I’m currently reading Scarcity by by Sendhil Mullainathan and Eldar Shafir. At this stage, I’m inclined to sympathise with the unnamed colleague who commented “There’s already a science of scarcity. It’s called economics”. So far, it’s mostly straightforward applications of the observation that time and attention are scarce resources, combined with some fairly familiar observations from behavioral econ on how people fail to optimise either the first-order problems of allocating a tight budget or the second order problem of allocating time and attention to the first-order problem (my terms here, not theirs). However, I’m only part way through, and the authors promise to show how their approach differs from the way in which economists would normally think about this kind of problem.
This post is about a specific and well known observation cited by Mullainathan and Shafir. Faced with paying $100 for an item that could be had elsewhere for $50, most people are willing to put in a fair bit of effort (say, driving for an hour) to get the lower price.[^1] On the other hand if the item costs $1050 and could be had for $1000, people with reasonably high incomes mostly pay up, instead of driving to the other store. This is obviously inconsistent with standard opportunity cost.
But, it seems to me that there is an obvious and justifiable heuristic at work here. I buy $50 items fairly often, so if I regularly end up paying $100 for them, the purchasing power of (a large section of) my income is halved. On the other hand, paying a 5 per cent surcharge on $1000 items won’t have a big effect. It follows that it’s worth a fair bit of mental and organizational effort to work out ways of paying the $50 price rather than the $100.
The critical assumption in this argument is that, having secured the low price on one occasion, it will be easier to do the same in future.
Once you see this, you can improve a bit on the heuristic. For once-off purchases, at whatever price, you may as well pay a premium price if the mental and physical effort involved in getting the low price outweighs the premium. So, if you’re in a strange city you don’t plan to revisit, and faced with a cab fare that’s double the fair price, it’s probably better to pay up than to work out how to reach your destination by bus.[^2] In your home town, it’s a totally different matter.
Update As Richard Coppola points out in comments at CT, xkcd is all over this. Check the infinite regress in the mouseover.
[^1]: Depending on your circumstances you may need to adjust the number. And one of the neat observations by Mullainathan and Shafir is that poor (US) people, to whom $50 is a significant cost at any time, are not only more willing to put in the effort, but less prone to change in response to a change in the total price.
[^2]: I haven’t always taken my own advice on this.
John Quiggin 
Yes, search costs and search capital do matter as does the frequency of purchase.
“So far, it’s mostly straightforward applications of the observation that time and attention are scarce resources, combined with some fairly familiar observations from behavioral econ on how people fail to optimise either the first-order problems of allocating a tight budget or the second order problem of allocating time and attention to the first-order problem (my terms here, not theirs).”
Unless the first order problem is mis-specified, as would be the case with of micro-economic decision making problems which belongs to complete markets models (eg MRS = p).
So my best guess is the authors are introducing micro-economic problems in the context of incomplete market models.
I disagree with this rationale. It’s not about the actual cost saving – it’s about the emotional cost of being ripped off. A 5% price hike compared with 50% is of much less concern.
Purchasing decisions are more about emotions than they are rational decisions – hence the propensity to buy stuff that’s marked down on the basis that “it’s a bargain”, rather than because you actually need it.