A necessary and sufficient condition

Josh Hendrickson makes the following observation:

Suppose that prior to the recession, I told you that I had a theory of the business cycle. My theory suggested that shocks to net worth were a significant explanation of the business cycle. Following a shock to net worth, consumption, investment, and hours worked would decline. Suppose that I also told you that increases in the monetary base by the Federal Reserve would keep inflation close to the implicit objective, but have little effect on consumption, investment, and real GDP. Given the events of the last four years, this theory seems to fit pretty well with what we have actually observed. . . .

Suppose instead that I told you that I had the following theory of the business cycle. My theory suggested that liquidity shocks were a significant explanation of the business cycle. Following a shock to liquidity, and without appropriate Federal Reserve policy, nominal and real GDP would decline and unemployment would rise. . . .

I have used these two examples to illustrate a couple of points. First, neither Bullard nor Sumner (or anybody else associated with similar views) are crazy. They have logically consistent ideas that are consistent with casual observation. Second, confirmation bias is dangerous. It is very tempting to have a theory, look at the world, observe events consistent with your theory, and conclude that your theory is correct. But that is just confirmation bias. What is necessary is to provide evidence to support your theory in light of other theories that have observationally equivalent observations. This is substantially harder to do “” even for those who realize it is necessary. Third, and perhaps most importantly, the ability to distinguish between these and other competing theories is incredibly important given the vastly different monetary policy implications.

This is a good observation, and it points to the need for not approaching macro with a blank slate, but rather coming in with a deep understanding of macroeconomic history.  What sort of models have consistently fit the facts, and which have failed to do so?

It turns out that the monetary approach to business cycles does an excellent job of explaining American recessions.  Even the 1974 “supply shock recession” (where data was distorted by the removal of price controls) saw NGDP growth slow from 11.7% to 7.7% (comparing 1972:3 to 1973:4 with 1973:4 to 1975:1.)

On the other hand the net worth model seems completely unable to explain the business cycle.  Where there seems to be a correlation (say after the 1929 stock market crash), the pattern is easily explained by reverse causation—perception of an oncoming depression led to lower equity prices.  In contrast, an almost identical stock market crash in 1987 had virtually no impact on GDP—indeed the economy boomed after the crash.  The difference of course is that monetary policy kept NGDP growing after the 1987 crash, whereas the Fed actually tried to reduce NGDP growth in 1930 by sharply raising the gold ratio (which is the only truly exogenous policy instrument under a gold standard regime.)

There’s a reason most economic forecasters were not predicting a recession as of June 2008; net worth models are useless.  The huge decline in house prices between 2006 would not be expected to cause a major recession, and indeed would not have caused one had NGDP not declined.

If Bullard had spent many years studying macroeconomic history he’d know that nominal shock models are incredibly powerful, especially in the US.  In contrast net worth models are essentially useless.  (Imagine a Venn diagram where two overlapping circles represent net worth shocks and NGDP shocks.  Recessions occur in the NGDP-only shocks, and the overlap, but not in the net worth only shocks.  Undoubtedly in other countries where supply shocks play a major role you can find exceptions, where an AS shock causes lower net worth without depressing NGDP.  But these models don’t seem to explain the US business cycle.  We are too diversified.

Sometimes when I read RBC economists they come across as almost taking pride in their ignorance of macro history, of wanting to approach the time series with their mind a blank slate.  Unfortunately, one must first know where to look.  And for that there is no shortcut for a deep immersion into macroeconomic history.  Robert Lucas’s familiarity with Friedman and Schwartz is what immunized him against the extreme forms of RBC theory.

PS.  In a previous post Josh made the following claim:

Optimal monetary policy within these frameworks is defined as the policy that minimizes fluctuations in utility around the steady state. By performing a second-order Taylor series expansion of the utility function around the steady state and some mathematical manipulation, it can be shown that the optimal monetary policy is one that minimizes the weighted average of deviations of inflation from its target and output from its “natural” level. (This, by the way, is contrary to the assertions by Scott Sumner and George Selgin that the welfare criteria in these models is ad hoc.) Woodford is obviously correct that in this context that flexible inflation targeting is the optimal monetary policy.

I never denied that inflation targeting was optimal in these models, but rather that there was no justification for the assumed welfare functions.  I am pretty sure that was also George’s criticism.

PPS.  The very first sentence in the post Josh links to said:

Many New Keynesians point out that there is a class of flexible inflation targeting rules that perform better (from a welfare perspective) than NGDP targeting.  This is true.

PPPS. In previous posts some commenters have desperately tried to find distinctions between 1929 and 1987 stock crashes.  Don’t bother; it’s a waste of time.  The graphs overlay almost perfectly.  One was followed by the Great Depression and the other by a boom with falling unemployment.  And I could find many other examples.


Tags:

 
 
 

35 Responses to “A necessary and sufficient condition”

  1. Gravatar of Saturos Saturos
    2. October 2012 at 07:31

    Spend one hour on Marcus Nunes’ blog and you’ll be disabused of the notion that anything else explains the facts anything like as well as MM does.

  2. Gravatar of Ryan Ryan
    2. October 2012 at 07:35

    The huge decline in house prices between 2006 would not be expected to cause a major recession, and indeed would not have caused one had NGDP not declined.

    In 2006, my net worth had significantly declined as a result of bank deleveraging in response to declining mortgage values. That same year, I attended a speech given by EDC economist Stephen Poloz, in which he warned us that the coming recession would produce conditions that wouldn’t normalize for something like ten years.

    I don’t remember his being in the minority POV at the time. In fact, he is about as mainstream an economist as Canada has produced. (And an excellent one, I might add.)

    Do we just hang out in different circles, or do we just remember 2006 differently?

  3. Gravatar of Greg Ransom Greg Ransom
    2. October 2012 at 07:37

    Hayek’s monetary theory / macroeconomics subsumes both of Josh’s models — offering explanatory power and unification of the sort which is valued in science, the sort of power & unification which Darwin brought to biology and which is valued by scientists in that domain.

    We need to explain what sociological forces and mechanisms of self-interest are encouraging economists to reject explanatory power and unification.

  4. Gravatar of Saturos Saturos
    2. October 2012 at 07:42

    Ryan, and why did Poloz think that people would work less if they were poorer?

  5. Gravatar of Ryan Ryan
    2. October 2012 at 07:59

    Saturos, I did catch that part of Sumner’s post. It’s a good point, for what it’s worth. I won’t speak on behalf of Poloz, because he is about 1,000x the economist I am.

    What I will say is that my personal experience with a sudden shock to my net worth shrunk my investment portfolio. If it happened to me, then it seems likely that it happened to others, too, and if such a thing can occur to an entire nation’s investors in aggregate, NGDP will fall without its being a story about employment at all.

    I will also suggest that when businesses experience a net worth shock, they often scale back on new projects at the margin. Through that process, employment rates can fall as contracts expire, if there are no new contracts to replace them.

    As you rightly point out, it’s seldom that people don’t want to work in response to tough times.

  6. Gravatar of Bill Woolsey Bill Woolsey
    2. October 2012 at 08:26

    Perhaps I don’t understand, but I believe that RBC models involve reduced productivity.

    Suppose the finanical system works less effectively. Some people consume more of their income now when it would be better used by other people now. Credit markets fix that with consumer loans.

    If you are able to measure this loss in utility in your model, this reduces real real income.

    Some of those who lend less now might work less. That is, they were working to save and saving to lend. Because finacial markets work more poorly, they work less, save less, and lend less. Less labor input, less current output.

    Suppose some firms retain earnings and rather than invest internally lend to other firms that can finance more productive projects. When the finanical system works poorly, firms use their retained earnings on less productive internal projects. Firms that have great opportuntities but no funds cannot exploit them. With less productive investment on the whole, total production falls.

    With less total production, it is certainly possible, and even likely that both consumption and investment fall. And if people who were working to save to lend are important, then employment might fall. (Of course, those who would have borrowed might want to work more to maintain consumption now that they cannot borrow. But we know this is unimportant, so we tune the model to make that effect small.)

    What does this have to do with net worth? People with higher net worth can borrow more because lenders are more confident they will be repaid. With less net worth, lenders are less willing to lend. And so, credit markets work less effectively. And then we have less efficient allocation of resources as above.

    Sumner instead focuses on people having less net worth so they work and save more to build it back up. This might require a reallocation of resources (fewer houses and more machines and equipment being built.) Nearly everyone else assumes that people spend less, so nominal and real expenditures fall, and so production and employment fall. Market Monetarists insist that with a proper monetary regime, there is no decrease in nominal expenditure, and just the reallocation of resources. But RBC types reject not only the Market Monetarist approach, but the conventional wisdom too.

    But if lower net worth means more saving and work, then result is less consumption but more investment and employment. Well, that explains less consumption, but we know that investment really fell and employment fell. This effect must be smaller than effect of lower net worth on the workings of credit markets. The model needs to be tuned so that people working more and saving more to rebuild net worth are small.

    The notion that total spending fell, and prices and wages contintued to rise, so real expenditure, real output and employment fell cannot be right. Why? Because prices and wages always adjust to clear markets. Rising prices and wages where there is a surplus of output and labor are inconsistent with the models.

  7. Gravatar of Felipe Felipe
    2. October 2012 at 08:26

    But these models don’t seem to explain the US business cycle. We are too diversified.

    Does this mean that the policy prescription (NGDPLT) might not apply for countries that are less diversified?

  8. Gravatar of Head Stomp Head Stomp
    2. October 2012 at 09:14

    It turns out that the temperature approach to human sickness does an excellent job of explaining my history of sickness. Even the 2006 “winter of misery” (where data was distorted by having no heating) saw body temperature rise from 198.6 to 102.3.

    Or maybe increasing body temperature doesn’t explain sickness and instead serves to fight off disease by reducing pathogen colonization; you know, kind of similar to how slowing NGDP growth after an unsustainable boom doesn’t doesn’t explain business cycles and instead serves to fight off depression by reducing capital wasting.

  9. Gravatar of Doug M Doug M
    2. October 2012 at 09:22

    RBC economists? Royal Bank of Canada? They always seemed pretty mainstream to me.

  10. Gravatar of Don Geddis Don Geddis
    2. October 2012 at 09:29

    @HeadStomp: but surely these causality hypotheses are easy to distinguish. Do you show symptoms of sickness prior to your temperature changing, or only afterwards? If you start getting sick, and we artificially force your temperature to remain stable, does your sickness get worse, or does it recede?

    In macro: what happens when there is an “unsustainable boom”, but NGDP growth doesn’t slow down? Does the recession happen anyway? Is it worse? Do the recession symptoms (e.g. unemployment) precede, or follow, the slowdown in NGDP growth?

    If you examine the historical data, you’ll find that your theory of an analogy between sick/temperature and recession/NGDP slowdown doesn’t hold. Your proposed causal theory is false.

  11. Gravatar of Sumner on Predicting the Great Recession Sumner on Predicting the Great Recession
    2. October 2012 at 09:49

    […] Scott Sumner writes: There’s a reason most economic forecasters were not predicting a recession as of June 2008; net […]

  12. Gravatar of Head Stomp Head Stomp
    2. October 2012 at 09:51

    Don Geddis: Symptoms of sickness precede body temperature rises and symptoms of business cycles precede falling NGDP. I’ve taken cold showers while sick and neither the sickness nor the fever went away. Maybe I should have used an ice bath until my temperature went down… and if that didn’t work then a dry ice bath… and if that didn’t work then a liquid nitrogen bath because clearly all that matters is forcing down my temperature.

    In macro: You kick the can down the road until a crack-up boom happens(or efficient fusion and replicators are invented, but I’m not holding my breath)?

    Having examined historical data, I believe MMs are, to steal the word choice of Murphy, kind of insane. Sure, not MMT insane, but still.

    Honestly, is the temperature/sickness metaphor so bad? It even leaves open room for an argument that at some point you have to force down temperature to prevent your brain getting fried like at some point you have to force up NGDP to prevent secondary deflation/depression.

  13. Gravatar of Don Geddis Don Geddis
    2. October 2012 at 10:21

    @HS: You can use drugs to force your elevated temperature down. But you’ll find that, on average, your sickness lasts longer if you do that.

    In 1987, there was a huge stock market crash. But NGDP growth remained stable. And there basically never was any effect on output or unemployment (for a good decade or two!). Why?

    US housing started crashing around 2006. Subprime mortgages around 2007. Again, no output or unemployment crash until NGDP plummeted in late 2007 / early 2008. Why was there no recession a year earlier?

    Your metaphor is bad, because in health, elevated temperature actually attacks the root causes of the sickness, the invading germs. So it helps to heal. But in macro, your theory seems to be that a recession (and lower NGDP growth) somehow “helps” a misallocated economy, somehow restructure. And yet all the historical evidence suggests exactly the opposite, that nominal recessions are pure self-inflicted pain, with all harm and no benefit.

  14. Gravatar of ssumner ssumner
    2. October 2012 at 10:54

    Filipe, Yes.

  15. Gravatar of Nic Johnson Nic Johnson
    2. October 2012 at 10:59

    I’m taking a class with Tim Kehoe right now, undoubtedly one of the most extreme RBC types. He has a book with Prescott (no need to mention how RBC he is) about macroeconomic history around the world through the lens of an RBC-type model: http://www.amazon.com/Great-Depressions-Twentieth-Century-Timothy/dp/0978936000

    Does that not count as macroeconomic history?

  16. Gravatar of marcus nunes marcus nunes
    2. October 2012 at 11:25

    And here´s a bit of evidence against “Net Worth” driven models:
    http://thefaintofheart.wordpress.com/2012/09/21/it%C2%B4s-all-in-the-timing-it-was-not-the-financial-crisis-but-the-drop-in-ngdp-that%C2%B4s-responsible-for-deleveraging/

  17. Gravatar of Peter N Peter N
    2. October 2012 at 11:37

    The difference between 1929 and 1987 as you say is the attitude of the federal reserve. I posted something here the other day on this, but it was at the end of the thread, and I doubt many people saw it given the number of articles you’ve been putting up, things get buried fast.

    So I’ll just repeat a small portion and give the link to the rest. The comment really fits better here than where I put it.

    “{Adolph C.] Miller (1935: 453) claimed critically that Strong’s policies in the presence of stock market speculation ‘proved to be unequal to the situation . . . in this period of optimism gone wild and cupidity gone drunk.’ The Federal Reserve Board’s ‘anxiety,’ he continued, ‘reached a point where [the Board] felt that it must assume the responsibility for intervening . . . in the speculative situation menacing the welfare of the country.’ Consequently, on February 2, 1929, the Board sent a letter, crafted mostly by Miller, to all the Fed Banks stating that the Board had the ‘duty . . . to restrain the use of Federal Reserve credit facilities in aid of the growth of speculative credit.” To accomplish this end, the Board initiated ‘he policy of ‘direct pressure’ [that] restricted borrowings from the federal reserve banks by those member banks which were increasingly disposed to lend funds for speculative purposes’ (p. 454).”

    http://www.themoneyillusion.com/?p=16610#comment-189152

    This is a change of policy due to the death of Benjamin Strong changing the voting balance at the Fed. Strong was an advocate of price stabilization – anathema to the Miller crowd.

    [N.B. that this stable price policy seriously distorted the traditional gold standard and its unwinding by Miller would be painful even if he stopped at the traditional gold standard, which he didn’t, instead distorting it even more in the opposite direction. It would be interesting to try to measure the cost of this whipsaw]

    The market went soft a few months after the Feb 2 move and the crash was inevitable as people trying to support the market were working harder and harder for smaller and smaller effects. It’s a classic pre-crash pattern.

    The first part of the crash was like 1987 though more protracted, but each rally was followed by a fall to a still lower level. The two charts part company at the 9 month point. One goes up almost monotonically for 3 years and the other goes down in similar style for 2 years.

    I believe this is incorrect:

    “sharply raising the gold ratio (which is the only truly exogenous policy instrument under a gold standard regime.)”

    The Fed had another tool as mentioned above, “direct pressure” – jawboning shut the discount window.

    “A bank not able to pass the “direct-pressure test” could
    not borrow from a Federal Reserve Bank at any rate, no matter how much “good” paper it had or how badly it needed “credit” to meet deposit withdrawals. To such a client bank, discretion by authority substituted for the Fed discount rate to ration Federal Reserve “credit.”

    But the Fed gold policy was a disaster:

    “By August 1931, Fed gold had reached $3.5 billion (from $3.1 billion in 1929), an amount that was 81 percent of outstanding Fed monetary obligations and more than double the reserves required by the Federal Reserve Act [which the Fed had the authority to waive, anyway]”.”

    The net worth theory has a serious problem. It is a theory based on an aggregate. In such cases, there is always the possibility that either you are aggregating away what you want to observe (perhaps the relevant distribution is bimodal, for instance) or that you have an effect analogous to the sum of the products not being the product of the sums (like aggregating nonhomothetic preferences).

    Here, however, you overstate your case or at least use an a questionable hypothetical:

    “The huge decline in house prices between 2006 would not be expected to cause a major recession, and indeed would not have caused one had NGDP not declined. ”

    I posted a link to a study a while back that concluded that the bursting of a housing bubble is about 3 times as damaging as the bursting of a stock bubble. There are any number of reasons this should be true. For instance most stock is in institutional hands and most of what isn’t is held by high net worth individuals. The demographics of housing are quite different.

    Also, you can’t really disentangle the housing bust from the damage to collateral liquidity from the revulsion toward securities based on non-standard mortgages (sub-prime and Alt A, negative amortization, balloon payments). As the prospect of conventional mortgages going under water increased and doubts grew as to the diligence of the ratings agencies, the revulsion spread. Repo collateral may be only a low opportunity cost form of inside money, but there was trillions of dollars of it and it had a very high velocity. When a form of broad (low opportunity cost, inside) money commonly used for some class of transactions is unavailable, either the transactions must cease or the parties must substitute a narrower form of money (more liquid, with a higher opportunity cost). I would expect this to propagate tightness up the money quality ladder.

    There is also a reduction in velocity due to a lower ratio of rehypothecation. And there is the classic tightening effect of wholesale unwinding of leveraged positions. You have to expend your best assets to support the unwinding. Nobody wants what has become trash through revulsion. This is the Lehman scenario.

    And, of course if credit is created during leveraging up, it is destroyed by leveraging down. We’ve seen in Europe how bad this can get. The ECB has the banks on life support. If they weren’t in the ECB’s ICU, most of them would be in receivership.

    The ECB is even accepting self-bonds (IOU’s, in effect) as collateral, and supporting a huge sovereign debt carry trade based on certain peculiarities of banking rules (sovereign Euro denominated debt can be treated as risk free when calculating assets at risk, so if it pays more than the ECB charges for loans….). This is an extraordinarily dangerous and potentially expensive free lunch. It has a encouraged marriage of a sovereigns with their banks, and exploits a loophole in the ECB charter prohibition of directly funding sovereign debt.

    Now the Fed comes back into the picture. Regardless of whether its prior behavior was defensible, once there was any sign that revulsion was so severe that there was a loss of faith in major banks as credit worthy counterparties, the Fed had to intervene. It should never have allowed the situation to deteriorate to the point that the commercial paper market seized up. At that point the damage had spread to the real economy in a serious way.

    A housing bust in a mortgage market financed the old fashioned way, might very well have been much more limited, but the Wall St. crowd created a massively correlated group of securities while getting them AAA ratings by over-collateralization, the sufficiency of which was calculated assuming uncorrelated risk.

    This is the same sort of stupidity that destroyed Long Term capital management a few years earlier – underestimating systemic risk.

    That these hotshots could then offload the risk onto the GSEs and AIG, and speculate against the securities they were selling made the whole thing a free lunch (for them. We paid for it).

    [It seems parenthetically that the mills of the attorney general of the state of New York grind slowly, but they do, in fact, grind. We’ll see about the exceeding fine part.]

    Where net worth seems to show up is in interest rates. Faced with losses and the prospect of more and faced with a huge shortage of safe assets (from all the revulsed mortgage securities and sovereign debt), money seeks safety in what remains, and we get massive distortions in interest rates.

    This looks like another expensive free lunch, though it’s not clear to me who gets to eat, who pays, when and how much.

  18. Gravatar of Head Stomp Head Stomp
    2. October 2012 at 12:04

    Don Geddis: Do you get that we define sickness in terms of the economy differently? To me, the sickness is the intertemporal discoordination investment; whereas, it appears you only consider the symptoms of recession/depression to be sickness. So, given usage of economic sickness, much like how targeting body temperatures with antipyretics can make an infection last longer, so too can NGDP targeting make the intertemporal discoordination of investment last longer.

    I don’t doubt that the can can be kicked. Do see how that historical anecdote cannot falsify either position?

    Why would I assume everything would happen at once? That isn’t the case with sickness either. That certain symptoms tend to precede or follow other certain symptoms is, I would think, not controversial nor damaging to either position.

    I think you are wrong but I wouldn’t be so naive or disingenuous as to say that all historical evidence suggests so. Hidden variables are a bitch.

  19. Gravatar of Peter N Peter N
    2. October 2012 at 12:31

    Consider this an addendum to my previous. It’s an IMF working paper on the effects of the loss of $4 trillion or so of collateral.

    “Large dealers are incredibly adept at moving collateral they receive that is pledged for re-use. The “velocity of collateral”””analogous to the concept of the “velocity of money”””indicates the liquidity impact of collateral. A security that is owned by an economic agent and can be pledged as re-usable collateral leads to chains. Thus, a shortage of acceptable collateral would have a negative cascading impact on lending similar to the impact on the money supply of a reduction in the monetary base. Thus the first round impact on the real economy would be from the reduction in the “primary source” collateral pools in the asset management complex (hedge funds, pension and insurers etc), due to averseness from counterparty risk etc. The second round impact is from shorter “chains”””from constraining the collateral moves, and higher cost of capital resulting from decrease in global financial lubrication.”

    http://www.imf.org/external/pubs/ft/wp/2011/wp11256.pdf

  20. Gravatar of Morgan Warstler Morgan Warstler
    2. October 2012 at 13:45

    Fixed it….

    “It is very tempting to have a theory, look at the world, observe events consistent with your theory, and conclude that your theory is correct. But that is just confirmation bias. What is necessary is to provide evidence to support your theory… WITHOUT ANY OTHER VARIABLES.”

    We have this theory called NGDPLT, and NOBODY can / will explain how and why it works without any make-up.

    Lars has written that he thinks it is more important than make-up.

    But Scott hasn’t explained why NGDPLT will work starting today at 4.5% with no make-up.

    Either they theory works ON ITS OWN or it doesn’t.

    Anyone? Buehler?

  21. Gravatar of Gene Callahan Gene Callahan
    2. October 2012 at 14:04

    “It is very tempting to have a theory, look at the world, observe events consistent with your theory, and conclude that your theory is correct. But that is just confirmation bias. What is necessary is to provide evidence to support your theory… WITHOUT ANY OTHER VARIABLES.”

    What the heck is that supposed to mean? Your theory can only use one variable?! Or that you can’t use variable from “outside the thoery”? But if you’re using them in your explanation, they are not outside the theory, are they!

  22. Gravatar of Matt Rognlie Matt Rognlie
    2. October 2012 at 15:06

    I never denied that inflation targeting was optimal in these models, but rather that there was no justification for the assumed welfare functions. I am pretty sure that was also George’s criticism.

    This statement doesn’t make much sense to me. Before it sounded as if you were complaining that the quadratic loss functions in New Keynesian models are ad-hoc. As Josh explains, this isn’t really true — the quadratic loss functions are derived as second-order approximations to the “true” welfare function in the microfounded model.

    And if you’re complaining that there is “no justification” for the parameterization of utility functions and production functions in the underlying model, then I think you’re still off base. The exact choice of these functions will not matter very much for the basic structure of the quadratic loss function (i.e. (x-xbar)^2 + c*pi^2, where x is the output gap, xbar is the efficient output gap, pi is the inflation rate, and c is some constant), because the internal logic behind the quadratic approximation to welfare is quite simple and will work with virtually any reasonable utility and production parameterizations.

    Basically, there are two key inefficiencies that are affected by monetary policy in the model: first, you can be producing the wrong total amount of goods, and second, you can be producing the wrong kind of goods.

    The first happens because the central bank sets demand at a level such that, from a social perspective, MPC*MPL does not equal -MUL: the marginal utility from having people work more and produce goods for consumption is not equated with the marginal disutility from labor. People are working too little or too less. This can be captured in terms of the “output gap” from the non-inflationary steady state of the model, with the caveat that the efficient level of the output gap may be nonzero due to monopolistic (or other) distortions in the economy. Anyway, in a second-order approximation to welfare, this distortion will produce a term that looks like (x-xbar)^2 pretty much no matter what functional forms you choose for utility and production.

    The second inefficiency, that we are producing the wrong kind of stuff, arises because there is unnecessary variance across goods in the ratio of prices to marginal costs. This arises from the combination of sticky prices and inflation — since firms are not constantly resetting their prices to the optimal levels, in the cross section prices will be somewhat out of alignment with their “correct” levels. This distortion, too, is pretty much inevitable regardless of the specific parameterizations of utility and production. You will get a term of the form pi^2 in the quadratic loss function.

    (Of course, the parameterization you choose will have a big impact on the relative weights of the (x-xbar)^2 and pi^2 terms in the loss function, and on the value of the xbar term, if there is any. And there is some room for arbitrariness here. But it sounds like your quibble is more because you think that there is no qualitative basis for this kind of loss function, which as discussed above isn’t really true.)

    All this is not to say that these quadratic loss functions are reasonable representations of what policy should try to achieve. But the problem there isn’t that the function is “ad hoc”, or that the underlying assumptions have “no basis”, it’s that the underlying model isn’t rich enough to capture the real losses from unemployment and inflation. In particular, the basic model doesn’t have involuntary unemployment, which is arguably the key reason that recessions matter — workers aren’t being nudged along their labor supply curve, they’re being thrown completely off it. The damage from a negative output gap is much more serious in the real world than it is in the basic New Keynesian model. The same is true for the pi^2 term. Although price dispersion might play some role in the losses from very high inflation, and it is an interesting metaphor for the general disruption of prices caused by inflation, it surely does not quantitatively account for the harms from inflation that most people think exist.

    But again, it isn’t that the model is “ad-hoc”, it’s that it fails to include the other imperfections (e.g. long-term nominal contracting) that make inflation truly a big deal. The problem is that if you want to improve the model, you need to have a clear idea about what those other imperfections really are.

  23. Gravatar of ssumner ssumner
    2. October 2012 at 16:43

    Peter, I agree with some of your comments, but a few suggestions:

    1. Fewer discount loans is equivilent to a gold ratio increase.

    2. Anything that reduces velocity tends to lead to tighter money, unless offset by the Fed.

    Matt, I don’t understand your comment. If I don’t think the model of the economy is correct, I obviously won’t have much faith in the welfare implications. The main problem with nominal shocks is not that the wrong type of output gets produced because inflation is variable (a cost I view as trivial) it’s that the labor market gets distorted because NGDP is unstable (actually because average nominal hourly wages are unstable.) Mankiw show that if wage stickiness is the main problem that a nominal wage target will often be superior to an inflation target. That’s also my view. The welfare implications of a model flow from the assumptions built into the model.

    BTW, All models should be “ad hoc,” I have no problem there. All my models are ad hoc.

  24. Gravatar of Felipe Felipe
    2. October 2012 at 17:16

    Scott, thanks for your reply. If I may continue on the theme: Why does NGDPLT require that an economy be diversified in order to be desirable? In other words, what problems arise with NGPLT if the economy is highly concentrated in an industry (say, a commodity exporter)?

  25. Gravatar of Peter N Peter N
    2. October 2012 at 21:16

    Scott, I agree that fewer discount carried to the extreme loans will result in a gold ratio increase. My points were that there are other ways they could have increased the gold ratio. The increase wasn’t their primary goal.

    Miller was a proponent of the real bills doctrine. He was determined to stamp out all lending for purposes of speculation. Any bad effects of this policy were the unfortunate but necessary cost of curing the evils of speculation. And, unlike Strong, he felt a policy of price stability was meddling of a sort the Fed should never do.

    I forgot to mention as a partial cause of the crash the Fed’s raising the rates on call money.

    “2. Anything that reduces velocity tends to lead to tighter money, unless offset by the Fed.”

    Absolutely, and the velocity of the repo market was quite high (a security could be rehypothecated 3 or 4 times and the low haircut based on the low information cost of the AAA repo collateral, made loans of a single day profitable. If the securities required anything other than trivial diligence, this wouldn’t be possible, but in the repo market diligence was concerned with the quality of the counter-party, not the security, which was effectively generic).

    There are those posting here in the past who have made a hard distinction between money and credit that would define Repo collateral as credit and thus NOT money. This in some way I don’t understand would mean that the destruction of $4 trillion dollars of it wouldn’t cause a tightening of “real” money.

    It seems obvious to me that if the transactions supported by the collateral are to continue, some other medium of exchange must be used, even if it has a higher opportunity cost, and that any transactions that had to be forgone would also cause a shortness of funds through the credit destroying effects of deleveraging.

    Whether a shortage of credit and a fall in financial asset prices is tight money or not, it’s still extremely damaging, when it affects everyday corporate funding.

  26. Gravatar of Major_Freedom Major_Freedom
    3. October 2012 at 00:13

    Wait, did I just read that correctly? The stock market collapsed in 1929 because investors perceived an upcoming Great Depression during the next decade and a half?

    This is not analyzing competing theories, it just hand waving at them.

    The stock market collapse of 1987 was followed by accelerating inflation, which postponed the inevitable recession until 1991. But the collapse on assets values are correlated with recessions, if you take into account a time delay factor.

  27. Gravatar of Bill Woolsey Bill Woolsey
    3. October 2012 at 03:57

    Matt:

    I think Scott has no real problem with seeing output gaps as bad.

    Market Monetarists don’t believe that all deviations of the inflation rate from target are bad.

    Returning the price level to its trend growth path after a deviation of nominal GDP from target is good. A shift in the inflation rate due to an aggregate supply shock is sometimes and somewhat good.

    It isn’t that the inflation is always bad, but we just have to put up with it to avoid excessively large output gaps. It is that not all inflation is bad. It is sometimes good–equilibrating.

    On the other hand, changes in the inflation rate due to shifts in spending away from target are bad. Whether a Calvo type model tells us much about why–buying too much of goods whose prices haven’t risen yet and too few of things whose prices have risen–is a bit doubtful.

    Scott:

    Do you really think the problem is “instability” of nominal wages? I think that the problem is that the trajectory of nominal wages is too stable. It fails to fluctuate enough to clear labor markets, and so a monetary regime that requires lots of adjustments in nominal wages to clear labor markets is bad relative to one that requires fewer such changes.

  28. Gravatar of ssumner ssumner
    3. October 2012 at 05:16

    Filipe, Suppose the price of oil suddenly doubled in Saudi Arabia. Would you want to drastically shrink the non-oil economy?

    Bill, I should have been more specific. If wages are sticky, then an unstable aggregate wage level shows that many wages are out of equilibrium. If you target aggregate nominal wages, then most wages will be close to equilibrium.

    Nic, I can’t comment, as I haven’t read that book.

  29. Gravatar of Felipe Felipe
    3. October 2012 at 07:58

    Scott: OK, that makes sense. As I understand it, the problem is when the economy is overly reliant on a single export, because the central bank has very little influence in the world demand of said export. However, I’m confused about which monetary policy regime would be better under these circumstances.

    * NGDPLT has the effect of choking the “inside” economy when the world demand is booming, and (possibly) overheating it when world demand is weakening.
    * Inflation targeting suffers from the same problem.
    * Exchange-rate targeting would cause lots of inflation during world demand booms, and may cause deflation during world demand weakening.
    * Some commodity standard. I think this would have similar effects to exchange rate targeting, not sure.

    Any other options?

    It seems to me that any sudden, large and positive shock to exports will tend to cause inflation and a shrinkage of the other sectors of the economy. NGDP targeting seems like the only one that lets the economy adjust by itself to the shock.

  30. Gravatar of Matt Rognlie Matt Rognlie
    3. October 2012 at 10:04

    Scott,

    I agree that the welfare loss term pi^2 from inflation arising from relative price distortions in the New Keynesian model is trivial. But the channel you’ve mentioned — instability in the labor market (and therefore output) arising from the same variations in monetary policy that are driving pi — is already in the New Keynesian model, in the (x-xbar)^2 term.

    This is, indeed, the whole idea behind what Blanchard and Gali (somewhat sarcastically, I think) called the “divine coincidence” of the New Keynesian model — the fact that even if the “relative price distortion” channel for the losses from inflation is absolutely trivial, so that the term pi^2 nearly drops out, it’s still optimal to target inflation because that’s also what results in stabilization of employment and output as well. There are many ways to break this result — for instance, if you enrich the model with shocks to a flexible-price commodity sector, the welfare calculations tell you that you absolutely should “accomodate” much of the resulting inflation. (In other words, it is appropriate for policy to make the kind of “headline/core” inflation distinction currently made by the Fed.) But it is a remarkable fact about the baseline model.

    The seminal paper introducing wage rigidities in the New Keynesian literature, Erceg, Henderson, and Levin (2000), tells us that in the presence of both price and wage rigidities, we should target a weighted average of price and wage inflation, with the relative weights depending on the amount of stickiness in each price. (If wages are much stickier than prices, then we should mainly target wage inflation.) This is in the context of a model that produces a slightly richer form of the quadratic loss functions that you’ve lambasted — see, for instance, equation (30) in the article linked above, where there are now three terms corresponding to the variance of the output gap, the variance of price inflation, and the variance of wage inflation.

    I guess I just don’t see what your complaint is about this aspect of the models. You seem to be reaching the same conclusion as the models, and qualitatively your reasoning is the same as the internal logic of the models — i.e. that the most serious consequence of nominal variability is the effect on real variability.

  31. Gravatar of ssumner ssumner
    4. October 2012 at 06:16

    Felipe. Perhaps the price of locally consumed goods. Or a local hourly wage index.

    Matt, You said:

    “The seminal paper introducing wage rigidities in the New Keynesian literature, Erceg, Henderson, and Levin (2000), tells us that in the presence of both price and wage rigidities, we should target a weighted average of price and wage inflation, with the relative weights depending on the amount of stickiness in each price. (If wages are much stickier than prices, then we should mainly target wage inflation.) This is in the context of a model that produces a slightly richer form of the quadratic loss functions that you’ve lambasted “” see, for instance, equation (30) in the article linked above, where there are now three terms corresponding to the variance of the output gap, the variance of price inflation, and the variance of wage inflation.”

    That sounds right to me. I have generally assumed that wage stickiness is the big problem (partly based on my own research) and hence a wage target is optimal. Since a wage target seems politically infeasible, I picked NGDP (per capita) as a second best proposal. NGDPLT also seems to be Woodford’s second best policy—so we agree!

  32. Gravatar of Major_Freedom Major_Freedom
    4. October 2012 at 06:38

    ssumner:

    Matt, You said:

    “The seminal paper introducing wage rigidities in the New Keynesian literature, Erceg, Henderson, and Levin (2000), tells us that in the presence of both price and wage rigidities, we should target a weighted average of price and wage inflation, with the relative weights depending on the amount of stickiness in each price. (If wages are much stickier than prices, then we should mainly target wage inflation.) This is in the context of a model that produces a slightly richer form of the quadratic loss functions that you’ve lambasted “” see, for instance, equation (30) in the article linked above, where there are now three terms corresponding to the variance of the output gap, the variance of price inflation, and the variance of wage inflation.”

    That sounds right to me. I have generally assumed that wage stickiness is the big problem (partly based on my own research) and hence a wage target is optimal. Since a wage target seems politically infeasible, I picked NGDP (per capita) as a second best proposal. NGDPLT also seems to be Woodford’s second best policy””so we agree!

    While you’re busying agreeing, I suggest you also look at the effect of inflation itself on price stickiness, rather than considering inflation as a cure without any negative consequences such as the very problem you’re trying to work around.

    Does inflation increase price stickiness? Does it exacerbate the stickiness of certain prices over others?

    It is well known that wage earners are risk averse, and prefer fixed incomes over variable incomes. Are wage earners therefore particularly sensitive to the risks of inflation such that they price in a premium to offset inflation even when the existing monetary conditions are deflationary and would otherwise see wage prices fall if full employment is to be had?

    If so, would inflation used to “fight” wage stickiness then be a case of a dog chasing its own tail, where the more inflation is used to deal with wage stickiness, the more sticky wages become, which then leads to perceptions of an increased value of inflation?

  33. Gravatar of Saturos Saturos
    4. October 2012 at 06:49

    MF, since you’re so smart, why don’t you publish a journal article arguing this point, instead of taking up comment space on this blog?

  34. Gravatar of Major_Freedom Major_Freedom
    4. October 2012 at 08:15

    Saturos:

    MF, since you’re so smart, why don’t you publish a journal article arguing this point, instead of taking up comment space on this blog?

    I don’t think I am so smart.

    If you feel so intimidated, why don’t you visit a psychiatrist instead of taking up comment space on this blog?

    Seriously, do you honestly believe that to be a proper commenter, I have to agree with its blog owner? It is HEALTHY to have varying viewpoints, and no, I don’t mean varying in terms of how many times you want Bernanke to hit CTRL+P, I mean true varying viewpoints that go to the fundamentals. If that pisses you off, if you desire an echo chamber, if you desire no serious debates, only debates around the edges, which never address your premises and keeps them nice and cozy, undisturbed, so that you feel safe, then I recommend you man up, so to speak, and not be so down on yourself.

    PhD economics departments in this age to a large extent filter out economists who are pro-market, because in this age economics is heavily influenced/managed/controlled by the state. The state finances most economics departments around the country, because to be frank, economics is just not that highly valued in the market, and the state needs court intellectuals to intellectually justify its naked aggression against its victims. I think economists deep down know this, and for many, it is insulting and intolerable to be told their profession is by and large worthless in the market, so much so that the combination of this, along with the filtration process alluded to above, has resulted in a steady stream of economics grads who are by and large anti-capitalist and pro-statist, who buy into the aggression and social control paradigm. I went through it. All my colleagues went through it. You probably went through it. Sumner went through it at Chicago, the economics department of which was started by the central banking establishment and continues to produce monetarists to this day.

    I’ll never forget this one experience I had in grad school when my colleague and I were having a lively discussion in the hallway, and one of our econ profs joined in, and for quite a while it looked like everything was fine. But then, my colleague (whom I greatly influenced to be more free market oriented) made a statement concerning how he thinks we should eliminate taxes on X, because the tax is generating Y undesirable outcome, and then he used the term “free markets!”. I then noticed that my prof got VERY nervous, he blushed, he started looking around as if he was being watched, and then he abruptly bolted, literally saying “I can’t be seen with you like this.”

    This is anecdotal of course, but it is a microcosm of how economics departments operate around the country. The reason why we see so little arguments against inflation as an institution (there are many papers that talk about “excessive” taxation and inflation, and “optimal” inflation rates and “optimal” taxation allocation, and whatnot, but hardly ever arguments against taxation and inflation per se) is because inflation is a tool of state power. Without it, the state is far more more limited in what statesmen can do. Invade a country in an unpopular war that the public will not want to finance via raised taxes? The state can just print off the dollars it needs to pay the difference. That’s exactly what happened for the Iraq invasion. Did you know that the NY Fed secretly (at the time) sent $40 billion to Iraq 2003-2008 to finance the invasion? This information would never have been revealed if it weren’t for Ron Paul’s (watered down) audit the Fed bill, which monetarists scoff at for obvious reasons.

    Inflation is not a tool for the people. If it were, aggressive force would not be necessary to maintain it. But aggressive force is being used against the people, via taxation laws which creates an artificial demand for dollars. That should prove to you who is actually benefiting from inflation. It isn’t Joe Schmo who is unemployed, contrary to the desperate pleas from inflationists that it is.

    Since self-interest is universal, even for statesmen, who want to expand and grow their operations, challenges to inflation as an institution are met with expected hostility and scorn from entrenched economists, and so we see accusations of dogmatism, ideology, and other vile verbiage thrown at those who dare argue that individuals should be free to opt out of it if they so choose.

    Some monetarists sleep at night believing that inflation is inevitable, so let’s deal with it like we deal with acts of nature. Let’s pretend that we cannot get rid of choices of violence, and believe violence is deterministic. How did this happen? Philosophical corruption, which is why I insist on including philosophy and ethics, the backbone of all human behavior.

    I know you would rather just take NGDP for granted every friggin day, and just talk about who said the term “NGDP” and then giggle like little schoolgirls because someone likes you. But I am here to shake things up and piss you off. Not because it is personal, but because that is the only way to wake sleeping people up. You can’t wake them up by blowing into their ears and whispering sweet nothings. You do it by shaking them, speaking loudly at them. That’s what I am doing. If you’re pissed, then considering your convictions, I am succeeding.

  35. Gravatar of Saturos Saturos
    4. October 2012 at 08:20

    You really do live in your own special world, don’t you MF…

Leave a Reply