A hedonic model of player wage determination from the Indian Premier League auction: Further evidence¹

1 Earlier versions of this paper were presented at: (i) Economics Society of Australia (Tasmanian Branch), Hobart, 18 March 2009; (ii) Staff Developmental Workshop, School of Economics and Finance, La Trobe University, 23 April 2009; (iii) Seminar, University of Canterbury, Department of Economics, New Zealand, 1 May 2009; (iv) Seminar, University of Queensland, School of Economics, 22 May 2009; and (v) 38th Australian Conference of Economists, University of Adelaide, 28–30 September 2009. The authors would like to thank the various participants of the seminars and workshops for their comments and suggestions, as well as Seamus Hogan, Michael Smiddy, Ray Stefani, and two anonymous reviewers.

Research highlights

► Hedonic price models allow estimation of ‘fair values’ of professional cricketers in a unique open-auction framework, allowing inferences on bidding patterns. ► Players from India, with an X-factor and previous Twenty20 experience attracted systematic premiums in the auction. ► There is evidence of an economic ‘superstar’ effect—overbidding for star players and underbidding for lesser players.

Abstract

A range of cross-sectional models are estimated with a view to establishing the factors that determine the valuation of professional athletes in a highly-specialised sport, with an application to cricket's Indian Premier League (IPL). We distinguish between personal characteristic and playing ability factors, and with respect to the former, between ability in different forms of the sport. We find a number of interpretable variables that have explanatory power over auction values, while decomposition according to batting and bowling specialisations produces very different results depending on the use of either Test or One-Day International (ODI) variables. There is also possible evidence of inefficient bidding, insomuch that overbidding was somewhat correlated with players with higher realised values.

JEL classification:

• C21
• D44
• L83

Keywords:

• Demand for sport
• Cricket
• Auctions
• Cross-section models
• Labour demand

Notes

1 Earlier versions of this paper were presented at: (i) Economics Society of Australia (Tasmanian Branch), Hobart, 18 March 2009; (ii) Staff Developmental Workshop, School of Economics and Finance, La Trobe University, 23 April 2009; (iii) Seminar, University of Canterbury, Department of Economics, New Zealand, 1 May 2009; (iv) Seminar, University of Queensland, School of Economics, 22 May 2009; and (v) 38th Australian Conference of Economists, University of Adelaide, 28–30 September 2009. The authors would like to thank the various participants of the seminars and workshops for their comments and suggestions, as well as Seamus Hogan, Michael Smiddy, Ray Stefani, and two anonymous reviewers.

2 It is important to note that while further auctions would be planned for future seasons, such auctions would largely involve only new or ‘uncapped’ players, while there would also be a transfer ‘window’ (like those in European football leagues), in which existing players could transfer from one IPL franchise to another prior to the commencement of the new season. There was a 3-year lock-in period for players auctioned in 2008 and their contracts were guaranteed by the BCCI.

3 To be eligible for a BCCI contract, foreign players required a ‘No Objection Certificate’ from their country's governing body. Since the 18 April to 1 June IPL season clashed with the English County Championship season (as well as a tour by New Zealand), the England and Wales Cricket Board (ECB) refused to release their players. Subsequently, there were no English players signed in the first auction, although non-international regular Dimitri Mascarenhas was bought in the second auction.

4 Only one player (James Hopes) sold for more than USD150,000 in the second auction. In the first auction, only three of the 80 players sold for less than this figure.

5 These players were allocated according to a draft system.

6 Bangalore, Jaipur and Kolkata each exercised this option.

7 Discretionary judgement was required in constructing some binary variables, such as player specialisation and international team status variables, and for some players, the values of these variables may differ slightly between Tests and ODIs.

8 Given the enormity of the correlation matrix, it is not reported here but is available on request.

9 Goodness-of-fit is reported in Table 6 via both the standard (R²) and adjusted $({\overline{R}}^2)$.

10 In the case of ‘Overall ODI’, the two regressions are the same, except that the stepwise-forwards procedure retained a statistically insignificant (at the 8% level) regressor. On the other hand, in the case of ‘Bowling Test’ the stepwise-forwards regression is far too parsimonious—it has only three regressors while the swapwise regression has 12 strongly significant and intuitively appealing regressors. Furthermore, its explanatory power is only about half of the swapwise regression (the adjusted-R² statistics are 0.33 and 0.63, respectively).

11 The use of a mean value for this comparative exercise is on the grounds that the various models differ quite markedly, and do not always include the same sample of players.

12 Shahid Afridi is also labelled and worthy of a special mention—he is just inside the lower-bound interval, and a bigger outlier than the other three in terms of absolute (as opposed to logarithmic) value.