A puzzling paper from the St. Louis Fed

Several commenters sent me a paper by Yi Wen of the St. Louis Fed:

Currently the U.S. real GDP is about 10% below its long-run trend (see Figure 2) and total asset purchases stand at $3.7 trillion (or less than 25% of GDP). Our model predicts that this level of asset purchases (even if permanent) would have little effect on aggregate output and employment even though it could reduce the real interest rate significantly by 2 to 3 percentage points. These predictions are consistent with the empirical evidence. Thus, based on our model the Federal Reserve’s total asset purchases must be more than quadrupled and remain active for several more years if the Fed intends to eliminate the 10% output gap caused by the financial crisis.

Lots of things puzzled me about this paragraph. How could there be a 10% output gap, if unemployment is only 7.2%? More importantly if the asset purchases are permanent, why wouldn’t the US experience hyperinflation when we exited the liquidity trap? It’s widely believed that that exit will occur within the next few years. If so, how can it be that a permanent QE would only reduce real interest rates by 2 to 3 percentage points?

The conclusion seems to answer these questions:

We provide a general equilibrium finance model featuring explicit government purchases of private debt to evaluate the efficacy of unconventional monetary policies. We show that when inflation is fully anchored, such policies can reduce the borrowing costs (the real interest rate) and relax borrowing constraints by raising the collateral value of fixed assets. However, to have a significant impact on real economic activities at the aggregate level, the scope of asset purchases must be extraordinarily large and the extent highly persistent. These predictions are not inconsistent with the empirical evidence provided in the Introduction.

.  .  .

Our model predicts that when inflation is fully anchored, LSAP can lower the real interest rate on both public and private debt and raise the collateral value of productive capital, thus increasing the number of borrowers and the aggregate quantity of debt. However, unless the extent of asset purchases is extremely large and highly persistent, CE through LSAP cannot effectively offset the negative impact of a financial crisis on aggregate output and employment.  [emphasis added]

Well now I’m totally confused. The assumption that inflation is fully anchored does explain the weak response of real interest rates. But since monetary policy affects economy via shifts in aggregate demand, the assumption that inflation is fully anchored is equivalent to an assumption the nominal GDP growth is fully anchored. In other words, this seems to be saying that if the monetary policy regime tries to avoid any increase in aggregate demand, it will succeed in avoiding any increase in aggregate demand. Yes, but if the goal was to boost recovery, why would they anchor aggregate demand?  Why not try to increase aggregate demand at a pace likely to yield the desired inflation/employment outcome?

Obviously I’m missing something. Can someone help me?

PS.  Some people must have thought I’d lost my mind when I endorsed Kevin Erdmann’s theory that tight money might have contributed to the housing bubble.  Gregor Bush sent me this article by Lars Svensson.

Leaning-against-the-wind monetary policy may lead to a Fisherian debt deflation, since it may lower prices below the anticipated level and therefore raise real debt above what was anticipated. This is what the Riksbank has done by keeping average inflation significantly below the inflation target for a long period. This has caused household real debt to be substantially higher than it would have been if inflation had been on target.

So is Svensson also crazy?

HT:  Saturos, Morgan, et al.


Tags:

 
 
 

39 Responses to “A puzzling paper from the St. Louis Fed”

  1. Gravatar of LK Beland LK Beland
    22. October 2013 at 06:53

    “How could there be a 10% output gap, if unemployment is only 7.2%?”

    The employment-to-population ratio is about 5 pp lower than its 2008 ratio, i.e. 63% (and 6.5 pp lower than the 2000 ratio).

    That amounts to an 8% output gap, if full employment is defined at 63% e-to-p and if that extra labor produces with mean productivity.

  2. Gravatar of Bill Woolsey Bill Woolsey
    22. October 2013 at 07:15

    Think of the “output gap” as being the difference between output and trend, not between output and potential. Assume that output is at potential, but potential is below trend.

    Why is potential so low? Not because of low aggregate demand at all. It is rather than financial market failures have reduced productive capacity.

    For example, suppose a financial market breakdown caused people to save less and consume more. The lower saving rate starts pushing us towards a lower per-capita capital economy and a lower growth path of potential output. If some of the extra consumption is leisure, there is less capital and less labor!

    Fully employment output, potential output, is on a lower growth path.

    To fix the credit markets, we need massive Fed intervention. (I guess.) And this will motivate people to work, save, and invest. I think they would have to pay higher interest rates on reserves to make this work. The private sector borrowers from the Fed at lower real interest rates (due to the Fed taking all the risk.) The Fed pays higher interest on reserves, which is passed on as higher interest rate on insured deposits, and so people save and also work more to save more.

    In other words, screwball real business cycle nonsense.

    Hey if all prices and wages must be perfectly flexible by assumption, so real expenditure always equals potential output, then there is never an “output gap” in the sense of less output than potential. But output, which is potential output, can shift to higher and lower growth paths, creating “gaps” between the current level and the past trend.

    Heck, Scott, you talk that way sometimes. Potential as trend.

  3. Gravatar of Steven Kopits Steven Kopits
    22. October 2013 at 07:31

    I’ll be making the entirely anti-Sumnerian focusing on the impact of oil on the economy. At Princeton on Thursday. Details below.

    “Oil and Economic Growth: A Supply-Constrained View”
    The presentation will discuss demand-driven (traditional) versus supply-constrained models with regards to oil markets, assess the data supporting each view, and discuss the implications for GDP growth, as well as implications for the oil business, particularly with respect to spending on exploration and production and related supply. Barclays and Citi Commodities will also be presenting their views. Thursday, Oct. 24th, 1:30 – 4:00 pm, Princeton University, Friend Center, Room 008. Open to the public.

  4. Gravatar of Steven Kopits Steven Kopits
    22. October 2013 at 07:32

    …”Sumnerian argument…” Still tortured, but a bit more comprehensible.

  5. Gravatar of Morgan Warstler Morgan Warstler
    22. October 2013 at 07:35

    My thing is how can having the least productive out of work lead to a big output gap?

    I know GI ?CYB increases consumption as the least productive work for each other to keep their prices down, but that’s not a lot of actual “productivity” – its just making welfare dollars go farther.

  6. Gravatar of ssumner ssumner
    22. October 2013 at 08:25

    LK, That might be the case if they were planning on rejoining the labor force, but I consider that highly unlikely.

    Bill, You said:

    “Think of the “output gap” as being the difference between output and trend, not between output and potential. Assume that output is at potential, but potential is below trend.”

    You lost me, I thought the output gap was the difference between output and potential. And why should we assume output is at potential?

    Nobody has anything on the Yi Wen paper?

  7. Gravatar of jknarr jknarr
    22. October 2013 at 09:02

    I’m now looking at RGDP to trend versus sex-participation-adjusted-and-unadjusted E-P ratios, and it has a pretty good 44% r-squared since the 1950s, and 95% since the year 2000. The data trend (even way back before 1947) also confirms a US RGDP undershoot to the tune of 10%.

    Wen is on the right track, and the temporary liquidity trap and disinflationary LSAPs are useful, though he is just describing the nonlinearity of base demand as it goes infinite at zero rates.

    His premises screw up his results, by focussing on LSAP asset purchases, and he tries to escape the noose by “anchored” tautologies. I don’t understand why smart people still focus on the composition of assets: the Fed buys only slender percentages of outstanding assets at market prices, while its liability creation dilutes every outstanding MOA: it’s clear which effect is more powerful. (The inflationary napalm stock is already there, in banking system reserves, and is just looking for a currency demand spark.)

    Scott, if Friedman can say that tight money was Japan’s problem, so can you for the US, and there is a clear parallel low-yield effect in asset markets, including real estate and equities: Japan certainly had bubbles. The supply- and demand- sides sort out: tight money-low rates-stretch for yield-strong currency-low base/NGDP, and serial bubbles in cash flow bearing assets.

    It’s not crazy in the slightest. You are on the right track: tight money since 1980 has caused huge demand for financial assets, tight money produces low yields and high debt, and tight money is an ongoing long-run policy choice.

    BTW, Yglesias has been crazy too.
    http://www.slate.com/blogs/moneybox/2012/12/04/the_fed_the_debt_interest_rates_and_easy_money.html

  8. Gravatar of TravisV TravisV
    22. October 2013 at 09:05

    Tim Duy on why the S&P 500 is up today…..

    “Bottom Line: Incoming data are not exactly consistent with the definition of “stronger and sustainable.” And even if the data shift upward in the next few months, it will take another three months to confirm that it is sustainable. That pushes any decision on asset purchases out to March at the earliest. If the Fed wants to taper sooner, policymakers will need to hang their hats on the unemployment rate alone, which seems unlikely. In the meantime, the one-way direction of financial markets will drive the “financial stability hawks” on the FOMC crazy, but at the moment they lack sufficient numbers to push forward their campaign to end asset purchases.”

    http://economistsview.typepad.com/timduy/2013/10/worth-the-wait.html

  9. Gravatar of Steve Steve
    22. October 2013 at 09:07

    “Some people must have thought I’d lost my mind when I endorsed Kevin Erdmann’s theory that tight money might have contributed to the housing bubble…So is Svensson also crazy?’

    I was told I was crazy when I posted this theory two years ago.

  10. Gravatar of Steve Steve
    22. October 2013 at 09:09

    Maybe someday someone will decide my crazy theory that high capital taxation (through tax-lock) contributes to bubbles, too. Then Alan Krueger’s head will explode.

  11. Gravatar of Bill Ellis Bill Ellis
    22. October 2013 at 09:30

    “How could there be a 10% output gap, if unemployment is only 7.2%?”

    Discouraged workers ? “U-6, total unemployed, plus all marginally attached workers, plus total employed part time for economic reasons, as a percent of the civilian labor force plus all marginally attached workers.”

  12. Gravatar of Bill Ellis Bill Ellis
    22. October 2013 at 09:31

    U6 is about 14%.

  13. Gravatar of dlr dlr
    22. October 2013 at 09:57

    Scott,

    Bill W. did address the paper — I’m not sure you followed. The underlying model in the paper is even more of an RBC model than the typical NK model; it is in fact a barter model — there is no money at all. Sticky prices don’t matter and thus AD doesn’t exist as you know it. The “output gap” can be seen as the difference between actual and potential output, but potential includes the possibility that a temporarily constrained financial supply side returns to previously trend. A constrained financial supply side in this case has to do with the allocative efficiency of credit. I don’t see unemployment in this model as in many RBC models — you’d have to model something else in — only less or more employment

    This model, like pretty much any RBC model, can add money and extend into a NK model, but to me this just highlights how much further away you are from the NK/Woodfordian viewpoint than you sometimes seem to think you are.

  14. Gravatar of Bob Murphy Bob Murphy
    22. October 2013 at 10:41

    Scott wrote:

    In other words, this seems to be saying that if the monetary policy regime tries to avoid any increase in aggregate demand, it will succeed in avoiding any increase in aggregate demand.

    Maybe it depends on how much weight you put on “fully” (in “fully anchored”), but Scott isn’t the author just saying that if we assume NGDP rises (mostly) because of growth in RGDP, not (much) because of price inflation?

    I thought this was what your whole plan was. 🙂

  15. Gravatar of Geoff Geoff
    22. October 2013 at 15:17

    “How could there be a 10% output gap, if unemployment is only 7.2%?”

    The relationship is not linear as you are assuming. A drop or increase of 1% employment does not have a drop or incease in 1% output.

    It is easier to grasp when the numbers are a lot larger. Thus, suppose unemployment rose to 90%, from an initial 0%. Output would certainly not be 90% less. Output would be close to 99.9% less.

  16. Gravatar of benjamin cole benjamin cole
    22. October 2013 at 17:02

    You know economists were never business operators. What Yi Wen just said is we could double QE to $7 trillion or so, make it permanent and hardly anything would happen.
    Holy Moly!
    Great Balls of Fire!
    Is anyone thinking this through?
    We can wipe out half the federal debt and nothing will happen.
    Why, please tell me, why would we not do this?

  17. Gravatar of ssumner ssumner
    22. October 2013 at 18:34

    Bill, I predict that three years from now the idea of a 10% output gap in 2013 will look absurd. We’ll be back to full employment, and output will have risen perhaps 1% per year above trend between 2013 and 2016. We will see.

    dlr, No, I certainly didn’t understand Bill’s comment. In any case, NK models do have sticky prices and nominal shocks do matter.

    But even if it’s a RBC model, it doesn’t really answer my question. QE is supposed to work by raising expected inflation (in NK models) So if you are going to argue QE doesn’t work, why assume the Fed is trying to avoid higher inflation?

    And if you argue QE is permanent, how can you assume inflation is anchored? Why would real interest rates only fall by 2% or 3%?

    Bob, I don’t see how that helps. The split is based on the slope of the SRAS. Whatever slope you assume, there is some inflation target that would get you a fast recovery. And how can he assume prices don’t rise very much, if the QE is permanent?

    Ben, Good point.

  18. Gravatar of Mark A. Sadowski Mark A. Sadowski
    22. October 2013 at 18:37

    Bill Woolsey and dlr hit the nail on the head. This is essentially an RBC model.

    The only real puzzle is why models of the economy that do not include aggregate demand (AD) continue to be so popular with the research divisions of the institutions charged with the responsibility of regulating AD, especially after the economy experienced the greatest negative AD shock in 80 years.

  19. Gravatar of Geoff Geoff
    22. October 2013 at 18:38

    “So is Svensson also crazy?”

    No, to the extent that individuals borrow more when interest rates are lower, ceteris paribus.

  20. Gravatar of ssumner ssumner
    22. October 2013 at 18:43

    Mark, Yes, but money is neutral in RBC models in the long run. So why no hyperinflation in this model?

  21. Gravatar of Geoff Geoff
    22. October 2013 at 18:50

    “Yes, but money is neutral in RBC models in the long run.”

    Money is neutral in the long run only if there is a one time change to money.

    If there are continuous changes to money, then we’re always in a non-neutral money world.

    Because there are continuous changes to money, it means we are in a non-neutral money world.

  22. Gravatar of Mark A. Sadowski Mark A. Sadowski
    22. October 2013 at 19:20

    Scott,
    “Mark, Yes, but money is neutral in RBC models in the long run. So why no hyperinflation in this model?”

    Because of the way the model is set up. Pages 34-35:

    “Our model provides an alternative explanation for the low in‡flation level. The Federal Reserve’s LSAP alone can depress infl‡ation near the liquidity trap: Once the real interest rate
    of …financial assets is low enough, QE induces ‡flight to liquidity because portfolio investors opt to switch from interest-bearing assets to money. Hence, the aggregate price level must fall to accommodate the increased demand for real money balances for any given target level of long-run money growth (or anticipated infl‡ation rate).24 Therefore, monetary injection through LSAP exerts downward pressure on the price level because it increases aggregate money demand
    for any given in‡flation rate. This de‡flationary effect is particularly strong at the liquidity trap, as shown in Figure 8. The U.S. economy has been in a liquidity trap since late 2008/early 2009 when the nominal short-term interest rate became essentially zero. Subsequent CE further reinforced the low in‡flation by keeping high pressure on real money demand.”

    Incidentally this model bears more than some resemblance to Stephen Williamson’s models, which isn’t surprising since it’s from St. Louis.

    The thing that I find so irritating about such research is that you can start with any conclusion and construct a model, no matter how unrealistic, which will support that conclusion.

  23. Gravatar of ssumner ssumner
    23. October 2013 at 05:30

    Mark, So just to be clear, is this model claiming that people and banks would willingly hold vast quantities of zero interest base money (say 50% of GDP) in a environment where T-bills yield say 5%? I still don’t get it.

  24. Gravatar of Mark A. Sadowski Mark A. Sadowski
    23. October 2013 at 08:21

    Scott,
    “Mark, So just to be clear, is this model claiming that people and banks would willingly hold vast quantities of zero interest base money (say 50% of GDP) in a environment where T-bills yield say 5%?”

    Given the way the model is set up, there’s absolutely no chance of T-bills ever yielding anywhere near 5%. LSAPs drive the real rate of return down until money is a perfect substitute for public and private debt. Page 22:

    “3.4.1 The Liquidity Trap Regime

    The real interest rate of public/private debt is bounded below by the real rate of return on money 1/(1+pi). At this lower bound, money is a perfect substitute for public/private debt since both forms of debt yield the same rate of return in real terms. Therefore, the liquidity trap regime is characterized by the conditions 1/(1+pi)=1+rg=1+rc< 1/Beta. The liquidity trap can arise, for example, if the supply of public/private debt is sufficiently low, which drives down the real interest rate to 1/(1+pi). Since the nominal interest rate of government bonds is given by (1+ig) = (1+pi)(1+rg), at the liquidity trap the nominal interest rate is automatically at its zero lower bound ig = 0. At the liquidity trap, further decreases in the supply of public debt or further increases in the demand for public debt have no e¤ect on the real interest rate rg, given the in‡flation rate."

  25. Gravatar of ssumner ssumner
    23. October 2013 at 08:45

    Mark, I understand that might be the case as a short run proposition, but he explicitly states there is little long run effect on the price level. That’s what I’m getting at. I’m sure the St louis Fed is not going to put out a paper suggesting we could fully monetize the entire national debt forever and there would be little or no inflationary consequences. Please don’t tell me I’m wrong.

  26. Gravatar of Mark A. Sadowski Mark A. Sadowski
    23. October 2013 at 10:52

    Scott,
    “Mark, I understand that might be the case as a short run proposition, but he explicitly states there is little long run effect on the price level.”

    No, that’s not what he says. Page 32:

    “In the engaged liquidity trap regime, if the central bank’s purchases of private debt increase, the real interest rate rc cannot fall further; instead, the demand for real money balances will rise. Given the total nominal money supply Mt, a higher demand for real balances is possible if and only if the price level becomes permanently lower. Therefore, the economy will experience a temporary de‡ation and then settle at a permanently lower price level.”

    In other words, in this model, not only does QE not raise the long run price level, it permanently *lowers* the price level.

    “I’m sure the St louis Fed is not going to put out a paper suggesting we could fully monetize the entire national debt forever and there would be little or no inflationary consequences. Please don’t tell me I’m wrong.”

    You’re wrong.

    Surely by now you realize St. Louis is a very very strange place. After all James Bullard is their President, and Steve Williamson is an economist there. And although David Andolfatto was nice enough to allow me to do a guest post on his blog once, I even have my doubts about him from time to time.

  27. Gravatar of ssumner ssumner
    23. October 2013 at 11:34

    Mark, Yes, I get that he thinks prices won’t rise in the LT. But the controversy here is whether he assumes liquidity traps last forever. Are you sure he does?

    Perhaps he assumes that after a huge one-time rise in the base, the base gradually falls at a rate that creates such low NGDP growth that the liquidity trap lasts forever. That’s possible, but would not happen if the central bank were trying to boost AD. And when we consider QE we normally assume the central bank is trying to boost AD. And that’s where my post comes in, I think his asusmption of a stable inflation target is the key to making the whole thing work. The sine qua non.

  28. Gravatar of Mark A. Sadowski Mark A. Sadowski
    23. October 2013 at 11:53

    Scott,
    I think your questions are all answered in a companion paper. Page 43:

    “However, our dynamic state-contingent policy analysis raises an important question: Suppose LSAP are large enough to raise aggregate output; would tapering or unwinding LSAP completely undo these positive effects? As far as we know, little existing work to date has attempted to answer this important question. Our model allows us to study the optimal timing and pace of exiting LSAP. In a companion paper (Wen, 2013), we apply our model to study exit strategies and show that the result is surprising: Even though the dynamics of our model are symmetric around the steady state and non-permanent QE appears to be ineffective in mitigating …financial shocks, tapering or unwinding of LSAP does not necessarily undo the gains under LSAP, depending on the exit strategies pursued.”

    I have not been able to find that paper (““Optimal Exit Strategies”) yet.

  29. Gravatar of ssumner ssumner
    23. October 2013 at 11:59

    Mark, If “non-permanent QE appears to be ineffective” does that imply permanent QE is effective?

  30. Gravatar of Mark A. Sadowski Mark A. Sadowski
    23. October 2013 at 15:05

    Scott,
    “Mark, If “non-permanent QE appears to be ineffective” does that imply permanent QE is effective?”

    It does not appear to me that the possibility of permanent QE is ever considered. It’s also not clear to me why exiting QE is ever desirable. Keeping real interest rates low increases real growth so why exit as long as inflation expectations are well anchored? I assume that these questions are answered by the companion paper.

  31. Gravatar of benjamin cole benjamin cole
    23. October 2013 at 17:10

    Mark S—
    I hope you are reading this…great work on your part. I am stupified at this…either the St. Loo Fed is koo-koo or we should obviously do another $5 trillion of QE on US debt…

  32. Gravatar of dlr dlr
    23. October 2013 at 17:13

    I think the model only says that it is possible to have well anchored inflation expectations amid permanent QE, i.e. a permanent liquidity trap. Because it is really an RBC model with flexible prices, it considers this the preferred scenario, at least as opposed to one where inflation is increasing amid QE, because it allows the portfolio shifting effects of QE to last the longest without the incurring welfare reducing costs of inflation (which, along with rising NGDP, don’t have any positive impact on output). I think the model can have relatively high inflation (but not very higher nor hyperinflation) without also relieving or manipulating the borrowing constraints it sees as underlying the output gap.

  33. Gravatar of ssumner ssumner
    24. October 2013 at 07:41

    Mark, Well he says he considers permanent QE in the quote I provide at the top of my post.

    dlr, Fair enough, but don’t you think most readers would assume the paper was telling us something useful about actual recent QE moves? Indeed doesn’t the paper suggest that? And yet obviously the Fed has not decided to intentionally create a permanent liquidity trap, nor do the markets have that expectation. In that case it seems to me that this exercise tells us nothing about current policy. Where am I wrong?

    BTW, I don’t care here about the output gap or the RBC aspects of the model. It makes nominal predictions, that’s what I care about.

  34. Gravatar of dlr dlr
    24. October 2013 at 08:16

    dlr, Fair enough, but don’t you think most readers would assume the paper was telling us something useful about actual recent QE moves? Indeed doesn’t the paper suggest that? And yet obviously the Fed has not decided to intentionally create a permanent liquidity trap, nor do the markets have that expectation. In that case it seems to me that this exercise tells us nothing about current policy. Where am I wrong?

    BTW, I don’t care here about the output gap or the RBC aspects of the model. It makes nominal predictions, that’s what I care about.

    Although I agree with you, it isn’t a slam dunk argument. You say that the Fed hasn’t intentionally decided to create a permanent liquidity trap (although sometimes I wonder, if we redefine intention as including a collectively unconscious derived from the intentions of a heterogeneous group of individuals), but the model as I read it hasn’t quite said that either. In fact, the purpose of the credit easing mechanism the model suggest could well be described as the Fed trying (largely without success) to escape from the liquidity trap.

    All the paper needs to argue to claim relevance on this issue is that the Fed/Fiscal Authority (they are consolidated in the model, but that doesn’t matter at all here) is not trying to escape from the liquidity trap by increasing inflation (or its nominal target). This is something I would imagine you have some hidden sympathy for, given that you think the Fed can escape from the liquidity trap whenever it wants yet is not doing so and in fact is accepting below target inflation.

    Now, it’s true that market expectations really don’t look like a permanent liquidity trap is expected (although I think this might be a more complicated question than it seems if you are a multiple equilibria fan). But the paper is really about offering a way to estimate the extent to which the credit easing is supposedly contributing to an escape from the trap and a return to output trend, not whether we might otherwise escape the trap. The point (as I read it) that QE wouldn’t even have much more effect if it was permanent amid well anchored inflation expectations is just a typical contingent point about implications of the model (i.e. it’s not having a big effect, and wouldn’t even if it was permanent amid anchored inflation). They could equally have noted that permanent CE doesn’t work well in their model even when inflation rises appreciably (they sort of did), which is another implication of their model.

    It seems awkward to read permanent QE as being agnostic about whether this includes inflation, but his line of LSAP literature is so used to an RBC perspective (i.e. nominal aggregates are incidental) that I don’t think the typical reader would be so surprised by the wording.

  35. Gravatar of Mark A. Sadowski Mark A. Sadowski
    24. October 2013 at 09:24

    Scott,
    “Mark, Well he says he considers permanent QE in the quote I provide at the top of my post.”

    He mentions permanent QE at various points but nowhere in this paper does he go into much detail on the effects of permanent QE.

    Again, this paper (although it is over 60 pages long) reads like half of a paper (presumably the companion paper finishes the effort). He sets up the model, discusses the effects of LSAP and then concludes. In fact there does not appear to be a monetary policy rule of any kind.

  36. Gravatar of ssumner ssumner
    25. October 2013 at 05:55

    dlr, You say:

    “All the paper needs to argue to claim relevance on this issue is that the Fed/Fiscal Authority (they are consolidated in the model, but that doesn’t matter at all here) is not trying to escape from the liquidity trap by increasing inflation (or its nominal target).”

    Of course I totally agree with this, but interpret it very different that you. Suppose I said “A bus driver that wanted to go to LA but set out on a road that he knew went to Chicago, would be unlikely to end up in LA.” So far so good. Suppose I then argued “This shows that bus drivers are unable to drive their buses to LA.” Someone might object that the second claim does not follow from the first.

    If the Fed sets out to do QE in such a way as to avoid any increase in NGDP, then yes, I agree it is unlikely to boost AD. An undergraduate would be able to tell that simply by looking at an AS/AD diagram. But it’s implications? There are none that I can see.

    He’s looking at the problem from the wrong direction. You start by pegging a NGDP futures contract at the level of NGDP growth you’d like to see. Then you ask how much the central bank would have to purchase to make the peg stick. Do they run out of eligible assets? Do they have to buy risky assets? But playing games with the policy goal “they sort of want to move to A, but they really don’t want to move to A” is just nonsense, it only confuses the issue. Bernanke is doing QE precisely because he wants to unanchor inflation, at least a tiny bit.

    Mark, I think dlr is right, it’s the assumption of a permanent inflation anchor, low enough to create a permanent liquidity trap, that drives the result. In other words he assumes Japan pre-Abe. If I’m right it tells us nothing interesting about “QE” as a policy tool, but rather QE as a response to elevated demand for base money. I.e. the the subject of my new post.

  37. Gravatar of jknarr jknarr
    25. October 2013 at 08:10

    I might add that the monetary base is not unitary: reserves are different than currency. A permanent expansion of reserves might have zero consequences if the economy does not lend and borrow at zero rates: the trap.

    The ZLB liquidity trap may be purely a product of debt saturation — too much debt supply, not enough debt demand=zero rates.

    In the ZLB trap, we ought to expect a substitution out of reserves and into currency. Bank’s business models fail and their counterparty risk grows at zero rates; and there is less incentive to store funds in demand deposits at zero rates: at the ZLB, bank cash is negative-yielding, while currency remains zero-yielding.

    Shifting the base out of reserves and into currency is the entire monetary stimulus path at the ZLB!

    So why isn’t it working? I really think that there is huge room to discuss the (inflation-adjusted) disincentives for currency withdrawals in the US. If the bank secrecy act of 1970 had a cap of $10,000 in currency as the original sum, that should be some $63,000 in current 2013 dollars.

    Bottom line, there are massive regulatory barriers that prevent base money from affecting the real US economy: instead, currency flows abroad, where it has no bank regulation oversight, and has higher utility as a result.

  38. Gravatar of dlr dlr
    25. October 2013 at 08:42

    Of course I totally agree with this, but interpret it very different that you. Suppose I said “A bus driver that wanted to go to LA but set out on a road that he knew went to Chicago, would be unlikely to end up in LA.” So far so good. Suppose I then argued “This shows that bus drivers are unable to drive their buses to LA.” Someone might object that the second claim does not follow from the first.

    You say you don’t care that this is an RBC, flexible price model at core, but then you seem to ignore the key implication of this fact. In this paper, Yi Wen does not agree with you at all that the bus driver wants to go to LA. He thinks the Bus driver wants to go from Houston (AS too low) to Winnipeg (AS higher), and doesn’t much care whether he passes through LA (high NGDP which would mean higher prices but not higher output) or Chicago (anchored inflation) along the way, though he thinks that Chicago would be probably the more efficient route. It is you who are imposing the LA destination.

    But playing games with the policy goal “they sort of want to move to A, but they really don’t want to move to A” is just nonsense, it only confuses the issue. Bernanke is doing QE precisely because he wants to unanchor inflation, at least a tiny bit.

    I think you are unfairly projecting your personal purity of motive onto the Fed, despite the fact that you normally seem to think the Fed has been profoundly unclear about exactly which city is in its GPS. After all, you would be the first to agree that if the Fed wanted to go to LA, it absolutely could. Yet not has it not gone there, and in fact is pointing the wheel less Westerly than it would if it were already in LA! (sorry, attempt to say inflation is below 2% target despite unemployment). Not only that, but it is not at all clear that Bernanke *only* believes the QE works by increasing AD. He has also written and spoken about supply side aspects of credit constraints and frictions (including using some flexible price assumptions at times) in an RBC-like context. And within the Fed the mix between nominalists and RBC proponents is probably even murkier than the mix within Bernanke himself.

    If Bernanke really wanted to unanchor inflation, he could (well, the Fed could). If he wants to just “a tiny bit”, as you say, it makes perfect sense for someone (who thinks this tiny bit is worthless anyway) to write down a model assuming that LSAP is mostly designed to keep inflation anchored but increase AS through various weird mechanisms. Where in the paper does the phrase “Aggregate Demand” appear?

  39. Gravatar of ssumner ssumner
    26. October 2013 at 07:02

    dlr, I guess I just see things differently than you do. If one wants to operate in a RBC framework that’s fine, but people like Bernanke and I focus on AD, and we are going to be impressed by anything that comes out of that sort of model. It tells us nothing about why QE is ineffective. For instance, suppose QE was used to peg the price of NGDP futures at the Fed’s policy target, what would the model say about that policy? As far as I can see the answer is nothing.

    There are two issues here, can monetary policy determine the path of nominal variables, and do nominal shocks influence RGDP?

    The answer to the second question is yes, which is why most economists tune out RBC models. But as far as I can see the RBC models also fail to provide a good analysis of what determines the path of NGDP. There is a huge difference between QE that represents an endogenous response to shifts in base demand, when the central bank is targeting a very low NGDP growth rate, and QE that is aimed at boosting NGDP growth. I don’t see him making that distinction.

Leave a Reply