We've had some pretty good comments from both Mark and Joshua in my opinion. I know Mark had problems over at Cullen recently-and of course, Mark and I recently had words over at Money Illusion. However, I certainly think the kind of comments he's put in on this have been a good contribution.
This obviously doesn't make him right but as long as things don't get personal like they did that one time at Money Illusion-where he suggested that I am 'unhnged' I certainly welcome his contributions. I've never been about anything than learning and understanding the economic world. Sumner doesn't like me and acts as if I have some nefarious motives but that's the truth.
As to the subject of GC it's a complicated subject which I clearly don't have the level of background that either Mark or Joshua have in it-however, please note that this doesn't mean, as Sumner seems to feel, that I should therefore just shut up and let the people who do have the background to do all the talking; you can't allow someone to intimidate you intellectually just because they may have a lot of formal training as they may be one of two things: wrong and/or using their knowledge advantage to trick you. Still, I feel better having them both here as at least I have more than one view of things to pick and choose from. Joshua is the perfect person for this as he's currently writing a major paper at his University-George Mason?-about GC.
Mark had yesterday answered Peter P's critical comments regarding his use of GC to show that the MB caused a number of postiive effects-the stock market, growth, etc. Joshua than had a reply for Mark's replay:
"A couple questions for you regarding Granger-causality since I'm currently working on a paper that focuses on the period from 1971 to 2008."
"1) In response to Peter P, you ask how "terms that aren't in the observation range are causing lagged terms of X within the observation range to be spuriously correlated with present values of Y." Let me propose two seemingly plausible answer."
"First, imagine that W causes changes in variables X and Y, however the response of variable X typically occurs with a shorter lag than Y. Wouldn't X exhibit unidirectional Granger-causality of Y, even though unidirectional Granger-causality would also exist from W to X and W to Y?"
"Second, imagine that expectations of W causes changes in X but only the occurrence of W causes changes in Y. The outcome described above could once again follow."
"2) You mention that "The only time one is likely to find that the monetary base Granger causes anything is when the monetary base is the monetary policy instrument." Does this imply that changes in the monetary base prior to 2008 were not expected to affect NGDP? If so, how does one reconcile this view with old Monetarist or Keynesian models that treated the money supply as an exogenous determinant of NGDP?"
"In any case if W causes Y then what you're really proposing is that X causes Y, but we already know that."
"2) I haven't shown that the monetary base Granger causes NGDP after December 2008. I'm using monthly data and NGDP is in quarterly frequency only. There aren't enough quarters since December 2008 such that Granger causality tests would be feasible."
"For what it's worth I've done Granger causality tests on the monetary base and NGDP from 1947 to present and find that there is bidirectional Granger causality between the monetary base and NGDP, and in fact bidirectional Granger causality between M1, M2 and MZM and NGDP."
"Suppose the central bank is targeting the inflation rate using the unsecured overnight interbank lending rate. We probably would find that commercial bank loans and leases Granger cause broad money (or the money multiplier, as has been found in some studies) and loans and leases Granger cause the monetary base. But if the unsecured overnight interbank lending rate is fixed, and the central bank is increasing the monetary base at ad hoc rates as in QE, then we may find that the monetary base Granger causes other quantities, no?"
"If so, how does one reconcile this view with old Monetarist or Keynesian models that treated the money supply as an exogenous determinant of NGDP?"
"I don't know. Why would anybody want to do that?"
If there's bidirectionality between NGDP and the MB at least by these three measures what does it prove-isn't it at most undecidable then? It seems clear to me that the effect of the expansion of the MB hasn't been what anyone seemed to expect previous to the Great Recession. Sumner complains about 'moving the goalposts'
http://www.themoneyillusion.com/?p=24636
but it seems to me that on this subject the goalposts have defintely been moved. I asked about inflation-one thing that expanding the MB has certainly always been expected to do is increase inflation-recall all the dire predictions back in 2009 of galloping inlfation-Bom Murphy, et. al.; I know Austrians it's about the Long Run right?
Mark argues that as low as inflation has been it would have been even lower otherwise:
Krugman has wondered out loud a few times why we didn't have accelerating deflation with a prolonged large output gap. But I think he has come to believe that the Phillips Curve is in fact a curve (as Phillips believed). That is, the Phillips Curve becomes more horizontal the higher the unemployment rate.
"However, is QE important because it expands the monetary base or because of what it does for confidence and signifies about future monetary policy which both Sumner and Krugman believe?"
"Why does QE work? Probably through expectations as both Sumner and Krugman believe, but there may be other more direct channels of which we are not aware."
This obviously doesn't make him right but as long as things don't get personal like they did that one time at Money Illusion-where he suggested that I am 'unhnged' I certainly welcome his contributions. I've never been about anything than learning and understanding the economic world. Sumner doesn't like me and acts as if I have some nefarious motives but that's the truth.
As to the subject of GC it's a complicated subject which I clearly don't have the level of background that either Mark or Joshua have in it-however, please note that this doesn't mean, as Sumner seems to feel, that I should therefore just shut up and let the people who do have the background to do all the talking; you can't allow someone to intimidate you intellectually just because they may have a lot of formal training as they may be one of two things: wrong and/or using their knowledge advantage to trick you. Still, I feel better having them both here as at least I have more than one view of things to pick and choose from. Joshua is the perfect person for this as he's currently writing a major paper at his University-George Mason?-about GC.
Mark had yesterday answered Peter P's critical comments regarding his use of GC to show that the MB caused a number of postiive effects-the stock market, growth, etc. Joshua than had a reply for Mark's replay:
"A couple questions for you regarding Granger-causality since I'm currently working on a paper that focuses on the period from 1971 to 2008."
"1) In response to Peter P, you ask how "terms that aren't in the observation range are causing lagged terms of X within the observation range to be spuriously correlated with present values of Y." Let me propose two seemingly plausible answer."
"First, imagine that W causes changes in variables X and Y, however the response of variable X typically occurs with a shorter lag than Y. Wouldn't X exhibit unidirectional Granger-causality of Y, even though unidirectional Granger-causality would also exist from W to X and W to Y?"
"Second, imagine that expectations of W causes changes in X but only the occurrence of W causes changes in Y. The outcome described above could once again follow."
"2) You mention that "The only time one is likely to find that the monetary base Granger causes anything is when the monetary base is the monetary policy instrument." Does this imply that changes in the monetary base prior to 2008 were not expected to affect NGDP? If so, how does one reconcile this view with old Monetarist or Keynesian models that treated the money supply as an exogenous determinant of NGDP?"
Here is Mark's answer to Joshua:
"1) Yes, but W *is* X. Maybe I should have been more explicit and said "terms *of X* that aren't in the observation range" but that's clear from the context of this conversation, right? (That is, W and X are both the monetary base.)."
"In any case if W causes Y then what you're really proposing is that X causes Y, but we already know that."
"2) I haven't shown that the monetary base Granger causes NGDP after December 2008. I'm using monthly data and NGDP is in quarterly frequency only. There aren't enough quarters since December 2008 such that Granger causality tests would be feasible."
"For what it's worth I've done Granger causality tests on the monetary base and NGDP from 1947 to present and find that there is bidirectional Granger causality between the monetary base and NGDP, and in fact bidirectional Granger causality between M1, M2 and MZM and NGDP."
"Suppose the central bank is targeting the inflation rate using the unsecured overnight interbank lending rate. We probably would find that commercial bank loans and leases Granger cause broad money (or the money multiplier, as has been found in some studies) and loans and leases Granger cause the monetary base. But if the unsecured overnight interbank lending rate is fixed, and the central bank is increasing the monetary base at ad hoc rates as in QE, then we may find that the monetary base Granger causes other quantities, no?"
"If so, how does one reconcile this view with old Monetarist or Keynesian models that treated the money supply as an exogenous determinant of NGDP?"
"I don't know. Why would anybody want to do that?"
If there's bidirectionality between NGDP and the MB at least by these three measures what does it prove-isn't it at most undecidable then? It seems clear to me that the effect of the expansion of the MB hasn't been what anyone seemed to expect previous to the Great Recession. Sumner complains about 'moving the goalposts'
http://www.themoneyillusion.com/?p=24636
but it seems to me that on this subject the goalposts have defintely been moved. I asked about inflation-one thing that expanding the MB has certainly always been expected to do is increase inflation-recall all the dire predictions back in 2009 of galloping inlfation-Bom Murphy, et. al.; I know Austrians it's about the Long Run right?
Mark argues that as low as inflation has been it would have been even lower otherwise:
Krugman has wondered out loud a few times why we didn't have accelerating deflation with a prolonged large output gap. But I think he has come to believe that the Phillips Curve is in fact a curve (as Phillips believed). That is, the Phillips Curve becomes more horizontal the higher the unemployment rate.
"However, is QE important because it expands the monetary base or because of what it does for confidence and signifies about future monetary policy which both Sumner and Krugman believe?"
"Why does QE work? Probably through expectations as both Sumner and Krugman believe, but there may be other more direct channels of which we are not aware."
I find this a little ironic-with all the shots MMers take at Krugman they're always very happy to use them to advance their own arguments. So Krugman's model that expected prolonged deflation is right?
It seems to me that the economy has acted here very similar to how it did in the recovery phase of the GD which also didn't have deflation-the deflation ended with the start of the recovery in 1933.
As for expectations that sounds right to me-QE is good for expectations. However, that means it's effect may not have much to do with the expansion of the MB. I notice that Mark's good friend Cullen has a new post about QE and-the virtually nonexistent-level of inflation.
That there is no correlation is not deniable. Of course, Mark's counterfactual is another question. Ironic that it depends on Krugman's model being right.
No comments:
Post a Comment