at the end of the paper.

It is useful to briefly describe variety data used in the study. General news, sports, localnews, business, and editorial commentary are the largest beats covered in most markets byvirtually all papers. Food, features, lifestyle and entertainment coverage is also very widelyavailable. Topics such as travel, education, fashion and health are less widely available,with coverage at about half of papers in 60% of markets. A large number of topics aremore specialized: coverage of gardening, family, agriculture, science, religion and bankingare available in fewer than half of MSAs at fewer than one third of papers. Specializedindustry, arts and policy topics are produced only by a small number of reporters in thevery largest markets. Staff allocations also differ at large and small papers, with smallpapers devoting about two thirds of reporting resources to the ten largest topics (GeneralNews, Sports, Editorial, Lifestyle, Business, Food, Fashion, Education, Features andRegional News.) Larger papers assign less than half of reporting resources to these areasand a greater fraction of reporters to topics outside of the top 10. These differences suggestthat, to the extent mergers and acquisitions produce larger papers, the number of topicscovered in a market may increase.

3.3. Measuring product differentiation

In addition to measuring content variety on an absolute scale, the distribution ofreporters is used to compute a measure of the distance between, or differentiation among,products in each market. The change in the measure following consolidation captureswhether or not mergers induce firms to spread products apart in product space. Charac-terizing each newspaper as a vector of shares devoted to each reporting topic, differentia-tion in a two paper market can be measured as the Euclidean distance between the vectors,or half the distance to the midpoint. In markets with more than two papers, it is the aver-age distance to the mean assignment

d ¼ 1

P

XP

i¼1

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðsi � �sÞ0ðsi � �sÞ

q; ð1Þ


where si = (si1, si2, . . ., sib), sib is the share of reporters assigned to beat b at paper i, P is thetotal number of papers in the market, and �s is a vector identifying the average share ofreporters assigned to each beat across all papers in a market.

To illustrate the intuition behind the distance measure, consider two markets with twopapers assigning reporters to Sports, News, and Entertainment. In the first market, assumethat both papers assign one reporter to each beat, with a vector of shares for each papergiven by: 1

3; 1

3; 1

3

� �. In the second market, consider the case where one newspaper assigns

one reporter to each beat and the other assigns two reporters to sports and one to news, witha vector of shares of 1

3; 1

3; 1

3

� �for the first paper and 2

3; 1

3; 0

� �for the second. Differentiation in

the first market is zero. In the second market, the average share of reporters assigned toSports, News, and Entertainment is 1

2; 1

3; 1

6

� �. Differentiation in the market is then 0.24. If

instead of two reporters devoted to Sports and one to News the second paper in the secondmarket assigned all three reporters to sports, differentiation would increase to 0.42.

An alternative, simpler approach to measuring product differentiation would be tocount the number of ‘‘unique’’ topics covered by only a single paper in a market. In theexample above, if both papers in a market allocate resources to Sports, News and Enter-tainment, the number of unique topics is zero. If one paper covers only Sports and News,Entertainment is a unique topic. Holding the number of papers in a market constant,fewer owners differentiating their products would produce a larger number of uniquetopics.

3.4. Measuring ownership concentration

The primary independent variable in the paper is ownership concentration, measured asthe absolute number of owners in a market and as the number of ‘‘owner-equivalents.’’The number of owner-equivalents is the inverse of the Herfindahl index, defined as theinverse squared sum of ownership market shares, 1PO

o¼1sh2

o

, where sho is the share of circu-

lation of owner o in a market. (Newspaper equivalents are similarly calculated with news-paper rather than ownership shares of circulation.) When owners in a market are equal interms of circulation, the number of owner-equivalents is equal to the number of owners.When circulation is not equal across owners, the number of owner-equivalents is less thanthe number of owners. In general, owner-equivalents are a better measure of concentrationin markets where circulation shares are not equal across firms. For this study, however,using the absolute number of owners provides some advantages. In particular, the numberof owner-equivalents depends on newspaper circulation, which is measured with consider-ably more error than the number of owners or papers in these data because Burrelle’s doesnot disaggregate circulation within and outside the home market. Also, because of itsdependence on circulation, the number of owner-equivalents in a market changes evenwhen the number of owners does not, introducing further error into the study of owner-ship changes. For these reasons, both the absolute number of owners and the number ofowner equivalents are presented throughout.

Another important consideration in calculating ownership concentration is the defini-tion of a newspaper market. The size of the market affects both the precision of coefficientestimates as well as their interpretation. If the market definition is too large, the effect ofownership concentration on variety is likely to be estimated with considerable error.Moreover, with a large geographic region, a repositioning of products that leads to an


increase or decrease in concentration may affect a large number of consumers. Consumerswho benefit from additional variety may not be the same as those that lose variety, intro-ducing distributional issues that complicate conclusions about welfare. With a smaller geo-graphic market, coefficient estimates are likely to be more precise and distributional issuesless of a concern. However, the market must be large enough to encompass potential sub-stitutes for most papers. If potential substitutes are omitted, then product repositioningthat occurs in response to ownership changes will not be captured.8

To illustrate, consider residents of a suburban area that subscribe to a county newspa-per. If the county newspaper should close, some residents may choose to read a paper pub-lished in an adjacent county, some may subscribe to the closest urban paper, and othersmay not subscribe at all. Owners of both the adjacent county paper and urban paper mightadd or alter content in response to closure of the county paper. Defining a newspaper mar-ket as a county would ignore repositioning of nearby papers, hence estimates of the effectof ownership concentration on variety would be meaningless. Defining the market toinclude all papers in the metro area and surrounding rural areas would make repositioningby two papers more difficult to detect, but would not otherwise bias results.

This study defines a newspaper market as a metropolitan statistical area (MSA) or con-solidated metropolitan statistical area (CMSA), typically a city and adjoining suburbs.This geographic area is large enough to take into account that the closest substitute formany local papers is the central city newspaper in the MSA. The primary disadvantageof using MSAs rather than a larger market is that about half of the approximately 1500daily newspapers published in the US are located outside of metropolitan areas, and theseregions have seen significant increases in ownership concentration. If mergers withinMSAs trigger repositioning of papers beyond MSA boundaries, the analyses here willnot capture these effects. However, the papers outside of MSAs are typically smaller thanmetropolitan papers. Excluding them thus helps ensure that the effects of consolidationmeasured here are not restricted to small papers at the fringes of a market. The robustnessof results to alternative market definitions are considered in Section 5.

3.5. Sample statistics

Table 1 summarizes market-level measures of variety, product separation, and owner-ship concentration as well as the underlying newspaper-level data on reporting topics. Thenumber of papers drops from 706 to 647 over the time period of the study, with the aver-age number of topics covered per paper dropping from 17 to 12. The smallest papers coverabout 3 reporting beats and the largest about 40 across all years. At the market level, themean number of topics covered per market drops from 24 to 19 over the period of thestudy while the average number of owners and papers drop only slightly, 2.23 to 1.95for owners and 2.68 to 2.45 for papers. Averages for owner and newspaper equivalentsare slightly smaller. The number of topics covered varies considerably across markets,ranging from 5 to 7 at the 5th percentile to more than 50 at the 95th percentile. The mean

8 The importance of overlapping markets in newspaper competition was a key conclusion in Dertouzos andTrautman (1990), which states: ‘‘Using the total price elasticities, it is easy to show that a 10% advertising andcirculation price increase would lead to smaller offsetting output declines of 15% for advertising volume and 8%for circulation. This implies that any definition of the relevant media market should include newspapers havingoverlapping markets even if they are not located in the same city.’’

Table 1Sample statistics

Year N Mean SD 5% 25% 50% 75% 95%

Paper level statistics

Topics 1993 706 17.17 10.06 3 9 16 24 36Topics 1999 664 16.63 12.03 3 7 14 23 42Topics 2004 647 12.43 11.06 2 5 9 15 38

Market level statistics

Papers 1993 263 2.68 4.09 1 1 2 3 7Paper equivalents 1993 263 1.61 1.01 1.00 1.00 1.17 1.92 3.00Owners 1993 263 2.23 2.65 1 1 1 2 6Owner equivalents 1993 263 1.47 0.88 1.00 1.00 1.00 1.68 2.71Topics 1993 263 24.37 10.93 7 15 24 33 41Distance 1993 263 0.09 0.11 0.00 0.00 0.08 0.18 0.26Unique beats 1993 263 15.05 8.29 4 9 14 20 31




distance among papers in a market is about 0.1 over all years, while the number of uniquebeats drops from 15 to 10 over the period of the study.

4. Empirical strategy

The goal of the empirical analyses is to measure the effect of ownership concentrationon the separation among papers and the number of topics covered in an MSA. Themost prevalent empirical approach in the communications literature is to regress contentattributes on market share or other observable market characteristics such as popula-tion, income, and education. Although cross-sectional analyses can in some cases beinformative, this approach is vulnerable to the concern that unobserved differencesacross markets affect both product characteristics and market observables, which couldbias results.

The method pursued in this study, which avoids this concern, is estimation with lon-gitudinal data before and after a period of intense merger activity. After remainingsteady through the late 1980s, acquisitions increased markedly beginning in 1993. Thenumber of papers changing hands jumped from an average of about 40 per year in


the early 1990s to about 100 per year for the remainder of the decade. Acquisition activ-ity dropped considerably in 2001 and has remained steady at about 40 per year since thattime. The acquisition period coincides with a period of overall economic expansion,which is taken here to be exogenous and unrelated to changes in preferences for contentvariety.

With observations in 1993, 1999 and 2004, fixed-effects regressions can be used to iden-tify the relationship between changes in concentration and changes in variety. If unob-served preferences for newspaper content across markets remain constant over time,this approach produces unbiased estimates of the effect of changes in the number of own-ers or owner-equivalents on changes in differentiation or variety. All of the relationshipsexplored in Section 5 are estimated using market fixed effects. The potential for ownershipchanges themselves to be endogenously related to preferences for newspaper content isconsidered in the discussion of robustness at the end of that section.

5. Results

This section proceeds in three parts. The first examines the relationship between own-ership concentration, product positioning, and product variety. The second considers thenature of the additional variety generated by increased ownership concentration. The thirdconsiders per capita newspaper circulation.

5.1. Ownership concentration, product positioning, and product variety

Theory suggests that because multi-product firms are better able to internalize business-stealing, owners want their papers to look different from each other and appeal to differentreaders. One way that newspaper owners might differentiate products is to shift the allo-cation of reporters assigned to particular beats. Recall from Section 3 that differences inemphasis among papers in a market can be measured as the average Euclidian distanceamong papers or as the number of unique beats covered in a market, holding the numberof papers constant. Table 2 shows the results of fixed-effects regressions of changes in the

Table 2Ownership concentration and product differentiation

Distance (1) Distance (2) Unique topics (3) Unique topics (4)

Owners �0.011 (2.34)* �1.97 (3.18)**

Papers 0.013 (3.35)** �0.25 (0.48)Owner equivalents �0.031 (2.91)** �2.88 (2.05)*

Paper equivalents 0.056 (5.45)** �4.55 (3.28)**

Year 1994 �0.012 (3.07)** �0.014 (3.73)** 5.61 (11.24)** 5.80 (11.45)**

Year 1999 �0.001 (0.37) �0.003 (0.72) 5.40 (11.16)** 5.53 (11.41)**

Constant 0.095 (11.43)** 0.063 (3.87)** 14.52 (13.27)** 20.78 (9.56)**

N 789 789 789 789MSAs 263 263 263 263

Notes: T-statistics in parentheses: + significant at 10% level; *significant at 5% level; **significant at 1% level.Dependent variable in columns (1) and (2) is mean distance to average topic assignment. Dependent variable incolumn (3) and (4) is the number of unique beats in a market. See text for details. All specifications include MSAfixed effects. Constants in fixed effects regressions represent the average value of the fixed effects.


differentiation measures on changes in ownership concentration. The dependent variablein the first two columns is distance and in the third and fourth columns is unique topics.The relationship is negative and significant in all cases. Holding the number of papers con-stant, a decrease of one owner in the market leads to an increase in distance among papersby 0.01 on a base of 0.1, with a slightly larger effect for owner equivalents. A decrease ofone owner reduces the number of unique beats by 2 on a base of about 14, again with aslightly larger effect for owner equivalents. Increased ownership concentration appears toincrease the distance between products.

Product positioning is interesting because predictions emerge directly from theory onfirm behavior in differentiated product markets. However, the total amount of varietyavailable in a market offers a more direct link to consumer welfare. As a first step inexamining the effect of ownership concentration on total product variety, it is usefulto ask how an increase in concentration affects the number of papers in a market. Thisis done by regressing changes in concentration measures on changes in the number ofpapers. Table 3 presents results. The first column shows the relationship in terms ofchanges in the number of owners. The second column show the effect in terms of ownerequivalents. In each case, the relationship between owners and papers is large in magni-tude and statistically significant. A decrease of one owner in a market reduces the num-ber of papers by 0.77, while a reduction in one owner-equivalent reduces the number ofpapers by 0.43.

A decline in the number of products is a first-order effect that would be expected todecrease variety. However, as discussed above, the full impact of concentration on varietydepends on how owners alter products that remain in the market. Turning to the primaryquestion of the paper, Table 4 shows the effect of changes in the number of owners andowner-equivalents on changes in the total number of reporting topics covered in a market.The first two columns show the effect of changes in owners and papers, while the third andfourth columns show the effect of owner- and paper-equivalents. All specifications includemarket fixed effects. The coefficients on ownership measures are negative and significant(in one case at 10%). On average, the loss of one owner in a market increases the numberof topics covered by about 2.5 beats on a base of 25 total topics, an increase of 10%.Results using owner-equivalents are similar. Ownership concentration appears to increasethe number of topics covered in a market.

It is worth noting that the effect of the number of papers is small and insignificant incolumn (2) but significant and close in magnitude to the ownership coefficient in column

Table 3Ownership concentration and newspaper closure

Papers (1) Paper equivalents (2)

Owners 0.776 (19.99)**

Owner equivalents 0.430 (10.71)**

Year 1994 0.016 (0.38) 0.058 (3.64)**

Year 1999 0.001 (0.01) 0.026 (1.68)+Constant 0.933 (11.51)** 0.916 (16.42)**

N 789 789MSAs 263 263

Notes: T-statistics in parentheses: + significant at 10% level; *significant at 5% level; **significant at 1% level.Dependent variable is number of papers or paper-equivalents in the MSA. All specifications include MSA fixedeffects. Constants in fixed effects regressions represent the average value of the fixed effects.

Table 4Ownership concentration and variety

Topics covered(MSA) (1)




Owners �2.752 (4.94)** �2.350 (3.17)**

Papers �0.518 (0.82)Owner

equivalents�2.673 (1.72)+ �4.903 (2.88)**

Paperequivalents

5.179 (3.09)**

Year 1994 6.149 (10.35)** 6.157 (10.36)** 5.688 (9.33)** 5.390 (8.80)**

Year 1999 5.678 (9.85)** 5.678 (9.85)** 5.548 (9.41)** 5.416 (9.24)**

Constant 24.358 (20.93)** 24.842 (19.06)** 22.609 (10.50)** 17.864 (6.80)**

N 789 789 789 789MSAs 263 263 263 263

Notes: T-statistics in parentheses: + significant at 10% level; *significant at 5% level; **significant at 1% level.Dependent variable is number of topics covered in the MSA. All specifications include MSA fixed effects.Constants in fixed effects regressions represent the average value of the fixed effects.


(4). This suggests that while reductions in ownership concentration increase variety, theeffect may be offset in markets where mergers lead to newspaper closure.

How robust are these results? As outlined in Section 3, one consideration is the defini-tion of a newspaper market. Using MSAs to define a newspaper market balances the need

Table 5Ownership concentration, product differentiation and product variety (DMAs)

Distance(DMA)(1)

Distance(DMA)(2)

Uniquebeats(DMA)(3)

Uniquebeats(DMA)(4)

Topicscovered(DMA)(5)




Owners �0.01(1.86)+

�1.02(2.36)*

�1.84(4.43)**

�1.50(2.88)**

Papers 0.01(2.30)*

�0.23(0.37)

�0.82(1.08)

Ownerequivalents

�0.02(1.69)+

�0.89(0.91)

�0.24(0.24)

�1.69(1.42)

Paperequivalents

0.01(1.53)

�0.31(0.30)

2.95(2.31)*

Year 1994 �0.06(9.67)**

�0.06(10.10)**

3.69(5.74)**

3.10(4.96)**

6.36(8.18)**

6.37(8.20)**

5.05(6.62)**

4.89(6.42)**

Year 1999 �0.03(4.46)**

�0.02(4.36)**

4.23(7.05)**

4.02(6.63)**

6.54(9.02)**

6.57(9.05)**

6.02(8.11)**

5.97(8.09)**

Constant 0.20(6.70)**

0.24(9.61)**

16.80(5.34)**

13.85(5.12)**

35.33(18.28)**

38.87(10.22)**

27.64(11.03)**

22.65(6.87)**

N 534 534 534 534 534 534 534 534DMA’s 173 173 173 173 173 173 173 173

Notes: T-statistics in parentheses: + significant at 10% level; *significant at 5% level; **significant at 1% level.Dependent variable in columns (1) and (2) is mean distance to average topic assignment. Dependent variable incolumn (3) and (4) is the number of unique beats in a market. Dependent variable in columns (5)–(8) is thenumber of topics covered in the DMA. All specifications include DMA fixed effects. Constants in fixed effectsregressions represent the average value of the fixed effects.


to include substitutes while not being so large as to introduce uncertainty over whetherthose affected by consolidation are the same as those affected by repositioning. It is worth-while to compare results, however, with those using a larger market. Table 5 repeats theanalyses in Tables 2 and 4 using designated market area (DMA) delineated by NielsenMedia Research instead of MSAs. DMAs generally correspond closely to MSAs, butinclude non-metropolitan areas that are excluded from MSAs. Coefficient estimates areclose in magnitude to those in Tables 2 and 4, providing some assurance that increasesin variety with MSAs are not offset by decreases outside the metropolitan boundaries.Standard errors are higher, particularly for owner equivalents, suggesting that expandingthe area of study introduces noise into the measurement of variety.

A second concern related to the definition of a market is that a change in the number oftopics covered has more impact at a paper with high circulation than at a paper with lowcirculation. One way of checking whether the results in Tables 2 and 4 are driven by smallpapers is to repeat the analysis with only larger papers. Table 6 presents differentiation andvariety measures for a sample that includes only papers with daily circulation greater than12,000 (the largest 70% of papers). Coefficient estimates are consistent with those in Tables2 and 4, with slightly larger coefficient estimates and smaller standard errors. Cutoffs of10,000 (top 75% of circulation) and 17,000 (top 65%) produce similar results The resultssuggest that the effect of ownership concentration on product positioning and productvariety is not driven by small papers.

A final consideration of robustness is the extent to which unobserved heterogeneitymight affect the results. All of the regressions in Tables 2–8 include market fixed effects,

Table 6Ownership concentration, product differentiation and product variety (large papers)

Distance(MSA)(1)

Distance(MSA)(2)

Uniquebeats(MSA) (3)

Uniquebeats(MSA) (4)

Topicscovered(MSA) (5)




Owners �0.023(5.55)**

�1.65(2.16)*

�4.114(6.04)**

�4.571(5.11)**

Papers 0.030(6.57)**

�1.72(1.98)*

0.796 (0.79)

Ownerequivalents

�0.036(3.97)**

�3.78(2.23)*

�3.292(1.77)+

�6.124(3.01)**

Paperequivalents

0.063(7.32)**

�5.71(3.52)**

6.353(3.26)**

Year 1994 �0.000(0.15)

�0.003(0.97)

6.50(12.88)**

6.73(13.17)**

5.653(9.62)**

5.615(9.52)**

5.133(8.41)**

4.803(7.83)**

Year 1999 0.001(0.25)

�0.000(0.03)

5.83(11.87)**

5.96(12.15)**

5.205(9.07)**

5.184(9.02)**

5.110(8.62)**

4.960(8.42)**

Constant 0.043(6.04)**

0.024(1.86)+

16.20(12.03)**

22.96(9.52)**

25.581(21.41)**

24.772(15.72)**

23.054(9.44)**

17.867(6.17)**

N 780 780 780 780 780 780 780 780MSAs 260 260 260 260 260 260 260 260

Notes: T-statistics in parentheses: + significant at 10% level; *significant at 5% level; **significant at 1% level.Dependent variable in columns (1) and (2) is mean distance to average topic assignment. Dependent variable incolumn (3) and (4) is the number of unique beats in a market. Dependent variable in columns (5)–(8) is thenumber of topics covered in the MSA. All specifications include MSA fixed effects. Constants in fixed effectsregressions represent the average value of the fixed effects.

Table 7Ownership concentration and newspaper characteristics

Topics Constant Owners

Agriculture 0.094 0.051+Arts – general 0.410 �0.029Arts – performing arts 0.251 �0.031Assignment and special issues 0.524 �0.056*

Auto 0.108 0.007Banking and finance 0.286 �0.010Book reviews 0.385 �0.051+Business – general 0.835 �0.023Business – international 0.349 �0.127**

Business – investments 0.346 �0.136**

Business – retail 0.212 �0.079**

Business – small business 0.056 �0.023Business – workplace 0.231 �0.094**

Calendar 0.378 �0.106**

Computers and internet 0.213 0.016Consumer interests �0.015 0.044+Courts 0.393 �0.016Crime 0.413 �0.090**

Defense 0.244 �0.087**

Economic conditions 0.396 �0.128**

Editorial and opinion 0.835 �0.015Education 0.638 �0.031Entertainment – general 0.617 �0.012Entertainment – movies 0.225 �0.018Entertainment – pop culture 0.019 0.002Entertainment – TV 0.218 �0.016Environment 0.296 0.002Family 0.236 �0.002Fashion 0.257 �0.032Features 0.837 �0.008Fitness and nutrition 0.256 �0.078**

Food and dining 0.584 �0.038Gardening 0.129 0.018Govt. and politics – General 0.513 �0.018Govt. and politics – Local 0.354 �0.016Health and medicine 0.559 �0.016History 0.003 0.002Hobby 0.180 �0.076**

Home 0.227 �0.021Humor 0.229 �0.065**

Industry – advertising 0.267 �0.094**

Industry – aviation 0.203 �0.076**

Industry – energy 0.348 �0.110**

Industry – entertainment 0.058 �0.027**

Industry – general 0.197 0.008Industry – healthcare 0.273 �0.089**

Industry – insurance 0.087 �0.035**

Investigative reporting 0.319 �0.102**

Labor and unions 0.061 0.008Lifestyle 0.676 �0.033Magazine 0.027 0.013Minority and diversity 0.277 �0.085**


Table 7 (continued)

Topics Constant Owners

Music – classical 0.202 �0.082**

Music – general 0.234 �0.012Music – pop 0.173 �0.051**

News – general 0.997 0.002News – international 0.067 0.002News – regional 0.867 0.000Obituary 0.240 �0.099**

Photography 0.521 �0.015Real estate 0.294 �0.011Recreation 0.438 �0.026Religion 0.380 0.019Science 0.125 0.002Senior citizens 0.177 �0.059**

Social policy 0.255 �0.062**

Society and gossip 0.091 0.020Sports – auto 0.120 �0.047**

Sports – general 0.967 �0.009Sports – local 0.518 �0.119**

Sports – teams 0.151 �0.015Sports – various 0.265 �0.089**

Technology 0.198 0.021Travel 0.485 �0.042Weekend 0.284 �0.091**

Women 0.011 0.039

Notes: T-statistics in parentheses: + significant at 10% level; *significant at 5% level; **significant at 1% level.Dependent variable is a binary variable indicating whether a topic is covered in the MSA. All specificationsinclude year dummies and MSA fixed effects. Constants in fixed effects regressions represent the average value ofthe fixed effects.


which control for market attributes that might be correlated with both the number of own-ers and underlying tastes for variety that do not change over time. However, it might bethe case that unobserved factors correlated with the change in ownership concentrationmay be correlated with changes in tastes for variety, biasing results. Without instruments,it is not possible to completely control for unobservables of this sort, so it is worth con-sidering particular market characteristics that are changing over time that might triggerconcerns about unobserved heterogeneity. One example is growth in alternative mediaoutlets such as the internet. Other factors might include general economic conditions.Fixed effects regressions show that changes in MSA internet penetration (as measuredby the 2000 and 2003 Current Population Survey) and changes in MSA unemployment(from the Bureau of Labor Statistics) are not correlated with changes in ownership con-centration. If ownership changes are more likely in markets with growing or failing papers,coefficient estimates may also be biased. Newspaper-level regressions of newspaper closureon whether or not an ownership change has occurred shows no relationship betweenchanges in ownership and closure. More generally, however, fixed effects regressions can-not rule out all sources of unobserved heterogeneity and thus represents a limitation ofresults in this paper.

Table 8Ownership concentration and circulation

Per capita sales(ABC) (1)




Owners �0.0021 (1.83)+ �0.0007 (0.46)Papers �0.0024 (1.38)Owner

equivalents�0.0011 (0.36) �0.0001 (0.02)

Paperequivalents

�0.0027 (0.78)

Year 1999 �0.0018 (2.16)* �0.0019 (2.21)* �0.0016 (1.81)+ �0.0017 (1.90)+Constant 0.0453 (17.61)** 0.0483 (14.27)** 0.0422 (9.53)** 0.0450 (7.92)**

N 498 498 498 498MSAs 249 249 249 249

Notes: T-statistics in parentheses: + significant at 10% level; *significant at 5% level; **significant at 1% level.Dependent variable is per capita newspaper sales in 1995 and 2000. All specifications include MSA fixed effects.Constants in fixed effects regressions represent the average value of the fixed effects. Markets excluded due tomissing circulation data are: Anchorage; Auburn–Opelika; Barnstable–Yarmouth, MA; Beaumont–Port Arthur;Birmingham; Dover; Janesville–Beloit; Jonesboro; Joplin; Lincoln; New York City; Oklahoma City; Tulsa; YubaCity.


5.2. Ownership concentration and newspaper characteristics

The increase in variety documented above is particularly surprising given that anincrease in concentration reduces the number of papers in a market, which, by itself, worksagainst more variety. It is interesting to consider the types of content that emerge withgreater ownership concentration. Does increased variety occur in the form of sports, enter-tainment and other ‘‘soft’’ news topics? Or does the reduction in duplication that comesabout with ownership concentration lead to greater allocation of reporters to regionaland national topics? To address this question, a fixed-effects regression of changes in own-ership concentration on the presence of coverage is run for each topic. In each case, adummy variable indicates whether or not a reporter is assigned to the topic in each marketeach year. Table 7 presents results. Virtually all of the coefficient estimates are negative,indicating that increases in ownership concentration increases coverage across many top-ics. About half of the coefficients are significant at a level of at least 10%. Most of the busi-ness and industry topics as well as the sports topics are large and significant. Most of thenews, entertainment, Arts and lifestyle topics are not significant. The pattern of coefficientbroadly suggests that the additional coverage that comes with increases in ownership con-centration does not favor ‘‘soft’’ news over ‘‘hard’’ news topics. Although the coefficientsfor sports, fitness and music are significant, many of the larger and significant coefficientsreflect traditional news topics such as crime, business, book reviews and investigativereporting. There is also no evidence of a shift toward more local or national news associ-ated with ownership concentration.

5.3. Ownership concentration and per capita circulation

Taken together, the results above suggest that although increased ownership concentra-tion reduces the number of papers in a market, the increase in concentration leads to


greater separation among papers. Moreover, concentration appears to increase total vari-ety across a range of news topics. This section considers the extent to which these changesaffect circulation.

Table 8 shows the results of a fixed-effects regression of changes in the number of own-ers and owner-equivalents on per capita newspaper sales. Sales data are taken from theAudit Bureau of Circulation (ABC) Circulation Data Bank, but are not available before1995. Zipcode level circulation from 1995 to 2000 are aggregated to the MSA level andmerged with Burrelle’s data for 1993 and 1999. ABC circulation is missing for majorpapers in a number of markets, which are excluded from regressions. National newspapersare also excluded.9 The effect of an increase in ownership concentration is negative in allcases, with only the first estimate approaching significance. The magnitude of the effect issmall, with the loss of one owner raising total per capita circulation by about 2 papers perthousand people on a base circulation of 45 papers per thousand. Although these resultsprovide only very limited evidence that increases in ownership concentration increase cir-culation, there appears to be no evidence that ownership concentration reduces circulation.Taken together, the results in this section suggest that the new content which emerges fromconsolidation does not reduce demand for newspapers.

6. Conclusion

The analyses above demonstrate that increases in ownership concentration lead firms todifferentiate products to a greater extent and cover a larger number of reporting topics.The additional coverage generated by increases in ownership concentration cover a rangeof ‘‘hard’’ and ‘‘soft’’ news topics. Moreover, this additional coverage does not lead toreductions in per capita newspaper sales.

It is worth considering the extent to which consumers benefit from this additional vari-ety. Revealed preference arguments hold that new readers entering the market are betteroff. Existing readers benefit from additional variety. These two factors suggest that con-sumers are not harmed by consolidation. However, because newspapers are bundledgoods, newspaper sales reveal no information about willingness to pay for particular top-ics and this analysis cannot show changes in infra-marginal utility. In other words, it mightbe that coverage changes resulting from increased ownership concentration provide read-ers with less utility per topic, such as might happen if consolidation decreased the qualityof writing or reporting. However, if the decreases outweighed the benefit of new coverage,sales would decline. Since it does not, infra-marginal changes are likely to be limited.

With respect to current policy, results presented here suggest that government policiesaimed at sustaining the number of newspapers and newspaper owners within markets maynot always benefit readers. To the extent that policy is concerned with aspects of diversityother than those associated with variety, these results identify a benefit of concentrationagainst which other potential costs should be weighed.

However despite evidence that ownership concentration has not harmed consumers, itis not possible to conclude from this research alone that current policies are misguided.

9 Additional excluded markets are: Anchorage; Auburn–Opelika; Barnstable–Yarmouth; Beaumont–PortArthur; Birmingham; Dover; Janesville–Beloit; Jonesboro; Joplin; Lincoln; New York City; Oklahoma City;Tulsa; Yuba City. Circulation at USA Today, the Wall Street Journal, the New York Times, and the Christian

Science Monitor constitute just under 10% of daily newspaper sales.


First, nothing can be said about aggregate welfare without taking into account how own-ership concentration affects advertising prices, hence policies limiting consolidation maystill be warranted when advertiser welfare is taken into account. It also might be the casethat increases in ownership concentration and coverage in newspaper markets in the 1990sare related to heightened competition with radio and cable television over this period.Although results in this paper are consistent with findings in the economics literature thatownership concentration in radio produces greater programming variety, little work hasbeen done to directly examine competition across media and it remains an important areafor further economic research.

In sum, regulation of media markets in the US and antitrust policy in particular pre-sume that more owners and more products lead to greater content variety. However theeffect of concentration on variety in differentiated product markets is an empirical questionthat depends on fixed costs and the value of consumers to advertisers. The analyses pre-sented here suggest that concentration in newspaper markets increases variety in a mean-ingful way. Policies that limit consolidation in media markets should reflect this result.

Acknowledgements

I am grateful to Joel Waldfogel for many thoughtful comments on earlier drafts. RachelCroson, Dennis Yao, Julie Wulf, Felix Oberholzer-Gee and participants in the WhartonApplied Economics Seminar also provided useful input. I am grateful to staff at Burrelle’sInformation Services for assistance in obtaining and interpreting data. All errors are myown.

References

Anderson, S., Coate, S., 2005. Market provision of broadcasting: a welfare analysis. Review of Economic Studies

72, 947–972.

Audit Bureau of Circulations, 1999. Circulation Data Bank, Audit Bureau of Circulations, New York.

Berry, S., Waldfogel, J., 2001. Do mergers increase product variety? Evidence from radio broadcasting. Quarterly

Journal of Economics 116, 1009–1025.

Blair, R., Romano, R., 1993. Pricing decisions of the newspaper monopolist. Southern Economic Journal 59,

721–732.

Bucklin, R., Caves, R., Lo, A., 1989. Games of survival in the US newspaper industry. Applied Economics 21,

631–649.

Burrelle’s Information Services. Burrelle’s Media Directory, 1994, 2000 and 2005 editions, Burrelle’s Information

Services, Livingston.

Chaudhri, V., 1988. Pricing and efficiency of a circulation industry: the case of newspapers. Information

Economics and Policy 10, 59–76.

Compaine, B., 1982. Who Owns the Media?: Concentration of Ownership in the Mass Communications Industry.

Knowledge Industry Publications Inc., White Plains, NY.

Compaine, B., 1995. The impact of ownership on content: does it matter? Cardozo Arts and Entertainment Law

Journal 13, 755–775.

Dertouzos, J., Quinn, T., 1985. Bargaining Responses to the Technology Revolution: The Case of the Newspaper

Industry. Rand Report: R-3144-DOL.

Dertouzos, J., Trautman, W., 1990. Economic effects of media concentration: estimates from a model of the

newspaper firm. Journal of Industrial Economics 39, 1–14.

Dixit, A., Stiglitz, J., 1977. Monopolistic competition and optimum product diversity. American Economic

Review 67, 297–308.

Genesove, D., 1999. ‘‘The Adoption of Offset Presses in the Daily Newspaper Industry in the United States.’’

NBER Working Paper 7076.


George, L., Waldfogel, J., 2003. Who affects whom in daily newspaper markets? Journal of Political Economy 11,

765–784.

Hicks, R., Featherstone, J., 1978. Duplication of newspaper content in contrasting ownership situations.

Journalism Quarterly 55, 549–569.

Hotelling, H., 1929. Stability in competition. Economic Journal 39, 41–57.

Lacy, S., 1987. The effect of intracity competition on daily newspaper content. Journalism Quarterly 64, 281–301.

Lacy, S., 1991. Effects of group ownership on daily newspaper content. Media Economics 35, 35–55.

Picard, R. (Ed.), 1988. Press Concentration and Monopoly: New Perspectives on Newspaper Ownership and

Operation. Ablex Publishing Corporation, Norwood, NJ.

Reddaway, W., 1963. The economics of newspapers. Economic Journal 73, 201–218.

Reimer, E., 1992. The effects of monopolization on newspaper advertising rates. American Economist 36, 65–70.

Rosse, J., 1970. Estimating cost function parameters without using cost data. Econometrica 38, 256–274.

Spence, M., 1976a. Product selection, fixed costs, and monopolistic competition. The Review of Economic

Studies 43, 217–235.

Spence, M., 1976b. Product differentiation and welfare. American Economic Review 66, 407–414.

Spence, M., Owen, B., 1977. Television programming, monopolistic competition, and welfare. The Quarterly

Journal of Economics 91, 103–126.

Steiner, P., 1952. Program patterns and the workability of competition in radio broadcasting. Quarterly Journal

of Economics 66, 194–223.

Waldfogel, J., 2003. Preference externalities: an empirical study of who benefits whom in differentiated product

markets. RAND Journal of Economics 34, 557–568.

Documents

Whats Fit to Print - The Effect of Ownership Concentration of Product Varieties in Daily Newspaper Markets - Lisa George