It was recently brought to my attention that Eliezer Yudkowsky regards the monetary theories of Scott Sumner (short overview) to be (what we might call) a “correct contrarian cluster”, or an island of sanity when most experts (though apparently a decreasing number) believe the opposite.
I would be interested in knowing why. To me, Sumner’s views are a combination of:
a) Goodhart’s folly (“Historically, an economic metric [that nobody cared about until he started talking about] has been correlated with economic goodness; if we only targeted this metric with policy, we would get that goodness. Here are some plausible mechanisms why …”—my paraphrase, of course)
b) Belief that “hoarded” money is pure waste with no upside. (For how long? A day? A month?)
If you are likewise surprised by Eliezer’s high regard for these theories, please join me in encouraging him to explain his reasoning.
To address your (a) comment, some countries have implemented close approximations to NGDP level targeting after the 2008 crisis, and have done well. They include most obviously Iceland (despite a severe financial crisis), and some less obvious instances like Australia, Poland and Isreal. One could point to the UK as the clearest counterexample, but just about everyone agrees that they have severe structural problems, which NGDPLT is not intended to address. And even then, monetary easing has allowed the conservative government to implement fiscal austerity without crashing the economy—this was widely expected to happen and there was a lot of public concern (compare the situation in the US wrt the “fiscal cliff” and “sequestration” scares. Here too, the Fed offset the negative fiscal effect by printing money).
As for (b), nobody argues that money hoarding is a bad thing per se. But it needs to be offset, because practically all prices in the economy are expressed in terms of money, and the price system cannot take the impact without severe side effects and misallocations. Inflation targeting is a very rough way of doing this, but it’s just not good enough (see George Selgin’s book Less than Zero for an argument to this effect). ISTM that this is not well understood in the mainstream (“NK”) macro literature, where supply shocks are confusingly modeled as “markup shocks”. I have seen cutting-edge papers pointing out that these make inflation targeting unsound (sorry for not having a ref here).
To address your (a) comment, some countries have implemented close approximations to NGDP level targeting after the 2008 crisis, and have done well.
None of the examples have targeted NGDP, which is what Sumner needs to be true to have supporting evidence. Rather, they had policies which, despite not specifically intending to, were followed by rising NGDP. The purported similarity to NGDPLT is typically justified on the grounds that the policy caused something related to happen, but there is a very big difference between that and directly targeting NGDP. And hence why it can’t demonstrate why targeting a metric (that, again, no one even cared about until Sumner started blogging about it) will have the causal power that is claimed of it.
As for (b), nobody argues that money hoarding is a bad thing per se.
I disagree; I have yet to see any anti-hoarders mention anything positive whatsoever about hoarding and take it as a given that eliminating it is bad. Landsburg says it better than I can: the very people promoting anti-hoarding policies lack any framework in which you can compare the benefits of hoarding to the hoarders against its costs, and thus know whether it’s on net bet. The best answer he gets is essentially, “well, it’s obvious that there’s a shortfall that needs to be rectified”—in other words, it’s just assumed.
To find an example of anyone saying anything positive about hoarding, you have to go to fringe Austrian economists, like in this article.
But it needs to be offset, because practically all prices in the economy are expressed in terms of money, and the price system cannot take the impact without severe side effects and misallocations.
But until you’ve quantified (or at least acknowledged the existence of) the benefits of hoarding, you can’t know if these supposed misallocations are worse than the benefits given by the hoarding. You can’t even know if they are misallocations, properly understood.
For once you accept that there’s a benefit to hoarding, then the changes in prices induced by it are actually vital market signals, just like any price. Which would mean that you can’t eliminate the price change without also destroying information that the market uses to improve resource use. I mean, oil shocks cause widespread price changes, but any attempt to stop these price changes is going to worsen the misallocation problem.
Toy example to illustrate the benefits, and important signal sent by, hoarding: let’s say we have a class of typical investors, with no special non-public knowledge about specific companies. So when they invest, they invest in the economy as a whole. (Let’s say they won’t even consider using this part of their money for consumption.) But! 70% of the economy’s investment venues are unsustainable and are actually destroying value in a way not currently obvious. In that case, it would be much better for these potential investors to hoard, rather than further advance this malinvestment. Sure, they’ll starve the good 30% of projects of funds, but they’ll also pull back on the bad 70%.
So I have yet to see any actual recognition of the benefits of hoarding among this group, which puts them in a ridiculous position. If holding money is bad, then the optimal situation is for any money received to be instantly spent on something else (whether consumption or investment). But this requires that you know what you’re going to spend the money on before you earn it—which just takes us back to barter! Thus, we see the benfit of hoarding/holding money: retaining the option value when you lack certainty about what you will spend it on. It thus signals consumers’ uncertainty that they will be able to enter sustainable patterns of trade, and cannot be costlessly squashed (as another another Econ School though of interest—that it could be zeroed without negative consequence).
None of the examples have targeted NGDP, which is what Sumner needs to be true to have supporting evidence.
I think my examples do constitute supporting evidence of some kind. Yes, it would be good to have examples of countries specifically targeting NGDP, to prevent spurious correlations or Lucas critique problems. But even so, Iceland and to a lesser extent, Poland—and, to be fair, the UK—specifically accepted a rise in inflation in order to sustain demand—it wasn’t a simple case of exogenously strong RGDP growth. (I think this might also apply to Australia, actually. Their institutional framework would certainly allow for that.) This makes the evidence quite credible, although it’s not perfect by any means.
Also, Sumner was not at all the first economist to care about NGDP as a possible target. He is a prominent popularizer, but James Meade and Bennett McCallum had proposed it first.
Your example of the “benefits of hoarding” doesn’t address the very specific problems with hoarding the unit of account for all prices in the economy, when prices are hard to adjust. Yes, money has a real option value, so money hoarding might signal some kind of uncertainty. However, you have not made the case that this “signaling” has any positive effects, especially when the operation of the price system is clearly impaired. By analogy, if peanuts were the unit of account and medium of exchange, then widespread hoarding of peanuts might signal uncertainty about the next harvest. But it would still cause a recession, and it wouldn’t actually cause the relative price of peanuts to rise (or rise much at any rate), which is what might incent additional supply.
Moreover, in practice, an uncertain agent can attain most (if not all) of the benefit of hoarding money by holding some other kind of asset, such as low-risk bonds, gold or whatever the case may be. It’s not at all clear that hoarding money specifically provides any additional benefit, or that such incremental benefits could be sustained without inflicting greater costs on other agents.
The comment is from hacker news thread about Bitcoin hitting $100. It would be cool to have him also expand more on bitcoin itself which he seems to regard as destructive but not necessarily doomed to fail? Here he entertains the idea about combining NGDP level targeting (which I don’t understand) with the best parts of Bitcoin. This all sounds very interesting.
I downloaded a Bitcoin client a couple weeks and was going to buy a few bitcoins, but the inconvenience of having to get a Mt Gox account or something made me keep on putting that off. Whoops. Hopefully this’ll teach me to be less of a procrastinator.
You think that’s bad? I considered buying a ~hundred Bitcoins after the last crash, when they were going for at less than $1, but could never be bothered. :-)
Bitcoin is a form of currency that’s supposed to be used. Now that so many people jump on the speculation train, and the rest mostly hold onto their bitcoins in the hope the price keeps rising, the practical viability of bitcoins can be called into question. I saw a stick of RAM for sale for the equivalent of over a thousand dollars. For a currency to rise so fast is terribly disruptive, and the rise (lack of supply in relation to demand) itself creates a vicious circle, since the faster it rises the more many bitcoin holders are tempted to further keep their bitcoins out of circulation.
If the price is then calculated by only the very small portion that’s up for sale (with the rest being held for often speculative purposes) - compared to a lot of prospective buyers like you who want to join the gold rush—what do you think could easily happen once some people start cashing in, the price drops, seeing the price dropping more people want to cash in, and suddenly there’s a very large portion up for sale, in conjunction with a loss of interested buyers (most don’t buy into a falling market, tautologically)?
My advice, which I may even follow myself: buy at 100, sell at 150, never look back.
Well… what I expected to happen when I downloaded the client was for the value of bitcoins to stay about the same (as it had done in the last couple months of 2012) or to rise by 10%-ish per week (as it had done in the first couple months of 2013). If I had bought some when they were at 50-ish, I would definitely be selling most of them now. And right now I don’t feel buying something that costs 20% more than it did literally yesterday.
And right now I don’t feel buying something that costs 20% more than it did literally yesterday.
Forgive me for stating the obvious: this sounds like the sunk cost fallacy. There’s a cost in that you did not buy coins when they were cheaper, and though this does affect how you feel about the issue, it shouldn’t (instrumental-rationally) affect your choices.
I did buy coins when they were at ~40$, and I was then regretting that I hadn’t bought more when two weeks earlier they were at 10$. When they were at 70$ I chose to buy some more—and I regretted not buying more when they were at 40$. But both my buy at 40$ and my buy at 70$ were good ones.
Now bitcoins are at around 141$ to 143$. Whether to buy or not buy at this point should depend on an estimation of whether the price is going to go up or down from here—and your estimation of how soon and how far the price of bitcoin is going to rise or crash from this point onwards. There’s always a risk and a chance.
I was going to reply “Actually, I meant it in the sense that given that their price has changed so quickly, I don’t trust their price to not fall by 40% while I’m sleeping”, but I’m afraid that that would just be a rationalization. (I might buy some if their price doesn’t change so much in the next couple days.)
Okay, then let me just warn people in general that transferring money from a bank to the usual bitcoin exchanges (mtgox, bitstamp, etc) may by itself take a couple days—they don’t tend to accept some of the faster methods like paypal.
This delay has both accidentally helped and hindered me in the past—it helped when it prevented me from buying in at 30 before the first crash in 2011; it had hindered me now, when I couldn’t buy in at 90 as I had wanted before the price rose to 140.
My bitcoin transactions during the last couple months have perhaps gotten me a 5000$ gain (or so) on the whole. It’s sad to think that I could have gained four times as much if I had sold one day earlier than I did; still I came out of this round benefitting, as I did back in 2011 (back then I had perhaps gotten a 2000 or 3000$ gain).
Now, I’m debating with myself whether to reinvest the money I got on bitcoin, or if the price is going to drop further… (it’s currently around 70$ in the exchanges which are still open like bitstamp.net)
You see that cute nice near-vertical drop around 15:00 UTC yesterday? That was while I was on the bus on my way home, when I had about 0.42 bitcoins on bitcoin-24.com (I’m not crazy enough to play with more money that that at the moment). #$%&. (By this morning, I had somehow managed to get back all of the value through sheer luck by repeatedly selling and buying at the right times. Now bitcoin-24.com is down, and I don’t know whether that happened before or after my offer to sell most of the bitcoins I had left at €80/BTC was accepted. (From this graph I guess I was lucky.)
It was recently brought to my attention that Eliezer Yudkowsky regards the monetary theories of Scott Sumner (short overview) to be (what we might call) a “correct contrarian cluster”, or an island of sanity when most experts (though apparently a decreasing number) believe the opposite.
I would be interested in knowing why. To me, Sumner’s views are a combination of:
a) Goodhart’s folly (“Historically, an economic metric [that nobody cared about until he started talking about] has been correlated with economic goodness; if we only targeted this metric with policy, we would get that goodness. Here are some plausible mechanisms why …”—my paraphrase, of course)
b) Belief that “hoarded” money is pure waste with no upside. (For how long? A day? A month?)
If you are likewise surprised by Eliezer’s high regard for these theories, please join me in encouraging him to explain his reasoning.
To address your (a) comment, some countries have implemented close approximations to NGDP level targeting after the 2008 crisis, and have done well. They include most obviously Iceland (despite a severe financial crisis), and some less obvious instances like Australia, Poland and Isreal. One could point to the UK as the clearest counterexample, but just about everyone agrees that they have severe structural problems, which NGDPLT is not intended to address. And even then, monetary easing has allowed the conservative government to implement fiscal austerity without crashing the economy—this was widely expected to happen and there was a lot of public concern (compare the situation in the US wrt the “fiscal cliff” and “sequestration” scares. Here too, the Fed offset the negative fiscal effect by printing money).
As for (b), nobody argues that money hoarding is a bad thing per se. But it needs to be offset, because practically all prices in the economy are expressed in terms of money, and the price system cannot take the impact without severe side effects and misallocations. Inflation targeting is a very rough way of doing this, but it’s just not good enough (see George Selgin’s book Less than Zero for an argument to this effect). ISTM that this is not well understood in the mainstream (“NK”) macro literature, where supply shocks are confusingly modeled as “markup shocks”. I have seen cutting-edge papers pointing out that these make inflation targeting unsound (sorry for not having a ref here).
None of the examples have targeted NGDP, which is what Sumner needs to be true to have supporting evidence. Rather, they had policies which, despite not specifically intending to, were followed by rising NGDP. The purported similarity to NGDPLT is typically justified on the grounds that the policy caused something related to happen, but there is a very big difference between that and directly targeting NGDP. And hence why it can’t demonstrate why targeting a metric (that, again, no one even cared about until Sumner started blogging about it) will have the causal power that is claimed of it.
I disagree; I have yet to see any anti-hoarders mention anything positive whatsoever about hoarding and take it as a given that eliminating it is bad. Landsburg says it better than I can: the very people promoting anti-hoarding policies lack any framework in which you can compare the benefits of hoarding to the hoarders against its costs, and thus know whether it’s on net bet. The best answer he gets is essentially, “well, it’s obvious that there’s a shortfall that needs to be rectified”—in other words, it’s just assumed.
To find an example of anyone saying anything positive about hoarding, you have to go to fringe Austrian economists, like in this article.
But until you’ve quantified (or at least acknowledged the existence of) the benefits of hoarding, you can’t know if these supposed misallocations are worse than the benefits given by the hoarding. You can’t even know if they are misallocations, properly understood.
For once you accept that there’s a benefit to hoarding, then the changes in prices induced by it are actually vital market signals, just like any price. Which would mean that you can’t eliminate the price change without also destroying information that the market uses to improve resource use. I mean, oil shocks cause widespread price changes, but any attempt to stop these price changes is going to worsen the misallocation problem.
Toy example to illustrate the benefits, and important signal sent by, hoarding: let’s say we have a class of typical investors, with no special non-public knowledge about specific companies. So when they invest, they invest in the economy as a whole. (Let’s say they won’t even consider using this part of their money for consumption.) But! 70% of the economy’s investment venues are unsustainable and are actually destroying value in a way not currently obvious. In that case, it would be much better for these potential investors to hoard, rather than further advance this malinvestment. Sure, they’ll starve the good 30% of projects of funds, but they’ll also pull back on the bad 70%.
So I have yet to see any actual recognition of the benefits of hoarding among this group, which puts them in a ridiculous position. If holding money is bad, then the optimal situation is for any money received to be instantly spent on something else (whether consumption or investment). But this requires that you know what you’re going to spend the money on before you earn it—which just takes us back to barter! Thus, we see the benfit of hoarding/holding money: retaining the option value when you lack certainty about what you will spend it on. It thus signals consumers’ uncertainty that they will be able to enter sustainable patterns of trade, and cannot be costlessly squashed (as another another Econ School though of interest—that it could be zeroed without negative consequence).
I think my examples do constitute supporting evidence of some kind. Yes, it would be good to have examples of countries specifically targeting NGDP, to prevent spurious correlations or Lucas critique problems. But even so, Iceland and to a lesser extent, Poland—and, to be fair, the UK—specifically accepted a rise in inflation in order to sustain demand—it wasn’t a simple case of exogenously strong RGDP growth. (I think this might also apply to Australia, actually. Their institutional framework would certainly allow for that.) This makes the evidence quite credible, although it’s not perfect by any means.
Also, Sumner was not at all the first economist to care about NGDP as a possible target. He is a prominent popularizer, but James Meade and Bennett McCallum had proposed it first.
Your example of the “benefits of hoarding” doesn’t address the very specific problems with hoarding the unit of account for all prices in the economy, when prices are hard to adjust. Yes, money has a real option value, so money hoarding might signal some kind of uncertainty. However, you have not made the case that this “signaling” has any positive effects, especially when the operation of the price system is clearly impaired. By analogy, if peanuts were the unit of account and medium of exchange, then widespread hoarding of peanuts might signal uncertainty about the next harvest. But it would still cause a recession, and it wouldn’t actually cause the relative price of peanuts to rise (or rise much at any rate), which is what might incent additional supply.
Moreover, in practice, an uncertain agent can attain most (if not all) of the benefit of hoarding money by holding some other kind of asset, such as low-risk bonds, gold or whatever the case may be. It’s not at all clear that hoarding money specifically provides any additional benefit, or that such incremental benefits could be sustained without inflicting greater costs on other agents.
Yes!
The comment is from hacker news thread about Bitcoin hitting $100. It would be cool to have him also expand more on bitcoin itself which he seems to regard as destructive but not necessarily doomed to fail? Here he entertains the idea about combining NGDP level targeting (which I don’t understand) with the best parts of Bitcoin. This all sounds very interesting.
I downloaded a Bitcoin client a couple weeks and was going to buy a few bitcoins, but the inconvenience of having to get a Mt Gox account or something made me keep on putting that off. Whoops. Hopefully this’ll teach me to be less of a procrastinator.
You think that’s bad? I considered buying a ~hundred Bitcoins after the last crash, when they were going for at less than $1, but could never be bothered. :-)
Bitcoin is a form of currency that’s supposed to be used. Now that so many people jump on the speculation train, and the rest mostly hold onto their bitcoins in the hope the price keeps rising, the practical viability of bitcoins can be called into question. I saw a stick of RAM for sale for the equivalent of over a thousand dollars. For a currency to rise so fast is terribly disruptive, and the rise (lack of supply in relation to demand) itself creates a vicious circle, since the faster it rises the more many bitcoin holders are tempted to further keep their bitcoins out of circulation.
If the price is then calculated by only the very small portion that’s up for sale (with the rest being held for often speculative purposes) - compared to a lot of prospective buyers like you who want to join the gold rush—what do you think could easily happen once some people start cashing in, the price drops, seeing the price dropping more people want to cash in, and suddenly there’s a very large portion up for sale, in conjunction with a loss of interested buyers (most don’t buy into a falling market, tautologically)?
My advice, which I may even follow myself: buy at 100, sell at 150, never look back.
Well… what I expected to happen when I downloaded the client was for the value of bitcoins to stay about the same (as it had done in the last couple months of 2012) or to rise by 10%-ish per week (as it had done in the first couple months of 2013). If I had bought some when they were at 50-ish, I would definitely be selling most of them now. And right now I don’t feel buying something that costs 20% more than it did literally yesterday.
Forgive me for stating the obvious: this sounds like the sunk cost fallacy. There’s a cost in that you did not buy coins when they were cheaper, and though this does affect how you feel about the issue, it shouldn’t (instrumental-rationally) affect your choices.
I did buy coins when they were at ~40$, and I was then regretting that I hadn’t bought more when two weeks earlier they were at 10$. When they were at 70$ I chose to buy some more—and I regretted not buying more when they were at 40$. But both my buy at 40$ and my buy at 70$ were good ones.
Now bitcoins are at around 141$ to 143$. Whether to buy or not buy at this point should depend on an estimation of whether the price is going to go up or down from here—and your estimation of how soon and how far the price of bitcoin is going to rise or crash from this point onwards. There’s always a risk and a chance.
I was going to reply “Actually, I meant it in the sense that given that their price has changed so quickly, I don’t trust their price to not fall by 40% while I’m sleeping”, but I’m afraid that that would just be a rationalization. (I might buy some if their price doesn’t change so much in the next couple days.)
Okay, then let me just warn people in general that transferring money from a bank to the usual bitcoin exchanges (mtgox, bitstamp, etc) may by itself take a couple days—they don’t tend to accept some of the faster methods like paypal.
That’s what prevented me from owning bitcoins yesterday afternoon while their price halved.
This delay has both accidentally helped and hindered me in the past—it helped when it prevented me from buying in at 30 before the first crash in 2011; it had hindered me now, when I couldn’t buy in at 90 as I had wanted before the price rose to 140.
My bitcoin transactions during the last couple months have perhaps gotten me a 5000$ gain (or so) on the whole. It’s sad to think that I could have gained four times as much if I had sold one day earlier than I did; still I came out of this round benefitting, as I did back in 2011 (back then I had perhaps gotten a 2000 or 3000$ gain).
Now, I’m debating with myself whether to reinvest the money I got on bitcoin, or if the price is going to drop further… (it’s currently around 70$ in the exchanges which are still open like bitstamp.net)
You see that cute nice near-vertical drop around 15:00 UTC yesterday? That was while I was on the bus on my way home, when I had about 0.42 bitcoins on bitcoin-24.com (I’m not crazy enough to play with more money that that at the moment). #$%&. (By this morning, I had somehow managed to get back all of the value through sheer luck by repeatedly selling and buying at the right times. Now bitcoin-24.com is down, and I don’t know whether that happened before or after my offer to sell most of the bitcoins I had left at €80/BTC was accepted. (From this graph I guess I was lucky.)
Sounds like maybe it wasn’t just timing. :-|
So do you have a to-do list that you wrote down “buy a few bitcoins” in? If not, maybe you didn’t actually procrastinate; maybe you just forgot.
I don’t.
More like, first the former, then the latter. ;-)