A critique of the “credit economy” hypothesis
Tyler Cowen recently linked to a post by Ashwin claiming that the US might now be a credit economy, and that this weakens the old quantity theory of money. This hypothesis is based on two misconceptions. Here Ashwin quotes Alex Leijonhufvud:
The situation that Wicksell saw himself as confronting, therefore, was the following. The Quantity Theory was the only monetary theory with any claim to scientific status. But it left out the influence on the price level of credit-financed demand. This omission had become a steadily more serious deficiency with time as the evolution of both “simple” (trade) and “organized” (bank-intermediated) credit practices reduced the role of metallic money in the economy. The issue of small denomination notes had displaced gold coin from circulation and almost all business transactions were settled by check or by giro; the resulting transfers on the books of banks did not involve “money” at all.
If we were moving to a credit economy then the demand for currency and base money would be declining. But it isn’t, indeed it is higher than in the 1920s. Admittedly this is partly due to the Fed’s decision to pay interest on reserves. But even the currency component of the base is larger than in the 1920s, even as a share of GDP! It is not true that the various forms of electronic money and bank credit are significantly reducing the demand for central bank produced money.
Here Ashwin quotes Claudio Borio and Piti Disyatat:
The amount of cash holdings by the public, one form of outside money, is purely demand-determined; as such, it provides no external anchor. And banks’ reserves with the central bank – the other component of outside money – cannot provide an anchor either: Contrary to what is often believed, they do not constrain the amount of inside credit creation. Indeed, in a number of banking systems under normal conditions they are effectively zero, regardless of the level of the interest rate. Critically, the existence of a demand for banks’ reserves, arising from the need to settle transactions, is essential for the central bank to be able to set interest rates, by exploiting its monopoly over their supply. But that is where their role ends. The ultimate constraint on credit creation is the short-term rate set by the central bank and the reaction function that describes how this institution decides to set policy rates in response to economic developments.
This is a common misconception, especially among Keynesians and MMTers. It is true that central banks often set a short term interest rate target, and that once this target is set short run changes in the base are endogenous. But if that’s all they did then the price level would become indeterminate. Instead, they use their monopoly control over the base to move interest rates around in such a way as to target the price level. Because money is neutral in the long run, a given change in the monetary base will produce a proportional long run change in the price level and NGDP. Borio and Disyatat did acknowledge that rates are adjusted to target macro goal variables, but they failed to see the implication of that observation.
This can best be explained with an analogy. In 1973-74 the OPEC oil cartel decided to increase the price of oil from $3 to $10 a barrel. At the new price, quantity supplied was completely demand determined, OPEC responded passively to consumer demand at that new price point. OPEC had no short run control over the supply of oil. But that’s not at all what economists think is “really going on.” OPEC was able to raise prices by virtue of its ability to sharply reduce would oil supply. When we teach this in class, we show a leftward shift in the world oil supply curve. Or we might show a monopoly diagram with OPEC picking the output point where MR=MC. In either case, they are only able to control prices by controlling quantity. Indeed they could have even decided not to set an official price, and instead merely set a sharply reduced output level—the effect would have been the same.
When the Fed eases monetary policy we generally notice them cutting their fed funds target. But the exact same effect would occur if they simply increased the base, and let the fed funds rate fall in the free market. Indeed they basically tell their trading desk to adjust the base as need to keep market rates at a desired level. But it doesn’t have to be that way—they could just as well tell the New York trading desk to adjust the base as needed to keep CPI futures contract prices at a 2% premium over the spot level. In that case no one would say; “the Fed controls the CPI by controlling the CPI.” They’d say; “the Fed controls the CPI by adjusting the amount of base money in circulation.” Today people say; “The Fed controls the CPI by controlling the fed funds rate.” In fact, they control the CPI by controlling the size of the monetary base in such as way as to produce a monetary base and interest rates that are expected to lead to 2% inflation.
There will probably always be money; a pure credit economy is unthinkable. Without money there is no price level, because the price level is defined as the average price of goods in terms of money.
PS. Some might object that a higher proportion of currency is now held overseas. But nothing in the QTM requires currency to be held in the country where it is produced. Double the currency stock and the price level will double, ceteris paribus, regardless of where currency is held.
Tags:
25. September 2011 at 08:29
Great analogy!
Here’s a graph for base money velocity. I’m I right that 1985 would be the peak of the ‘credit economy’ for USA? And, if so, do we know why?
http://i.imgur.com/Pc3G8.png
25. September 2011 at 08:35
I agree, I guess that central banks modulate monetary base, so it never move exactly up to where the demand is.
For example, central banks rarely permit the full allotment of all the quantity demanded by banks; or they have banks reserves to sterilize some of the monetary base , as FED and ECB now.
I suppose that that has always established a important distance between official statment and facts.
My own experience is that central banks have always feared the full consequences of their decisions.
25. September 2011 at 08:40
I don’t get this. Credit economy? Are we talking member banks vs. non-banks?
“Double the currency stock and the price level will double, ceteris paribus, regardless of where currency is held.”
1/2 to 2/3 of all U.S. currency is held overseas. That skews things. Also, currency doesn’t turn over as fast as DDs.
“But the exact same effect would occur if they simply increased the base, and let the fed funds rate fall in the free market”
Changes in FFR’s impact the economy very very slowly, whereas changes in legal reserves are almost immediate.
“once this target is set short run changes in the base are endogenous”
Since the FED started targeting interest rates in 65, the bankers know that no matter what lines of credit they extend, they can always honor them, since the Fed assures the banks access to costless legal reserves, whenever the banks need to cover their expanding loans – deposits.
With interest rate targeting, the Fed has pursued a policy of automatic accommodation. That is, additional, costless, reserves, & excess reserves, are made available to the banking system whenever the bankers & their customers see an advantage in expanding loans.
The member banks, lacking excess reserves, would just bid up the federal funds rate to the top of the bracket thus triggering open market purchases, bank reserves, more money creation, larger monetary flows (MVt), higher rates of inflation – and higher federal funds rates, more open market purchases, etc.
25. September 2011 at 08:55
Because money is neutral in the long run, a given change in the monetary base will produce a proportional long run change in the price level and NGDP.
Doesn’t that only happen if the additional base money actually gets out into the real economy to become part of the broader monetary aggregates and play a causal role in bidding up prices? And doesn’t whether that happens or not depend on endogenous factors?
25. September 2011 at 09:37
Related to this and being discussed at the same time: David Glasner did not believe in the “thin air” definition of money, whereas Nick Rowe did. Glasner seemed to think you were closer to his opinion of money as being based on supply and demand. Is that correct?
25. September 2011 at 12:44
I don’t know whether or not we will have a “pure credit” economy, but it wouldn’t make the price level meaningless.
Private banknotes are part of a “pure credit” economy, and so privatizing hand-to-hand currency would be sufficient to be rid outside currency.
Checkable deposits are part of a “pure credit” economy, and their sum total can be defined as the quantity of money, even if there is no hand-to-hand currency. It is just private.
I quote a price of a dollar. It means I expect a check (or electronic payment) of a dollar. This is credited to my deposit account, and I can write a check for one dollar, purchasing a good or service with a quoted price of a dollar.
It is not even necessary that the price of the deposits have fixed dollar prices. They can all be mutual funds.
The banking system does require some kind of clearing mechanism, but it doesn’t have to based on settlement with central bank reserve balances. A private clearinghouse with the balances matched by financial assets. Maybe even money market type balances. I would certainly count this as money, and even as base money, but it is still inside or credit money.
Now, the price level isn’t tied down in such a system, and so you need some kind of redeemability. Index futures convertibility, for example.
25. September 2011 at 12:52
Kervick:
I don’t think you should focus so much on base money “getting outside” the banking system.
The actual analysis is “given the demand for base money, an increase in the quantity leads to a proportional increase in the price level.” You are implicitly assuming that the banking system increaces its demand for base money. If there is an increase in the demand to hold base money by anyone, even the banking system, and the quantity rises an equal amount, the price level doesn’t change at all.
25. September 2011 at 15:26
Remember the old idea of GDP bonds? Imagine that price were set as a function of the moving average of real GDP level… No money, in the sense of an independent store of value that has “stickiness” to it and a separable quantity. Prices are a fraction of units of real economic output (however that’s measured), and credit/debt is measured accordingly. As real GDP level goes up faster than target rate, “prices” go down. As real GDP level fails to hit target rate, “prices” go up. In a sense, solves the AD problem, but too many practical implementation issues. Besides, if you believe the strong version of money neutrality, we’re nearly there already.
BTW, if we’re in a non-money based economy, the swiss intervention should have failed miserably by now… Oh, wait, look…
25. September 2011 at 15:46
I think I’m just repeating what other commenters have noted, but you say “Because money is neutral in the long run, a given change in the monetary base will produce a proportional long run change in the price level and NGDP.” Isn’t this only true if the base multiplier and/or velocity stay the same. And aren’t these impacted by the FF rate and the expectations from suppliers and consumers of credit.
25. September 2011 at 17:56
Scott, Mish makes the strongest argument, this is what I want you two to debate, because debt / credit matters:
“Definition of Terms
Before discussing terms one must define them. I have on numerous occasions defined mine, and my definition was the basis for the bet.
Inflation
Inflation is a net increase in money supply and credit, with credit marked-to-market.
Deflation
Deflation is a net decrease in money supply and credit, with credit marked-to-market.
Hyperinflation
Complete loss of faith in currency.
The first two definitions have nothing to do with prices per se, the third does (by implication of currency becoming worthless).
Price Myopia
Many if not most economists, especially Keynesians, think of inflation in terms of prices.
In contrast, Austrian-minded economists generally have definitions similar to mine except most of them fail to properly include credit in their analysis. Austrians in general look at money supply alone, and that is a huge mistake.
Role of Credit in Inflation
Failure to include credit in the definition of inflation and the analysis of economic activity causes many problems. Credit influences consumer prices, jobs creation, and asset prices. The mark-to-market value of credit influences the ability and willingness of banks to lend.
People tell me all the time, “all I care about is prices”. If they really mean it, they are fools. Without credit expansion there is little hiring. Without hiring and money to pay for things, consumers cannot pay back loans and asset prices in general, crash.
Trillions of dollars in debt-inflated (thus imaginary) wealth have been wiped out in housing and the stock market because of falling credit, loss of jobs, and inability to service debt. Many homes fell in price from $500,000 to $200,000 (or equivalent percentages).
This is far more important than the price of gasoline hitting $4 or the price of carrots rising 50% to $2 a bunch. Yet, inflationists constantly fret about prices, ignoring far more important credit conditions.
Price myopia has other problems. Both Greenspan and Bernanke ignored an explosion of credit that fueled housing. Thus, a focus on prices induced errors on the way up and on the way down.”
http://globaleconomicanalysis.blogspot.com/2011/08/yes-virginia-us-back-in-deflation.html
25. September 2011 at 18:24
Peter, It depends on the ratio of taxes to interest rates. That ratio was low in the 1980s, hence low demand for currency.
Luis, Our current one sure seems timid.
Flow5, Not sure how all that relates?
Dan, In the long run an injection of money should have no impact on things like the money multiplier.
Becky, I left a comment over there, I really couldn’t understand their dispute–they seemed to be talking about different things.
Bill, I agree.
Statsguy, I don’t follow—is that NGDP targeting?
dtoh, Yes, but not in the long run.
Morgan, If “inflation” is what you call a rising money supply, what do Austrians call rising prices?
25. September 2011 at 20:47
Scott,
I guess the question I have is what is the mechanism for actually bumping up NDGP. If the net effect of open market operations just results in financial institutions, corporations or individuals switching one asset for another asset (i.e. cash in a current account) then you just get a drop in velocity which perfectly offsets the increase in the monetary aggregate and there is no impact on NDGP or price level.
Don’t there need to be additional (or related) mechanisms (e.g. wealth effect, credit expansion, or changed expectations) which induce economic players to make expenditures rather than holding assets. Just shifting the asset to money alone won’t have any impact.
25. September 2011 at 21:54
Hi Scott.
I very much appreciate your blog. I have to say, though, that I find your explanation of how the world works difficult to reconcile with observed facts.
As Kyddland and Prescott identified more than 20 years ago, shifts in the monetary base significantly lag shifts in credit-driven components of the money supply. Specifically, through observations of data from 1959 through 1989, they demonstrated that the monetary base lagged movements the unique components of M2 by three quarters or more.
You write:
“When the Fed eases monetary policy we generally notice them cutting their fed funds target. But the exact same effect would occur if they simply increase the base.”
My understanding, though, is that central banks around the world tried that experiment back thirty years ago or so and found that, in fact, they could not effectively manage both interest rates and money supply by focusing on the monetary base.
Of course, given Kydland and Prescott’s findings, the failure of this approach would not be the least bit surprising.
To be perfectly honest, I find Ashwin’s explanation far more plausible than the explanation you offer here (i.e. I can see very clearly how his explanation actually fits with observed facts).
26. September 2011 at 02:35
Scott – Thanks for the comments. What Borio and Disyatat identify as close to a credit economy has a few features, none of which are novel insights:
Reserves don’t constrain “inside credit creation” by banks. Even the impact of a small penalty on excess reserves can be mitigated by banks via fees on deposits etc. The best case scenario of a significant penalty on excess reserves is unwise lending by banks for purposes they would not consider creditworthy otherwise and into asset price inflation. In the current scenario, the relation between asset price inflation and investment/employment/wage inflation is weak.
The assets bought now via QE/Operation Twist have perverse effects and reduce consumption in many cases as people are forced to make up for negative real rates by saving more to meet fixed real savings goals in the future.
The core problem is lack of investment which requires financing,not saving – I subscribe to a Schumpeterian/Minskyian view on this matter.
I am a skeptic on the efficacy of NGDP targeting – primarily because I do not believe in the validity of the strong-form efficient market hypothesis (I believe markets can’t be arbed but prices don’t need to be right). But even if we target NGDP, the main point of my post is that we need to move away from buying assets to simple direct transfers into individuals’ accounts. The primary impact of current policy is to transfer wealth from the taxpayer to monetary “Kremlinologists” (as Steve Waldman put it) who can anticipate the Fed’s actions and buy up assets in advance, as well as propping up incumbent corporate crony interests rather than stimulating new business growth which is the primary engine of employment growth.
26. September 2011 at 04:42
dtoh, That’s right, you also need the hot potato effect.
Brett, The relationship between base changes and M2 depends on the monetary regime. In the Great Contraction the fall in the base preceded the fall in M2. In the big 1921 recession it occurred at about the same time. In 1937 base growth slowed before the recession.
But I think you misunderstood my post. I wasn’t arguing that control of the base is a good idea, I said you could affect interest rates without targeting them. I don’t think anyone disputes that.
I happen to favor abandoning interest rate targeting and replacing it with NGDP futures targeting.
BTW, It is another myth that the Fed engaged in money supply targeting during 1979-82—they did not do so consistently.
Ashwin, I certainly agree with your criticism of current Fed policy, but don’t agree that NGDP futures targeting depends on the strong form of efficient markets being correct. Rather it depends on any NGDP risk premium not being massively unstable, which is certainly a plausible assumption according to futures market experts. The worst case is I’m wrong, in which case I get rich. I don’t think it’s likely I’ll get rich.
I don’t agree that the assets purchased by the Fed are particularly important, or that people are getting rich off those purchases. In any case I’d like to go back to the pre-2008 regime where the Fed bought very few assets, so no one could get rich off them even if you and Waldman are right.
Negative IOR wouldn’t make banks take unjustified risks, they’d simply buy Treasury securities if there were no good loan prospects out there.
There is only one policy that will work–faster NGDP growth. It’s a necessary and sufficient condition for recovery. Everything else is gimmicks.
26. September 2011 at 04:43
Ashwin, I should add that if market inefficiency is a problem, the Fed could target its own internal NGDP forecast, and do level targeting. That would work almost as well.
26. September 2011 at 05:10
Scott, it’s NGDP targeting without the money, in the sense that “fiat” currency is directly defined as a % of next quarter’s GDP (or some weighted average of GDP over an interval, possibly including recent past and near future, or even distributed in a bell shaped curve over time).
If the market is anticipating a recession, the value of your currency goes down, and you buy today. If the economy is anticipating a boom, the value of your currency goes up and you delay purchase. Debt crises auto-adjust, the system is inherently counter-cyclical, and balances to make optimal use of capacity. Investment time window can be controlled by altering the weights on past/future GDP.
Consider the connection to Nick Rowe’s and Krugman’s barter economy examples. It’s an exchange economy model without the N+1th good. When people sell goods for dollars, they’re buying equity shares in the US economy.
Won’t ever happen, though. But if you deploy an NGDP target and bind it with an equillibrium condition constraint on expectations, you might get the same thing.
26. September 2011 at 12:13
Thanks for your response Scott.
Yes. I find your idea of targeting NGDP attractive and I think the various rationales you have offered are compelling.
I still think, though, that if you are going to be successful at fostering growth (particularly under current conditions) you have to acknowledge the role that bank/consumer-driven debt-creation plays in allowing growth in aggregate demand–what Ashwin terms “inside-credit-creation by banks”.
I think Ashwin and I are on exactly the same page on this, and I think Kydland and Prescott’s analysis proved that (under most conditions at least) bank/consumer-driven credit creation is THE major source of money creation. This doesn’t mean the Fed doesn’t have (and exert) tremendous influence. What it does mean, though, is that central banks probably need to revisit the mechanisms by which they exert influence.
As I put it to people I talk to, the central challenge is this:
(First) What if you acknowledge that, when it comes to the nation’s total household debt, there are only three possible states? In all instances, the household debt must be (1) increasing, (2) decreasing, or (3) holding steady.
(Second) Even a brief glance at history will tell you that only one of these states is consistent with a healthy and expanding economy. Household debt must be increasing in BOTH real and nominal terms.
(Third) What if you recognize that real household debt cannot expand forever? Yes, debt is something we owe ourselves; but if you want to understand the world, you cannot lose sight of the fact that household debt is owed by borrowers (borrowers with limited real capacity to borrow).
If you believe that this logical progression holds, then you cannot avoid a conclusion that (1) debt-creation and debt-destruction phases are inevitable and (2) the tools central banks use to manage the economy in debt-creation phases should probably be very different than the tools they use in debt-destruction phases.
My take is this is precisely what Ashwin is arguing, and I suspect you are not so far from believing it too.
26. September 2011 at 12:19
Scott,
What is your understanding of how the hot potato effect works.
Ashwin – If the Fed has a consistent policy which is not on again off again, then no one will get rich even if the Fed is making asset purchases.
26. September 2011 at 12:59
Brett – Yes, we’re on the same page. The best expositor of this perspective is Minsky.
dtoh – Yes, consistent written-in-stone policies are better to avoid biases but the very act of buying up existing assets (especially if we move beyond treasuries to corporate bonds or equities) is biased in favour of existing corporate interests.
27. September 2011 at 07:23
Brett,
“If you believe that this logical progression holds, then you cannot avoid a conclusion that (1) debt-creation and debt-destruction phases are inevitable and (2) the tools central banks use to manage the economy in debt-creation phases should probably be very different than the tools they use in debt-destruction phases.”
You slipped an assumption in there, namely that the expansion of debt in real and nominal terms outpaces income (GDP). Debt /GDP can hold steady in an expanding economy whilst debt increases in nominal and real terms. If it holds steady I do not see a reason why you should have a debt-crisis. Unless of course distribution matters, but you have not mentioned that and why the distribution would be such that a debt-crisis would be inevitable.
27. September 2011 at 09:26
Martin.
I totally agree the second step in the logical progression is the key. I wouldn’t characterize it as an assumption, though, but rather an observation of historical correlations.
If you look at US data on household debt you will see only a handful of instances where the ratio of household debt to GDP is falling for any prolonged period. Almost universally, these periods correspond with periods of recession.
Specifically, if you look at changes in household debt/gdp on a year-over-year basis from 1946 to 2007, you see significant and steady expansion. The average year-over-year growth for the period was 1.4%; there were only ten instances when hh_debt/GDP decreased and those instances correlate VERY strongly with periods of slow growth or outright recession; and there was only one instance where hh_debt/GDP decreased by more than 1.4% (1981).
Things get even more interesting if you hone in on rates of change of household debt. If you look at accelerations and decelerations on a quarterly basis, you see that eight out of the last nine recessions were preceded by decelerations in household debt creation. (The exception was the recession that followed the events of 9/11.)
I put together a chart that shows this pattern, and it is jaw-dropping. You have to adjust for seasonality for the correlations to really pop, but anyone who mapped these data and looked at the chart in early 2007 would have been terrified by what was coming.
If you’re interested, I wrote a piece on all this that I posted back in January, and the chart I just mentioned can be found on page 9.
http://integratedeconomics.com/attachments/File/Time_to_Re-Think_How_Economics_has_Failed_America_012111.pdf
27. September 2011 at 10:44
Brett,
Fair enough, but my comment was more directed at why debt/gdp has to rise rather than why it has to fall. I understand the latter if the former is true, but I do not understand why the former is inevitable.
This: http://research.stlouisfed.org/fredgraph.png?g=2sQ is what puzzles me. I believe your piece has a similar graph.
Also, Koo makes a similar distinction between the two phases of debt-creation and debt-destruction, but I do not see why that is a distinction in kind when it comes to monetary policy.
I understand why debt-destruction might be a distinction in degree, and that if the destruction phase is a deep structural problem and as a result you’d get quite a bit of inflation, but I do not see why that would make monetary policy less effective. Less desirable perhaps, less effective, not too sure. Sure interest rates are less effective, but I don’t think interest rates are the sole standard form monetary policy can take.
28. September 2011 at 09:43
Martin.
It seems like, at this point, you have three questions:
1. Why does hh_debt/gdp have to rise for the economy to be healthy?
2. Why is there a difference in kind regarding debt-creation and debt-destruction phases? and
3. Related to question #2, why would a debt-destruction phase make monetary policy less effective.
The answer to your first question is involved, but at its core the answer revolves around two fundamental ideas:
(1) a recognition that household demand is the force that drives economic expansion (five years ago you would have found lots of economists who contested that idea, and I am sure there are a lot out there who are loath to give up their position, but these days only the most dogmatic are willing to state their position flat out).
(2) a recognition that (under the current economic structures) the principal mechanism that allows demand to grow (to allow demand to exceed what it was yesterday) is the expansion of household debt.
In short, growing economies are driven by growing household demand, and growing demand is realized through real increases in household debt.
As you have acknowledged, if the above is true, and if debtor households have some limit to their capacity to borrow, this means that debt-creation and debt-destruction phases are inevitable.
Now… on to your second and third questions: What makes a debt-destruction phase so different that it makes traditional monetary policy ineffective?
The answer is that, at its core, traditional monetary policy is all about managing debt creation. In times when households are able to expand their debts, central banks use increases and decreases in interest rates to manage how much debt-creation (how much purchasing-power creation) happens at any given point.
Back in the early 1990s, when the economy slipped into a recession (i.e. when households slipped into a modest debt-destruction mode), inflation was still high enough, and households still had enough capacity to take on more debt, that the Fed could lower interest rates and entice them to do so. Voila, the household debt creation switched from shrinking to growing and the economy switched back to a growth track.
If what I am saying is true, you would expect to see a marked long-term trend: As households came closer and closer to exhausting their capacity to take on debt, you would see progressively slower recoveries from recessions.
Matt Yglasius has a great blog post that documents just this, showing very clearly that the recovery from the 1981 recession was slower than preceding recoveries, the recovery from the 1990 recession was slower than 1981, the recovery from the 2001 recession was slower still, and finally we arrive at the point the current situation, where in many respects the recovery doesn’t happen at all.
http://thinkprogress.org/yglesias/2011/08/05/289658/the-era-of-weak-recoveries/
Another long-term trend you would expect to see is a long, down-trend in underlying inflation and nominal interest rates.
So the problem is this: When you finally get to the point where you exhaust your ability to create more debt and switch from a debt-creation phase to a debt-destruction phase, the central bank finds that its arsenal of tools for managing the speed of debt-creation are of no use.
They can pump lots of dollars into the monetary base, and they can push interest rates down all the way to zero, but if households need to destroy debt, even a very low interest rate is not going to entice them to change course and begin expanding their debts again.
In theory, the central bank could push interest rates below zero (i.e.they could pay people to take out loans). But at that point they might as well just cut to the chase and start dropping money out of helicopters.
This problem of running into the “zero lower bound” is the so-called “liquidity trap” that Paul Krugman talks about all the time. And my reading is that the fear of the zero lower bound is also one of the considerations that leads Scott to advocate NGDP targeting.
A true helicopter drop is exactly what the Fed should be doing. Unfortunately, given how much bad economic thinking there is out in the world, and given how much leverage lenders have over the political process, actually executing a true, meaningful helicopter drop is much easier said than done.
The last thing that needs to be said is that, when you are a big economy (for which net exports represent a very small share of the economy), debt-destruction phases are very painful (think the Great Depression or Japan’s current long-running deflationary stagnation).
Demand remains sluggish; real and nominal incomes fall for many households; with falling spending power and high debt loads, households get crushed by their outstanding debts; asset values for things like homes and equities tend to fall; banks get crushed because of lots of defaults and a lack of willing and qualified borrowers (banks make money by making loans), etc.
Anyway… I hope this helps you understand where I (and Ashwin and Minsky and Koo and many others) are coming from.
28. September 2011 at 10:32
Brett – You can get around the zero bound problem by allowing the FED to flexibly set capital reserve ratios for banks. Under such a regime, the Fed could set a negative capital reserve ratio for housing loans or loans to consumers. Even when the FF rate is approaching 0, rates on household loans are considerably higher and there is still a lot of room to push these rates down by forcing the banks to expand household loans through manipulation of capital reserve ratios.
28. September 2011 at 14:13
dtoh – Yes, I think you’re right. I would submit, though, that this might less a way around the zero lower bound as much as an illustration that, if we get creative, we can squeeze a bit more out of the system before the lower bound becomes truly constraining.
I think, though, that it is useful to step back and consider all of this more broadly?
It seems to me that for at least the last thirty years we have aggressively innovated new ways to expand the amount of debt that households can take on. We have held interest rates at very low levels for a longer and longer periods of time; we have lowered transaction costs; we have created mechanisms that allow banks and other originators to shift default-risk for the loans they originate onto others (which made it more attractive to originate loans for entire populations who would have been deemed unacceptably risky under old structures)…
I bet we could could all have a lot of fun adding to this list.
In broad terms, I would argue that all of these innovations are mechanisms we devised to forestall the day when debtor households cannot take on any more debt and the debt-destruction phase begins. We did a great job of forestalling that day, and that fruits of that success are real household debt burdens that tower over the last peak (on the eve of the Great Depression).
Can we push mortgage rates even lower? I think we definitely can and should–less because it entices more people to expand their debt, more because it may be the closest thing the FED has to a unilateral, politically-palatable helicopter drop (at least for as many households as are able to refinance their mortgage).
It seems to me, though, that the fundamental question is still this: Do you believe that, sooner or later, a debt-destruction phase is inevitable and necessary? I believe it is,and as a result, I think we have to seriously consider what good policy looks like (monetary and otherwise) in a debt-destruction phase.
28. September 2011 at 14:44
Statsguy, I’ve learned to try to avoid saying something will never happen. We are headed for an all electronic money system in the late 21st century, which will radically change the set of feasible monetary policies.
Brett, I don’t think the monetary authorities need to pay attention to debt, Just focus on NGDP expectations, and adjust the monetary base as needed. I also favor fiscal reforms that would radically increase the savings rate, for unrelated reasons.
dtoh, Mostly through expectations of future changes in NGDP, and also current changes in asset prices. In a simple economy with no financial system it would work by directly raising AD for goods and services.
28. September 2011 at 20:17
Scott – Sorry to be persistent, but isn’t it – Expectations effect asset prices which then effect aggregate demand. And if so… what is the mechanism by which asset prices impact AD. I assume it is not just a wealth effect, but that higher prices lead to creation of new assets.
29. September 2011 at 17:23
dtoh, Higher home prices cause more home building. Higher commercial RE prices cause more commercial RE construction. Higher commodity prices cause more commodity production. Higher stock prices cause more construction of corporate assets (the Tobin Q effect). Higher used car prices cause more people to buy new cars. Etc. etc.
29. September 2011 at 18:20
I think we should forget “debt creation / destuction” and say:
“taking loans on hard assets”
“paying off or losing hard assets”
Hard assets stick around, so I suspect Brett is concerned with people losing their hard assets.
But if people don’t lose hard assets on loans they can’t pay off… other people can’t buy them up for pennies on the dollar.
loans are not important, who gets to own the hard assets is paramount.
30. September 2011 at 08:21
Again, just to clear, the central idea is a simple logical progression.
1. When it comes to household debt, there are only three possible states: it must be growing, shrinking, or holding steady.
2. Even a glance at history shows that only one of those states is consistent with a healthy and expanding economy: household debt must be increasing (in both real and nominal terms).
3. Household debt cannot expand forever.
This is a pretty clear-cut formulation of logic and observation, and I think it forces us to reach the conclusion that debt-destruction cycles are both inevitable and extremely painful.
So… if you don’t believe it, I am curious which observation, or which logical step, you have trouble with.
1. October 2011 at 16:52
Brett, Number 3.
1. October 2011 at 21:22
It is possible for household debt to equal 100 times GDP?
1. October 2011 at 21:59
Sorry. I should clarify. In specific terms, point 2 should read:
2. Even a glance at history shows that only one of those states is consistent with a healthy and expanding economy: household debt must be increasing (in both nominal and real terms [i.e. the ratio of household debt to GDP must increase]).
3. October 2011 at 15:57
Brett, I don’t agree with number 2. “Glances at history” show almost nothing.
4. October 2011 at 10:31
Scott, now you’re just being silly.
As you know perfectly well, when correlations are very strong you tend to be able to see them at a glance.
For example, if I look at volumes of rainfall in the mountains near my home and at river levels in the valleys below in the days that follow, even at a glance I can see that there is virtually an absolute correlation between the two.
If you want to take a look at the data, I threw some charts up on a blog I created last week:
http://failureofeconomics.wordpress.com/2011/10/04/useful-charts/
(Mine is an admittedly minimalist affair, and while I’m thinking of it I have to commend you on both your energy and the generosity with which you share your time and talents with readers of your blog.)
So… I think we can agree that strong correlations can be seen at a glance, and I think we can agree that there is a strong correlation between expanding nominal and real household debt and economic health.
Just to hit some highlights…
In the last 90 years, there were only two instances when year-over-year private debt went from positive to negative growth in nominal terms. Those years were 1929 and 20008.
If you look outside the US, you can see another instance–it happened in Japan in 1997 (according to McKinsey’s data, anyway).
In terms of household debt, in the Post-WWII years for which we have good data, the only time when year-over-year household debt growth went from positive to negative in nominal terms was 2008.
In real terms (i.e. in terms of the ratio of debt-to-gdp), since 1952 the only instances where the ratio fell were either associated with a time of war (i.e. the end of the 1960s and beginning of the 1970s) or in periods closely associated with recessions.
In fact, other than the recent downturn in household debt creation, there was only one period where household-debt-to-gdp decreased in a significant way, and that was in the recession that began in 1980.
So, again, I think most objective people will agree that the strong correlation exists.
That brings me back to my basic logical progression:
1. When it comes to household debt, there are only three possible states: it must be growing, shrinking, or holding steady.
2. Even a glance at history shows that only one of those states is consistent with a healthy and expanding economy: household debt must be increasing (in both real and nominal terms).
3. Household debt cannot expand forever.
When I consider this progression I am forced to conclude that debt-destruction phases (or at the very least, debt-stabilization phases) are inevitable. I am also forced to believe that, unless we figure out some new way to mitigate them, they will be exceedingly painful.
4. October 2011 at 16:02
Brett, You are the one being silly, as correlation doesn’t prove causation. I don’t have the data in front of me, so I won’t comment on correlation, but if NGDP grows at a steady 5%, then the business cycle will get smaller regardless of what debt does. I agree that debt cycles would probably also get much milder in that situation, but my response is “So what?”
It’s also true that unemployment always rises in recessions, does that mean unemployment causes the recession?
5. October 2011 at 10:35
Hi Scott.
Please don’t take offense at this, because I sincerely do not mean it in a negative way at all, but I don’t think you have seriously considered the things I’ve been saying.
That’s OK. I do the same thing. When I think I understand something much more deeply than the person I’m conversing with, I am slow to seriously consider the things they are saying. It is a natural, efficiency-maximizing scheme that allows us, in most instances, to spend our intellectual energies wisely.
I do appreciate your willingness to continue this discussion. I honestly believe, though, that you might want to put a bit more energy into considering the things I am saying. In particular, if you want to spend only five additional minutes thinking about this, I would recommend seriously looking at the second chart I posted on my blog-page.
http://failureofeconomics.wordpress.com/2011/10/04/useful-charts/
Now… on to the points at hand…
Of course I agree with you that correlation does not prove causation. I actually think there is a strong case to be made for causation, which I will get to in a minute, but before I get to that I want to focus on another point.
When we want to understand how to make something happen and we look at correlating conditions in the past, there are two things we need to worry about: (1) causation and (2) conditions that are necessary-but-not-sufficient to achieving our goal.
For example, you picked the example of employment…
Because employment tends to lag shifts in other measures of economic growth, it is probably safe to believe that employment shifts do not provide the initial impetus for, say, accelerations in economic output. However, it may well be that employment growth is a necessary condition to maintaining output expansion once the acceleration happens.
If, for some reason, employment growth is stymied, we might very well worry that any acceleration in output that occurs would not translate to a robust phase of economic growth.
By the same logic, if real increases in household debt correlate with expansions in output, but we have reason to believe that real increases in such debt are not sustainable, we might well worry that we lack a necessary condition for achieving broad economic growth.
To make matters worse, since household debt is a leading correlating factor (rather than a lagging one like employment), we might worry that constraints on household debt would block an acceleration in output before it ever happens.
This last paragraph, of course, re-introduces the question of causation. Specifically, if accelerations in the creation of household debt correlate with, and in fact, LEAD all other accelerations in indicators of economic growth (which seems to be the case), then it seems we ought to have some strong suspicions that causality may exist.
I hate to keep pointing to Kydland and Prescott’s 1990 study: Business Cycles: Real Facts and a Monetary Myth, but it is a treasure-trove of information about what leads what.
As you know, their headline finding was that changes in the monetary base lag changes in measures of money that derive from debt-creation (by a significant period of time). Of equal interest to me, though, is their analytic finding that the one economic indicator that seemed to lead accelerations in all others was Consumer Durable Investment.
At one level, this isn’t all that surprising. Most people believe that the principal lever central banks use is their influence over interest rates. All else being equal, lower interest rates increase the enticement and ability of households to increase their levels of debt””most of which debt goes to purchasing consumer durable investments.
To circle back, though, this finding does underscore the importance of considering potential limits in how much debt households can take on. If expanding household-debt-to-GDP ratios is either a causal or necessary factor for economic growth, but such real expansion cannot go on forever, I think it is really important to recognize how this system works.
5. October 2011 at 18:57
Brett, I get 100s of comments so I don’t have time to give each one a lot of thought. But don’t assume I don’t understand because I disagree. How would you interpet the boom of 1964-70, when the ratio of credit to GDP was falling. Didn’t you say that ratio always rises during booms? But that was one of the great booms of all American history.
I don’t agree that changes in the base lag changes in M2. Sometimes they do, but other times like 1929-30, they lead. But in any case, as you know I don’t consider the monetary aggregates to be good indicators of monetary policy. So the K-P findings have no bearing on my views.
I don’t agree that monetary policy works through the lever of interest rates.
Looking at leads and lags is not a good way to establish causation.
In my view stable NGDP growth will tend to stablize credit/GDP, and this will allow for more stable RGDP growth, even if credit does have a causal role.
6. October 2011 at 15:44
Hi Scott.
Good. Now we’re getting to the real conversation.
Don’t worry. I don’t have any illusions that you will be easily swayed, but I do appreciate your willingness to continue the conversation.
If I can just take a moment to consolidate some common ground…
-We both acknowledge that continuous growth in household-debt-to-GDP is not sustainable over the long term.
-We both acknowledge that, for a long period of time now, that was precisely the path the economy had been on (until we hit the deceleration of household debt creation that began in 2006).
Where we differ at this point is this:
You think that, notwithstanding historical correlations, real growth in household debt is not a necessary condition for sustainable economic growth. You believe that NGDP targeting will allow the economy to grow while stabilizing levels of real household debt.
(Please correct me if I am wrong on any of the above.)
I am a big fan of the idea of NGDP targeting, but the benefits of NGDP targeting notwithstanding, I believe that the only way to allow the economy to grow while stabilizing real household debt levels is to drastically rethink how monetary and/or fiscal policy is pursued.
In your post, you asked how I would interpret the period from 1964 to 1970, where GDP experienced robust growth while real household debt levels remained stable, and even decreased by a small amount?
In order to test the model I am working from, I would rephrase that a bit and ask:
How was it that DEMAND in the economy increased without an accompanying real increase in debt?
When I think about in those terms, I can think of a few theoretical explanations:
1. It could be that there were expansions in non-household debt. When I look at the data, I do see that real, non-financial corporate debts did expand during the period, and when one adds household and corporate debt together, the combined real debt loads did expand consistently over the period. However, I am skeptical that the addition of corporate debt alone explains what happened.
2. It could be that, by the mid-1960s, households became less conservative in terms of accumulating wealth. If they steadily shifted to spending more and more of their earned purchasing power, then that would translate to increasing demand without increasing debt. When I look at those data, I do see that there was a significant slowdown in the growth of household balance sheets (Time and Savings Deposits), and I think that might be part of the explanation.
3. It could be that the distributional effect of the war on poverty put increasing purchasing power in the pockets of people who were sure to spend it (moving dollars out of the pockets of those who would be more likely to save/lend). Certainly, the plunge in the number of people in poverty coincides perfectly (beginning in 1964 and ending in 1970). Certainly, too, the ratio of federal-tax-receipts-to-GDP went up during the ’64-’70 period.
4. Finally, it could be that many households were seeing real expansions in spending power because they were steadily capturing a larger share of the GDP pie. If this was the case (and these households were not saving more) then one would expect to see steady increases in their animated purchasing power. When I look at the data, this seems like a potentially powerful factor. The period 1964 to 1970 aligns precisely with an unprecedented period when strong economic growth was combined with a massive reduction in the share of GDP that was captured as corporate profits.
In 1964, corporate profits reached a post-war peak of more than 12% of GDP. From 1964 to 1970, profits as a share of GDP experienced a steady, rapid fall until they bottomed at 7.5% of GDP in 1970.
In my mind, it seems plausible that factors three and four (with a little help from factors one and two) played key roles in allowing household demand to increase on a year-over-year basis without expanding household debt.
On a time-frame precisely aligned with the period in question, poor and middle-class households significantly and steadily expanded the portion of the GDP pie they were capturing, which in turn would have allowed consumer demand to expand without resorting to a real expansion of debt.
Certainly, there was SOMETHING different about the period from ’64 to ’70, and these seem like very plausible candidates.
To the extent we find this explanation compelling, it seems to me this might be one of those instances where we have an exception that proves the rule. We see that it is possible to put a run of six years together where events conspire to avert the necessity of driving growth through expanding household debt. BUT… all of these dynamics were transitory, and it seems a run of six years might be the best you could ever hope for.
Real household debt expansion is transitory too, but it is possible to put runs together that last much longer than six years.
Now… switching gears…
When you write that you think stable NGDP growth will tend to stabilize credit/GDP, I have to say I find your statement extravagant.
We can look back as far as we can see (hundreds of years if you look at the evidence offered by the Elliott Wave folks and others) and what we see is evidence that expanding economies are characterized by real accumulations of debt. And now you say that making a relatively modest change in how we manage our economy will make that dynamic go away. When I put those two things side by side, I just find your statement hard to swallow.
When someone tells you something that has been around for hundreds of years will just go away, I think there has to be a pretty substantial burden of proof before you should believe it.
As I’ve said before, I really like the idea of NGDP targeting per se. In particular, I think it solves a lot of problems we have witnessed lately.
In my mind, though, the fundamental challenge remains: One way or another, end-demand today needs to exceed what it was yesterday to drive economic growth. Historical evidence suggests that the way this generally gets done is through expansion of hh-debt/GDP.
Now… If the Fed targeted NGDP by writing checks to every household in the country when they wanted to inject purchasing power into the system… THAT would be a game-changer.
8. October 2011 at 17:02
Brett, I think you are misunderstanding what the data tell you. You notice that debt ratios usually expand during above normal NGDP growth and usually fall during below normal NGDP growth (or RGDP, if you prefer.)
And then you assume that these patterns tells us something about what you’d expect if NGDP growth was neither above nor below normal. I don’t think it tells us anything. But if forced to guess then I’d guess that if above normal growth is associated with rising debt ratios, and below normal growth is associated with falling debt ratios, then normal growth might be associated with stable debt ratios. Why is that so far-fetched?
10. October 2011 at 12:05
Hi Scott.
I have to confess, there are so many places I would like to go here that I find myself at a bit of a loss.
I want to delve much deeper into the question you pose above, but I also want to delve into a bunch of other things. So, for today, I will just make a quick observation about your statement and then move on.
You write:
“You notice that debt ratios usually expand during above normal NGDP growth and usually fall during below normal NGDP growth (or RGDP, if you prefer.)
And then you assume that these patterns tells us something about what you’d expect if NGDP growth was neither above nor below normal.”
My quick response is that, when I look at the data, I do not see ANY meaningful correlations of the sort you describe. Specifically, I do not see a correlation between any given level of positive NGDP growth and declining HH_Debt/GDP.
I see one period where HH_Debt/GDP stabilized for six years and actually fell a bit (1964-1970), but that was a during a period of very high NGDP growth. I see other periods with similar levels of NGDP growth but relatively rapidly increasing HH_Debt/GDP.
I see two periods of more moderate levels of NGDP expansion (the mid-to-late 1990s and 2003-2006) and one of those periods corresponds with a period of modest-but-still-substantial HH_Debt/GDP growth (the 1990s) and the more recent period corresponds with HH_Debt/GDP going through the roof.
So, again, I just don’t see where the correlation you describe exists.
You go on to write that:
“…if forced to guess then I’d guess that if above normal growth is associated with rising debt ratios, and below normal growth is associated with falling debt ratios, then normal growth might be associated with stable debt ratios. Why is that so far-fetched?”
I guess my response would be that, in the absence of any data, this certainly sounds like a plausible description of how the world works. In fact, I genuinely hope that it is true. Again, though, I just don’t see ANY sort of robust evidence to support it.
With the exception of the late 1960s (which I think is really interesting and want to revisit), when I look at the data, the only things I see are:
1. I see that the only periods of declining HH_Debt/GDP occurr during recessions. On the chart I posted, prior to 2008 and other than the late 1960s, the only places you can see meaningful declines in HH_Debt/GDP were in 1957-58, 1960, 1974, 1980-81, 1983, and 1991-92. (Needless to say, these instances correlate almost perfectly with periods of recession.)
2. A very-short-term flattening out of HH_Debt/GDP in the year immediately following every recession from 1958 through 1991 (the point at which (1) the denominator in our ratio is accelerating and (2) lagging Corporate_Debt/GDP-growth kicks in to carry the load).
Like I said, I want to delve more deeply into this, and I am particularly interested in what we might learn from looking at three periods (the late 1960s, the early 1990s, and 2002-2006)””but that will have to wait.
For now, though, I’d like to shift gears and focus on comparing what you and I have been talking about with things that folks like Matt Rognlie, Paul Krugman and many others have been saying.
When Paul Krugman talks about our current position his term of choice is “debt-overhang.” Krugman and Eggertsson signaled that this term is a good descriptor of their understanding of the world in their deleveraging work. In setting up their model, they envisioned a condition where the world was operating at some appropriate level of debt, but an external shock rendered that level too high so a debt-destruction cycle began.
Likewise, in Rognlie’s recent post on Deleveraging and Monetary Policy, the way he framed the discussion suggests that he is working from a very similar understanding of the world. In his post, Rognlie frames the issue as one of the economy needing to return to the “correct level of leverage.”
In the context of our discussion here, there is a clear problem with this understanding: When we look at the history of household leverage, what exactly is the “correct level” of debt?
One can look at the history of, say, household WEALTH in America (i.e. household net worth) and it is easy to argue that the ratio of Wealth/GDP is pretty darned stable. For all of the last 60 years, wealth/GDP has ranged between roughly 300% and 450%. And, in fact, until we got to what I would describe as bubbles in high-tech stocks and housing values, the ratio of Wealth/GDP was far more stable than that, ranging from roughly 300% to 360% (a level to which it has now returned). This, I would argue, means that it is entirely reasonable to refer to a “correct” level of net household wealth.
http://www.calculatedriskblog.com/2008/12/total-household-net-worth-as-percent-of.html
As you and I have observed, though, one cannot say the same thing about HH_Debt/GDP. If anything, an observation of history would suggest that the “historically correct” state of HH_Debt/GDP is for it to be ever expanding. Certainly, over the past 40 years, there are no periods of more than one year in which HH_Debt/GDP was stable or falling and the economy was not in a recession.
http://failureofeconomics.wordpress.com/2011/10/04/useful-charts/
You and I have both acknowledged this dynamic. You and I have both observed that there is no such thing as an historically-typical ratio of HH_Debt/GDP. And you and I have both agreed that such a trend is not sustainable.
Frankly, I don’t think it is possible to overstate the importance of the things you and I have already agreed upon, and how much it is missing from current economic thinking.
In your response above you hazarded a guess that normal growth might be associated with stable debt ratios.
Personally, the evidence I see in the past 40 years makes me doubt that your guess is correct, but that is something we can delve into and debate. The reality is, though, that at this point you and I are way ahead of the game.
Specifically, you recognize that if we pursued NGDP targeting tomorrow and it turned out your guess was wrong, you would recognize the problem and need to think long and hard about what that means.
I respect many of the things that Rognlie and Krugman are saying, and I think Krugman and Eggertsson’s recent work on debt-destruction cycles is really important (putting numbers behind things that I have said in theoretical terms about the forces at play in such periods), but honestly, Rognlie, Krugman and the vast majority of economists seem to be totally oblivious to the path we have been on for at least the last 40 years.
If Rognlie and Krugman had spent much time considering the data, I have to think they would be much less cavalier in using terms like “debt overhang,” or referring to some “correct” level of debt, and much more respectful of the risks posed of every-expanding HH_Debt/GDP.
Anyway… enough for now.
11. October 2011 at 18:21
Brett, You are misunderstanding the evidence. When you look at cyclical patterns you aren’t seeing what happens with expanding economies, you are seeing what happened during above average growth.
With NGDP targeting you no longer have the cyclical pattern you describe, so you can’t make any assumptions about hiow the debt ratio will change over time. There’s no more cyclical expansions, just long run growth. It might well stabilize the ratio. After all, over the very long run debt ratios don’t show much change. And those very long run patterns are what is relevant for a NGDP rule.
13. October 2011 at 10:16
Hi Scott.
I don’t think I’m misunderstanding. I think you’ll agree that when we look at cyclical patterns we are looking at oscillations””oscillations that translate to growth rates that sometimes exceed your “average” rate; sometimes fall below your average rate; and in moving between those two states, must at some point equal your sustainable rate (at least for a moment in time).
So…for example… If we focus on the ten years prior to 2005, we see rates of NGDP growth that range from a quarterly low of 0.2% to a high of 8.8%. During that period, we also see a period when for eight consecutive quarters, NGDP growth averaged 3.4% (and never exceeded 4.8%). I’m assuming 3.4% NGDP growth would be below an optimal target level, but for that entire period HH_Debt/GDP was expanding rapidly.
My earlier point was simply that when I look at the data, the only strong correlation that exists is between recessions and falling HH_Debt/GDP. Excluding recessions, I don’t see any correlation between any given rate of NGDP growth and HH_Debt stabilization.
I think, perhaps, that what you are really arguing is that the oscillation process itself precludes HH_Debt/GDP from ever stabilizing. This is something that is certainly possible, but tough to argue from a theoretical perspective. You would have to come up with a theory that simultaneously explains (1) a strong correlation between contracting HH_Debt/GDP and contracting RGDP, but (2) no strong correlation between stabilizing HH_Debt/GDP and any given level of positive RGDP growth””even during prolonged periods of soft growth.
15. October 2011 at 09:50
Yes, but when you aren’t in recession then growth is generally above average, which was my point.
Put it another way. With steady 5% NGDP growth you would presumably no longer see a steady upward tredn in non-recession years, as there would be no recessio nyears. Unlesss recessions occurred with steady 5% NGDP growth, which is possible, but likely to be very rare.
20. October 2011 at 12:25
[…] Anyone who is interested in these questions may want to check out our back-and-forth in the comments section of this posting: http://www.themoneyillusion.com/?p=11077 […]