Bad arguments in favour of good ideas are a pernicious issue.

So many arguments for so many things come across our radars that objecting to an argument in favour of a conclusion we support seems a waste of breath — no matter how silly it may be.

I’d not intervene, for example, if I heard someone argue that voting for Donald Trump is a bad idea because he is Illuminati. Whatever. Fine.

But the cost argument against the same-sex marriage plebiscite offers a wonderful opportunity to stir us from this complacent state and say “no”. Even though a same-sex marriage plebiscite is a bad idea and would involve all sorts of costs, these iffy calculations from PwC ought not stand.

For context, PwC (the professional services brand formerly known as accounting firms Price Waterhouse and Coopers & Lybrand) this week released modelling showing it would be very expensive to hold a standalone national plebiscite to determine if Australia should permit same-sex marriage.

The firm added up the following to arrive at “total economic costs” of $525 million:

  • $158 million to facilitate the vote;
  • $66 million for the community to fund the “for” and “against” campaigns;
  • $281 million in lost productivity as people take time out to vote; and
  • At least $20 million in costs associated with the impact on the mental health and wellbeing of Australian citizens.

I’m not saying there’s anything bad about this particular analysis. PwC even acknowledges at the end of the report that aspects were left out. It’s just the same approaches are used whenever a social issue comes up for economic analysis — providing the illusion of precision where none is possible.

Here are the main problems.

  1. Arbitrary selection of modelled costs. What gets bundled in is what PwC can a) think of, and b) measure in a way that is, if not accurate, at least defensible.
  1. Counting some economic activity as costs and others as benefits. In this case, government expenditure on polling places and money spent on advertising firms are counted as a cost to society. In other cost-benefit analyses, that sort of expenditure would count on the benefit side, as “stimulating economic activity”.
  1. Adding cash costs and non-cash costs. These exercises routinely add up the opportunity costs of not doing something and money changing hands. Doing so relies on economic models of marginal cost being appropriate.

For example, PwC models the cost of voting at $17 an hour, based on the minimum wage. The reasoning comes from economics: the minimum payment you will accept to give up leisure for work must equal the marginal value of your leisure time.

The question of the opportunity cost of voting is an especially fascinating little walnut to crack open.

Applying conventional economic theory, it follows naturally that if the minimum wage is $17 an hour, voting costs at least $17 an hour (presumably voting is costless for the retired and really dear for the rich).

But then, if you apply conventional economic theory, voting makes little sense at all. The risk-weighted probability of your vote mattering is close enough to zero that there is no instrumental justification for voting. So the only reason to vote is to avoid a fine. You wouldn’t even enrol in the first place.

So if you assume “voting is like work”, you can’t really explain why people vote. So that assumption is questionable. Instead, some of the best contemporary economic theories of voting suggest voting is expressive, i.e. something we want to do. PwC ignores that.

Of course, cost-benefit analyses have their place. But it is in those parts of the community that are substantially market-based. If you are a firm planning to build a new factory, do the analysis. If you want to build a toll road, yes, perhaps a cost-benefit analysis will shed some light onto the process. But the further we stray from the realm of the purely economic, the more we should be suspect of their use.

To summarise: we should vote to endorse same-sex marriage in Parliament because it’s a human right that’s rather overdue. We should not believe reports that generate “total economic costs”. And we should be especially suspicious of their application in fields where markets are not at play.