I think that Oxoby's conclusions* are right on:
The question as to who was a better singer, Bon Scott or Brian Johnson, may never truly be resolved. However, our analysis suggests that in terms of affecting efficient decision making among listeners, Brian Johnson was a better singer. Our analysis has direct implications for policy and organizational design: when policymakers or employers are engaging in negotiations (or setting up environments in which other parties will negotiate) and are interested in playing the music of AC/DC, they should choose from the band’s Brian Johnson era discography.
But, if I were a JPE referee, I'd ask: where are the suggestions for future research? Here is a start:
Experimental economics has long played a role in improving the survey design of the contingent valuation method (CVM) (Cummings, Brookshire and Schultze, 1986). Future research should examine whether other AC/DC songs/singers, or the music of other hard rock bands (e.g., Led Zeppelin), could be used in other types of experiments and surveys. For example, it is quite possible that a good listen to You Shook Me All Night Long or Black Dog could mitigate hypothetical bias in willingness to pay experiments, especially in conjunction with a cheap talk treatment, an open keg and the burning of a big fat one. Implementation of hard rock music treatments would not be especially difficult to implement with telephone, internet or in-person CVM surveys.
*Note: Oxoby claims that the paper is a joke.