These were the posted prices at my local gas station this morning:
(Note the 5 cent discount for cash on regular and premium, as opposed to the 65 cent (!!!) discount on plus.) Unless this was just a mistake, my complete inability to explain it makes me question whether I should be allowed to teach economics.
In case it helps, “premium” is 93 octane with 10% ethanol, whereas “plus” is 91 octane with no ethanol. This raises the subsidiary question of who buys “plus”. Ideologues who will pay extra for the privilege of not participating in the subsidy? And, incidentally a whole lot extra for the privilege of not using government-issued cash? Or what?
Are they always like this, with a big premium for credit card payments on purchase of ‘plus’ ?
Perhaps you should review Occam’s Razor….
jerry is right, Occam says the simplest explanation is the 9 should be a 3: $4.39
‘Who buys plus’? Any consumer who has a car manufactured to the European E5 (5% ethanol) standards and is hoping NOT to void the car’s warranty:
http://www.fcai.com.au/publications/all/all/all/3/can-my-vehicle-operate-on-ethanol-blend-petrol-
Hahaha! Nice to see that even expert economists trip up over the simplest explanations. (Or perhaps it’s precisely your expertise that hinders you from seeing the obvious solution).
Cheers and love to you, Steve.
Do they have an option for any other number of tenths other than 9/10?
I suspect there is some connection here with the previous posts. If we were to add the percent charged for credit on regular to premium and plus we would end up with 418.4 and 440.7 respectively. They actually charge 417.9 and 439.9 (assuming comments 2 and 3 are correct). This could be because the charge is different for some financial reason – unlikely – or the garage is willing to forego the extra in order to arrive at a price that ends in 0.9.
My guess is that the consumer is not reacting rationally to the prices, and is interpreting the price ending in 0.9 as cheaper than it actually is. The garage believes it will make more by reducing the price in this way.
Similarly to people being uncertain about the expected utility in the jar case, they are uncertain about the expected utility even with simple prices.
The matter of who would buy premium – well it is not quite as bad as it looks. The energy density of ethanol is quite a bit lower than gasoline (30 against 34.4 MJ/L). E10 has an energy density only 97% of gasoline, so the price difference is not as great as it looks. If you want to know why hydrogen has not taken off, this is one of the reasons. Liquid hydrogen energy density is only 8.5 MJ/L, and that requires significant cooling (critical temperature -240°C). Without cooling, even at 690 bar the energy density drops to 4.5 MJ/L. This can be ameliorated by greater efficiency of fuel cells, but we will still need a big tank to get us as far as a petrol car.
I can think of four explanations, mistake/misunderstanding, price discrimination, increased transaction costs.
A mistake or misunderstanding will be corrected by the employees or the owner when it is noticed or does not get the desired results.
Price discrimination might be occurring if the customers who tend to buy plus are less price sensitive. This does beg the question of why the cash versus credit price difference for the other two fuel types is exactly the same. For this to be valid, there would need to be some sort of geographic/brand factor. Do other nearby stations offer the same plus product? How do they price? Would a tourist or traveler passing through use the other stations? Price discrimination is unlikely due to the likely existence of other substitutes for customers.
While credit cards do have an increased direct transaction cost from the credit card companies over cash, the 5 cent difference on the other fuel types is likely not enough to fully cover that cost. However, it may be the case that collections are riskier with the most expensive type of gasoline. Are customers required to prepay for cash sales? If so, they are likely less risky.
Customers planning on engaging in drive offs or card fraud of some sort may select the most expensive variety in preference to the others. It’s unlikely to be reflective of cost differences due to fraud, as that is a short term solution. As the customer mix among tenders and local stations changes, the charge required to cover the increased fraudulent % as other customers switch away from credit plus purchases at this station will increase. That would likely require an even larger premium unless the steady state based on price discrimination including factors like station convenience and brand has already been reached.
Overall, I don’t really like any of the explanations other than temporary aberration, but I can tell some stories that might justify the cost difference. Time will tell as the station adjusts to the market’s reaction. If these persist, then there indeed is something underlying the difference in list price.
Apparently, I can’t count. I meant to say three explanations. I was going to talk about owner fraud as an additional possibility, but decided against it. Cash purchases might enable some fraud by the owner and have additional value.
ie Pump and replace (with fraudlent cash purchases from a suppler to balance the books) or simply inflate reported sales to realize additional subsidies, without having to buy additional gasoline. This might be easier to hide if there’s a larger volume of cash sales to keep the fraud under a certain % sales threshold. I couldn’t think of a reason why this applies to only plus and edited it out (can’t fake deliveries of the other types?), but missed the leading number.
Isn’t it usually the other way around, that premium is the one with NO ethanol?
I’m guessing mistake. Since it’s your local station, can we assume this is the first time this has happened, since it’s the first time you’ve posted about it?
Here’s one possible (albeit unlikely) guess: the station is running out of premium. It’s a temporary situation. The station has three options: (a) run out of premium; (b) raise the price of premium; or (c) raise the price of premium on credit only.
(a) might lose customers who now feel they can’t count on the station to have the product they want.
(b) might lose customers who feel that the station is trying to rip them off by only advertising the price of regular, but charging more than traditional “premium” on the 93 gas, which doesn’t show up until you get to the pump.
(c) would only lose customers who don’t have cash in their pockets. Those, for whatever reason, might be the less profitable ones anyway. Or, those might be the ones who would just switch to regular for that one fillup without really caring that much.
Farfetched, perhaps. I think (c) would still be a losing strategy because some customers would pay by credit without looking at the price, realize later that they’d paid 60 cents more than they expected, and curse you forever.
Smaller engines cannot handle ethanol, so people have to buy the ‘plus’ for lawnmowers, cbainsaws etc. Link : http://news.consumerreports.org/home/2013/03/gas-with-ethanol-can-make-small-engines-fail.html
Um, I don’t suppose one might not just go up and ask them?
Mistake.
David A, in economics that would totally be cheating! (It’s also assumed to be unreliable.)
I vote for mistake. The proof will be that they fix it.
I always see mislabeled gas prices. Assuming this is not mislabeled, you could say that plus customers are more likely not to carry cash and are also not shopping around for the lowest price they can get. For the plus customer, the more expensive the better, just like the vodka drinker.
That is why they call it “Plus”.
It is probably a mistake, but maybe the gas station operator is counting on “plus” people not paying close attention.
The discussion of fuel grades has revealed a hole in Steve’s knowledge. The three grades, regular, plus and premium are apparently based on octane rating only.
http://www.path2usa.com/gasoline-grades
According to Wikipedia, all gasoline must contain at least 5.9% ethanol in most states in the USA.
I agree with @Harold #18. Plus has nothing to do with ethanol, and only with octane rating. However, in South Carolina at least, there are stations that sell gasoline without ethanol. A friend of mine has computed that difference in mileage his Hyundai gets from gas with no ethanol and gas with ethanol is slightly greater than the price premium paid for no ethanol gas, so he purchases gas without ethanol.
Unfortunately, GasBuddy.com doesn’t have a price for his local station today, so I can’t demonstrate the comparison.
I don’t recall Occam’s razor ever being an acceptable explanation in Steve’s books.
This is the closest I’ve seen to a live-action performance of the old joke at the top of this article:
http://www.indexuniverse.com/sections/research/123.html
As the article goes on to tediously explain, mistakes by economic actors in efficient markets are just relatively rare, self-defeating, and hard to exploit, but not non-existent.
I it is just a mistake.
I agree it’s probably a mistake.
Regular should be 3.149 and 3.529 for premium.
Harold:
The three grades, regular, plus and premium are apparently based on octane rating only.
Nevertheless, it is in fact the case, at this particular gas station at this particular time, that premium is 93 octane and 10% ethanol, while plus is 91 octane and 0% ethanol.
#24: I stand corrected. I was looking at the general case and you have details of the specific case. It is quite difficult to know from the name what you are getting.
If presented with a ethanol-free gas option, I always choose it. Part of the reason is that, yes, I am spiteful and against the ethanol program — but I also have a rational reason. A quick back-of-the-envolope comparison revealed to me that it is at least no more expensive than the ethanol options, and I go further on a tank of gas with no ethanol. Thus, I have to fill up less often and that saves time. Time is valuable to me, and I would actually be willing to pay a premium for this option. (Pun unintended…)
In order to determine if ethanol-free gas would cost or save money, I would have to make many representative trips using both kinds (since I prefer an emperical approach), take careful notes, then compare the results. That I have not done. I don’t need to, because as noted, as long as it is close, I prefer ethanol-free gas.
Steve, your example is probably just a mistake, but Paul Krugman has argued that people seem to act irrationally in their gasoline buying habits. For instance, they switch from full serve to self serve when the price go up, and the switch what grade of gasoline they buy. See his post here:
krugman.blogs.nytimes.com/2012/07/22/mental-accounting-and-the-microfoundations-wars/
Alternative possibility: some of the numbers are generated by an algorithm and the algorithm happens to have reached a cutoff point for some but not for other.
For instance, imagine the algorithm “Put up a new cash price every day. If the existing credit price less than or equal to the new cash price, then set the credit price equal to the cash price plus some amount.” If this algorithm was used, it may be that Plus was bumped up yesterday and the other two were not.
Unless you have fun trying to untangle gordian knots, why don’t you just ask the owner or manager of the station?
I would guess that the cost of transacting the payment on CC for premium is higher, though I don’t know why absolutely but can only speculate. Either that or ethanol mixtures (which I think I’ve read before) taints the mileage per gallon as opposed to pure octane.
The station manager ran out of 3’s. This won’t be a problem after the price for regular reaches $4.
Yes, the proof of the pudding will be in how they behave over the longer term. Is Plus+Credit always significantly more expensive than Plus+Cash?
The answer would need information like : who are the typical customers who buy from this station? Especially, who buys Plus+Credit?
There’s a possible answer here : http://conversionxl.com/pricing-experiments-you-might-not-know-but-can-learn-from/ Perhaps the $4.99 for Plus+Credit is to encourage cash sales of Plus?
“Once you’ve seen a $150 burger on the menu, $50 sounds reasonable for a steak. At Ralph Lauren, that $16,995 bag makes a $98 T-shirt look cheap.”
And once you’ve seen Plus advertised for $4.999 a gallon, paying $4.349 cash looks like a good deal.
#31- I found the “magic of number 9” interesting, and related to my comments above. Chicago University sent out 3 identical mail order catalogs except one item was priced at either $34, $39 or $44. The $39 item sold the most. I can’t think of a “rational” reason for this.
Did they fix the error yet?
Bradley #33 – yes.
And now there is a crowd of “plus” consumers outside Steven’s house with torches and pitchforks.
Thinking about Coase made me think about the language used in the post. Steve refered to a 5c discount for cash, but equally this could be a 5c mark-up for credit. If there are only two prices, how can we say which is the “normal” price?
It’s been a – long – while since I’ve been to the US, is this “cash price” vs. “credit price” a recent development?
As with every posted price, they charge what they can get. As they should.
PG: This rather begs the question of “why is this what they can get?”
I am happy that demands are subjective. I used to look at contents of other peoples’ shopping carts.
PG: You’re perfectly welcome to not be interested in this puzzle, but I’m a little baffled as to why you felt the need to tell us so. There are a lot of people on the Internet talking about things I’m not interested in, but I don’t ordinarily interrupt their conversations to mention my lack of interest.
I was unclear. Some demands are mysterious. This is why many sellers find a price only after they go through a process of trial and error adjustments.
@Harold #35, 10 states have laws against credit card surcharges (see here for the list: http://www.fivecentnickel.com/2011/10/27/states-that-dont-allow-credit-card-surcharges/). In addition, the credit card companies standard agreements with retailers forbid credit card surcharges, as part of a 2010 agreement with the Justice Dept. However, the 2010 agreement and all 10 states allow discounts for cash. From the perspective of the consumer there wouldn’t seem to be a difference between a credit surcharge and a cash discount, but legally there is.
Interestingly, the answer to whether a debit transaction counts as credit or cash varies by chain, at least in South Carolina. Most chains treat debit transactions as cash, presumably because of the lower transaction fees and the lower risk due to the need for a pin as part of the transaction. However, Citgo treats debit transactions the same as credit.
Yeah. If you don’t like it, feel free to go read a game theoretical explanation of monogamy.
‘Cuz I thought it was cool, but couldn’t think of how to weave it into the conversation — that’s why!
Mistakes are more likely to persist on the stuff that is purchased less often since there are fewer trials to discover the error.
#37, #38, perhaps they can’t sell Plus at 4.999, but listing such a price increases sales of Plus at 4.349?
What are the prices today?
AIV #42 – Is there any way for the authorities to distinguish between a discount for cash and a surcharge for credit – other than the way it is described? This one always seemed a stupid law. If the credit card companies impose a cost on the garage, why should they not be allowed to pass it on to the consumer? Could be something to do with lobbying power of the big credit card issuing companies managing to pass themselves off as “helping the consumer”.
#45 – the article you linked to earlier describes this as the anchoring effect. The experiment described was asking people what percentage of African countries were members of the UN. One group was primed with the number 65. They guessed 45%. Another group was primed with the number 10, they guessed 25%.
Open up a gas station next door. Buy premium gas from this guy using cash and sell it to people using credit cards at only a $0.60 premium.
If it’s not simply a mistake, and #11 is correct that purchases of Plus tend to be for small engines, it could be that credit transactions are costlier to the station for what tend to be very small purchases (e.g. filling a 1-2 gallon portable tank). Some retailers have a flat minimum purchase amount for credit cards. #4 gives a contrary account though.
#32 – the “magic of 9” is discussed in a book called “Why Popcorn Costs So Much At the Movies and Other Pricing Puzzles” by Richard McKenzie. The upshot is that mental rules like “rounding down” / ignoring of less significant digits by time-constrained consumers to combat information overload is NOT necessarily ‘irrational’. And if buyers come to expect the 99-cent suffix, sellers may become “trapped” into following this custom (charging a round price risks selling fewer units while charging something less like 90 cents just gives away profit per unit). Also different odd endings like .99, .97 or .93 can provide an instant ‘sale type’ code to cashiers for processing returns (e.g. WMT’s “rollback savings”?) Even an interesting historical anomaly – British or other imported goods used to be considered superior in the US and due to currency conversion ended in odd numbers so US sellers started to emulate this. (I’ve also heard in the early days of cash registers forcing cashiers to make change was a way to prevent them from pocketing $.) Such customs may no longer apply but helped reinforce the ‘baking in’ of odd prices. Not sure any of it explains why $39 would be preferred to $34 however.
As the title suggests, the book also expands on the discussion of popcorn in TAE.
@48: ” “rounding down” / ignoring of less significant digits by time-constrained consumers to combat information overload is NOT necessarily ‘irrational’.” I agree that it is rational to have a strategy that minimises time taken to process information. This is similar to my suggestion for why people choose jar C – it is the application of a “shortcut” when information is complicated. This however switches the market failure from “irrational” to “uninformed”. Where we are talking about a cent here or there it doesn’t matter very much.
There are all sorts of explanations for the 99c pricing, and some may have validity. I can think of none that can explain $39 selling more than $34 other tha some psychological irrationality.
If you apply your knowledge of consumer behaviour to extract more money from me by charging $4.999 for Plus, or raising the price of a shirt from $34 to $39, or printing the prices in smaller fonts, does that mean I am irrational?
Let me ask the same question a different way – if you beat me in a game of chess, does that mean I am irrational?
#50. We cannot know for certain if someone is rational from this behaviour alone. We must know something about how they derive their utility, or their goals. If my goal is to get rid of all my money, then spending more for the same item is not irrational, in the same way as if my goal is to mangle my hand then it is not irrational to put it in the blender. However, we pretty much know from revealed preference, what people say, and from our own self knowledge, that most people do not increase their utility by spending more for the same thing.
Taking shortcuts to save time may be rational – if my time is worth $1 a minute and it will take me 2 mins to save $1 by rounding up rather than rounding down, then it is rational for me to round down. I cannot think of any such shortcut or reasonable goal that fits in with what we know about people that would rationally explain more people buying the $39 item than the $34 item.
So in short, if you sell a shirt to me only by raising its price than I am (probably) irrational. You however, are not. I don’t see the comparison with chess.
#51 — Well here’s a candidate: another well-known mental rule is using price as an indicator of quality, with people often choosing a middle option – you know, for “balance” :) (“I don’t want to waste my $ but I want something decent.”) Of course retailers play on this too, e.g. by offering high end items as a way to migrate people up from the cheapest. And of course assuming price is in fact correlated with quality may rest on a further assumption that the seller cares about reputational value (e.g. views itself as a going concern). The question with your mail-order example is whether people had enough context to view $39 as a “middle” price. Naturally hard to put too much stock in one example / study as well.
I agree there may often be a fine line separating time-constrained heuristics from irrationality, but it’s usually more enlightening to not jump straight to the latter conclusion. Also with respect to several comments on this post, to paraphrase David Friedman in “Hidden Order”: in order to get very far in economics one does typically need to assume that people have understandable goals and take reasonable measures to achieve them. Otherwise it devolves into “why did I stand on my head with a burning $10 bill between my toes? Because I *wanted* to stand on my head with a burning $10 bill between my toes”.
P.S. Where is “Chicago University”? I think you mean U of C
I did indeed mis-appropriate the work to a mythical Chicago University – it was U of C.
The consumers had only one catalog to look at, with one of the three prices. Those viewing the $39 item may have interpreted it to be of higher quality than those viewing the same item at $34, but they were wrong. I don’t see any reason why if price was the guide to quality, and high quality is desirable, that the $39 would be more attractive than $34 but less attractive than $44. By any normal use of the term, the rational choice would be to choose the lowest price among identical items – or in this case, that the lowest priced identical item would sell more. If this does not generally hold true, then we can forget supply / demand curves, and I dont think we want to do that just yet.
The heuristic you mention also leads to irrational choices, with people choosing options based on factors unrelated to the choice made – such as the price of a different item.
We are forced to use such heuristics, but that does not make the individual choice any more rational in teh economic sense, only more understandable.
“heuristics…[do] not make the individual choice any more rational in the economics sense”
Well McKenzie would disagree: “Under time constraints and mental limitations to process information…there is a “cost to thinking”. Buyers can decide, very rationally, to ignore rightward, least important digits…this also means buyers will be rationally informed about prices of what they buy — and, at the same time, rationally ignorant or rationally inattentive — and better off for being so.” He refers to a number of studies that produced mixed results. The one you cite is indeed puzzling; if not just anomalous, one additional possibility that comes to mind is that if people have in fact become accustomed to prices ending in 9, an ‘odd’ ending like 4 may seem to signal the item has been marked down, perhaps taken as a sign (again subconsciously by time-constrained consumers) that it hasn’t been selling well, perhaps taken as a sign that it is of questionable quality. Just a thought. Or we can always jump to the conclusion, in McKenzie’s words, that “perhaps consumers of practically all goods are stupid (as distinguished from rationally ignorant)”.
#54 I think it is partly about semantics really. It may be rational overall to be “rationally ignorant”, but you still end up with a non-efficient choice. If we look at a demand curve, the lower the price the higher the demand. If this fails, it doesn’t matter too much if it because of lack of information or irrational.
The problem arise if we are unable to adjust our heuristics. It may make sense to ignore small digits if prices were randomly distributed – overall the loss of efficency is a good as we can get in practice. If prices are selected to take advantage of the heuristic, then we may be in a situation where the heuristic is no longer working. If we cannot adjust, then there will be an overall loss of efficiency.
The U of C study mentioned needs a bit more looking into to see if it is really significant.
I don’t know why the difference in the discount, but the reason it costs less to pay cash is because the vendor has to pay a fee to the credit card companies for each transaction. Usually they spread this fee out to both cash and credit customers, but some prefer to reward customers who make doing business less expensive for them.
Late addition. The paper from MIT and U of C about $9 “charm prices” is here.
http://www.kellogg.northwestern.edu/faculty/anderson_e/htm/personalpage_files/Papers/Effects_of_9_Price_Endings_on_Retail_Sales.pdf
Looks pretty good to me. The figures cited in the article are a tiny part of the whole study. In all, catalogs were sent to over 100,000 people over 3 studies, and over 100 prices were changed in different ways. The $9 ending consistently increased sales and was statistically significant.