Interesting post (I love Steven Levitt) but there's a great lesson for developers here: Efficiency is only one consideration. What about usability? Do you really want to wait in line at McDonald's or Starbucks while someone (maybe a high school dropout) counts on his fingers to make change in 71 & 37 cent coins? The overhead here (an extra coin in your pocket) is well worth the added convenience. Aggregated over the whole economy, those extra seconds would literally be worth billions of dollars a year..
How about a lesson for lateral thinking? I'd still prefer to do without the coins and go for a all-virtual, all-plastic system. Efficiency and Usability.
Dropping hard currency raises the bar to entry and kills a lot of low end business. No more street vendors, cash only local business, friendly card games or bets, informal services, or convenient tips. It would also give the plastic channel immense power over businesses. You might not use cash very often, but dropping it altogether would not be a good idea.
Additionally, I like to pay with cash when I don't want my purchasing habits to be collected by the store or my creditor. I find it disturbing how our modern economy makes the task of collecting information on one's interests and identity so effortless. I wouldn't want it to become all the easier.
The plastic equivalent of cash would not necessarily have your personal information attached to it. Think of it like a prepaid phone. You could go to a Post Office and buy (using your credit card or check) a card with a predetermined value.
A gift card also lets the store track your purchases (unless of course you give it away).
The anonymity advantage of cash is that it cycles through the register in a non-deterministic manner. If I buy a $20 bill (serial number 123) from an ATM, and it later winds up at PotMart, that is not compelling evidence I went to PotMart. If I bought a prepaid cash card and it was spent at PotMart, that indicates with decent probability that I purchased some weed.
It would be easy enough to identify regular customers when they in their home town. These days it isn't all that difficult to pair someone's spending habits with their location an do a bit of online research to find out who they are.
If one was careful about it they could probably stay relatively anonymous though.
I have serious doubt about "killing a lot of low end business". I'd imagine that we'd start seeing lots of alternate, local currencies. Card games and bets would probably have other medium of exchange (poker chips) or in-kind value (a 12-pack of beer).
Probably. Hell, we could all just start using Euros (or whatever) for cash, and then exchange our Euros for plastic-only US dollars at changing booths (which would suddenly become a lot more common).
This, of course, doesn't make life any easier for anyone, except the jerks who already use cards for everything (aka the jerks I always get stuck behind in line while I wait for their little receipt to print and get signed).
The overhead for taking credit cards in a small business is huge. A small bookstore owner I knew said the card processing fees were more per month than the rent or utilities.
For that matter why not do away with the card as well and use some sort of biometric technology to tie into your bank account?
I think that this highlights an interesting parody. To get increased efficiency you need to buy more expensive equipment. A cash register full of bills and coins doesn't cost much and after the initial investment you are pretty much done except for the expense of receipt paper.
However, credit cards require more complex card readers, systems to tie the card into central servers, etc.
In turn biometrics would require even more: pupil scanners, etc.
So the drive for efficiency and usability would appear to lead to even more complex devices, rather than the simple usability of handing a metal disk from one person to another person.
Body-tagging as a means of identification for payment is an old idea, almost 2000 years old at least, perhaps longer ...
From the New Testament, Revelations, chapter 13: "Then I saw another beast, coming out of the earth ... He exercised all the authority of the first beast on his behalf ... He also forced everyone, small and great, rich and poor, free and slave, to receive a mark on his right hand or on his forehead, so that no one could buy or sell unless he had the mark, which is the name of the beast or the number of his name."
Equipment and I.T. systems to enforce biometric/body-tagging systems are costly, but then coins are costly too, and are subject to counterfeiting.
The ultimate rationale for a plasticless, cashless system though will be less in transaction costs than in control and elimination of counterfeit currency, money-laundering, payment in currency for corporate and political corruption. In such a system all money flows are visible to someone at some level, and corrupt transactions will become for-barter only. Of course, the central overseeing authority will become corrupt, which the Revelations passage hints at.
Of the three, what you know is by far the most secure (in theory). What you are is the least secure.
What you know can never be stolen from you, at least not without you being aware of it. What you have is hard to steal, but not impossible. What you are is trivial to steal.
Biometrics is the least secure method of verifying someone. It is trivial to clone, and even worse you can clone it at a distance without the person even realizing it was cloned.
So no, biometric tech is not a good way to paying someone.
However, what you know is subject to people forgetting, and it's hard to have a very long pass phrase.
What you have wins easily. And the real world bears this out - every security system everywhere uses it, credit cards, keys, drivers licenses.
Computers use what you know only because it's hard to do 'what you have' remotely.
Where do you want the complexity -- one time, upfront, or during every single transaction?
The existence of Hacker News requires a whole lot of complicated technology. Having this conversation via smoke signal requires a much smaller investment. And yet...
Clearly without a study it would be impossible to determine whether the upfront complexity of designing and installing specialized credit card equipment would be more or less than the comparatively small complexity of counting out coins.
Walmart, for instance, now uses a system that dispenses the coin change automatically so that the person at the cash register doesn't even have to do it themselves.
Just speaking rhetorically.. don't mean to generalize. I'm sure Bill Gates would have dropped out sooner (in high school, instead of college) if opportunity came knocking
sooner.. You're obviously an exception & probably dropped out for good reasons :)
Obama economic advisor Austan Goolsbee once recommended eliminating the 1-cent penny -- by having the government declare pennies by fiat to be the new 5-cent nickel:
It's quite a neat idea, getting rid of the current seignorage loss and inefficiency of 1-cent pennies, without incurring the wrath of the penny materials lobby -- or even worse, voters with giant penny jars.
Given the reasoning here, perhaps pennies should become 3-cent pieces.
Rebasing small coins to be worth more isn't quite dropping cash from helicopters, but it should still have a mildly stimulative and redistributive effect. Something for everyone!
I love this idea. Not so much because of its efficiency, but because it'll make so many people say "Wait, you can do that?". Too many people haven't internalized what having a fiat currency really means. Having the government flex its muscle in this manner would be a fantastic civics lesson.
I'm not sure it would really succeed. For a long time after the change many people would still refuse to accept a penny as 5 cents. Vending machines, toll booths and the like would all also have to be reworked.
There are many different (and perhaps more practical) criteria than average coins per transaction. For instance, I don't like a lot of coins in my pocket at once. And, I always want to get rid of coins when I use them! So, what are the least number of coins I can carry to guarantee that I come back with at most the number of coins I started with (after change is given)?
On the other hand, sometimes I only carry coins to get rid of them! With the current system, I need to carry nine coins to guarantee exact change -- what system maximally reduces this?
“Do you know what this country needs today? A seven-cent nickel. Yessiree, we've been using the five-cent nickel in this country since 1492. Now, why not give the seven-cent nickel a chance? If that works out, next year we could have an eight-cent nickel. Think what that would mean. You could go to a newsstand, buy a three-cent newspaper and get the same nickel back again. One nickel carefully used would last a family a lifetime!”
That's all fine and dandy in theoretical world, but I'd dispute that the probability of a transaction resulting in value v is uniform. I'd guess that there's some fairly prominent banding due to psychogical pricing at or just under 'round' figures.
That's a perculiarly American problem - everybody else factors tax into the sticker price.
It's only America that seems to add it at the register. I assume it's to take advantage the famously high level of math education among Americans who can easily add 4.5% state and 1.8% city sales tax to a coffee in their head.
Benford's law, also called the first-digit law, states that in lists of numbers from many (but not all) real-life sources of data, the leading digit is distributed in a specific, non-uniform way.
Benford's law is about the first digit of the number, it has nothing to do with the current problem. The pennies are the two last digits.
However, IMO the assumption you quoted is indeed false but due to other reason:
In real life, prices are not uniformly distributed because usually they are rounded to multiple of 5 cents or commonly they are .99 or .89 or s.th. like that.
I am wondering what happens if we put non-uniform prior on the probability of the pennies.
Another flaw in this research is forgeting that we often are getting change back. So maybe problem is to find combination not only to give exactly money, but giving and receiving change back. For example:
Using 1,5,10,20,50 you can make exact 4c only by 1,1,1,1, (4 coins). But getting change you can do 5, -1 (2 coins).
while i agree with the premise that the prices aren't uniformly distributed, i think its less of an issue than people think.
there are undoubtedly uneven humps that favor certain areas. but i think that the actual result of real life retail transactions is a more level curve than you would expect due to things like multiple, varied item purchases and the possible inclusion of various types of sales tax.
Given the number of retailers that end prices in 9 or 5, one could probably calculate a better model of efficiency based on real world distributions. The secondary problem would be that tax is applied after the nominal price, and local taxes vary considerably.
This does not apply here since we are looking at the last digits, and not the first, so different rules apply. The bigger your price is, the more uniform the cents will be (aside from psychological factors like: 00, 95, 99 being very common)
On the subject of coin usability, I'd say one requirement is that any coin denomination should divide cleanly into 100. Surprising that they left that out.
I completely agree. What do you do when you have 3 37-cent coins?
Theoretically, it wouldn't change much for the end-user if s/he was always trying to use the coins they already have (which probably doesn't happen that much). But for banks, stores and everything else, it would be a mess, because there's no easy way to put the coins in rolls of "easy" values of a round number of dollars.
An alternative definition of efficiency is total number of coins you need to carry in your pocket to cover a single transaction of 1..99 cents. Interestingly all the good 4-coin solutions in the article, including (1, 3, 11, 37) as well as our current (1, 5, 10, 25), require a total of 10 coins.
There are 12 other 4-coin solutions, (1, 3, 9, 25) being the most reasonable looking, that only require a total of 9 coins.
According to Wikipedia (I haven't bothered to check the math), a ternary currency system would maximize the likelihood that a customer would get exact change.
How about a coin that is worth 1/3000th of a US dollar? That's what we have in Russia :) Yes, it is as useless as it sounds.
By the way, the real price distributions are far from uniform. You mostly encounter .99, .49, etc.
Also, in Russia some retail chains do not deal with pennies - the prices are mostly with pennies, but at checkout any pennies in the total sum are chopped off automatically.
Ignoring the extreme inconvenience, this might be an interesting way to tackle our nation's (U.S.) problem of students with poor performance in mathematics.
