Select Committee on Culture, Media and Sport Appendices to the Minutes of Evidence


Memorandum submitted by Dr Mark Baimbridge, University of Bradford



  A recent development in the UK television industry has been the emergence of satellite, cable and latterly digital broadcasting. This memorandum extends the familiar model of sport attendance to examine the impact of the BSkyB satellite network into the coverage of the First Division of the Rugby Football League over the 1993-1994 season. The entrance of satellite channels broke the traditional terrestrial competition between the BBC and the ITV companies. The analysis in this memorandum indicates that live television transmission of games exerts a negative influence on attendance, although the net financial consequences are positive for Super League teams. There are, however, a number of potentially harmful externalities arising from the switch to non-terrestrial broadcasters. These comprise issues such as upward pressure on transfer fees and salaries, together with reduced exposure of the game through smaller television audiences.


  1.  This memorandum extends the analysis of Baimbridge et al (1995) which examined the influence of satellite television on attendance at First Division Rugby League matches in the 1993-1994 season[3]. It finds that satellite television has a significant net negative effect on attendance. Until 1990 Rugby League coverage was only available on regional television channels, with Granada and Border televised live weekend matches, while Yorkshire Television featured edited highlights. Live coverage of the Silk Cut Challenge Cup continues to be shown on BBC1 in all regions of the UK. The first instance of satellite channel broadcasting, by BSkyB Television, began in 1992.

  2.  There have been numerous studies of sports attendance by economists. However, one of the major, but largely neglected, issues is the influence of television on the demand for tickets, which has potential significance for the finances of individual clubs. In a system where facility fees are paid to the host club there is a risk of spiralling inequality of income distribution. The rich may become richer with the consequence that the distribution of more talented ("star") players becomes increasingly skewed. It could make the outcome of individual matches and league competitions fall below the optimal level of uncertainty (Quirk and El-Hodiri, 1974; Jennett, 1984; Peel and Thomas, 1988; Cairns, 1990).

  3.  The broadcast media has become an important factor in the promotion and transmission of sport as a social product (McPherson et al, 1989). Similarly sport has become an influential part of the entertainment industry (Hughes and Coakley, 1984; Neallunsford, 1992), while television has become intimately involved in the growth, production and control of sport (Birrell and Loy, 1979; Greendorfer, 1981 and 1983; Powers, 1984; Rader, 1984; Whannel, 1984).


  4.  The only previous study of Rugby League (Burkitt and Cameron, 1992) used pooled cross-section time-series data of annual average home team attendance. Consequently, the broadcasting issue has not previously been addressed. Consequently, the literature with respect to broadcasting, predominately refers to professional football in the UK, together with American football, ice hockey and baseball.

  5.  The literature has largely ignored the impact of television despite token recognition of its importance (Sloane, 1980; Cairns et al, 1986). The Chester Report (1968) considered television to have a major but complex impact on football, as a substitute, per se, for people's leisure time. Similarly Hart et al (1975) and Peel and Thomas (1988) suggested that televised sporting events are specific alternative attractions. Hence, UK studies had to ignore the impact of live coverage because, for a long time, the majority of matches were highlights only. However, terrestrial television has increasingly shown live matches and the coverage of satellite live showings has increased rapidly[4].

  6.  Wiseman (1977) argued that edited highlights would raise expectations and hence reduce attendance since actual games had significant periods of inactivity. Bird (1982) acknowledges the popularity of televised football highlights, but declined to incorporate it, asserting that opinion in the game regarded television coverage as having a detrimental effect. The second Chester Report (1983) concluded that coverage increased ground advertising revenue, yet the focus on First Division matches might lessen attendance at other lower level games. However, no statistical evidence was provided to support these assertions.

  7.  Sunday league cricket provided an opportunity for Schofield (1983) to test for a television effect as matches were covered on BBC2 at the time of this study. The initial hypothesis was that advance announcement of the match selected for television coverage would adversely affect attendance. However, the variable was ultimately omitted from the preferred model when coverage proved to be consistently insignificant. In contrast, Hynds and Smith (1994) when examining demand for test match cricket, were precluded from analysing the influence of television due to saturation coverage.

  8.  The first study to explicitly examine the impact of live and edited highlights coverage of English FA Premiership football in the 1993-94 season (Baimbridge et al, 1996), found that attendance fell by 15.2 per cent for Monday evening games although there was no evidence of declining attendance for either satellite televised Sunday matches, or those featured as edited highlights on terrestrial channels.

  9.  Subsequently, Kuypers (1997) reviewed the potential future implications of pay-per-view television, whilst a number of more general studies (Horsman, 1998; Morrow, 1999; Szymanski and Kuypers, 1999), have sought to review the inter-relationship between television and sports within the growing business orientation of professional football in the UK.

  10.  In the US[5], Siegfried and Hinshaw (1979) found that the television status of a NFL match had no significant effect on non-attendance on advance ticket holders (no-shows). Arguably, however, the opportunity cost for such patrons is higher than for spectators in general. Zuber and Gandar (1988), concluded that no-shows were insignificant for the League as a whole, but increased in likelihood for less successful teams. The study of Siegfried and Eisenberg (1980) suggested that telecasts will not have a significant role in explaining differences in attendance, but fails to provide supporting empirical evidence. Thomas and Jolson (1979), on the basis of a survey, reported that baseball was regarded as a substitute good for television. With respect to televising baseball matches in the home team's city, Hill et al (1982) again found no influence on demand. For American college football, Kaempfer and Pacey (1986) reported that telecasts and attendance are complementary. Fizel and Bennett (1989), examining the same sport over a longer period, find that an increase in television appearances is detrimental to demand, whilst a complementary relationship existed between the number of historical broadcasts and attendance.


  11.  The media have for most of this century played a central role in the development of sport. Following the cinematic newsreels of the interwar period, the establishment of the British Broadcasting Corporation (BBC) brought with it both radio and television coverage of sport. The significant technological advances of the late 1960s not only proved important for television in general, but in particular for sporting events. Moreover, even the establishment in 1955 of commercial independent television (ITV) failed to dislodge the BBC from its position of pre-eminence in sports coverage until 1984 (Williams, 1994). Indeed the position in relation to major sporting activities had been determined by government statute declaring top-listed events to be the property of the whole nation[6].

  12.  The historical relationship between professional sports' governing authorities and broadcasters has been problematic, taking the view that televising either UK or even continental matches would significantly reduce attendance. This difficult relationship was illustrated with respect to professional football when in 1978 an exclusive agreement for television rights was reached between London Weekend Television (LWT) and the Football League for £4.9 million. However, following writs against both LWT and the Football League by the BBC claiming breach of previous non-exclusivity arrangements the League signed a four year contract for £9.2 million jointly with the BBC and ITV (Sloane, 1980).

  13.  This apprehension on the part of sports' governing bodies is likely to be a consequence of the net effect of television being theoretically indeterminate. It raises the relative price of attending a match which induces substitution away from the game. Additionally there is an income effect. A fanatical supporter could, for example, watch a live match and record the televised game. For a sport of minority interest there is also the possibility of drawing in new live audiences through initial television viewing (Kaempfer and Pace, 1986; Fizel and Bennett, 1989; Ingham, 1992). The effect of switching from predominately regional terrestrial television to BSkyB, as with Rugby League, is thus complicated whilst the potential television audience is expanded for the casual fan through continuous national exposure replacing regionally based coverage, simultaneously it is contracted for the regular aficionado unless they bought a satellite receiver[7]. Buying a satellite dish, together with the monthly channel subscription, would create additional income and substitution effects which would be expected to mitigate against attending live matches. The potential future development of a pay-per-view system for individual matches is likely to exacerbate these forces. Moreover, it is possible that general sponsorship may be dependent on the total viewing audience, thus the increased revenue from the move to satellite/cable/digital broadcasting may be outweighed by lost income from a net fall in audience.


  14.  The specification of the basic demand equation for match attendance (ATT) is a well trodden path[8]. Although a basic price variable is included, the unemployment rate has to be used as a proxy for income. The other non-economic variables represent variously, expected quality of the match, constraints on attendance and expected level of partisan interest, in addition to those variables with which we are mainly interested.

  15.  To analyse the factors influencing attendance the hypothesised determinants are grouped into four categories. First, is a series of economic variables. The preferred measurement of ticket prices was that of average price (PRICE) derived from unweighted nominal ticket prices for each game, together with season ticket prices, weighted in relation to the number of such tickets. However, previous studies have shown limited success when incorporating this variable resulting in both reported positive and negative influences upon attendance (Cairns et al, 1986).

  16.  A further economic variable which is frequently utilised as a proxy for differences in economic activity between areas is unemployment. In this study only the home team's regional unemployment rate (UNEMP) is entered with a hypothesised negative estimated coefficient.

  17.  Given the presence of brand loyalty associated with Rugby League club support, the main potential direction of substitution will be towards other leisure opportunities rather than cross-substitution between clubs. Thus the variable SUBSPORTS measures such possible alternatives within the club's catchment area. These constitute the other major winter sporting activities of professional football, Rugby Union and speedway.

  18.  The second category of hypothesised determinants are those related to demographic and geographic factors. Attendance will be potentially a function of the market size the club encompasses, thus a measurement of the population from Census data for local authority areas for the home team (HOMEPOP) is included. The away teams' likely travelling support is determined by the average of their home support weighted by the distance between the clubs in miles (AWAYSUPP). The initial geographical factor is the distance in miles between the home and away teams (DISTANCE). This is taken to indicate the factors such as the opportunity cost of time (Becker, 1965) and the potential additional difficulties relevant to away match attendance. However, this specification assumes a monotonic relationship with respect to the dependent variable. An alternative hypothesises that for relatively short distances match attendance will be higher, being indicative of derby matches with traditionally vociferous rivalry. To accommodate this feature of professional sport, the variable DERBY is included to represent matches when teams are situated less than 20 miles apart.

  19.  A third series of explanatory variables focus upon the relative attractiveness of the match identified by its prospective quality and overall importance. The former of these is tested through incorporating the number of special quality or star players in the home (HOMESTAR) and away (AWAYSTAR) team. Such a player is defined as either an overseas player or one that has appeared in an international match during the last three seasons.

  20.  With respect to the question of seasonal uncertainty the technique of Jennet (1984) based on the ex ante championship/relegation significance of each game using ex post information is problematic. It utilises information that is unavailable to the agents whose behaviour is being modelled, namely the eventual points total of the winning/relegated clubs (Cairns et al, 1986). Moreover, a team is assumed to be constantly within a mathematical possibility of the championship or relegation until the mathematical certainty is greater than zero. Thus no account is taken of the team's anticipated performance in relation to its own form and that of others.

  21.  In this study three techniques are employed to examine the question of seasonal uncertainty. First, two dummy variables represent when both teams were either in the top four league positions (TOPPOS) and hence in direct competition for the championship and in the bottom four places (RELPOS) and hence in relegation battles. Second, the home team's league position prior to the match (HOMEPOS), since the majority of the crowd is likely to constitute home supporters. Finally, the visiting team's league position (AWAYPOS) is included to further proxy for likely significance of the match.

  22.  The fourth general category of explanatory variables are a miscellaneous series of exogenous determinants. An intertemporial dimension to attendance is measured by a dummy variable which is hypothesised to represent residual support following the club's performance in the previous season (POST-1). The age of each club relative to the 1993-94 season is examined to differentiate between established and newly formed teams to allow for the effect from a build-up over time of both tradition and support, with the variable FORMATION indicating if a club is less than 20 years old.

  23.  Specific exogenous match-day variables for weekday games (WEEKDAY), evening matches (EVENING) and those played on public bank holidays (BANKHOL) are estimated in relation to the traditional fixture time. Further, improved measures of weather by using three separate types of variables for warm or cold air temperature (WARM), high wind (WINDY) and rain (WET), respectively, were derived from match newspaper reports and general weather summaries. There is a paucity of evidence for weather effects (Bird, 1982; Janssens, 1982; Peel and Thomas, 1998), although Cairns (1990) reported a strong negative impact of rain and weak evidence of a positive influence of sun, whilst Hynds and Smith (1994) found that rain was detrimental to cricket attendance. However, the use of local/regional rather than site specific data questions the reliability of even these findings. Furthermore, in the context of this study, the use of inadequate proxies for weather conditions may cause serious specification error thus biasing the estimate of the television effect. For instance, the potential attender who is on the margin of deciding to watch a match on television is going to find climatic conditions of crucial significance (Thornes, 1977). There are two aspects of prevailing weather conditions, first they alter the comfort of the attender, and second, they may also influence the quality of play for an open-air sport such as Rugby League. Moreover bad weather will also increase the risk of match cancellation.

  24.  As previously discussed the net effect of television is theoretically indeterminate. To examine these hypothesised effects the variable BSKYB is incorporated to indicate those matches televised live by satellite by BSkyB during the 1993-94 season.


  25.  Table 1 shows the multiple regression results[9].

Table 1


Dependent variable: log (ATT)
Independent variables:
Adj R20.85
F (22,217) 63.90**

  Where: ** indicates significance at 1 per cent level; * indicates significance at 5 per cent level.

  26.  The price and income variables do not perform well which is not unknown in sport demand studies. The income proxy is insignificant while the price variable is wrongly signed according to orthodox economic theory. This could result from either a price discrimination policy adopted by clubs relative to the importance of an individual match, its specific timing within a holiday period, or merely reflecting that the price elasticity of demand (attendance) is particularly inelastic (unresponsive) for some clubs, thereby allowing them to set admission prices in excess of the league norm without suffering a consequential decline in attendance, indicating significant supporter loyalty.

  27.  Of the second category of four geographic and demographic factors, three are found to significantly influence match attendance. Both the home teams' local population and the away teams' weighted travelling support positively impact upon attendance, whilst the distance between the clubs exerts the expected negative influence with each mile between the participating teams reducing attendance by 0.2 per cent. However, the anecdotal suggestion that "local derby" games attract a more fanatical crowd might be true in the volume of feelings expressed, although the actual attendance is not significantly different from non-derby games.

  28.  The penultimate series of explanatory variables attempted to analyse the key issues of match quality and seasonal uncertainty. As with previous studies (Schollaert and Smith, 1987; Jones and Ferguson, 1988) the quality indicators of star players were significant for both home and away teams, adding 11 and 3.9 per cent respectively to attendance. The two indicators of actual league position are significant, with each single movement of the home team affecting demand by 17.3 per cent, whilst that of the visitors alters attendance by 1.1 per cent.

  29.  Of the nine variables in the final category of match specific exogenous factors, six are found to be statistically significant. For those examining games played on alternative occasions to the traditional time, weekday matches result in an average fall in crowds of 21.6 per cent, although Bank Holiday games raise attendance by 11.9 per cent.

  30.  The negative relationship between the clubs formed less than 20 years ago and attendance indicates that relatively newly constituted teams suffer from a slow build-up of support. Given that their crowds are some 58.1 per cent below those for clubs established for more than 20 years, this would suggest that if Rugby League is to prosper through developing a wider geographical base of teams and thereby support, newly formed clubs might require particular nurturing until attendance and financial sponsorship enables them to compete with more established teams.

  31.  This analysis has striven to produce improved measures of the prevailing weather, whilst most previous studies have utilised a single dummy variable for climatic conditions (Cairns, 1990). In contrast to a contemporaneous study of attendance at Premiership football matches (Baimbridge et al., 1996), this analysis indicates that both cold and windy conditions adversely affect crowds by 9.4 and 10.8 per cent respectively. A possible explanation for this is that Rugby League stadium facilities remain more spartan than their top-flight counterparts in professional football, which have improved through greater investment and as a secondary influence of the Taylor Report (Cm. 962, 1990).

  32.  Finally, with respect to the transmission of matches by BSkyB, the scheduling and corresponding live broadcast leads to a reduction, certis paribus, of an estimated 25.1 per cent in attendance. Thus in comparison to a normally timetabled fixture, a quarter of the usual spectator audience is lost through simultaneous satellite coverage.


  33.  Although caution must be exercised when analysing the results from a limited study, the estimation techniques employed, together with examination of television coverage would appear to be complementary extensions to previous studies. It is, however, the hitherto neglected influence in UK studies of television coverage that this memorandum has primarily sought to focus upon. Findings indicate that live transmission of matches result in a significant decline in demand, certis paribus.

  34.  The fee for a satellite televised match on BSkyB, usually on a Friday night, for the period studied was £20,000, approximately equal to 1/7 or 1/8 the price of a "star" player. The initial question, however, is whether the lost revenue is compensated by the fee. Given average ticket prices and attendance figures, gate receipts for a non-televised match would approximate to £33,000. With live satellite transmission of the game we would expect, on average, that receipts would fall to some £24,700, a loss of £8,300. However, once the match fee is taken into account the average club will set a net gain of £11,700. Given the specification used here, a club with lower attendance, ceteris paribus, stands to gain more as it loses ticket sales.

  35.  Therefore the general implication of the analysis is that Rugby League clubs have secured a lucrative contract from their arrangement with BSkyB. However, the increased income derived from television rights has resulted in several fundamental repercussions. Although the additional revenue has been diverted into team building not only to maintain Super League status, but also to mount successful challenges for the lucrative international club competitions; a limited supply of talented players has led to an inevitable spiral in transfer fees.

  36.  A further corollary is the effect of smaller television audiences following the movement to satellite and away from terrestrial broadcasters. Although such figures appear not to have been detrimental to the Super League's income from sponsorship and advertising, the potential long-term impact might be more fundamental since clubs require the continued commitment of the public on two key fronts. First, as spectators to provide the rationale for league and cup competitions as a spectator sport. Second, as the basic raw material for the game itself in terms of aspiring players. The danger is that through screening out the majority of the potential audience these two factors fundamental to Rugby League's prosperity could be weakened.

  37.  The critical finding of this memorandum relates to the first evidence of the consequences of live broadcasting upon demand for Rugby League. For BSkyB coverage estimated attendance fell by 25.1 per cent in the 1993-94 season. However, there is an overall gain for clubs in terms of both gate receipts and the broadcast rights fee. Moreover, any future deal between Rugby League's governing bodies and broadcasters is likely to exploit a greater breadth of secondary options such as pay-per-view, again substantially raising the sport's income. However, the distribution of satellite coverage is skewed in favour of the more successful clubs, which tend to be wealthier[10]. Therefore, while the general implication of the analysis is that clubs are currently receiving a good deal from televised matches, on the current pattern of live coverage there exists an inherent risk of widening income inequality between clubs. The relationship between television and sport therefore possesses profound consequences for those professional sports which embrace the financial rewards offered by the new satellite, cable and digital television channels.

June 1999

3   Williams (1994) examines two perspectives of the relationship between satellite television and sport. The contract between BSkyB and the Football Association for coverage of the English Premier League and the role of sport in constructing national identities. Neither aspect therefore directly relates to this study. Back

4   Satellite television began in the UK at the end of the 1980s, initially offering three channels dedicated to sport. Deregulation, with the abolition of the listed events clause, raised concerns that satellite companies would outbid terrestrial stations for major sporting events (Whannel, 1992). Back

5   For a review of the history of sports broadcasting in the UK see Hochberg and Horowitz (1973), Horowitz (1974) and Pacey (1985). Back

6   The Television Act of 1954 contained a non-exclusivity agreement such that no broadcasting organisation was permitted to obtain exclusive rights for the following events: the Boat Race, Test Match cricket, Wimbledon, the FA Cup Final, the Grand National and the Epsom Derby (Whannel, 1992). However, the Broadcasting Act of 1990 stated only that these events shall not be part of any pay-per-view scheme. Thus satellite, cable and digital companies would not be prevented in future from bidding for these events. Back

7   They could now purchase into cable or digital coverage, but that was not possible at the time covered by the data in this study. Back

8   See, for instance, Hart et al (1975); Hill et al (1982); Bird (1982); Janssens (1982); Schofield (1983); Borland and Lye (1992); Hynds and Smith (1994) and Baimbridge et al (1995, 1996) for details of the usual model specification in demand studies of professional team sports. Back

9   It is estimated by OLS following a logarithmic transformation of the dependent variable which renders the variables interactive. For dummy variables the coefficients are biased estimates of the percentage between group differences (Halvorsen and Palmquist, 1980; Kennedy, 1981). Following the Breusch and Pagan (1979) Lagrange multiplier test for heteroskedasticity, the reported results show t-ratios derived from heteroskedastic-consistent standard errors. Given the semi-logarithmic functional form of the estimated model, the inclusion of quadratic regressors and relatively small t-statistic for the intercept term, additional tests such as RESET (Ramsey, 1969) for specification error and non-linearities are not reported. Indeed, the RESET test although designed to detect missing regressors is also an effective test for non-linearity. Consequently, this severely reduces its attractiveness, since rejection of a model could be due to either non-linearity or an omitted independent variable (Kennedy, 1992). Back

10   For the 1993-1994 season of First Division matches covered in this study the highest rates of satellite coverage were for Wigan (7), St. Helens (7) and Leeds (5), whilst the lowest levels were Leigh (3), Castleford (2) and Featherstone (1). Back

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries

© Parliamentary copyright 1999
Prepared 14 December 1999