The relation between media expenditure and general economy.
In 1972, when McCombs formally proposed the Principle of Relative Constancy (PRC), he did not expect that it would open a new avenue to media economics research on media expenditure. From then on, a lot of following research on PRC existed. On one hand, theoretical studies offered different conceptual contributions (Noh & Grant, 1997; Dimmick, 1997), while on the other, empirical examinations resulted in mixed conclusions. Some of the empirical studies support the PRC (McCombs, 1972; McCombs & Eyal, 1980; Furlton, 1988; Werner, 1986), while other results were against the PRC (Son & McCombs, 1986; Son, 1990; Wood & O'Hare, 1991); the rest of empirical studies provided mixed landscape (Wood, 1986; Dupane, 1994; Dupagne, 1997). During its development over 30 years (1972-2005), though there were several scattered evaluative reviews of PRC (McCombs & Nolan, 1992; Lacy & Noh, 1997; Dupagne, 1997), it turned out that there was a lack of systematical critical reviews on it. This paper's purpose is to fill this gap by summarizing the previous literature and implying future research dimensions; therefore, expecting to bridge the previous research and the future exploration of media expenditure under the circumstances of new media, mainly represented by the Internet.
The Origin of PRC
The history of the study on the relationship between media expenditure and the general economy dates back to 1884, with the earliest exploration on the relationship between the media industry and the general economy, the "law of newspaper growth" by S.N.D. North suggested the existence of a constraint on growth of the press (as cited in McCombs, 1972, p.7). There was a lot of following related research, most claiming the positive relations between them (for a general review, see McCombs, 1972; McCombs & Nolan, 1992; Son, 1990; Greco, 1998). The explicit idea of a constant relation between the two, first stemmed from a descriptive analysis of media expenditure from 1929 to 1957 by a newspaper publisher Charles E. Scripps, the chairman of the Board of Scripps-Howard Newspapers. The following words can be viewed as the origin of the idea of the following-up PRC research.
If we may suggest one broad generalization, it is that in spite of the increasing complexity of the mass communication with the advent of new media, the pattern of economic support has been relatively constant, and more closely related to the general economy than to the various changes and trends taking place within the mass media field itself.
The consistency evident in the pattern of economic support for the mass media seems significant. It suggests that mass communication has become a staple of consumption in our society much like food, clothing, and shelter. Its stability in times of economic stress indicates that consumers feel that mass communication is a necessity of life, although their selection of media may vary (p.6).
However, he didn't give any in-depth analysis on it and expected someone else do the following work. This task was fulfilled 13 years later by McCombs (1972) in a monograph "Mass Media in the Marketplace", stimulated by this fresh idea.
In his pioneering study, McCombs (1972) formally put forward the hypothesis of PRC. The author employed 40-year data on media expenditure (including both consumer and advertiser spending on media) and the general economy in the U.S. from 1929 to 1968 and found the following conclusions: the first is "the level of spending on mass media by consumers and advertisers is determined by the general state of the economy, and any change in the level of the economy causes a parallel change in spending on mass media" (p.10). The second conclusion is that there is a tremendous ebb and flow of money among the various media. The first conclusion, constancy assumption, and the second one, functional equivalence assumption, collectively are called the Principle of Relative Constancy (PRC).
Constancy assumption was supported mainly through a partial correlation between time and media expenditure controlling inflation- the household. There only existed a slight positive trend in the mass media expenditure (for both total media spending and consumer expenditure) after controlling both population growth and inflation (p.9, 17). The study also found the total media expenditure accounted for 5.24% of total personal income across 40 years with a standard deviation of 0.71% (at actual dollars, p.10), and this proportion is similar to the GNP trend but with a moderately declining trend. As for the proportion of consumer media spending in total consumption expenditures, the average proportion across 40 years was 3.14% with a standard deviation of less than 0.20% (the author did state at constant or current prices, p.17). To support functional equivalence, the most obvious proof is the decline of motion picture accompanying the introduction and diffusion of the television.
As a pioneer in PRC research, McCombs and his colleagues conducted a series of studies to justify the hypothesis he put forward in 1972. McCombs and Eyal (1980) extended the data period from 1968 to 1977 and confirmed the same conclusions they got 8 years ago. PRC still holds during this 10-year period; the decline of books coincided with the increase of periodicals. Cautious attention should be paid here because ironically, 12 years later, McCombs and Nolan (1992) interpreted the findings of 1980 contrary to the theory of relative constancy in another paper.
According to constancy hypothesis, the percentage of media expenditure in the general economy would remain constant over time. Is it still true in acuate changing media circumstances? Fullerton's (1988) empirical answer is yes, but he found a somewhat short-term departure. He studied the diffusion pattern of television under the umbrella of PRC from 1948 to 1962 in detail and found the U-shaped curve of non-TV media spending, which suggested TV adopters attracted money from other media; this is consistent with McCombs' (1972) finding. The findings partially support the bell-shaped diffusion curve of TV spending, considering a short-term spike in 1950-1951. The author interpreted that innovators and early adopters may contribute to the temporary exceptions to the PRC, which distinguished a revolutionary technological change from an ordinary evolutionary one. The author added that the spike did not mean the departure from the PRC, but instead, the quick return to the normal pattern indicated the economic constraint to media spending.
Wood (1986), the first scholar who criticized the PRC proposed by McCombs (1972), found long-term constancy but short-term departures in two of the five decades in each of the two models with the data from 1929 to 1981. Then, he proposed three problems in the previous research on PRC in the pioneering study of McCombs and his colleagues. The first is the presence of serial correlation in their correlation analysis. The second is zero correlation may disguise the great fluctuations over time, instead of holding constancy. The third is the inaccuracy of using personal income (PI) instead of disposable personal income (DPI) because DPI only comprises the private consumption and savings. However, PI includes the taxes which cannot be consumed. As an economist, he made some methodological revisions from the perspective of econometrics. The author proposed two basic models of PRC: time-trend model
[M.sub.1] = [B.sub.0] + [B.sub.1] [D.sub.1] + [B.sub.2] [Y.sub.dt] + [u.sub.t] and income-share model [M.sub.t] = [A.sub 0] + [A.sub.1] Y.sub.dt] + [u.sub.t], where [M.sub.t] is the media expenditure at time t and is DPI. The time-trend model is the same as partial correlation used by McCombs and his colleagues, and the income-share model was first proposed by him. Since then, these two models are widely used (Wood, 1986; Wood & O'Hare, 1991; Son & McCombs, 1993; Dupagne, 1994), and the income-share model tended to be employed with some modifications when testing the advertising PRC (Demers, 1994; Kim, 2003; Chang & Chan-Olmsted, 2005).
In a follow-up article, Wood and O'Hare (1991) analyzed the two models Wood (1986) proposed, especially the income-share one. Then they extended the data to a longer period of 1929 to 1988, lasting six decades (the longest research period in PRC), to examine the PRC in the face of the "video revolution" they coined. Both the time-trend and income-share models confirmed long-term constancy, but the income-share model reported short-term departures (there is no report on short-term results of the time-trend model). The long-term constancy masked short-term departures as Wood (1986) found. The authors addressed the increasing proportion of media expenditure from 1979 to 1988 due to the popularity of video technologies.
Son and McCombs (1993) revisited PRC under the circumstances of restless media market with the entrance of pay cable, VCRs, videodisc players, and videotext, especially the first two successful media. In this latest PRC study by McCombs and his colleagues, it appears that they accepted the critique of Wood (1986) and Wood and O'Hare (1991). In this article, they employed disposable personal income (DPI) instead of personal income (PI), regression instead of correlation, and constant prices (1982) instead of current prices, as well as the Durbin-Watson test for autocorrelation to examine the media expenditure from 1929 to 1987 with historical emphasis on the period from 1975 to 1987. Both the time-trend and income-share models were employed to examine the PRC with aggregate data as well as each kind of media expenditure data. Only two of the four pairs of results between the two models (two of the time-trend model and three results of the income-share model cannot be run for autocorrelation) were in agreement: both models concluded that PRC held for all media expenditure from 1929 to 1974 and from 1929 to 1987. However, inconsistency between the results of the two models existed, which lies in the conclusions on the recent two decades: the time-trend model supports the hypotheses that total consumer spending on mass media and total consumer spending on all audio and audiovisual media have increased from 1929 to 1987, but the income-share model doesn't support these hypotheses. The authors argued that it appeared that the income-share model is out of step with previous research (McCombs & Son, 1986). However, the authors ignored that the time-trend model was the same as the research by using partial correlation by McCombs and Son (1986). In a similar doctoral dissertation (Son, 1990), she examined the impact of new electronic media on economic support for mass media from an audience from 1975 to 1987 in the long historical context from 1929 to 1987, mainly for methodological considerations. The author argued that the influence of TV and VCRs and cable TV on existing media is quite different; the former is the competitor of existing media while the latter, the complement. Even though the study follows the method proposed by Wood (1986), its conclusion wasn't consistent with Wood's. It may be due to disparities in the research period (Wood studied the media expenditure from 1929 to 1981 while this research was from 1929 to 1987) as the author pointed, and this research may have reflected real picture of media expenditure because the data were more recent. It was also one of the rare studies on audience support for mass media both on money and time (also see Ogan & Kelly, 1986). Another contribution of the study is that it was one of only three papers (also see Werner, 1986; Noh, 1995) that studied the media at the individual levels of audience support. The authors found that low-income respondents spent the largest percent of both money and time on the media, which is consistent with the findings of the other two.
In his pioneering monograph, McCombs (1972) proposed PRC is mainly based on the economic constraint of the spending on mass media. McCombs also mentioned the limit of time but without further analyzing it. Ogan and Kelly (1986) first took a look at both the time and money spent on media, with special emphasis on the impact of new media, such as VCRs, satellite dishes, pay-TV, and personal computer, on existing media.
Results found little impact of new communication technologies on media expenditure and no relationship between the time and money spent on media. Though it is a valuable exploration on both the time and money spent on media, it can only be a probe leading to further in-depth research on this topic. The too small sample size of 276, the unrepresentativeness of the sample in Indianapolis, the accuracy of the self-report media expenditure, and even the quality of this telephone survey with only a 55% response rate should be improved in future studies. Methodologically, the study is a sheer description without any research questions or hypotheses. Son (1990) extended the PRC to spending time with mass media: the relationship between the total time and time spent on mass media is constant, which was called time share constancy hypothesis by the author, but this cautious conclusion that time share constancy held in the adoption of cable TV didn't get much justification because of the limitation of only descriptive statistics on the secondary data. These two studies should be viewed as a new direction and expansion on PRC research.
In a PRC study of only advertiser expenditure, Demers (1994) went beyond the traditional model of PRC, arguing that the income alone couldn't predict media expenditure for the complexity of social change. He added a composite index of structural pluralism as an independent variable besides the index of the general economy (e.g., GNP or NPI), which was an index of the total U.S. resident population, the number of civilians employed in nonagricultural jobs, and the number of cities with 100,000 populations or more in the decimal data from 1850 to 1990 and the first two variables in yearly data from 1929 to 1989 because of the lack of third. He found that the proportion of advertising in the general economy increased from the 19th to 20th century instead of staying constant in the decimal data, but with some short-term departures, while there is no significant trend in the yearly data, but still with several minitrends --due to the changes in technology, competition, and economic and political conditions (p.46). Advertising correlated positively with the structural pluralism. Even more, the former increased exponentially, instead of linearly, with the increase of the latter. The author concluded that the underlying assumption of PRC is that the time people spend on media is fixed is erroneous.
Although it is quite right that the author argued that income alone cannot predict the media expenditure, other variables should be employed. The author made it too abstract to coin the concept of structural pluralism just as Dimmick (1997) criticized. In fact, it will be straighter to use the original indexes (e.g., population, nonagricultural labor force).
Glascock (1993) examined the trend of consumer and advertiser expenditures on media from 1978 to 1990, a critical period for cable diffusion (penetration rate from 17% to 56% during this period), with a special emphasis put on the effect of cable on mass media spending. The author found that both advertiser and consumer spending on media as a percentage of general economy (GNP, personal income) were increasing during the period examined, suggesting that cable attract money from other resources besides the existing media, which is not in agreement with PRC. Advertiser spending trends of all media are positive or constant, especially all advertising expenditures of subcategories of cable (local, national, and total) exhibited positive trends. Consumer expenditures on traditional media were declining, but both consumer spending on cable and all media including cable were positive. All the trends were presented after controls of population, inflation, and personal income.
If PRC is a principle, it should be universal at least in market-oriented societies (see Son & McCombs, 1993). Werner (1986) took the first step in examining PRC outside the U.S. She tentatively concluded that PRC held in Norway from 1958 to 1982; it is also one of the few studies to analyze media expenditures at the individual level of audience (also see Organ & Kelly, 1986), and she found that low-consumption audiences spend a higher percentage of total consumption than high-consumption counterparts. Some fresh ideas on the study of media expenditure are offered, but it should be noted that Werner only used the data of six surveys instead of annual ones. Nor did she use any hypothesis testing with regression; the result is a sheer description of percentage of media expenditure. Finally, she was not the only PRC scholar to use expenditure data at current prices instead of constant prices (e.g., McCombs, 1972; Demers, 1994). The tentative conclusion of constancy should also be questioned when viewing the great variations on the proportion of media spending to total consumer expenditure: from 3.5% to 5.6% in only six surveys. In fact, even the author herself did not keep consistent when interpreting the variance of percentages in this monograph: though she argued constancy (p.255), she also stated the increasing percentage of media consumption to total consumption (p.256) and regarded media as luxury expenditure instead of stable (1957). So there seems to be no clear conclusion in this monograph, even though the author and the following quoters treated it as the support of constancy (e.g., Dupagne, 1997a). Finally, the study of relatively constant percentage of various media expenditure, of course, can shed more light, but it is not necessarily the assumption of PRC, which was misunderstood by the author.
In a study of PRC in the United Kingdom, Dupagne (1994) concluded a long-term constancy from 1963 to 1989 but with a short-term increase from the constancy during the 1980s owning to video hardware and software. The analyses of inter-media variations and constancy indicated a two-step evolution of mass media expenditure in UK: early constancy and late increase. Though nothing innovative in methodology and theories this study sheds more light on media expenditure, as the first formal PRC examination outside the Unites States.
Almost all existing mass media expenditure research before 1997 has focused on the relationship between consumer mass media expenditure and general economy (e.g., income) within the framework of PRC proposed by McCombs in 1972. Dupagne was the first scholar on media consumption who went beyond the traditional PRC framework. Dupagne (1996, 1997) went deep into the media expenditure evolution from 1953 to 1991 in Belgium. In a study testing the Principle of Relative Constancy in Belgium, Dupagne and Green (1996) concluded that PRC didn't hold in the long run. Also in this study the authors performed tests with both aggregate media expenditure and individual media expenditure, which was different from only examination of aggregate data of media expenditure in the previous tests. In doing so, the study could shed more light on the functional equivalence assumption, which was more neglected by previous research. The results revealed that some categories of media spending were constant over time, while most were not, and the aggregate media spending data didn't exhibit constancy either. In the end, the authors concluded that PRC is not universal.
In an influential article (Dupagne, 1997b), he employed multiple regressions and found that price and population, other than income, are better determinants of consumer mass media expenditure, and lagged variables account for a great amount of variations of media expenditure changes. However, it still left unsolved why DPI was not significant for all individual medium expenditure, but significant for aggregate mass media expenditure. There were following innovations in his study on Belgium: the first was the inclusion of five regressors (price, population, unemployment, interest rate, and DPI) instead of only one variable of general economy (e.g. income) or income and time in traditional PRC studies. The second was that he first used the dynamic model which contained both lagged independent variables and lagged dependent variables. The third was both the first-differencing and logarithmic transformation of the raw data. Logarithmic transformation is a usually-used procedure to remove problems of heteroscedasticity, is first-differencing (e.g., subtracting observations at [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] from those at t), and can solve the problem of serial correlation in the spurious relationships between non-stationary variables. Finally, the author regressed media expenditure both aggregately (mass media expenditure) and individually (e.g., newspapers, books, cable ...). The result indicated that differences existed for the determinants among different media, but the author didn't further analyze them. It is also a pity that the author didn't include advertising expenditure in the analysis in this in-depth study.
In a study of the impact of new media on existing media expenditure from 1953 to 1991 in Belgium, Dupagne (1997c) assessed the collective impact of three communication technologies (color TV, cable TV, and VCRs) on mass media spending under the theoretical framework of functional equivalence. The result showed that consumers expanded their spending on mass media due to the introduction of the new communication technologies. The proportion of consumer mass media spending to DPI rose from 1.8% in 1970 to 3.4% in 1991, and three new communication technologies collectively contributed 56% of the rise of this percentage. The author analyzed the joint impact of three communication technologies instead of only one as done in previous studies. The author argued the necessity of constancy and functional equivalence assumptions in media consumption. The author also called for the extension of the research to other new technologies and software-based explanation of the departure from functional equivalence assumption. Though it was a new way to test the joint impact of three new communications, it should be more instructive to test the individual impact of new communication technology due to the different characteristics of each communication technology and their distinguishing evolution process.
As the only study of PRC in a developing country to our knowledge in English literature, Kim (2003) tested the PRC from 1961 to 1993 in Korea, with special emphasis on the effect of VCR on the media market. Unlike the traditional empirical examination on whether the PRC held or not, the author employed the income-share model to test the PRC and also used market scale (Media Markets/GNP), market elasticity ([MEDIA Media Markets/[DELTA]GNP), and cross-elasticity analysis among various media ([DELTA] Media Markets/ [DELTA] VCR Market). The study found that the media market in Korea saw a positive trend with the advent of the VCR. The long-term PRC failed to be supported in Korea during 33 years but was supported until the introduction of the VCR in 1982. The VCR increased the media expenditure, unlike the television drawing money mainly from movies (McCombs, 1972; Fullerton, 1988). Generally,
consumer spending on mass media varied according to the time span. However, there were some limitations in this study. First, the data were not complete; media expenditure data only contained the software spending on four major media: newspaper, television, movie, and VCR. Also, it is not appropriate to include both subscription and advertising data in one dependent variable. The author didn't explain whether the data are based on current or constant prices and didn't employ other indicators except GNP as independent variables either; there were some misunderstandings of the author in interpreting the income-share model. It is the intercept, not the coefficient of the GNP as the author often used, that determines the trend of media spending. Even sometimes when the author used the intercept, she also misunderstood the relation of the sign and the trend (see Wood, 1986; Wood & O'Hare, 1991). It is a good idea to compare the differences of new media expenditure evolution between Korea and the U.S. (Wood & O'Hare, 1991), but it cannot be generalized that the growth of the VCR was faster in Korea than in the U.S. (p.16) due to the different independent variables (GNP for Korea but DPI for U.S.) and different time periods examined. In the cross elasticity analysis in both the long-term and short-term, the author didn't explain the specific time period of the two terms, nor did she explain how to get the long-and-short term cross elasticity because the elasticity just compares the change between a specific year (t) and previous year (t-1) (pp.12,17).
In the only large-scale cross-national examination of PRC, Chang and Chan-Olmsted (2005) examined PRC empirically across more than 70 nations from 1991 to 2001, using advertising expenditure as the dependent variable, GDP as the independent variable, and population, foreign direct investment, economic freedom, and press freedom as control variables. In the static PRC examination on cross-sectional level in 2001, though constancy didn't hold for total advertising expenditure, the advertising expenditures for some media such as television, magazines, cinema, and the outdoors did keep constant relative to GDP across 81 nations, and this suggests that the economic impact on advertising expenditure relies on the types of media. In the dynamic analysis on the longitudinal level, the authors tested the PRC with both the time-trend and income-share models in 70 nations during the period of 1991-2001, and the results showed that only 7 nations passed both tests. This suggests that PRC was not an international phenomenon. The authors further found that PRC tended to hold in developed countries. Though it is a good idea to test the PRC on a cross-sectional large scale, PRC cannot be compared across different nations because of the different economic and political conditions, different media system, and categories of data. So, it is not an appropriate way to conduct the static examination across nations at a fixed time point (2001). Also, it is meaningless to compare absolute advertising data across countries between PRC nations and non-PRC ones (p.352) because PRC is about relative spending on media.
Contrary to Chang and Chan-Olmsted (2005), Lacy (1987) tested the PRC on a local level. He examined the effect of the radio industry on newspapers among the largest 72 metropolitan cities in America from 1929 to 1948 with discriminant analysis in dividing the cities to those of competitive and non-competitive newspaper industries. With a few exceptions, the development of radio had little overall effect on the decline of newspaper competition, which was against the assumption of PRC. The study did not use a random sample of cities; the three retained independent variables only explain 18% of the variance, and the function correctly classified 63.9% of the 72 cities, implying that other variables should be included. Compared to the conclusion, what is more important is that the study was a significant attempt to examine the PRC on a local level instead of traditional national level.
Two case studies on PRC offered quite opposite explanations. Carroll (2002) found that the recipients of daily email news updates from the furniture industry trade publication Furniture/Today did not reduce time spent on its print edition overall, based on the survey of 678 recipients. A majority of respondents claimed "no real change" in their reading habit since they began receiving eDaily, suggesting a complementary relationship between the print edition and electronic one. The examination of time spent on media opens a new door to traditional PRC research because it is just the limit of time and money which result in the relative constant of expenditure on media (also see McCombs, 1972). However, it should be noted that the examination of time is just an extension of classic PRC assumption. The author in this study misunderstood the original meaning of PRC, arguing that "consumers spend a relatively fixed amount of money on and time with media". Though the results from this study may generate some fresh implications on time consumption on media, it is only based on a biased sample which cannot generalize the findings. The sample of recipients of eDaily magazine on the furniture industry and the low response rate (13%) will be big problems in its representation. Even more, among 678 respondents, only 383 of them are the subscribers of the print edition, which is a very small valid sample. Also, the emailed version was on a daily base while the print edition was weekly, making the two incomparable. In another case study, Edge (2004) employed PRC to analyze the failure of an experimental tabloid in Singapore, Project Eyeball. Though there were some market factors contributing to its failure, such as overpricing and competition, the author argued its disappearance from the market was partially due to the economic constraint restricted by PRC.
Traditional PRC research focused on the aggregate media spending, while Greco (1998) studied the media spending of only one medium- books from 1984 to 1994 under the umbrella of PRC. The result indicated that there was a sharp increase in consumer book sales during this period. The percentage of consumer books in the total print rose from 38.25% in 1984 to 48.00% in 1996, but the author didn't answer the key research question of whether the percentage of Domestic Consumer Expenditure (DCE) for consumption of books in the U.S. remain "constant" he proposed at the beginning of the paper.
The aforementioned empirical examination of PRC turned out in mixed results; some studies (McCombs, 1972; McCombs & Nolan, 1980; Werner, 1986) supported the PRC, more studies (Wood, 1986; Wood & O'Hare, 1991; Son & McCombs, 1993; Dupagne, 1994; Dupagne, 1997; Kim, 2003; Demers, 1994) offered mixed results, PRC held in the long term and failed in the short period. PRC examination of a more recent period under the circumstances of new media (e.g., Glascock, 1993; Noh, 1994; Noh, 1997) tended to generate results of increase instead of constancy in the media expenditure trend.
Theoretical & Methodological Explorations
The discovery of the reality (what) is not the only goal of science, but maybe the more important work is the investigation of the underlying rationale (why). Compared to the empirical tests, however, the theoretical explanations tend to be less.
In his original proposal of PRC, McCombs (1972) employed the concept of scarcity, both of income and time, to explain the economic constraint on media spending, even though he didn't clearly state the scarcity. He wrote: "He sips his martini, scans his newspaper, and listens to the stereo simultaneously, but there must be some limit. The decisive factor (of the limit) can instead be a resource on the consumption side, namely time" (p.62) (similar statements also see Son & McCombs, 1993). However, other studies showed the time Americans spent on leisure is on rise instead of constancy (Robinson, 1981; Hornik, 1981; Demers, 1994).
Lacy and Noh (1997) offered a theoretical review of the PRC. They argued that even though PRC seemed theoretically perfect on the surface, there existed a big problem: the great difference in the measurement between percentages and numbers. The authors questioned the accuracy of the definition of constancy; if it is constant, it should have zero variation on percentages, but a slight variation on a percentage would represent a big difference on the number- thus, they postulated the problem of the measurement of the constancy. Though there was only a variation of 0.19% in the original study by McCombs (1972), it represented $10 billion. Furthermore, there is no mechanism underlying the constancy of media spending. The authors suggested that existing economic theories can shed some light to existing PRC research. Among them are the demand theory, industrial organization (IO) model, video windows, opportunity costs, and zero-sum game. Considering the lack of advertising data in PRC studies, the authors also pointed out that the advertising should be considered in PRC research although it was not fully supported for PRC in the original study by McCombs (1972).
Unlike Lacy and Noh's (1997) focus on theories related PRC, Dupagne (1997a) provided a comprehensive critique on both theories and methodology the same year. As for the theories, the author argued that the most fundamental weakness of the PRC is the lack of any economic justification. The author tried to fill this gap by employing such economic concepts as time ceiling, scarcity, Engel law, constancy, and functional equivalence. Even though they shed some light on the research on the media expenditure, none of them could infer that consumers will devote a constant proportion of income on media over time.
Dimmick (1997) employed the theory of the Niche to analyze the case of the "Video Revolution" put forward by Wood and O'Hare (1991), which means that the majority of households owned media of video technologies since 1980s. The theory of the Niche is a bioecological metaphor with three key concepts- niche dimension, the breadth, and overlap. The author employed the concepts of gratifications and gratification opportunities as utilities or choice dimensions to explain the consumer mass media spending.
Each innovation of media offers a great opportunity to the study of PRC, such as television (Son & McCombs, 1993; Fullerton, 1988) and VCR (Son & McCombs, 1993; Wood, 1986; Wood & O'Hare, 1991). Noh and Grant (1997) explained the departure from the PRC because of the diffusion VCR with the concept of media functionality (also see Noh, 1994). For the comparison between VCR and television, the authors pointed out the different patterns of the increase of spending on them: the television was a functional competitor of the existing media (e.g., movie), while the VCR was both a functional competitor and a complement to the existing media- at the early stage of the diffusion of the VCR it was mainly a functional complement, while in the latter, it mainly became a functional competitor to the existing media, especially for some interpersonal communications. The authors also attributed no decrease on existing media spending by the concept of functional duality- something different with Dimmick's (1997) explanation with Niche: VCRs and existing media were interdependent on each other to survive. Maybe both of their interpretations can explain this collectively. The authors further argued that the PRC could only hold when the new media offered competitive functions to the existing media which represent the whole needs and functions of conventional media, while PRC can't hold with the introduction of media with functional complement. The media expenditure is a nonzero-sum game instead of zero-sum game because most of the new media have ambiguous boundaries and can offer both competitive and complementary functions to established media. Finally, the authors concluded that the PRC becomes PRV- the Principle of Relative Variation, in media evolution full of all kinds of new technologies. Furthermore, they argued that two types of departures from PRC (a revolutionary technological change and an ordinary evolutionary one (see Fullerton, 1988) may be determined by media functionality- the former was because of the diffusion of a functional competitor and the latter a functional complement, which was a crucial point in examining the departures from the PRC. Media functionality can be viewed as the same explanations with different concepts with Dimmick, but it is a more powerful and clearer explanation than Dimmick's Niche concept. They also proposed two new concepts: the index of constancy (I[C.sub.(t)]) and the degree of contribution to the media spending increase for a medium i (%) (D[C.sub.i(t)]) beyond the mere test or presence of increase on media spending of traditional PRC framework.
Demers (1994) argued that the willingness for information will increase when the society becomes more and more structurally pluralistic in the future, which increased the media expending relative to the general economy. Unfortunately, the author failed to explain the willingness.
There was also great improvement in the methodology of PRC. In McCombs' (1972) original study, PRC was mainly examined through correlation, especially partially correlation, and PRC was tested whether or not zero correlation existed between time and media expenditure controlling other relative variables such as population, inflation. Dupagne (1997a) gave a comprehensive critique on methodology besides the aforementioned theoretical critique. The author argued that the different findings of PRC were not only due to the contextual explanation but also to methodological artifacts. The first big methodological problem was the lack of a primary model, especially since sometimes the income-share and time-trend model produced different results (see Son & McCombs, 1993; Wood, 1988; Wood & O'Hare, 1991). Four other problems may contribute to the disparities: DPI versus PI, regression versus correlation, short-term constancy versus long-term constancy, and treatment of serial correlation. It was discussed above that DPI was more appropriate than PI, while meanwhile, Wood (1986) said that the regression analysis was superior to the correlation analysis because regression could solve the problem of serial correlation, which usually happened in many time-series studies. McCombs and Nolan (1992) criticized that Wood's (1986) 10-year sample was too small and divided arbitrarily without any contextual consideration of mass communication, but ironically, he himself did it with 10-year sample one year later (Son & McCombs, 1993). Also, Son (1990) argued that the analysis with only 10 observations may lead to different findings than running the same analysis with 50 observations. Statistically, analyses with only 10 observations will present unreliable and weak conclusions. The problem is that the availability of data decides the period of research without much consideration to historical context in secondary analyses. The different treatment of the serial correlation may get different results (compare McCombs, 1972; McCombs & Eyal, 1980; Wood, 1986). For example, the result of the time-trend from 1929 to 1988 confirmed constancy by Wood and O'Hare (1991), but the same result from 1929 to 1987 concluded an increase instead of constancy by Son and McCombs (1993).
Until now, most of the critiques and recommendations haven't been adopted in the following research. For example, the test of autocorrelation wasn't even conducted by the author in the following research (e.g., Chan-Olmsted, 2005).
Future Research Directions
From the aforementioned literature, the conclusion can be drawn that since the proposal of PRC in 1972, PRC research prospered in 1980s and 1990s, but it was on decline in 21st century. PRC was a hot topic in the past and it should have its value in the future. This paper tries to put forward some possible directions for the future research.
The biggest problem in PRC research is its lack of economic theories (Dupagne, 1997a; Dimmick, 1997). The foundations of the theories may include time ceiling and scarcity, Engel law, budget lines and indifference curve, income elasticities of media products (for a general review, see Dupagne, 1997a). The beliefs that media spending took a relatively constant proportion in the general economy or there is a strong underlying foundation for the constancy should be further studied. The Engel coefficient is on decline, which means expenditures on goods other than food, including media products, is rising. It seems that the constancy tends to go away in the circumstances of changing media technologies, though there should be a limit on the increase in current society, where work, sleep, and dining are necessary parts of everyday life.
Methodologically, there should be a primary model. Dupagne and Green (1996) put forward a primary econometric model of PRC based on Wood's (1986) income-share model but with greater internal validity after log-transformation and first-differencing: Because after log-transformation and first-differencing:
[DELTA]ln MASS = [[beta].sub.0] + [[beta].sub.1][DELTA] ln DPI + e,
[[beta].sub.0] = 0, [[beta].sub.1] = 1. Because after log-transformation and first-differencing, the variables and the error term tend to be stationary, and this primary model can avoid the inconstancy in results with the same data between time-trend and income-share models. The income-share model will make more sense because PRC states media expenditure covaries with general economy (e.g., income), and McCombs (1972) also said income created a constraint to media spending. Wood (1986) also argued that the income-share model has been more widely used with aggregate data in economics. In more recent studies of PRC, especially those with emphasis on advertising, the income-share model was more employed (e.g., Demers, 1994; Kim, 2003; Chang & Chan-Olmsted, 2005). It is the income-share model that put emphasis on the impact of the economy. The future research of PRC could be based on this primary model, with some modification when necessary (e.g., addition of some controlling variables such as population, price, or change of independent variable to GDP when testing advertising spending) if there is no better one. All expenditure or income variables should be expressed in constant prices.
As Figure 1 shows, if the constancy holds, the regression line would be the Line A that goes through the origin of coordinates with the slope of 1; Line B and Line C are the examples of deviation of constancy.
[FIGURE 1 OMITTED]
Another question related to the primary model is the definition of the core concept of PRC: constancy. There was much confusion in the definition of constancy, mainly because of the inconsistency of the dependent variable and independent variable(s). It is no doubt that the constancy should mean the proportion of media expenditure in the general economy over time, so what is media expenditure and what is general economy? For example, in McCombs' (1972) original study of media expenditure from 1929 to 1968, this figure is 3.14% over 40 years when calculating the proportion of consumer spending on mass media in total consumer spending at current prices. In the same monograph, this constancy was 5.24% when reckoning the percentage of both advertiser and consumer spending in personal income at current prices over 40 years. In another occasion (Son & McCombs, 1993), this figure was 2.08% when calculating the proportion of consumer spending on media in DPI from 1929 to 1974 at constant prices. It will vary greater when using the proportion of adverting spending in GDP, total consumer and advertiser spending in GNP, expenditure at constant price instead of current price. Consistent with the primary model, the constancy should mean the proportion of consumer spending on mass media in DPI at the constant price, and other expressions should be viewed as its extension.
Future research should expand the scope of mass media expenditure. Geographically, PRC research should go beyond the Western world, especially the U.S., to explore the patterns of media expenditure in developing countries. Existing literature of PRC seldom touched the developing countries; only Kim (2003) and Edge (2004) shed some light on two advanced countries: Korea and Singapore. Given data available, the examination of PRC will bring different results, considering the immaturity of media industries in these nations, which will enrich the literature of PRC research. We expect the PRC doesn't hold in developing countries because both the economy and media industry within these countries enjoyed great growth in past several decades. Also, PRC can be studied at both the local and international level. For local level study on media expenditure within a city or state, or a group of states or cities, would produce some different and interesting results, which would offer a more practical value for media investment. New knowledge will be offered when PRC is conducted among several countries or many countries, which can be analyzed in combination with different media systems. Lacy (1987) made an effort at the local level while Chang and Chan-Olmsted (2005) did in an international fashion. Future research should also explore the media expenditure among various levels of audience because just as other expenditures, media expenditures should vary among different groups of people, especially those with different economic characteristics (see Werner, 1986; Noh, 1995; Ogan & Kelly, 1986). Also, different media has different sources of economic support; some are advertiser-financed, such as television, and some rely more on support from consumers, such as motion picture. They have different characteristics in nature (e.g., media durables-radio and TV sets, media non-durables-newspaper and magazine, and media service-cable). So, it is too general to say that media is a staple like food and shelter in McCombs' (1972) original statement of PRC. In fact, Wood (1986) used data from 1929 to 1981 and revealed that food was luxury instead of staple. Different media have different characteristics (e.g., Werner, 1986; Wood, 1986; Wood & O'Hare, 1991; Dupagne & Green, 1996; Chang & Chan-Olmsted, 2005). So the difference of expenditures on different media also should be explored in the future, but until now, only Greco (1998) made an effect in this direction.
It is a fast changing era full of new communication technologies, such as the Internet, iPod, HDTV, and interactive TV. So, this will be a challenge for media expenditure study: what is the scope of media? McCombs wrote in an email in the end, "Is there still a Principle of Relative Constancy? That is an open question" (personal communication, December 6, 2007). Though PRC studies included the impact of almost all major new media on media expenditure (e.g., television, VCR, cable TV, for general reviews of impact of each new media on the existing media marketplace, see Son, 1990), it never included the Internet, the most revolutionary innovation in media. Advertising is one of the dual markets of media. For example, in 1968 consumers took up 48% of total media spending while advertisers took up the rest at 52%, but compared with consumer spending, advertising was less studied in the previous literature.
As a social product, the media will interact with other industries especially those media-related ones. For example, dollars lost from non-media sources could have helped to finance TV. In 1949, consumers spent $800 million less on jewelry, watches, houses, and furniture than they had in 1947 before the advent of television (McCombs, 1972). The relationship between expenditures on media and other related products should be analyzed. Lacy and Noh (1997) also suggested a broader definition of media that allowed other nonmedia products and services to compete with media at a macro-level analysis beyond the income-media expenditure model.
From this perspective, we should question the existence of PRC research. If there is constancy in media expenditure, then this principle should also exist in staple expenditures, such as food, housing, and clothing. For example, McCombs (1972) analyzed food and housing spending from 1929 to 1957 in the U.S. and found that they demonstrate constancy, but in fact, economics, where we mainly borrow theories from in media expenditure studies, would seldom hypothesize a relatively constant or proportionate relationship between expenditure on a specific good and income over time (Dupagne, 1997a). PRC can only be a good summary of the past long-term trends but has little to do with predicting the future for media expenditure (Wood, 1986).
All existing theories in related economics cannot explain the constant relationship between media consumption and the general economy. As Demers (1994) pointed out, "there is no theoretical reason to justify the arguments that spending on mass media should remain constant through time, even if empirically it appears to have been that way."
More importantly, we should develop an alternative model, besides the PRC, to media expenditures. To my knowledge, media expenditure studies since the 1970s didn't go beyond the framework of PRC. Most of the aforementioned literature of PRC only produced more partial supports of PRC, especially in more recent periods with the introduction of new communication technologies, the percentage of media expenditure tended to increase (Son & McCombs, 1993; Wood, 1986; Wood & O'Hare, 1991; Dupagne, 1994; Dupagne, 1997c; Dupagne, 1996; Kim, 2003). Even the pioneer of PRC, McCombs and his colleagues rejected PRC in their latest paper on PRC (Son & McCombs, 1993). Now is the time for us to shift expenditure research from PRC to a more promising model. Dichotomy of the constancy/non-constancy of media expenditure would be too simple, considering the complication of media expenditure. Generally, the research on PRC should be transformed into that of media expenditure in the future.
There are four types of trends in media expenditure listed in Figure 2 (Dupagne, 1994). Pattern A indicates constancy, in which different media counterbalances each other to keep the overall media expenditure level constant relative to the general economy over time. Pattern B represents positive or negative trends in media expenditure over time. With the introduction of new technologies, increasing trends would be more frequent. Pattern C may better reflect the reality of media expenditure; media expenditure will keep at a level over fixed periods. With the entrance of new media, the equilibrium will be broken through to a new level, at least temporarily. Then, before the next innovation comes into being, it will return to the new equilibrium. Media expenditure would evolve dynamically but with a fixed level in the long run. Pattern D is a random one which fluctuates year by year without constancy, which makes the media expenditure study impossible. It rarely happens in reality in the circumstances of media maturity. Meanwhile, the selection of periods studied should be considered under the social context, especially related to media development.
In the future study of PRC (in fact, let me call this the media expenditure study for the moment) more variables should be employed because media expenditure would not be influenced by income alone. Also, lagged variables (t-1, t-2) should be included in the model to determine whether consumers respond to income changes immediately, which the mechanistic Keynesian income-consumption model assumes or behaves in a way of inertia (see Dupagne & Green, 1996; Dupagne, 1997b). Different variables should be considered in different contexts (e.g., fiscal expenditure, literacy rate, and urbanization may be significant in China).
[FIGURE 2 OMITTED]
The constancy and functional equivalence are two inseparable assumptions in PRC, and functional equivalence is more fundamental that the constancy assumption that comes from it (Noh & Grant, 1997). On the contrary, there were fewer studies on it, compared to the constancy assumption. Functional equivalence became an accessory to constancy assumption in previous PRC research, and constancy became the synonym of the PRC. As a theoretically driven assumption (Dupagne, 1997a), functional equivalence should be emphasized more.
More than 30 years ago, the pioneer of PRC, McCombs, asserted, "Minimally, it [constancy hypothesis] challenges others to put forward a more accurate and useful conceptualization of behavior in the mass media marketplace" (McCombs, 1972, p. 30). It is the case in previous research, but now it is the time for PRC scholars to take this challenge.
Renmin University of China
933 Building Pinyuan
59 Zhongguancun Street, Haidian District
Beijing 100872, P.R. China
Carroll, B. (2002). Newspaper readership v. news emails testing the principle of relative constancy. Convergence, 8(3), 78-96.
Chang, B.H. & Chan-Olmsted, S.M. (2005). Relative constancy of advertising spending: A cross-national examination of advertising expenditure and their determinants. The International Journal for Communication Studies, 67(4), 339--357.
Demers, D.P. (1994). Relative constancy hypothesis, structural pluralism, and national advertising expenditures. The Journal of Media Economics, 7(4), 31-48.
Dimmick, J. (1997). The theory of the niche and spending on mass media: The case of the 'video revolution'. The Journal of Media Economics, 10(3), 33-43.
Dupagne, M. (1994). Testing the relative constancy of mass media expenditure in the United Kingdom. The Journal of Media Economics, 7(3), 1-14.
Dupagne, M. (1994). The evolution of consumer mass media expenditures in Belgium from 1953 to 1991, Ph.D. dissertation of Indiana University (full text).
Dupagne, M. (1997a). A theoretical and methodological critique of the principle of relative constancy. The Journal of Media Economics, 10(1), 53-76.
Dupagne, M. (1997b). Beyond the principle of relative constancy: determinants of consumer mass media expenditures in Belgium. The Journal of Media Economics, 10(2), 3-19.
Dupagne, M., & Green, R. J. (1996). Revisiting the principle of relative constancy: consumer mass media expenditures in Belgium. Communication Research, 23(5), 612-635.
Dupagne. M (1997c). Effect of three communication technologies on mass media spending in Belgium. Journal of Communication, 47(4), 54-68.
Dutta-Bergman, M. (2004). Complementarity in consumption of news types across traditional and new media. Journal of Broadcasting and Electronic Media, 48(1), 41-61.
Edge, M. (2004). The failure of Project Eyeball: A case of product overpricing or market overcrowding? The International Journal on Media Management, 6(1&2), 114-122.
Fullerton, H.S. (1988). Technology collides with relative constancy?The pattern of adoption for a new medium. The Journal of Media Economics, 1(3), 75-84.
Galbi, D.A. (2001). Some economics of personal activity and implications for the digital economy, Retrieved from http://www.galbithink.org/ activity.htm
Glascock, J. (1993). Effect of cable television on advertiser and consumer spending on mass media 1978-1990. Journalism Quarterly, 70(3), 509-517.
Greco, A.N. (1997). The market for consumer books in the U.S.: 1985-1995. Publishing Research Quarterly, 13(1), 3-40.
Greco, A. N. (1998). Domestic consumer expenditures for consumer books: 1984-1994. Publishing Research Quarterly, 14(3), 12-28.
Kim, S. (2003). The effect of the VCR on the mass media markets in Korea, 1961--1993. Journal of Asian Pacific Communication, 13(1), 59--74.
Lacy, S. (1987). The effect of growth of radio on newspaper competition, 1929-1948. Journalism Quarterly ,64(4), 775-781.
Lacy, S., & Noh, G. Y. (1997). Theory, economics, measurement, and the principle of relative constancy. The Journal of Media Economics, 10(3), 3-16.
Lerner, D. (1958). The passing of traditional society: Modernizing the Middle East. New York: Free Press.
McCombs, M. E. (1972). Mass media in the marketplace. Journalism Monographs, 24. Thousand Oaks, CA: Sage Publications.
McCombs, M.E. & Eyal, C.H. (1980). Spending on mass media. Journal of Communication, 30(4), 153-158.
McCombs, M.E., & Nolan, J. (1992). The relative constancy approach to consumer spending for media. The Journal of Media Economics, 5(2), 4352.
Nguyen, A., & Western, M. (2006). The complementary relationship between the Internet and traditional mass media: the case of online news and information. Information Research, 11(3).
Noh, G. Y., (1994, August). New media departure in the principle of relative constancy: VCRs. Paper presented at the annual convention of the Association for Education in Journalism and Mass Communication, Atlanta, GA.
Noh, G. Y. (1995). Consumer support for mass media and functional relationship. Unpublished master's thesis, University of Texas at Austin.
Noh, G. Y., & Grant, A.E. (1997). Media functionality and the principle of relative constancy: An explanation of the VCR aberration. The Journal of Media Economics, 10(3), 17-31.
Ogan, C., & Kelly, J. (1986, August). Time and money spent on the mass media in an age of new communication technologies: A market study. Paper presented at the annual convention of the Association for Education in Journalism and Mass Communication, Norman, OK.
Picard, R.G. (2001). Effects of recessions on advertising expenditures: An exploratory study of economic downturns in nine developed nations. Journal of Media Economics, 14(1), 1-14.
Rogers, E.M. (2003). Diffusion of innovations (5th Edition). New York: Free Press. Scripps, C.E. (1959). The economic support of mass communications media 1929-1957. N. Y.: Scripps-Howard Research.
Son, J. (1990). The impact of new electronic media on audience support for mass media, Ph.D. dissertation of The University of Texas at Austin, (full text).
Son, J., & McCombs, M.E. (1993). A look at the constancy principle under changing market conditions. The Journal of Media Economics, 6(2), 23-36.
Werner, A. (1986). Mass media expenditures in Norway: The principle of relative constancy revisited. In Margaret McLaughlin, ed. Communication Yearbook 9. Beverly Hills, Calif.: Sage.
Wood, W. C. (1986). Consumer spending on the mass media: the principle of relative constancy reconsidered. Journal of Communication, 36(2), 39-51.
Wood, W.C. & O'Hare, S.L. (1991). Paying for the video revolution: Consumer spending on the mass media. Journal of Communication, 41(1), 24--30.
Yang, F. & Shanahan, J. (2003). Economic openness and media penetration. Communication Research, 30(5), 557-573.
Renmin University of China
|Printer friendly Cite/link Email Feedback|
|Publication:||China Media Research|
|Date:||Jul 1, 2010|
|Previous Article:||Conceptual challenges to the paradigm of communication research in national and local contexts: a critical analysis of the existing studies in Taiwan.|
|Next Article:||China's regulations on Internet cafes.|