The bulk of this paper discusses a series of surveys of attendances at cultural performances. These survey data are compared with ticketing statistics which cover mainly major performing companies. The attendance surveys are published by the Australian Bureau of Statistics (ABS), and the ticketing data by Live Performance Australia (LPA), the peak body of the entertainment industry. The two sources provide different types of data, both valuable for the understanding of the developing musical performance markets.
Among the performance categories covered by the full period of the ABS surveys (1991-2006), the number of people attending classical music performances represent the only category increasing more rapidly than the underlying population growth. Attendances at popular music concerts have increased significantly more slowly than population growth, and so to a lesser extent have attendances at theatre and dance performances.
The category falling most short of keeping up with population growth is the combined group of musicals and operas. Supplementary evidence from the surveys suggests that opera has borne the brunt of this, while attendance at musicals has improved in recent years.
This paper refers to an external statistical file.
There are two main sources: the Australian Bureau of Statistics (ABS) periodic surveys of attendances at cultural venues and events, and more recent ticketing statistics compiled under the auspices of Live Performance Australia (LPA). Because it is important to develop reliable and consistent attendance statistics, the apparent strengths and weaknesses of both sources are discussed in some detail in this section. Readers may wish to skip this technical section, which is possible through the hyperlinks in the contents list above.
The Australian Bureau of Statistics has conducted surveys of attendances at cultural venues and events since 1991. The full statistics are published in a series of publications under the ABS Catalogue number ("Cat") 4114.0. This source is recommended for further consultation and analysis, though summary data are available in Cat 4172.0 which gives a statistical overview of a number of aspects of arts and culture in Australia.
The venues and events most relevant here are classical music concerts, popular music concerts, musicals and operas, plus theatre and dance performances, other performing arts, and cinemas for perspective. Musicals and operas are generally treated as a group but some separate data are available (discussed in the ‘operas versus musicals’ section below).
The other attendance data in the publications refer to art galleries, museums, zoos and aquariums, botanic gardens, and libraries. These venues are not discussed here.
The ABS surveys that appear to be sufficiently comparable to warrant presentation as a time series provide data for four 12- month periods ended June 1991, March 1995, April 1999, and June 2006, respectively (but see below for the ABS’s own comments on compatibility). For convenience, the four surveys are labelled by the calendar year in which the twelve months end: for example, 2006 rather than 2005-06.
The following discussion of coverage and validity is largely based on the explanatory notes in the 2005-06 version of Cat 4114.0.
The ABS conducts a monthly Australia-wide Multi-Purpose Household Survey (MPHS) as a supplement to its Labour Force Survey. It consists of a number of special-purpose surveys including, every four years or so, the survey of persons attending cultural venues and events. The 2005-06 survey was based on the MPHS, while the three previous surveys were based on the compatible Monthly Population Survey, which has existed since 1960.
Comparisons with data collected during 1995 and 1999 supplementary surveys to the Labour Force Survey are shown (and comparisons of 1991, 1995 and 1999 in the previous publication), but the ABS notes that “the methodology used in these surveys differed between years, as well as to the 2005–06 MPHS, and this may affect the validity of comparisons. It is not possible to determine the extent to which the differences between the 2005–06 MPHS and the 1995 and 1999 survey methodologies have contributed to the different results.”
Additional attendance surveys were conducted in 2002 and 2006 as part of yet another general ABS survey, the General Social Survey (GSS). The 2002 survey has been used previously in the measurement of trends since 1991, but the explanatory notes in the 2005-06 publication now state that it is not compatible with the other attendance surveys and should not be used in the comparisons.
Table 9 in the statistical appendix compares the 2002 and 2006 GSS-based surveys. It shows higher attendance rates for 2006 than the 2005-06 Multi-purpose Household Survey (MPHS), even after adjusting the latter to exclude 15-17 year olds who were not part of the GSS. Attempting to explain the difference, the notes to the GSS publication suggest “context effects, as well as different collection methodologies (face-to-face interview in the GSS versus telephone interview in the MPHS). However, it has not been possible to determine the extent to which the differences between the MPHS and GSS methodologies may have contributed to the different results.” (p 97)
For the time being, then, we have two different time series but our main focus is on the four surveys between 1991 and the MPHS-based 2006 survey. These surveys basically cover the population aged 15 years and over living in private dwellings.
The 2005-06 MPHS was based on a sample of 18,846 private-dwelling households, which yielded 14,219 full responses (88%). An individual aged 15+ was randomly selected from each household, the results aggregated, and each population group weighted and ‘benchmarked’ against known population statistics to yield representative total population estimates for each group.
As in all surveys, the results are subject to statistical sampling errors, but additional non-sampling errors may arise from non-response and a range of other factors. The ABS notes that “every effort is made to reduce the non-sampling error to a minimum by careful design of questionnaires, intensive training and supervision of interviewers and effective processing procedures.”
These publications enumerate people who attended a cultural venue or event at least once during the 12 months before interview in 2005–06. It is noted that “the data presented cannot be compared with any ‘total admissions’ administrative data held by cultural venues, since details in relation to attendees under 15 years of age were not part of the survey. In addition, total admissions data includes multiple attendances while the MPHS provides data on how many times a respondent visited a cultural venue or event.”
The ABS does not note that asking people about their visits to cultural venues and events over 12 months may provide a bias, since few would be able to recall exactly whether attendances were within an exact period, let alone how often they visited during that period. It is difficult to determine whether this bias exaggerates or diminishes the number of estimated attendances.
LPA is the peak body for Australia’s live entertainment and performing arts industry. Since December Quarter 2003, it has been conducting a survey described as “the cornerstone of the case which we make to government, the media and the general public about the economic, social and cultural value of the Australian live performance industry.”
The survey provides ticketing and revenue information for ten different live performance categories, of which six are reasonably comparable with the ABS categories. Each LPA category is carefully defined in the survey publication. The six categories are shown in the reference given (musicals and operas are combined in most of the ABS tabulations).
The four remaining LPA categories are children’s performances, special events, multi-category festivals, and single-category festivals. Apart from special events (exemplified by wrestling contests), these other categories would contain some musical performances.
The data sources are eight major ticketing companies (two national, and one from each State and Territory except Victoria and Tasmania), and the 28 members of the Australian Major Performing Arts Group (AMPAG), whose ticketing statistics is organised through the Australia Council for the Arts. Data have been collected quarterly since December Quarter 2003, and we understand that results for 2006 will soon be available. The explanatory notes in the survey publication include the following observations:
The differences are taken up in connection with the concluding statistical analysis (concentrating on Table 11) at the end of this paper. Basically, the LPA survey concerns ‘major’ events but not the plethora of gigs which take place in local pubs and clubs, small festivals and many other live performances.
The next sections describe the main ABS attendance data. Please click here for the statistical file, which opens in a separate window. Readers are advised to keep the file open and refer to it as the story proceeds.
About 1.5 million persons aged 15 years and over stated that they had attended one or more classical music concerts according to the 2006 survey of attendance at cultural venues and events during the past 12 months (Table 1). With a total population of 16 million aged 15+, the attendance rate was 9.4% as shown in the right-hand half of the table. However, the attendance rate was only about 6% for young people below the age of 25, rising to a peak for 45-64 year olds (about 12.6% for the two age groups combined). It then declined but still managed to stay slightly above the 9.4% average for people aged 75 years plus.
The overall attendance rate for classical music concerts was the lowest of the performance categories shown in Table 1. Apart from cinemas, included for perspective and showing that 65% of the population aged 15+ said they had been to the pictures at least once in the past 12 months, the highest attendance rate was for popular music concerts (25.2%), followed by theatre (17%), other performing arts (dominated by circus) with 16.6%, musicals and operas as a group (16.2%), and dance performances (10.2%).
Age distributions differed markedly with the highest proportions of younger people going to popular music concerts (especially the 18-24 year group). However, attendances were significant across the age groups and only dropped off markedly from age 65 or so.
The third main music category, musicals and operas, had two peaks: one in the 15-17 year group and one between 45 and 64. It is possible that the distributions would differ if we were able to show musicals and opera separately. Unfortunately the publication doesn’t show this detail. However, there were similarities with theatre goers showing similar peaks, and to some extent for people attending dance performances (though the second peak was in the 35-54 year groups rather than 45-64). All these categories (musicals and operas, theatre and dance) showed marked declines between the 15-17 and 18-24 year groups, possibly because of the influence of school performances for the younger group.
One significant change over the 15 years shown in Table 1 is the aging of the audience, especially among classical music concert goers (Chart 1). In 1991, the highest attendance rates were among the 35-44 and 45-54 year age groups, both at 10.2%. By 1995, the peak had shifted to the 45-54 year group, by 1999 to the 45-54 and 55-64 year groups, with a further aging peaking on the 55-64 year age group in 2006. Looking across each age group, the relative shift increased with age: the 15-17, 18-24 and 25-34 year attendance rates shifted much more modestly upward than was the case in the 55-64, 65-74 and 75+ groups.
Classical music is unique among the categories covered in the ABS surveys of attendance at live performances, in showing a significant increase in overall attendance rates, from 8% in 1991 to 9.4% in 2006 (relative increase + 18%). Ignoring the intervening overall rates for 1995 and 1999, the comparable relative changes (all declines) were:
The age patterns were also very different. There was an almost continuous decline in attendance rates for popular music performances from the 18-24 year group on, though the curve was much flatter in 2006 than in 1991, and attendance rates were actually slightly higher in 2006 than in 1991 for age groups 45 to 74 (Chart 2). Popular music comprises a range of styles from pop, rock, hiphop, rap and heavy metal, to blues, jazz, country and folk. Some of the latter genres (such as jazz) may attract older people. We don’t know if these genres have become relatively more popular, but if so it would help explain the aging of the total popular music audience. There is also some evidence based on physical observations that older people are turning out for the increasing number of tours by the pop stars of their youth.
Attendance rates at musicals and operas (Chart 3) were lower in all age groups in 2006 than in 1991, with a definite shift towards the older groups. Attendance rates were relatively low among younger people but rose towards the middle age groups, peaking in the 45-54 year group in 1991 and in the 55-64 year group in 2006.
Attendance rates for theatre goers in 1991 showed a generally falling trend through the ages (except for low rates in the 18-24 group). By 2006, however, there was a strong recovery in the three decadal age groups from 45 to 74, to above the attendance rates fifteen years earlier (Chart 4). Somewhat similar trends were observed for dance audiences, except that the recovery in the older age groups was weaker and largely limited to the 35-44 and 45-54 year age groups (Chart 5).
In summary, the factors affecting total attendances over the fifteen years are population growth, changing age distributions, and changing preferences within age groups. The bottom panel of Table 1 shows the total population aged 15+ in each age group in 1991, 1995, 1999 and 2006, rising from 12.9 million in 1991 to just above 16 million in 2006 (+23.4%). Clearly, the population has been aging as shown by the right-hand part of the bottom panel. The total population in the four youngest age groups (15-44) increased by 7% from about 7.9 million to 8.5 million. In stark contrast, the number of 45-54 year olds increased by 54% between 1991 and 2006, 55-64 year olds by 58%, the 65-74 year group by 23% and people 75+ by a staggering 84% - another striking sign that the population is aging. The relatively modest increase in those who were 65-74 years old in 2006 is a legacy of the depression years of the 1930s, when birth rates declined drastically.
We can now analyse the two main elements of total growth in attendances, of which total population growth accounted for 23.4% between 1991 and 2006 (Table 2). The second element is attendance rates, which consist of changes in age distribution and changes in the apparent popularity of particular performance types. The only category for which the latter factors acted in a positive fashion over the full period was classical music, increasing by 17.8% over and above the population growth factor. The attendance rate component was negative for the four other categories for which full 15-year data exist, with the worst effect for musicals and operas followed by popular music concerts. To the extent that the latter are especially favoured by young people, it would be relatively strongly affected by the changing age distributions. This is aggravated by young people seemingly becoming relatively less attracted to rock and pop, which seems to have been the case in the 15-24 year age groups according to Table 2. Would that be due to the development of technologies such as DVD and iPod, competing with live performances?
The right-hand half of Table 2 shows the same analysis between 1999 and 2006. During that period, the total population aged 15+ increased by 7.3%, but there was a particularly strong recovery in audiences at dance performances followed by a 7.2% increase for classical music concerts taking total growth for these performances to 15% for the most recent seven-year period. There was also a modest additional growth in attendance rates for theatre performances as well as a near-zero growth in attendance rates for musicals and operas. Attendance rates for popular music concerts declined marginally for popular music concerts, shown by Table 1 to be concentrated on the age groups up to 44 years, whereas attendance rates for all older age groups all rose significantly. This may represent a general shift towards live performances of country, folk and jazz away from rock and pop, as well as older people being attracted to the pop idols of their own youth touring in increasing numbers.
Table 3 shows the changing distribution of the population aged 15+. The postwar baby boom shows up in strong increases in the shares of 45-64 year olds, in contrast to declining shares of younger age groups. The share of 65-74 year olds declined between 1995 and 2006 as the proportion of those born in the depression years of the 1930s increased. The oldest age group, 75+, increased at the highest rate of all those represented in Table 3.
The median age of all audiences increased between 1991 and 2006 (Table 4). It was strongest for theatre performances (from 36.6 to 44.3 years, or a relative change of 21%), followed by popular music concerts (19%), other performing arts including circus and variety shows (19% since 1995), and classical music concerts (16%. Classical concert audiences had the highest median age throughout (49.3 years in 2006, compared with 36.7 year for popular music concerts).
Female attendance rates in 2006 were almost without exception higher than male attendance rates for all audiences (Table 5). The only exceptions were the 25-34 and 65-74 year age groups for popular music concerts, and 65-74 year olds for classical music concerts, which may or may not be statistically significant findings. The relative difference in attendance rates over all age groups varied quite markedly:
These ratios are affected by changes in the overall number of females relative to males, shown in the bottom panel of Table 5. Apart from the top age group of 75+, where the number of women exceed the number of men by 33%, the differences are generally modest (the total number of females 15+ exceeded males by 3%).
The surveys also explore other characteristics of respondents (see details in the publications themselves):
Attendance rates for classical music performances in 2006 were significantly higher than the overall average of 9.4% in the Australian Capital Territory (13.3%), higher in Western Australia (10.8%) and marginally higher in New South Wales and South Australia (9.7%). The lowest attendance rates were in Queensland and the Northern Territory, at 8.2% and 7.9%, respectively (Table 6). Classical music concert attendance rates in Australia as a whole were significantly higher in the six State capital cities (10.6%) than in the rest of Australia (7.4%) according to the second panel of Table 6. The relative difference (43%) may explain some of the differences in total geographical attendance rates: the capital city component of the ACT was practically 100% according to the bottom panel of the table, and Western Australia had the second-highest capital city proportion, while Queensland and the Northern Territory had the second- and third-lowest capital city proportions (only Tasmania showed a lower proportion). So the ratio of capital city to total population provides part but not all the explanation of differences between attendance rates for classical music audiences.
The capital city ratio is less likely to influence differences between popular music attendance rates as the overall attendance rate was only 7% higher in the six State capital cities than in the rest of Australia. The attendance rate was again way above the Australian average of 25.2% in the ACT (34.1%) and Western Australia (30.8%). The lowest rates were in NSW (23.9%), Victoria (23.6%), and Tasmania (24.1%).
The geographical audience pattern for musicals and operas is different again from the other categories in Table 6. While attendance rates in the six State capitals exceeded the rates in the rest of Australia by 43%, the pattern across States and Territories only partly reflected the relative dominance of the capital city. The overall attendance rate of 16.3% was exceeded in the ACT (19.1%) and Victoria (18.4%), both areas with a relatively high proportion of the population living in the capital city. But attendance rates in South and Western Australia, which both have an even greater concentration in the capital city than Victoria, were relatively low.
The main classification of performance categories combines operas and musicals, but there are supplementary tables that enable the two to be separated. Table 7 shows the number of people aged 15 and over who attended either opera or musicals or both during the past year. Table 8 shows the associated attendance rates. The three panels of each table refer to the surveys in 2006, 1999 and 1995, respectively.
The left-hand column of each table shows the main ‘opera or musical’ question included in other tables (2,614,000 persons in 2006). This excludes persons who went to both musicals and operas, and therefore would add to the total number of attendances if musicals were split from operas. The next three columns show who went to musicals, who went to operas, and the total of these two. In 2006, this total amounted to 2,807,700 persons, which exceeded the main ‘operas or musicals’ total by 193,700.
The inference is that these 193,700 went to both operas and musicals, and should therefore be counted in both categories if separated from each other. Hence, out of the total 2,402,100 people who attended musicals according to Table 7, that number less 193,700 (or 2,208,400) must have attended musicals but not opera, and 211,900 must have attended operas but not musicals. The sum of 193,700, 2,208,400 and 211,900 equals the main total ‘operas or musicals’ figure of 2,614,000 persons in 2006.
Because these statistics are available for three years, we can now infer that attendance at opera performances showed a steady downward trend between 1995 and 2006. Those attending both operas and musicals fell from 307,300 in 1995 to 217,000 in 1999 and 193,700 in 2006 (a total fall of 37%). Those attending opera only declined from 327,100 in 1995 to 293,500 in 1999 and 211,900 in 2006 (a fall of 35%). Meanwhile, attendance at musicals but not operas fell in the 1995-99 period but then showed a substantial increase in the period to 2006, leaving the total number of people attending musicals at approximately the same level as in 1995.
In terms of attendance rates, eliminating the impact of population growth, there was a fall from 17% for musicals in 1995 to 14.3% in 1999, followed by an increase to 15% in 2006 (total relative decline 1995-2006: 12%). For operas, the decline was continuous from 4.5% in 1995 to 3.4% in 1999 to 2.5% in 2006 (total relative decline 1995-2006: 44%). The two components making up total opera attendance rates both showed large relative declines over the 11 years: those attending both operas and musicals by 45% from 2.2% to 1.2%, and those attending opera only by 44% from 2.3% to 1.3%.
The downward trends in opera attendance rates were generally evident across Australia, but Tasmania and the Northern Territory appeared to buck the less violent overall downward trend in attendance rates for musicals (though these rates are subject to relatively high statistical sampling error).
It was mentioned in the introduction to this paper that attendance statistics have also been collected as part of the ABS General Social Survey. This evidence is summarised in Table 9, showing overall results for males, females and total persons in 2002 and 2006, and a valuable comparison showing that these new results are higher than shown by the main surveys analysed above.
The main result is a general increase in attendance rates for all live performance categories between 2002 and 2006, ranging from 9% to 12.4% for classical music performances and 26.4% to 32.4% for popular music performances, to an increase from 18.7% to 23.2% for musicals and operas. All the 2006 rates are way above the comparable MPHS survey results described in previous sections. As described in the introductory section, the ABS rather disconcertingly is unable to explain why the results are so different. For the time being, we recommend using the results from the main surveys described in previous sections, though they may err on the conservative side.
Respondents to the main surveys described in previous sections were asked to state the number of times they had attended particular events and venues during the previous 12 months (Table 10). In the 2006 survey the question was quite specific for low frequencies (once, twice, three times, four times, five times), and then grouped into 6-10 times, 11-20 times, and 21 and more times. As a result, we can infer a total number of attendances in the 12 months ended June 2006 by assigning reasonable average figures to the grouped 6-10, 11-20 and 21+ questions.
For instance, 1.5 million persons attending classical music performances appeared to have generated some 4.5 million attendances, or an average of about three concerts each. Four million people attending popular music performances appeared to attend almost 14 million performances, or about 3.4 concerts each. And so on for the other performance types shown in Table 10.
These results can be compared with the Live Performance Australia ticketing survey discussed in the early part of the paper. Bearing in mind that this survey concentrates on elite performances by major performing arts organisations and those performances administered by major ticketing companies, we would not expect anything like full agreement. The coverage is too different.
Table 11 bears this out. For dance, classical music, popular music, and theatre performances, the estimated attendances from the ABS survey as reported in Table 10 are between 3.3 and 4.1 times higher than the LPA survey results. The multiple is smaller for musicals and operas (presumably because a larger proportion is performed by major organisations) but still a substantial 1.76.
It is important to reconcile these differences as far as possible. One problem with the ABS survey is that of exact recall over 12 months. When did I last go to the opera? Not easy to remember, let alone how many times I went during the exact past 12 months. Against this, the exclusion of children is the most important omission in the ABS survey.
It is difficult to estimate how this affects the ABS findings. It is also very hard to determine the coverage of the LPA survey. Clearly it encompasses the major events put on by the AMPAG companies or otherwise covered by the major ticketing bureaux, and the major participating festivals (mentioned but not specified in the LPA survey publication). But a large number of events is not covered, which would have drawn responses in the ABS survey, including:
One obvious line of research is to develop the above points into a comprehensive list of what is and what is not included in the ABS and LPA surveys, respectively. Both approaches provide important results and should be retained but not rigidly compared. The ABS survey provides valuable numbers, including time series, on participation in cultural events. The main time series from 1991 to 2006 provides a comparison which makes sense.
The ticketing survey, summarised in Table 12, provides an important view of major live performance events, which may be extended but which it is difficult to imagine ever being developed to cover all live performances in Australia – which is not really its intention anyway. The data provide a new dimension of knowledge about the music and adjacent sectors of the Australian economy, and one cannot but encourage its continuation.
Hans Hoegh-Guldberg. Last updated 3 January 2008.