Hasan Bakhshi

Who they are

Bakhshi's paper is the best measured case this publication holds for what mediation adds, and it asks the displacement question in the one place where somebody had the transaction data to answer it properly. National Theatre Live launched in June 2009, broadcasting performances from the National Theatre into cinemas. The obvious worry was that a cinema screening would take the audience a theatre needed. The obvious hope was that it would introduce the art form to people who would then buy tickets. Bakhshi and Whitby tested both.

The data came from The Audience Agency's Audience Finder, funded by Arts Council England, with screening locations supplied by the National Theatre. About 16 million transaction records for 54 performing arts organisations across England, representing 44 million tickets, from early 2009 to late 2013. After cleaning and coding by art form, about 12 million records representing 28 million tickets remained, 32 per cent of them theatre, with 21 of the 54 venues in London. The method was ordinary least squares on log visits, with two-way fixed effects for geographic area and for organisation by year, controls for population and Land Registry median house price, and treatment coded as distance from a participating cinema in one-kilometre bands against a control region of five kilometres or more.

The result falls away with distance in the shape a real local effect should. Theatre visits were about ten per cent higher within a kilometre of a cinema that had shown a broadcast in the previous calendar year, seven per cent at one to two kilometres, three per cent at two to three, and nothing beyond. A single three-kilometre dummy gives a 5.3 per cent average effect, 6.4 per cent in London and an insignificant minus 0.1 per cent outside it. Their conclusion is that they find no evidence of cannibalisation at a broad spread of English venues, with some evidence of complementarity, and that caution is nonetheless required in interpreting the results.

The reason to cite them rather than merely to quote them is what they do next. Their own second hypothesis had predicted that the effect should be stronger on theatre than on other art forms. It was not. Non-theatre visits showed a similar pattern, about seven per cent within a kilometre. They offer two readings in their own words: that the broadcasts promote arts attendance generally, or that the analysis suffers from an omitted variable correlated with the treatment. Then they name the candidates themselves, including cinemas self-selecting into the programme in anticipation of a growing high-cultural-participation demographic. They also warn against reading aggregate area-level results as individual behaviour. A paper that states what would undo its own headline is worth more than one that does not.

Two figures from their introduction are the plainest available statement of what mediation adds to reach. The National Theatre's first live broadcast, Racine's Phèdre on 25 June 2009, allowed it in one night to double the overall audience for the play across its three-month run at the National. On 27 February 2014, 120,000 people watched War Horse in United Kingdom cinemas, against the 1,024-seat capacity of the theatre it was staged in.

Key papers

Bakhshi is an economist working in a research foundation, and the document that matters here is a working paper rather than a peer-reviewed article. This publication has verified no book of his and says so rather than guessing at one.

Bakhshi, H., & Whitby, A. (2014). "Estimating the Impact of Live Simulcast on Theatre Attendance: An Application to London's National Theatre." Nesta Working Paper No. 14/04, June 2014. JEL codes Z11 and O33, ISSN 2050-9820. The distance-band panel analysis, the no-cannibalisation finding, the anomaly stated against their own hypothesis, and the reach figures.

Bakhshi, H., & Throsby, D. (2014). Journal of Cultural Economics, 38(1), 1-8, cited by the working paper above as having reached the same conclusion for the National Theatre's own box office. This publication has not read it and lists it only as the working paper's own reference.

Key ideas

Simulcasting live theatre into cinemas showed no sign of taking audiences from English theatres across 44 million tickets between 2009 and 2013.

The local effect is real and short-range: about ten per cent more theatre visits within a kilometre of a participating cinema, falling to nothing by three kilometres.

The effect is concentrated in London, at 6.4 per cent on a three-kilometre measure, and is statistically indistinguishable from zero outside it.

The anomaly is theirs to report and they report it: the same pattern appeared for art forms the broadcasts had nothing to do with, which their own hypothesis had not predicted, and they name omitted-variable candidates that would explain it.

---