There’s so much wrong with this article. It’s clear the author knows very little about opioid use disorder or public health, but treating opioid addiction with a hicksian demand curve is flat wrong. There’s so much that distorts the price signal - in fact, opioid overdoses sometimes increases consumption in areas. the entire Hicksian framework assumes a stable utility-maximizing agent, which is exactly what addiction destabilizes. Secondly, it assumes opioid demand is elastic, which is a hell of an assumption. Drug addiction is mainly about mitigating withdrawal, and therefore price is typically way more inelastic than presumed here. At the end of the day this is just a moral hazard argument covered over in sleight of hand price theory. Just look at the empirical data that already widely disproves the potential conclusions here. It’s irresponsible to ignore that while pondering that your conclusions “narcan may actually increase opioid deaths”. For shame.
I think the assumption that demand is actually elastic, in Hendricksons example, is actually grounded well.
I this essay he measures the “tax” on opioids as the percentage chance of death per unit of consumption. This implies something about price being in relation to the percentage chance of death. Which means assuming an elastic quantity demanded isn’t as dubious as it may seem. I don’t blame you for thinking this though, it is confusing as we are taught drugs to be inelastic in every intro micro course. This is just a poor example.
I think everything else you said is well put and probably true!
This is another interesting piece--you guys are doing a great service. I think it's related to the "Peltzman Effect" of offsetting behavior, no? Seatbelts, anti-lock brakes, etc. allow/encourage people to drive less carefully? My pal Bob Chirinko introduced this to me (me being a poor benighted macroeconomist/international type) via an old NYT op-ed: "As Cars Get Safer, Drivers Take Risks," By Robert S. Chirinko, April 10, 1994. Your expenditure function analysis helps flesh out the intuition. I had a vague notion that the wide availability of Narcan had been credited with the sharp drop in opioid deaths from 2023-2024. I suspect--based on my very limited knowledge of addictive behavior-- that this might be an important feature for drug policy.
Incredible post sir! I’m just finishing up my first year Econ PhD, currently preparing for core exams. This post reminds me of a bunch of Micro I topics.
Two questions,
1) How much weight do you think the horizontal supply curve is carrying in this application to narcan?
2) If we are taking the probability of an overdoses death per unit of consumption as the implicit tax rate on opioids, what does this do to the interpretation of the elasticity of demand for opioids? This seems crucial for the interpretation of the “cost of the tax”. Are we interpreting the price of consuming opioids as the probability of death as well? This is confusing, because drugs are often used in early micro courses to describe goods of high INelasticity; If I’m addicted enough to a drug -> I’ll still buy it when the price increases. But, (I think) in your example, if you imply prices as the percentage chance of death, of course this should be much more elastic. I hope this makes sense.
The horizontal demand curve is doing some work here because it allows us to think about the entire cost being on the consumer. As the supply curve becomes more elastic, the burden of the tax shifts toward the producers. The underlying logic remains the same, but the total cost of the "tax" to the consumer is smaller.
I think that if you think of the price as being the opportunity cost of using opioids, which is inclusive of the probability of death, then yes you would have a more elastic demand curve.
In a subsequent post, I referenced estimates that we have of the price elasticity of demand for heroin. The Marshallian elasticity is around -0.8. The estimated income elasticity is 0.11. By the Slutsky equation in elasticity form, this implies that the Hicksian elasticity (what we need for the analysis) is
e = -0.8 + 0.11s
where s is the share of income that is spent on heroin. If we assume that s is small, then it will save lives, but the effect on lives saved will be small. However, the bigger than s becomes, the more lives that would be saved.
Great post. Sometimes the post stands on its own merits and sometimes it stands on the merits of the comments, which often demonstrate inelasticity of ideology.
Starting an examination of data with a fixed conclusion (or outcome) is always bad science. Shame on some of your readers!
There’s so much wrong with this article. It’s clear the author knows very little about opioid use disorder or public health, but treating opioid addiction with a hicksian demand curve is flat wrong. There’s so much that distorts the price signal - in fact, opioid overdoses sometimes increases consumption in areas. the entire Hicksian framework assumes a stable utility-maximizing agent, which is exactly what addiction destabilizes. Secondly, it assumes opioid demand is elastic, which is a hell of an assumption. Drug addiction is mainly about mitigating withdrawal, and therefore price is typically way more inelastic than presumed here. At the end of the day this is just a moral hazard argument covered over in sleight of hand price theory. Just look at the empirical data that already widely disproves the potential conclusions here. It’s irresponsible to ignore that while pondering that your conclusions “narcan may actually increase opioid deaths”. For shame.
I think the assumption that demand is actually elastic, in Hendricksons example, is actually grounded well.
I this essay he measures the “tax” on opioids as the percentage chance of death per unit of consumption. This implies something about price being in relation to the percentage chance of death. Which means assuming an elastic quantity demanded isn’t as dubious as it may seem. I don’t blame you for thinking this though, it is confusing as we are taught drugs to be inelastic in every intro micro course. This is just a poor example.
I think everything else you said is well put and probably true!
This is another interesting piece--you guys are doing a great service. I think it's related to the "Peltzman Effect" of offsetting behavior, no? Seatbelts, anti-lock brakes, etc. allow/encourage people to drive less carefully? My pal Bob Chirinko introduced this to me (me being a poor benighted macroeconomist/international type) via an old NYT op-ed: "As Cars Get Safer, Drivers Take Risks," By Robert S. Chirinko, April 10, 1994. Your expenditure function analysis helps flesh out the intuition. I had a vague notion that the wide availability of Narcan had been credited with the sharp drop in opioid deaths from 2023-2024. I suspect--based on my very limited knowledge of addictive behavior-- that this might be an important feature for drug policy.
Incredible post sir! I’m just finishing up my first year Econ PhD, currently preparing for core exams. This post reminds me of a bunch of Micro I topics.
Two questions,
1) How much weight do you think the horizontal supply curve is carrying in this application to narcan?
2) If we are taking the probability of an overdoses death per unit of consumption as the implicit tax rate on opioids, what does this do to the interpretation of the elasticity of demand for opioids? This seems crucial for the interpretation of the “cost of the tax”. Are we interpreting the price of consuming opioids as the probability of death as well? This is confusing, because drugs are often used in early micro courses to describe goods of high INelasticity; If I’m addicted enough to a drug -> I’ll still buy it when the price increases. But, (I think) in your example, if you imply prices as the percentage chance of death, of course this should be much more elastic. I hope this makes sense.
The horizontal demand curve is doing some work here because it allows us to think about the entire cost being on the consumer. As the supply curve becomes more elastic, the burden of the tax shifts toward the producers. The underlying logic remains the same, but the total cost of the "tax" to the consumer is smaller.
I think that if you think of the price as being the opportunity cost of using opioids, which is inclusive of the probability of death, then yes you would have a more elastic demand curve.
In a subsequent post, I referenced estimates that we have of the price elasticity of demand for heroin. The Marshallian elasticity is around -0.8. The estimated income elasticity is 0.11. By the Slutsky equation in elasticity form, this implies that the Hicksian elasticity (what we need for the analysis) is
e = -0.8 + 0.11s
where s is the share of income that is spent on heroin. If we assume that s is small, then it will save lives, but the effect on lives saved will be small. However, the bigger than s becomes, the more lives that would be saved.
Here is the link to that paper, by the way: https://www.sciencedirect.com/science/article/pii/S0167629615000090
Josh,
Great post. Sometimes the post stands on its own merits and sometimes it stands on the merits of the comments, which often demonstrate inelasticity of ideology.
Starting an examination of data with a fixed conclusion (or outcome) is always bad science. Shame on some of your readers!