Newspaper quality, Pulitzer Prizes, and newspaper circulation.

The study includes a large number of variables which have been demonstrated to be determinants of newspaper circulation or provide alternative sources of news as controls. The many controls are to assure the robustness of the results regarding newspaper quality and circulation and also estimate parsimonious models of circulation based on elimination of individually and jointly insignificant variables. One variable not included is price. Newspaper subscription prices are relatively low and exhibit little variation cross-sectionally, in part because newspapers receive a majority of their revenues from advertising. In addition, Thompson [1989] found that price was not a significant determinant of demand. The other independent variables are as follows (unless otherwise stated, the source for a variable is the 1990 Census):

POPULATION is the natural logarithm of the estimated 1997 population of the paper's MSA. A larger population should increase circulation [SRDS Circulation '98].

INCOME is the natural logarithm of median household income. A positive sign for INCOME indicates that news is a normal good, which seems most probable. However, wealthier households have a higher value of time, so newspapers might suffer from cost disease [Baumol, 1967], offsetting the income effect on circulation.

CABLE is the percentage of households in the MSA's television market subscribing to cable television. Cable provides national and local news channels and public affairs programming, which can substitute for daily newspapers, so a negative sign for this variable is expected [Broadcasting and Cable Yearbook, 1998].

NETWORKS is a dummy variable which equals 1 if the television market to which the MSA belongs has an affiliate of each of the three over-the-air networks (ABC, CBS, .NBC). This variable measures the availability of both news and entertainment programming from the major networks, which could compete with newspapers for the audience's time, so a negative sign for this variable is expected [Broadcasting and Cable Yearbook, 1998].

TV STATIONS is the number of over the air televisions stations in the market to which the MSA belongs. Television stations compete for the audience's attention, so a negative sign for this variable is expected [Broadcasting and Cable Yearbook, 1998].

PAPERS is the number of daily (or Sunday) papers published in the MSA. According to the umbrella theory of competition [Lacy and Simon, 1993], even small suburban dailies can provide competition for a major metropolitan paper. Competition should reduce the circulation of a paper, so a negative sign for this variable is expected [SRDS Circulation '98].

COLLEGE is the percentage of MSA residents over age 25 who hold a degree from a four-year college. Education is generally correlated with greater interest in news and public affairs, so a positive sign for this variable is expected.

ENGLISH is a measure of language proficiency. It equals one minus the fraction of the population age five and older listed in the Census category that does not speak English well. Almost all of the papers in the data set are published in English. Residents with poor English skills face a higher cost of reading these papers, which also may not feature extensive coverage of minority communities, so a positive sign for this variable is expected.

MALE is the percentage of men in the population and is included as a control.

SAMEMSA is the percentage of residents over age five that lived in the same MSA in 1985 as in 1990. Newspapers in the U.S. typically offer extensive coverage of local news, which may be of less interest to individuals who have recently moved into the area and who may instead consume a product with more national or international news. A positive sign for this variable is expected.

OVER25 is the percentage of adults in the population and is included as a control.

WHITE describes the racial composition of the population and is included as a control.

TURNOUT is the total number of votes cast in the 1996 Presidential election divided by the MSA's estimated 1997 population. This variable measures interest in politics, so a positive sign is expected [Scammon, McGillivray, and Cook, 1998].

CLINTON is the percentage of the major party (Republican and Democrat) votes in the MSA received by Bill Clinton in the 1996 Presidential election. Goff and Tollison [1990] contend that liberals may have greater demand for news and find evidence consistent with this hypothesis when examining state newspaper circulation. Sutter [2002] finds that total newspaper circulation is higher in MSAs with a higher 1996 Democratic vote share. This variable is included as a control [Scammon, McGillivray, and Cook, 1998].

Table 1 provides summary statistics for the variables used in this study. The population of MSAs in the sample range from 57,000 to 18.3 million. The average Daily circulation is 90,300, with a range from 5,200 to 968,000 (Los Angeles Times), while average Sunday circulation is 114,000, with a range of 5,200 to 1,340,000. The newspapers in the sample face substantial media competition, with an average of 9.08 for TV STATIONS, 67.7 percent for CABLE subscribers, 3.42 for Daily papers, and 2.69 for Sunday papers.

Results

Table 2 presents analysis of Daily and Sunday circulation of the sample of newspapers using a binary Pulitzer Prize variable. All of the results presented in this paper report White's heteroskedasticity-consistent standard errors. Quality matters for circulation, as the Pulitzer Prize dummy variable is significant at better than the 1 percent level for both Daily and Sunday circulation in the full specifications. The quantitative impact of quality is also significant. Papers which received a Pulitzer Prize had 55 percent higher Daily circulation and 43 percent higher Sunday circulation. As expected, population is positive and highly significant. (4) The other consistent, significant determinant of circulation is PAPERS, which is negative and significant at better than the 1 percent level. Most of the other control variables are insignificant in the full specification, with only ENGLISH attaining significance at conventional levels for Daily circulation. As expected, language proficiency increases circulation. Overall, the specification performs well and the model explains a slightly higher portion of Sunday circulation.

Table 2 also presents parsimonious specifications, omitting some of the insignificant control variables. The parsimonious specifications serve as a robustness check for the main result concerning the impact of Pulitzer Prizes and provide further insight on the determinants of circulation. The parsimonious specifications presented were chosen based on Adjusted [R.sup.2), the Akaike Information Criterion, and the Schwarz Criterion. All of the variables dropped from specifications were always jointly insignificant. The parsimonious specifications confirm the significant effect of Pulitzer Prizes on circulation, with the quantitative impact virtually identical to the full specifications. POPULATION and PAPERS remain highly significant in the parsimonious specifications, and several other control variables attain significance at the 10 percent level or better. SAMEMSA, COLLEGE, ENGLISH, OVER25, and WHITE increase Daily circulation, while INCOME, ENGLISH, and OVER25 are positive determinants and MALE a negative determinant of Sunday circulation.

Table 3 presents analysis of Daily circulation employing an integer Pulitzer Prize variable equal to the number of prizes won by each paper between 1987-97. Column (a) presents a full specification, including the Pulitzer integer variable, while column (b) adds a quadratic Pulitzer variable to allow for a possible declining marginal effect of Pulitzer on circulation. Column (c) presents a parsimonious specification with a quadratic term buy excludes individually and jointly significant variables. Winning additional Pulitzers increases Daily circulation. In specification, (a) PULITZER is significant at the .5 percent level and each Pulitzer Prize increases circulation by about 5 percent. In specification, (b) PULITZER increases circulation, [PULITZER.sup.2] decreases circulation, and both variables are significant at the 1 percent level. The quantitative impact is substantial as well. The first Pulitzer increases circulation by 17 percent in (b). The marginal impact of a Pulitzer on circulation declines and is maximized at about 10 Pulitzers during the previous decade (although only three papers in the data set won this many prizes). Again, POPULATION and PAPERS are significant (at the 1 percent level) determinants of circulation, with POPULATION positive and PAPERS negative. The only other significant control variable in the full specifications is ENGLISH, which increases circulation. In the parsimonious specification (c), the PULITZER and [PULITZER.sup.2] variables remain significant at the 1 percent level, with positive and negative signs and similar point estimates as in the full specification (the first Pulitzer a paper wins increases circulation by 19 percent). POPULATION and PAPERS remain significant at the 1 percent level as well. ENGLISH and OVER25 attain significance at the 1 percent level in (c), while SAMEMSA now attains significance at the 10 percent level and circulation is higher when the population is less mobile. (5)

Table 4 examines the effect of the number of Pulitzer Prizes won on Sunday circulation. Column (a) presents a full specification with the integer Pulitzer variable, while (b) adds a quadratic term, and (c) presents a parsimonious specification with the quadratic term. The pattern of significance for the Pulitzer variable is similar to Table 3. PULITZER is positive and significant at the 5 percent level in (a) and positive and significant at better than the 1 percent level when the quadratic term is included, while [PULITZER.sup.2] is negative and significant at the 1 percent level. Additional Pulitzer Prizes increase circulation--the first Pulitzer increases Sunday circulation by 16 percent--but the marginal effect on circulation again declines and becomes zero at about 10 prizes. Among the control variables, POPULATION is positive and highly significant in each specification and PAPERS is negative and highly significant in each specification. In the full specifications, only ENGLISH, OVER25, and TV STATIONS attain significance among the control variables. ENGLISH and OVER25 are positive and significant at the 10 percent level, while TV STATIONS is negative and significant at the 10 percent level in specification (b). In the parsimonious specification (c), these remain the only significant variables. However, OVER25 is now significant at the 1 percent level, while ENGLISH becomes positive and significant at the 5 percent level. (6)

Print Get the Mag Weekly Updates