13 ms·

Alchian–Allen effect

122 points by henryaj 8y ago

dfee 8y ago
> when the prices of two substitute goods, such as high and low grades of the same product, are both increased by a fixed per-unit amount such as a transportation cost or a lump-sum tax, consumption will shift toward the higher-grade product. This is true because the added per-unit amount decreases the relative price of the higher-grade product. > Another example is that Australians drink higher-quality Californian wine than Californians, and vice versa, because it is only worth the transportation costs for the most expensive wine.
- ashrk 8y ago
  My unscientific observation has been that the opposite holds for French wine. They seem to ship the US their worst, but it hits our shelves marked up 4x what it'd be in France. Maybe it's a branding thing, and French wine's got enough power in that department that they don't need to send us good stuff to move product, at least at the lower end. Maybe with some French writing on the label low-price-shopping consumers are equally likely to buy it whether or not it tastes good, so there's no reason to send anything but the worst they have. Other European countries seem to have no trouble shipping us nice wine in the lower price tiers, but then they're not France.
  - zwaps 8y ago
    In reality there's a lot more in a product than just good and bad. Especially with wine, where it is not actually so clear that a true quality ordering exists, all sorts of other effects are at play. For example, we know how the signaling value of price overrides some true quality variables. In economics, its important to know these simple models and their outcomes, but it is equally important to know when other effects are at play, or when assumptions do not hold
andyjohnson0 8y ago
I don't see anything in the article about the extent to which this effect has been observed in the real world. Is it another example of economists assuming that people are idealised rational actors (who would prefer a 1.6x coffee bean premium to a 2x, to take the example in the article), or is there some hard evidence that it explains the real behaviour of real people?
- dyeje 8y ago
  They give 3 examples at the end: "The effect has been studied as it applies to illegal drugs and it has been shown that the potency of marijuana increased in response to higher enforcement budgets,[2] and there was a similar effect for alcohol in the U.S. during Prohibition.[3] This effect is called Iron law of prohibition or Cardinal rule of prohibition. Another example is that Australians drink higher-quality Californian wine than Californians, and vice versa, because it is only worth the transportation costs for the most expensive wine.[4]"
  - andyjohnson0 8y ago
    Thanks. I did read the article and noticed the examples that you referred to. To clarify: I just don't see where the evidence for any of these claims is accessible. Four of the six references don't appear to be available on-line. Of the two that are, one is a brief summary that leads to a paywalled FT article that I can't read, and the other is a ~200 pdf economics textbook. I'm happy to accept that a price differential changes when a constant is added to both prices. But I do question whether people in the real world will switch to a different type of coffee bean because the differential goes from x2 to x1.6. And I wanted to know more about the real-world circumstances in which it had been observed. I might be pursuaded that the market for illegal drugs/alcohol could exhibit this effect, but it would be interesting to know how how, given that illegal products are probably going to be subject to other effects not shared by legal products, biases and selectivities were compensated for. Edit: Made first paragraph clearer.
    - zwaps 8y ago
      Here is a study in trade on this phenomenon https://krannert.purdue.edu/faculty/hummelsd/research/apples.pdf https://krannert.purdue.edu/faculty/hummelsd/research/apples...
pwaivers 8y ago
> The effect has been studied as it applies to illegal drugs and it has been shown that the potency of marijuana increased in response to higher enforcement budgets... If the price of all marijuana goes up, then I bet overall usage of it goes down even if the individual potency goes up.
- cbsmith 8y ago
  Depends... if quality goes up, sometimes that grows the demand for a product. Kraft Parmesan convinced me I didn't like Parmesan for the longest time...
- trhway 8y ago
  one of the reasons in play for the fentanyl and carfentanyl trafficking observed these days. >If the price of all marijuana goes up, then I bet overall usage of it goes down in high margin (and supply limited) goods i think it works in the opposite direction - the price goes up in response to the demand going up.
cbsmith 8y ago
I didn't realize they had a name for it... it seems so obvious.
- azeirah 8y ago
  Why shouldn't obvious things have names?
  - pc86 8y ago
    They weren't saying obvious things shouldn't have names. They were saying they didn't realize it had a name.
  - cbsmith 8y ago
    Usually if it is an obvious thing, you don't have a name tied to the people who discovered it... you know, because it predates recorded history.
- jldugger 8y ago
  > it seems so obvious. I'm not sure the effect is logically sound -- before the price hikes you weren't willing to pay a dollar more for name brand (or premium drugs etc), but after the price hike you are? I'm not sure if the authors have a mathematical model supporting the decision, or are just observing illegal drug markets and generalizing.
  - TezlaOil 8y ago
    > before the price hikes you weren't willing to pay a dollar more for name brand (or premium drugs etc), but after the price hike you are? Imagine that 10 normal people and Bill Gates want to buy bottles of wine. Bill goes for a bottle of $6000 Richebourg Grand Cru, and the 10 normal people always go for Shitscreek Red, sold for $4. So about 9% of the buyers are willing to pay extra for the name brand. After the Great Wine Tax Bill of 2021, the prices are both increased by a fixed per-bottle amount of $6000. So now the Richebourg costs $12000, and the Shitscreek costs $6004. The 10 normal people migrate to beer, and Bill Gates still buys Richebourg. So about 100% of the buyers are willing to pay extra for the name brand.
tlb 8y ago
The reverse effect works too. When an overhead cost drops, people consume crappier versions. Free online news is a striking example. If you walked down to the news stand and paid $0.75 for a physical newspaper and the content was all Clickhole listicles you'd never buy it again. But people consume plenty for free.
- srtjstjsj 8y ago
  That's not the reverse effect, that's supply and demand, and people forgetting opportunity cost of time. Newsstands sell plenty of junk like US Weekly.
  - pc86 8y ago
    It is literally the opposite of the effect described in the article.
    - gpvos 8y ago
      That being said, both the Alchian-Allen effect and its opposite are a fairly direct consequence of basic supply and demand effects.
- mikeash 8y ago
  There was and still is plenty of Clickhole-level crap on newsstands next to the newspapers. People buy it.
  - stonecraftwolf 8y ago
    In fact IIRC people buy printbait considerably more than they buy actual newspapers.
gnulinux 8y ago
> the Alchian–Allen theorem Uhm, I'm not an economist but how is this a "theorem"? What is the theory, what are the axioms, where is the proof? This seems like an empirically observed trend.
- shitgoose 8y ago
  there is some basic arithmetic in the article. imagine the cost was raised by $100, then prices of beans would be $101.5 and $103, i.e. pretty much everyone would go for higher quality. gradually decrease the premium, and you will gradually decrease the number of people who prefer the higher grade product in favor of lower price. this is pretty much it. quite simple and neat observation, actually. "the harder the punishment, the harder the drug" is another very important non-trivial observation, missed by most people.
  - gnulinux 8y ago
    > quite simple and neat observation, actually. Did you read my comment? That's exactly what I wrote. This is not a "theorem". Theorems live inside theories, they're logically implied by axioms of the theories. That basic arithmetic proves nothing of the sort.
    - shitgoose 8y ago
      "Theorems live inside theories" - sorry, didn't get that. axioms: * given same price, consumer goes for higher grade product. * as relative price difference decreases, some consumers start going for higher grade product. * rules of arithmetic. in any case, it seems that we have appreciation for different things. you prefer detailed verbosity, i prefer a simple idea that can be easily understood. great and profound ideas are not necessarily complex, many of them are quite simple. let's keep them that way.
    - zwaps 8y ago
      Then it is a theorem, as it is tautologically implied by certain axioms imposed on preference relations or choice correspondences, such as completeness, transitivity and a notion of continuity. Since the whole theory won't fit in one wikipedia article, I suggest you read the first chapters or either a) Mas-Collel, Whinston and Green or b) Kreps, Microeconomic Foundations I (not the old undergrad book from him) This would provide you with the theory to understand this result in a more mathematical correct way.
tr352 8y ago
Similar to an even exchange in chess. If you’re behind it’s a good deal, if you’re ahead it’s bad. Of course evaluating value taking into account positions (and thus determining whether an exchange is really “even”) is not straightforward.
- daxat_staglatz 8y ago
  This seems wrong: an even exchange does not change absolute balance but magnifies relative balance, which you do not want if you're behind. Let's say your "strength" is simply the sum of the value of your pieces. The losing side has a strength of l and the winning side has a strength of L. The losing side loses by l-L in absolute terms, or (l-L)/(l+L) in relative terms. An even exchange of value k makes it go to (l-k)-(L-k) = l-k in absolute terms (no change); and to (l-L)/(l+L-2k) in relative terms. (l-L)/(l+L-2k) < (l-L)/(l+L). To take an example, if the situation is white : two pawns and black: one pawn, going to white: one pawn and black: nothing is a bad deal for black (the losing side)!
  - tr352 8y ago
    You're right, I meant the exact opposite: an even exchange is good if you're ahead, bad when you're behind.
- murftown 8y ago
  I don't consider myself a good chess player at all, but I always thought the opposite way: When I'm ahead, I'm willing to whittle down both sides with even trades until they are left with no options and can be checkmated. When I'm behind, an even trade feels scary - I'm looking for a way to catch up. Am I missing some reason why it would be the opposite?
  - tr352 8y ago
    Yes you're right, I meant the exact opposite.
robrenaud 8y ago
> Another example is that Australians drink higher-quality Californian wine than Californians, and vice versa, because it is only worth the transportation costs for the most expensive wine. As a person who loves Belgian beer, I came to this realization when I was on a bit of a beercation there. For the most part, the good Belgian beer makes it here. When I'd try beer from breweries I hadn't heard of while I was there, it was mostly worse than the Belgian beer I had tried at home.
anonytrary 8y ago
I thought this was just common sense. Suppose you're buying necklaces. Today: Gold $10; Silver $0.10 Tomorrow: Gold $1010; Silver $1000.10 By adding a constant to both prices, the price ratio no longer remains aligned with the underlying value ratio. Assuming you have to get one necklace each day, there is less of a decision tomorrow than there is today, because silver is overpriced relative to gold. The correct prices for tomorrow should have been "Gold: $1010, Silver: $10.10", so you're effectively getting more scammed by going with silver.
- jetrink 8y ago
  I'd agree that once presented with as stark an example as yours, it is common sense, but it takes some insight to create such a thought experiment in the first place.
lurquer 8y ago
So that's why I go broke on vacations... I want to eat somewhere cheap and the wife says, "No, let's go to the expensive joint... we've already spent $x getting here, so an extra $50 won't make a difference!" Now I have a name for the phenomena.
- peterkshultz 8y ago
  I would call what you're experiencing the sunk cost fallacy [1]. [1]: https://en.wikipedia.org/wiki/Sunk_cost#Loss_aversion_and_the_sunk_cost_fallacy https://en.wikipedia.org/wiki/Sunk_cost#Loss_aversion_and_th...
  - lurquer 8y ago
    No. The sunk cost comes into play in explaining why we've gone to Disney World three time too many simply because we bought some multi-day package years ago. I hate Disney. She hates Dianey. But, we gotta use up the stinking tickets.

122 points by henryaj 8y ago

dfee 8y ago

> when the prices of two substitute goods, such as high and low grades of the same product, are both increased by a fixed per-unit amount such as a transportation cost or a lump-sum tax, consumption will shift toward the higher-grade product. This is true because the added per-unit amount decreases the relative price of the higher-grade product. > Another example is that Australians drink higher-quality Californian wine than Californians, and vice versa, because it is only worth the transportation costs for the most expensive wine.

ashrk 8y ago
My unscientific observation has been that the opposite holds for French wine. They seem to ship the US their worst, but it hits our shelves marked up 4x what it'd be in France. Maybe it's a branding thing, and French wine's got enough power in that department that they don't need to send us good stuff to move product, at least at the lower end. Maybe with some French writing on the label low-price-shopping consumers are equally likely to buy it whether or not it tastes good, so there's no reason to send anything but the worst they have. Other European countries seem to have no trouble shipping us nice wine in the lower price tiers, but then they're not France.
- zwaps 8y ago
  In reality there's a lot more in a product than just good and bad. Especially with wine, where it is not actually so clear that a true quality ordering exists, all sorts of other effects are at play. For example, we know how the signaling value of price overrides some true quality variables. In economics, its important to know these simple models and their outcomes, but it is equally important to know when other effects are at play, or when assumptions do not hold

andyjohnson0 8y ago

I don't see anything in the article about the extent to which this effect has been observed in the real world. Is it another example of economists assuming that people are idealised rational actors (who would prefer a 1.6x coffee bean premium to a 2x, to take the example in the article), or is there some hard evidence that it explains the real behaviour of real people?

dyeje 8y ago
They give 3 examples at the end: "The effect has been studied as it applies to illegal drugs and it has been shown that the potency of marijuana increased in response to higher enforcement budgets,[2] and there was a similar effect for alcohol in the U.S. during Prohibition.[3] This effect is called Iron law of prohibition or Cardinal rule of prohibition. Another example is that Australians drink higher-quality Californian wine than Californians, and vice versa, because it is only worth the transportation costs for the most expensive wine.[4]"
- andyjohnson0 8y ago
  Thanks. I did read the article and noticed the examples that you referred to. To clarify: I just don't see where the evidence for any of these claims is accessible. Four of the six references don't appear to be available on-line. Of the two that are, one is a brief summary that leads to a paywalled FT article that I can't read, and the other is a ~200 pdf economics textbook. I'm happy to accept that a price differential changes when a constant is added to both prices. But I do question whether people in the real world will switch to a different type of coffee bean because the differential goes from x2 to x1.6. And I wanted to know more about the real-world circumstances in which it had been observed. I might be pursuaded that the market for illegal drugs/alcohol could exhibit this effect, but it would be interesting to know how how, given that illegal products are probably going to be subject to other effects not shared by legal products, biases and selectivities were compensated for. Edit: Made first paragraph clearer.
  - zwaps 8y ago
    Here is a study in trade on this phenomenon https://krannert.purdue.edu/faculty/hummelsd/research/apples.pdf https://krannert.purdue.edu/faculty/hummelsd/research/apples...

pwaivers 8y ago

> The effect has been studied as it applies to illegal drugs and it has been shown that the potency of marijuana increased in response to higher enforcement budgets... If the price of all marijuana goes up, then I bet overall usage of it goes down even if the individual potency goes up.

cbsmith 8y ago
Depends... if quality goes up, sometimes that grows the demand for a product. Kraft Parmesan convinced me I didn't like Parmesan for the longest time...
trhway 8y ago
one of the reasons in play for the fentanyl and carfentanyl trafficking observed these days. >If the price of all marijuana goes up, then I bet overall usage of it goes down in high margin (and supply limited) goods i think it works in the opposite direction - the price goes up in response to the demand going up.

cbsmith 8y ago

I didn't realize they had a name for it... it seems so obvious.

azeirah 8y ago
Why shouldn't obvious things have names?
- pc86 8y ago
  They weren't saying obvious things shouldn't have names. They were saying they didn't realize it had a name.
- cbsmith 8y ago
  Usually if it is an obvious thing, you don't have a name tied to the people who discovered it... you know, because it predates recorded history.
jldugger 8y ago
> it seems so obvious. I'm not sure the effect is logically sound -- before the price hikes you weren't willing to pay a dollar more for name brand (or premium drugs etc), but after the price hike you are? I'm not sure if the authors have a mathematical model supporting the decision, or are just observing illegal drug markets and generalizing.
- TezlaOil 8y ago
  > before the price hikes you weren't willing to pay a dollar more for name brand (or premium drugs etc), but after the price hike you are? Imagine that 10 normal people and Bill Gates want to buy bottles of wine. Bill goes for a bottle of $6000 Richebourg Grand Cru, and the 10 normal people always go for Shitscreek Red, sold for $4. So about 9% of the buyers are willing to pay extra for the name brand. After the Great Wine Tax Bill of 2021, the prices are both increased by a fixed per-bottle amount of $6000. So now the Richebourg costs $12000, and the Shitscreek costs $6004. The 10 normal people migrate to beer, and Bill Gates still buys Richebourg. So about 100% of the buyers are willing to pay extra for the name brand.

tlb 8y ago

The reverse effect works too. When an overhead cost drops, people consume crappier versions. Free online news is a striking example. If you walked down to the news stand and paid $0.75 for a physical newspaper and the content was all Clickhole listicles you'd never buy it again. But people consume plenty for free.

srtjstjsj 8y ago
That's not the reverse effect, that's supply and demand, and people forgetting opportunity cost of time. Newsstands sell plenty of junk like US Weekly.
- pc86 8y ago
  It is literally the opposite of the effect described in the article.
  - gpvos 8y ago
    That being said, both the Alchian-Allen effect and its opposite are a fairly direct consequence of basic supply and demand effects.
mikeash 8y ago
There was and still is plenty of Clickhole-level crap on newsstands next to the newspapers. People buy it.
- stonecraftwolf 8y ago
  In fact IIRC people buy printbait considerably more than they buy actual newspapers.

gnulinux 8y ago

> the Alchian–Allen theorem Uhm, I'm not an economist but how is this a "theorem"? What is the theory, what are the axioms, where is the proof? This seems like an empirically observed trend.

shitgoose 8y ago
there is some basic arithmetic in the article. imagine the cost was raised by $100, then prices of beans would be $101.5 and $103, i.e. pretty much everyone would go for higher quality. gradually decrease the premium, and you will gradually decrease the number of people who prefer the higher grade product in favor of lower price. this is pretty much it. quite simple and neat observation, actually. "the harder the punishment, the harder the drug" is another very important non-trivial observation, missed by most people.
- gnulinux 8y ago
  > quite simple and neat observation, actually. Did you read my comment? That's exactly what I wrote. This is not a "theorem". Theorems live inside theories, they're logically implied by axioms of the theories. That basic arithmetic proves nothing of the sort.
  - shitgoose 8y ago
    "Theorems live inside theories" - sorry, didn't get that. axioms: * given same price, consumer goes for higher grade product. * as relative price difference decreases, some consumers start going for higher grade product. * rules of arithmetic. in any case, it seems that we have appreciation for different things. you prefer detailed verbosity, i prefer a simple idea that can be easily understood. great and profound ideas are not necessarily complex, many of them are quite simple. let's keep them that way.
  - zwaps 8y ago
    Then it is a theorem, as it is tautologically implied by certain axioms imposed on preference relations or choice correspondences, such as completeness, transitivity and a notion of continuity. Since the whole theory won't fit in one wikipedia article, I suggest you read the first chapters or either a) Mas-Collel, Whinston and Green or b) Kreps, Microeconomic Foundations I (not the old undergrad book from him) This would provide you with the theory to understand this result in a more mathematical correct way.

tr352 8y ago

Similar to an even exchange in chess. If you’re behind it’s a good deal, if you’re ahead it’s bad. Of course evaluating value taking into account positions (and thus determining whether an exchange is really “even”) is not straightforward.

daxat_staglatz 8y ago
This seems wrong: an even exchange does not change absolute balance but magnifies relative balance, which you do not want if you're behind. Let's say your "strength" is simply the sum of the value of your pieces. The losing side has a strength of l and the winning side has a strength of L. The losing side loses by l-L in absolute terms, or (l-L)/(l+L) in relative terms. An even exchange of value k makes it go to (l-k)-(L-k) = l-k in absolute terms (no change); and to (l-L)/(l+L-2k) in relative terms. (l-L)/(l+L-2k) < (l-L)/(l+L). To take an example, if the situation is white : two pawns and black: one pawn, going to white: one pawn and black: nothing is a bad deal for black (the losing side)!
- tr352 8y ago
  You're right, I meant the exact opposite: an even exchange is good if you're ahead, bad when you're behind.
murftown 8y ago
I don't consider myself a good chess player at all, but I always thought the opposite way: When I'm ahead, I'm willing to whittle down both sides with even trades until they are left with no options and can be checkmated. When I'm behind, an even trade feels scary - I'm looking for a way to catch up. Am I missing some reason why it would be the opposite?
- tr352 8y ago
  Yes you're right, I meant the exact opposite.

robrenaud 8y ago

> Another example is that Australians drink higher-quality Californian wine than Californians, and vice versa, because it is only worth the transportation costs for the most expensive wine. As a person who loves Belgian beer, I came to this realization when I was on a bit of a beercation there. For the most part, the good Belgian beer makes it here. When I'd try beer from breweries I hadn't heard of while I was there, it was mostly worse than the Belgian beer I had tried at home.

anonytrary 8y ago

I thought this was just common sense. Suppose you're buying necklaces. Today: Gold $10; Silver $0.10 Tomorrow: Gold $1010; Silver $1000.10 By adding a constant to both prices, the price ratio no longer remains aligned with the underlying value ratio. Assuming you have to get one necklace each day, there is less of a decision tomorrow than there is today, because silver is overpriced relative to gold. The correct prices for tomorrow should have been "Gold: $1010, Silver: $10.10", so you're effectively getting more scammed by going with silver.

jetrink 8y ago
I'd agree that once presented with as stark an example as yours, it is common sense, but it takes some insight to create such a thought experiment in the first place.

lurquer 8y ago

So that's why I go broke on vacations... I want to eat somewhere cheap and the wife says, "No, let's go to the expensive joint... we've already spent $x getting here, so an extra $50 won't make a difference!" Now I have a name for the phenomena.

peterkshultz 8y ago
I would call what you're experiencing the sunk cost fallacy [1]. [1]: https://en.wikipedia.org/wiki/Sunk_cost#Loss_aversion_and_the_sunk_cost_fallacy https://en.wikipedia.org/wiki/Sunk_cost#Loss_aversion_and_th...
- lurquer 8y ago
  No. The sunk cost comes into play in explaining why we've gone to Disney World three time too many simply because we bought some multi-day package years ago. I hate Disney. She hates Dianey. But, we gotta use up the stinking tickets.