How did small losses cause the Great Recession?

Here’s a thought experiment: what would happen if gas stations suddenly started requiring payment in cash? Would there be a recession?

Of course not. It wouldn’t be a big deal: you’d probably carry around a little extra cash for when you needed to buy gas, but your life would stay mostly the same. It’s hard to imagine that the hassle would cause you to change your driving habits, much less your other consumption—even at high interest rates, the forgone interest from carrying a few hundred dollars in cash to prepare for fill-ups is minimal next to the cost of the gas itself.

So why am I mentioning a recession? How could such a trivial issue possibly have any serious macroeconomic consequences?

It wouldn’t—but understanding why is more subtle than you think.

Imagine a world where the Fed, rather than targeting interest rates, changes the size of the monetary base (the sum of cash and bank reserves) according to some rigid rule. Perhaps it increases the base at a rate of 5% a year. In this world, the sudden imposition of a requirement that motorists pay cash would cause an increase in the demand for base money—but no corresponding increase in supply. Since the short-term demand for base money is incredibly inelastic with respect to the nominal interest rate (minimal fluctuations in the supply of base money are enough to implement massive changes in the federal funds target), even a modest increase in the demand for cash would be enough to produce a large, contractionary spike in the federal funds rate. And thus a seemingly irrelevant change in the method of payment for a tiny part of the economy would be enough to send the United States into a deep recession.

Of course, central banks around the world realized long ago that this would be a problem—that’s why they target interest rates, not the total supply of base money. But there’s still a lesson here: somehow, a small change in the demand for one asset (cash) has the potential to bring a $15 trillion economy to its knees, if the Fed doesn’t respond in the right way.

Anyone trying to understand the crisis of 2007-2008 should be paying close attention. The central mystery of the crisis (and, indeed, many other crises) is how a relatively minor initial loss—$500 billion to $2 trillion, of the same order of magnitude as the change in the value of equities during a weak month—somehow led to a much larger collapse in the economy as a whole. As Brad DeLong puzzled in late 2008:

And our models for why the risk discount has taken such a huge upward leap in the past year and a half are little better than simple handwaving and just-so stories. Our current financial crisis remains largely a mystery: a $2 trillion impulse in lost value of securitized mortgages has set in motion a financial accelerator that we do not understand at any deep level but that has led to ten times the total losses in financial wealth of the impulse.

Now imagine that we have a shock to demand for liquidity in general rather than cash. For whatever reason, investors want to park more of their portfolio in T-bills and repos, and less in illiquid bonds and equities. At the same time, imagine that the supply of liquidity is actually decreasing—the “AAA” tranches of mortgage-backed securities are no longer acceptable collateral, and in general it’s much harder to transform real, risky assets into liquid, riskless ones.

What happens? Remember that with a shock to cash demand, nothing happened as long as the Fed targeted the federal funds rate: it would supply whatever additional cash was necessary to keep the interest rate at its target level. But with shocks to liquidity supply and demand, the world isn’t so rosy. At least traditionally, the Fed doesn’t target the liquidity premium. (Though QE/QE2 may be viewed as attempts to do just that.) Classic “open-market operations” are just trades of base money for T-bills—one liquid asset for another—leaving supply unchanged. As a result, if short-term supply and demand for liquid assets are inelastic, even a small shock can produce a large spike in the liquidity premium, and a rise in interest rates in general. In exactly the same way that a few gas station owners demanding cash could raise the price of credit across the economy, a small decline in supply or increase in demand for liquidity can force up the interest rates paid by all businesses and consumers, leading to a severe contraction.

Now, if the Fed realizes what’s happening, it can offset the effect of an increase in the liquidity premium by decreasing the federal funds rate. But this standard reaction may be delayed, and eventually it becomes impotent: the federal funds rate hits zero, and exchanging money for T-bills can do no more.

At this point, the fate of all assets in the economy is determined by the idiosyncrasies of the market for liquidity—which, like the market for cash, may be very small relative to the economy as a whole. A few pension funds deciding to move $100 billion (<1% of GDP) from real estate into liquid assets can—by increasing demand in an inelastic market—cause a noticeable runup in all interest rates, interest rates that govern almost $50 trillion in household financial assets. On the supply side, we’ll see the same effect from a change in market expectations that renders useless $1 trillion in repo collateral.

So we have a way that small losses in one market can lead to far vaster consequences: they cause a disproportionate change in the liquidity premium, which (once the Fed has dropped rates to zero) is responsible for determining every single interest rate in the economy.

How is this possible? Why should the whims of a few players in the market for liquidity make such a difference to macroeconomic outcomes? Doesn’t this mean there’s some kind of externality?

Indeed there is. Remember that money is the unit of account. It’s not hard to imagine why there is an externality in a world where all prices are denominated in, say, Mountain Dew. By changing the supply of the standard by which all goods and services are priced, your decision to pop open a can will have economic consequences far more profound than your momentary buzz. Similarly, in a world where the Fed fails to adjust the monetary base, gas stations’ decision to require cash affects economic outcomes in a way that belies their small role in the economy. And once the economy has reached the zero lower bound—and money is just another liquid asset—minor shocks to the supply and demand of liquid assets can have catastrophic consequences.

We saw this in 2008. Even as the Fed eventually brought the federal funds rate down to zero, the rates that actually mattered in the economy shot upward.

What does this all mean?

For one, it suggests that the quasi-monetarists were right. As David Beckworth argues, there was a great liquidity shock in 2007-2009 that made monetary policy effectively very tight. This doesn’t mean that monetary policy can fix every problem—not all zero lower bound episodes are caused by a liquidity shock, and in those cases zero is a very real barrier. But when a shortage of liquid assets is the main issue, the Fed has a simple remedy: trade money for illiquid assets. If it does enough, it’ll bring the liquidity premium down to earth, and allow the interest rates faced by businesses and consumers to decline to levels consistent with economic stability.

It’s unintuitive to think that such a severe recession was provoked by shocks to a single market, shocks whose total dollar value was trifling in comparison to the subsequent downturn. But after a little reflection, there’s a natural connection to monetary economics. We already know how a quantitatively trivial shock can devastate an economy: the formidable Volcker tightening, which led to the highest unemployment rate since the Great Depression, was just a small blip in the monetary base. We understand all this with the market for money. Yet somehow we haven’t managed to make the connection to the market for liquidity.

Someday we will.

About these ads

39 Comments

Filed under macro

39 responses to “How did small losses cause the Great Recession?

  1. foosion

    The naive observer might think that an increase in demand for currency would result in a corresponding increase in the velocity of money, and all would be well.

    The naive observer might also think that the increase in Baa yield represented a risk premium rather than a liquidity shock. “It’s risky to hold debt of a business when demand for its products are falling.”

    How would you respond?

    • The naive observer might think that an increase in demand for currency would result in a corresponding increase in the velocity of money, and all would be well.

      What’s the channel for this? If I suddenly demand $20 billion more in currency, and the currency stock stays the same, what prompts everyone else to hold less currency? (and use money at a higher velocity to conduct their transactions…)

      The only channel I can imagine is the nominal interest rate: as interest rates increase, it becomes more costly to hold currency, and everyone disposes of it more quickly once they obtain it. But it might take a very big swing in the nominal interest rate for this to happen, especially in the short run—as I mentioned near the end of the post, in the early 80s a deviation of base money from trend that is barely even noticeable in the data managed to cause a massive increase in the nominal interest rate and the worst recession since WWII. And that change in the nominal interest rate has consequences outside the money market.

      The naive observer might also think that the increase in Baa yield represented a risk premium rather than a liquidity shock. “It’s risky to hold debt of a business when demand for its products are falling.”

      I agree that this was surely part of the explanation. But this is a secondary effect—why was demand for businesses’ products expected to fall so much in the first place? First you need some mechanism for the initial impulse (which was not exceptionally large) to cause a massive deterioration in projected demand and profits. I think that a liquidity shock is a very plausible candidate for this mechanism. Of course, once expectations deteriorated, there were other accelerators at work compounding the decline—for instance, the classic “financial accelerator” of Bernanke-Gertler-Gilchrist. But these don’t fully explain why losses in 2007-2008 were so much more damaging to the economy than losses in 2000-2001—that’s why I think the liquidity shock is so important.

  2. Scott Sumner

    Your comparison of interest rate targeting and monetary base targeting involves a sort of non-sequitor. You point out that a fixed growth rate of the monetary base leaves the econ0my exposes to nominal instability, if the demand for base money is unstable. But the exact same applies to interest rate targeting. Indeed with a stable fed funds rate the price level becomes completely indeterminate–hyperinflation or hyperdeflation are likely to occur.

    Of course those who favor interest rate pegs also tend to favor discretionary monetary policies, where interest rates are adjusted to keep AD on target. But some economists who favor MB targeting (i.e. Bennett McCallum) also favor changes in the MB growth rate to keep AD on target. For instance, he advocated a policy of using changes in the MB to offset changes in the velocity of circulation.

    Of course there may be a policy lag problem with McCallum’s proposal. Thus Aaron Jackson and I published a paper in 2006 (Ec. Inquiry) where we advocated adjusting the base to offset changes in expected future velocity (via a velocity futures market.) During each period the base would be set at a level that was expected to produce on-target NGDP growth over the next 12 months. Whatever fed funds rate that occurred in the free market with that base setting, would also be the fed funds rate setting that would produce on-target expected NGDP growth if interest rates were the policy instrument. So there is a symmetry between using the base and the fed funds rate as a policy instrument. Indeed, neither interest rates nor MB targeting is stable unless the instrument is adjusted to offset changes in the macroeconomy.

    If the Fed had maintained 5% expected NGDP growth during the financial crisis, it is very unlikely that nominal rates would have fallen to zero, and hence very unlikely that the Fed would have been forced to buy illiquid assets. That’s because at positive interest rates the demand for ERs is trivial, and the demand for cash is not very responsive to changes in the opportunity cost of holding cash. I’m not saying it’s impossible–you could imagine a banking crisis so severe that T-bill yields fell to zero despite on-target NGDP growth expectations. In that case the base might have to increase a lot–although I doubt the demand for base money would exceed the national debt. If it did exceed the national debt, then so be it–buy some illiquid assets such as German government bonds or corporate AAA bonds.

    In the real world nominal rates fall to zero because NGDP growth falls way below target—every single time.

    Of course the other way to overcome the policy lag problem (and the zero bound) is to do level targeting. There’s too much focus on the mechanical aspects of monetary policy at the zero bound–the problem is 100% about inflation/NGDP growth expectations. Solve that and the liquidity trap disappears.

    • Scott,

      Thanks for the comment and link. I’m afraid that I wasn’t very clear in that part of the post. My main goal was to highlight the similarities between an extremely simplistic base money target with zero discretion (a policy so ridiculously bad that none of us would ever advocate it) and what seems to be the actual policy (save for some ad-hoc QE programs) for managing the supply of liquid assets.

      I agree that it is unfair to use my reasoning as a criticism of all base-targeting schemes. At most, my argument only demonstrates that a completely rigid base target would be a bad idea—which we all already know. Using interest rates as an intermediate target helps avoid the specific problem that I highlight (sudden changes in money demand leading to a contraction), but a poorly designed interest rate target will result in other problems like indeterminacy, as you mention.

      While we’re on the topic, though, I think that market-driven schemes for adjusting the base to meet some target (like the NGDP target) would probably use interest rates as an intermediate indicator, even if the government wasn’t explicitly doing so. As long as the Fed isn’t conducting unconventional operations, the stance of monetary policy is fully captured by the set of expected state-contingent levels for the nominal interest rate. If market participants saw that the current level of the base was leading to interest rates high enough to be contractionary, they would sell NGDP futures until interest rates were brought into line.

  3. Pingback: TheMoneyIllusion » More links

  4. Suppose Mountain Dew was the medium of account, and prices were fixed in terms of Mountain Dew. That is economically equivalent to the price of Mountain Dew being fixed by law, but indexed to the CPI. In both cases, an increased demand for Mountain Dew would require an increase in the relative price of Mountain Dew to maintain equilibrium, but that relative price cannot adjust, so there’s an excess demand for Mountain Dew.

    But Mountain Dew is not the medium of exchange. Mountain Dew has a market of its own. There will be an excess demand in the market for Mountain Dew, and nothing else will happen. It will not cause a Great Recession. Rent controls on New York apartments cause an excess demand for New York Apartments. They are not what caused the Great Recession.

    You speak of “the market for money”. There’s no such thing. Every market is a money market. Every market is a market for the medium of exchange. That’s why an excess demand for the medium of exchange, unlike an excess demand for Mountain Dew or New York apartments, disrupts every single market in the economy, and can cause a Great Recession.

    Again, is your implicit (Woodfordian?) macro model a model of a monetary exchange economy or a model of a costless barter economy? If it’s a barter economy, then why don’t the involuntarily unemployed hairdresser, manicurist, and masseuse, simply barter each others’ services and get back to full employment that way?

    • Well, to make the analogy more fitting, let’s say that people get “Mountain Dew services” from holding on to Mountain Dew, even if they don’t drink it themselves (but these services are not related to any value as a medium of exchange—maybe everyone enjoys looking at the shiny green cans).

      The amount of Mountain Dew in the economy determines the marginal value of Mountain Dew services (if there are too many cans, people get sick of looking at them), which pins down the nominal interest rate (denominated in cans of Mountain Dew).

      Now suppose that I drink a lot of Mountain Dew. I still demand Mountain Dew services, so I need to replenish my stock of the drink. The problem is that the marginal value of Mountain Dew services changes dramatically after even a small movement in the stock of Mountain Dew—so when I try to replenish my stash of Mountain Dew, I bring down others’ Mountain Dew levels to the point where they obtain very high marginal utility flows from keeping it. This forces up the nominal interest rate, denominated in units of Mountain Dew. Expected inflation increases less than one-for-one, which means that the real interest rate shoots up and there is a contraction in the economy.

      This is possible even if the market for Mountain Dew is vanishingly tiny compared to the market as a whole—thus my “unit of account externality”. And it doesn’t rely in any way on Mountain Dew being a medium of exchange—in fact, I’m stipulating that it’s not, and that it’s just another “liquid” asset.

      Now, in the implicit Woodfordian model I’m using, the contraction is necessarily accompanied by an increase in markups. You ask why individuals don’t barter each others’ services. While this is an interesting critique, I’m not sure how relevant it is to the recessions we see—personal consumption expenditures on services barely dipped at all, and were only a very small contributor to the 2007-2009 recession. And I’m sure that expenditures on the kinds of services that could be bartered—haircuts, manicures, and so on—dropped by even less than expenditures on services in general. The goods whose production declined the most were produced by profit-maximizing companies, companies that list sticky prices and rig their markups to maximize expected profit, not employment or production. A disgruntled Boeing employee can’t barter with United to produce planes in exchange for free airfare.

      • Matt:
        “The problem is that the marginal value of Mountain Dew services changes dramatically after even a small movement in the stock of Mountain Dew—so when I try to replenish my stash of Mountain Dew, ….”

        I find I can’t replenish my stash of Mountain Dew, because there’s an excess demand for Mountain Dew, so I am on the long side of the market for Mountain Dew, and am rationed in my purchases, and there is nothing I can do about it. So the process stops right there.

        But if Mountain Dew were the medium of exchange, there *would* be something I could do about it. I would stop spending the Mountain Dew I earn as income, and replenish my stash that way. And when I stop spending my Mountain Dew, that reduces demand for all other goods and services. And if everyone else does the same, that causes a fall in Aggregate Demand and a recession.

        “A disgruntled Boeing employee can’t barter with United to produce planes in exchange for free airfare.”

        If barter were easy, he and Boeing would indeed want to barter planes for free airfares with United and its pilots, and United would accept the deal. It’s a mutually advantageous trade. Again, your (Woodfordian?) model is economically inconsistent because the conclusions only make sense if barter is hard and people have to use monetary exchange, and yet there is no medium of exchange in the model.

  5. Lee Kelly

    Your thought experiment is the reason why banks should be allowed to issue their own banknotes. Withdrawals of currency are not a drain on reserves when currency is not base money, but they would be a mere switch of liabilities from chequing accounts to banknotes. The macroeconomic consequences of gas stations demanding payment in currency would be trivial. But in the present circumstances, the Federal Reserve must adjust the quantity of reserves to compensate for changes in the demand for currency; this just adds one more complication to an already difficult task.

  6. Are we really so sure that an excess demand for liquidity is the problem here?

    Here’s another story. For whatever reason (possibly the Chinese currency peg, but perhaps not) the US is importing more than it’s exporting and it’s not possible to close the trade gap quickly (say because of supply lines are in place and there are time to build frictions preventing an immediate switch to domistic producition of the same basket of goods.)

    We thus find that net exports (X-M) is always negative and thus at any income level Y we have Y < C+I+G. Further at suppose that the trade deficit (X-M) is an increasing function of Y (for whatever reason, probably to do with consumption of manufacutured goods)

    Thus, in aggregate (public and private) we are always dissaving and in order to increase Y we must be induced to increase our debt to income levels (actually to increase the rate of increase of our debt to income level). One way to do it is have the government decide to take on ever more debt buying stuff but this is undesirable because the government can't be counted on to buy stuff we want.

    So, can we induce C and I to increase? Perhaps the only way to do induce the private sector to keep increasing it's debt to income ratio at a fast enough pace is to maintain high, and ever increasing, asset prices.

    This would explain why the only full employement episodes of the past 20 years went with asset bubbles and imply that our problem is that asset prices are too low. It would aslo imply that the supply of or demand for money is not at all the issue, nor is a lack of liquidity the problem.

    http://canucksanonymous.blogspot.com/2010/07/why-we-keep-having-bubbles.html

    • I think this was probably one of the other mechanisms at work, but the sudden acceleration in the contraction that occurred immediately after the financial crisis was probably the result of a liquidity shock.

      Let me restate your case to make sure that I understand it. The Chinese are sending a lot of capital to the US. This capital needs to go somewhere (there’s an acute shortage of “stores of value”), and without some sink for capital like the housing bubble it will force real interest rates to be very low, probably negative. Since the housing bubble absorbed so much capital, it allowed us to avoid a liquidity trap even though we had low inflation, because the equilibrium real interest rate was propped up enough by the bubble that we could hit it without encountering the zero lower bound on nominal interest rates.

      Once the bubble collapsed, however, there was no place for all the excess capital to go, and the equilibrium real interest rate fell dramatically. It became negative enough that the prevailing level of inflation was not high enough to allow us to avoid the zero lower bound. Thus we had a contraction.

      I think that this is the fundamental story for why there are recessions after bubbles collapse—the economy suddenly finds itself lacking a store of value for all the capital sloshing around, and the equilibrium real interest rate falls to a level where the Fed cannot avoid contractionary policy. This is a very, very important narrative. At the same time, though, I think we need to understand why there was an extra kick from this recession following the financial crisis—and there, a “liquidity shock” story seems most plausible.

      • Well yeah, my point really was that 4 years after things started going wrong, with all the policy interventions, I don’t see that the problem is still ” an excess demand for liquidity”, or money.

        Also, I tend to think that when you write down a model with only money and bonds as assets it’s clear that what you mean by “money” is all liquid assets and by “bonds” you mean illiquid, risky, yielding assets.

        It doesn’t mean money as base money.

  7. Lee Kelly

    Matt, I want to agree with you. I almost agree, I think. But I think the core problem is the excess demand for money, i.e. desired money balances exceed actual money balances. Do you agree with that? A change in the supply and demand for liquidity is not really important, or at least I think it is a misleading way of conceptualising the situation. I’ve written this example elsewhere and people seemed to understand, so maybe it will work again here.

    An excess demand for money is both necessary and sufficient to suppress aggregate demand. An excess demand for safe and liquid assets can only cause a recession by first causing an excess demand for money. But in saying this safe and liquid assets are no different from any other commonly traded good or service: in this sense, an excess demand for oranges could also cause a recession.

    Normally, an excess demand for oranges does not suppress aggregate demand, because prices adjust. But even with rigid price controls, an excess demand for oranges still does not suppress aggregate demand. People who are unable to purchase oranges just demand apples instead, e.g. the excess demand for oranges might be said to “spill over” into apples. The allocation of resources will be very much as though people never preferred oranges in the first place.

    One cannot enjoy money the way one enjoys oranges. If people are frustrated in their attempts to purchase oranges, chances are they will quickly spend the money on something else that roughly satisfies the same end, e.g. apples. Money is not a good substitute for oranges.

    The situation for even the nearest of near monies is just the same except that near monies are, by definition, good substitutes for money. Sometimes the next best thing to satisfy the same end as near monies is money itself. If there should be an excess demand for near monies because of the zero-nominal bound, demand is likely to “spill over” into money. That is, when people are frustrated in their attempts to purchase near monies, chances are they will choose to hold additional money balances instead.

    The difference here between oranges and near monies is one of their degree of substitutability with money. In theory, an excess demand for either oranges or near monies could “spill over” into money, but in practice one is much more likely to than the other. But whatever its cause, when the demand for money rises and is not accommodated by a proportional increase in supply, aggregate demand will fall. If this occurs quickly, then the general level of prices will be unable to absorb the change and production will be cutback and workers let go.

    Once the excess demand for money starts, it can snowball; previously safe and liquid assets become less so. With curtailed nominal incomes, otherwise safe debtors can begin to look risky (even the U.S. Government). Moreover, an asset is only liquid when their is a ready and willing buyer, but to purchase an asset requires money and money is in short supply. The liquidity position of portfolios dries up, intensifying the demand for assets like T-bills, which in turn causes even more spillover into the demand for money.

    The Fed’s response is simple. Safe and liquid assets with a near zero rate of interest are extremely close substitutes for money; purchasing these assets bonds may increase the money supply, but it is also likely to increase money demand in proportion, and so any excess demand for money will remain unchanged. Purchasing more risky and illiquid assets with higher interests rates, however, exchanges quite different things. Households and firms that sell these assets are unlikely to hold their new money (or else they wouldn’t be holding these assets in the first places), but instead they will begin spending.

    Of course, the Fed should clearly signal a long term growth path for NGDP and promise to do whatever is necessary to hit its target. That would go a long way to preventing such crashes in aggregate demand in the first place.

    • Matt, I want to agree with you. I almost agree, I think. But I think the core problem is the excess demand for money, i.e. desired money balances exceed actual money balances. Do you agree with that?

      I don’t see why this is true. Nick Rowe also says that excess demand for money is the problem, because money is the medium of exchange. But what kind of “money” are we talking about here? Since you’re drawing a distinction between “money” and other liquid assets, I assume that you’re talking about base money.

      That said, here is the “excess demand” story as I understand it: base money is the medium of exchange. When there is excess demand for base money, everyone will be trying to obtain more of it, which is impossible in the aggregate. One way to obtain more money is to spend less than you take in—again, if everyone tries to do this, no one succeeds, but we do see a “general glut” and a recession.

      This is all very bizarre to me. If you want more base money, I have a very easy solution: go to the ATM and withdraw some. I’m serious. Everyone with a bank account (i.e. virtually everyone with any spending power) has the ability to do this, and it is infinitely easier than changing your consumption decisions.

      Maybe I’m missing something basic, and this certainly sounds glib on my part, but I just don’t get it—if you want more base money, just go to the ATM and get some. It’s that easy.

      • Lee Kelly

        Matt,

        Withdrawing currency from an ATM is not an increase in money demand. Crediting your chequing account and debiting your currency account for the same amount is an exchange of one type of money for another, because both chequing accounts and currency serve as media of exchange. In this example, aggregate desired money balances remain unchanged; the demand for money remains unchanged. (Perhaps some of the repo agreements created by the shadow banking sector were also serving as media of exchange in a limited capacity.)

        An increase in money demand occurs when people attempt to build up their cash balances. Maybe they try to finance these larger cash balances by reducing consumption, but more likely (especially in the case we’re talking about) they finance them by reducing spending on financial assets. But remember, it is not because they really want more media of exchange; it’s because they want to hold safe and liquid assets. But they are in short supply, and when we hit the zero-nominal bound, that frustrated demand spills over into demand for money.

        When the excess demand spills over into money, and the money supply does not increase enough to satisfy the new demand, then we get an excess demand for money, turning a panic into a recession. The excess demand for money is crucial for explaining how the changes in relative spending become a sudden fall in aggregate demand.

      • Matt: “Nick Rowe also says that excess demand for money is the problem, because money is the medium of exchange. But what kind of “money” are we talking about here? Since you’re drawing a distinction between “money” and other liquid assets, I assume that you’re talking about base money.”

        I’m talking about something wider than base money. I’m talking about medium of exchange. (Admittedly, I’m not exactly sure where to draw the line in practice.)

      • David Beckworth

        An excess demand for money problem is an excess demand for all assets that serve as the medium of exchange. Thus, it isnot limited to base money. Most of us don’t use much base money (i.e. cash) as the medium of exhange in our daily transactions whereas we do use alot of checking funds, saving funds, money market funds. That is not to say the demand for base money is unimportant at all. Obviously for banks it is very important. In short, the money demand problem is when collectively there is an increased desire to hold more of all of these money assets. (This is one reason why I think it is useful to look at the expanded equation of exchange, BmV=PY, since it provides an indicator of the demand for base money–via the money multiplier–and for broad money–via velocity.)

        Now it is true that the Fed can only directly influene the moneary base, but appropriately setting expectations and properly managing the moneaery base it can also influene the supply of these other money assets as well. Thus, the need for an explicity nominal GDP level target.

      • Lee Kelly

        Note: when I say “cash balances” I mean money balances. In accounting it is customary to lump all money in the “cash” account; that’s why I sometimes write “cash balances” like in my previous comment.

        It’s a stupid habit: I am not even an accountant.

  8. Scott Sumner

    Nick, Are you sure the analogy is correct? With a medium of account we normally assume the market for that special good reaches equilibrium. Thus we assume the quantity supplied and demanded for Mountain Dew are always equal. The price level adjusts to maintain equilibrium–it’s the only way the relative price of Mountain Dew can change. In that case an increase in the demand for Mountain Dew must cause deflation. And I assume you agree that anything that causes deflation is also likely to cause a recession.

    The New York rent control analogy seems wrong. If these apartments really were the medium of account, then their supply and demand would always be in equilibrium, and all other prices would adjust. But the market for apartments is not in equilibrium, so the legally set “price” is not a market price, not the “value” of New York apartments. Indeed the true price is the nominal (legal) price plus queuing costs.

    I do agree that a non-barter economy is probably a necessary condition for this sort of macro disequilibrium. As I recall you insist that the importance of non-barter is something more than simply that barter presumably causes price stickiness to break down. (I seem to recall it depends on whether we use a sticky price model with monopolistic comp, or sticky wages with perfect comp.) I won’t address that issue here, as I fear my comment may have already misinterpreted your remarks.

    One other comment. Under the gold standard the government adjusted the supply of currency so that the gold market was in equilibrium at a fixed nominal price. The Fed does not adjust the supply of currency so that the market for NYC apartments is in equilibrium.

  9. Scott: “Nick, Are you sure the analogy is correct? With a medium of account we normally assume the market for that special good reaches equilibrium. Thus we assume the quantity supplied and demanded for Mountain Dew are always equal. ”

    I would disagree with that assumption. If prices are sticky in terms of Mountain Dew, the market for Mountain Dew may not clear.

    “And I assume you agree that anything that causes deflation is also likely to cause a recession. ”

    Anything that causes a fall in the *equilibrium* price level, where the actual price level does not fall immediately to that new lower equilibrium, is what causes a recession. Let P be the actual price level and P* be the equilibrium price level. It’s not the fall in P, or even the fall in P*, that causes a recession. It’s the fall in P* relative to P.

  10. To understand why relatively “small” losses on financial assets made by banks have such a large effect on the economy it is important to have a correct theory of money. Read The Riddle of Money for a model of the economy that incorporates endogenous money and the creation of money by bank lending (and its reverse: the destruction of money by a contraction of bank lending).

    There are two multipliers at play here. One is the multiplier that operates from the monetary base to the total bank lending. Effectively, banks that had a large hole blown in their balance sheets cut back on lending in a big way. The last graph in the above article shows that the effect on money caused by the fall in bank lending was not small. There was a 20% fall in money supply (in my case it is Corrected Money Supply, a new aggregate) from the peak at the beginning of 2006 to the trough in the third quarter of 2009.

    The second multiplier is the one from money supply to total income. Assuming a savings rate of 5% this multiplier amounts to 20 in the model.

    Taken together the two multipliers in the above model explain fully why relatively small bank losses (think of the size of TARP) lead to such large effects on the economy (think of the total stimulus measures taken by the US government).

  11. Lance

    Going back to the original point of the post, it seems small changes in the monetary base* could have pronounced macroeconomic effects if modeled within a model that has financial accelerator effects via credit markets and demands for liquidity. The work of Bernanke, Gertler, and Gilchrist come to mind.

    *I believe it was Ohanian and Cole who said that such a small change in the monetary base could not have precipitated the 1937-1938 recession which Friedman & Schwartz attributed to increased reserve requirements.

  12. Rob Parenteau

    Wondering if you have just rediscovered Keynes liquidity preference theory of asset prices? Keynes theory of interest rates relied upon this. Interest rates did not clear loanable funds market or equilibrate saving and investment flows in Keynes theory by the time he got to the General Theory. Interest rates equilibrated portfolio demand for liquid assets with relatively stable money values with the stock of available liquid assets. Less liquid asset prices must fall when liquidity preference increases unless the stock of liquid assets is increased. Seems like you are not saying much more than that, which is reiteration of Keynes in the GT.

  13. Scott Sumner

    Nick, I still don’t follow. The US used gold as a medium of account for decades. There was never any surplus or shortage of gold, despite sticky prices in gold terms. In contrast, there is a shortage of NYC apartments, and other goods for which the government fixes the nominal price, BUT DOES NOT ADJUST THE MONEY SUPPLY TO MAKE AN EQUILIBRIUM PRICE.

    Matt, you said;

    “As long as the Fed isn’t conducting unconventional operations, the stance of monetary policy is fully captured by the set of expected state-contingent levels for the nominal interest rate.”

    Yes, but that doesn’t necessarily mean people care about nominal interest rates–they may care much more about indicators of what you call the “state” (say CPI futures prices, or stock prices, or commodity prices, or the foreign exchange rate for a small country.) I’m also puzzled by why you make an exception for unconventional monetary policies. Does that include QE? (which is arguably just an OMO.) If QE affects the economy at all (in the Keynesian model), it’s presumably because it affects the future expected state-contingent path of interest rates. The main problem with interest rates is the state-contingent part. A sharp rise in interest rates over the next 5 years could be either an ultra easy or an ultra-tight monetary policy. We don’t directly observe the “states” in a way that allows us to do more than come up with crude estimates via very imperfect indicators like the Taylor Rule. In a perfect world where we could directly observe the Walrasian natural rate, I’d agree with you. But we are far from that world–making nominal interest rates an exceedingly dangerous indicator. For instance, many proponents of using nominal rates have assumed that money has been “easy” ever since 2009.

    Finally, aren’t state-contingent exchange rates (or other financial asset prices), also indicators of the stance of monetary policy?

    • The main problem with interest rates is the state-contingent part. A sharp rise in interest rates over the next 5 years could be either an ultra easy or an ultra-tight monetary policy. We don’t directly observe the “states” in a way that allows us to do more than come up with crude estimates via very imperfect indicators like the Taylor Rule.

      I agree that in practical terms, an NGDP target is far more practical than trying to lay out exactly what state-contingent interest rate rule would achieve the target path of NGDP. But since I think that the interest rate rule (and expectations of it) is the only lever through which conventional monetary policy works, even an NGDP target has to be working at some level through its effects on interest rates.

      Here is what I imagine would happen with a market-based NGDP targeting regime. With a sufficiently low level of base money, the nominal interest rate would be very high, at a level far too contractionary to achieve the NGDP target. Seeing this, investors would sell NGDP futures to the central bank, which causes more base money to come into the system. This would continue until the level of base money was sufficiently high that the nominal interest rate returned to a level consistent with hitting the NGDP target.

      Since the effect of monetary policy inevitably must work through interest rates, investors enforcing an NGDP targeting rule will be observing interest rates themselves when they decide whether to buy or sell.

      I’m also puzzled by why you make an exception for unconventional monetary policies. Does that include QE? (which is arguably just an OMO.) If QE affects the economy at all (in the Keynesian model), it’s presumably because it affects the future expected state-contingent path of interest rates.

      I think that affecting the state-contingent path of interest rates probably was the main effect from QE2, but it is possible that “unconventional” monetary policy will work in other ways. Conventional monetary policy works because the Fed has complete monopoly control over the supply of base money—it therefore has virtually unlimited power to adjust this supply so that the yield on overnight debt in terms of base money hits whatever target it desires.

      If we broaden our reach to the supply of “liquid assets” more generally, the Fed is no longer a monopoly supplier. But it’s still a big enough supplier (or, at least, it can be) that it has market power, and therefore it can affect the yield spread between liquid assets and illiquid ones. As long as the federal funds rate is above zero, it makes little sense for the Fed to attempt this, since it is far easier to affect nominal yields throughout the economy by using its monopoly power on base money to work at the “money/liquid assets margin”, rather than using its much weaker market power to work on the “liquid assets/illiquid assets” margin. But once it hits the zero lower bound, it is possible that it may want to try affecting the latter margin as well.

      Finally, aren’t state-contingent exchange rates (or other financial asset prices), also indicators of the stance of monetary policy?

      They are indicators, but they are just capturing the effect from the expected trajectory of nominal interest rates. I am not strong on international macroeconomics, but I believe that if you rule out significant effects from sterilized interventions, the central bank can only affect nominal exchange rates by changing the expected yield. Again, in practice the “state-contingent nominal interest rate rule” is a very elusive beast, and it may make sense to think about indicators like NGDP or exchange rates when discussing policy… but at some fundamental level, expected interest rate policy is doing all the work.

      (By the way, I’m not sure whether I’ve mentioned this before, but nominal exchange rates are a distinct channel through which monetary policy can have an effect. The others are altering real interest rates—which is what I believe most of the superficially separate “transmissions mechanisms” boil down to—and affecting real wages in an environment with sticky nominal wages.)

  14. Scott Sumner

    Matt, You said;

    “Maybe I’m missing something basic, and this certainly sounds glib on my part, but I just don’t get it—if you want more base money, just go to the ATM and get some. It’s that easy.”

    The most fundamental part of monetary theory is that individuals determine the real demand for base money and the Fed determines the nominal supply. Any individual can “get more money” but only from someone else–in nominal terms. Of course everyone can simultaneously get more money–even without the Fed producing any more—but only in real terms, i.e. through deflation. Without the fallacy of composition–sometimes called the “hot potato theory”–all of monetary economics collapses, there’s be no reason to consider money an important part of price level determination. If the Fed doubled the monetary base, and people could just “return the money they didn’t want to ATMs,” then there’d be no inflation.

    • The most fundamental part of monetary theory is that individuals determine the real demand for base money and the Fed determines the nominal supply. Any individual can “get more money” but only from someone else–in nominal terms. Of course everyone can simultaneously get more money–even without the Fed producing any more—but only in real terms, i.e. through deflation.

      I agree. The “go to an ATM” story was a glib way for me to emphasize that there is clearly not a “shortage” of base money—in theory, there can be, but no one is currently being stopped from costlessly trading other liquid assets for base money. In practice, if there was a shortage of base money, nominal interest rates would simply increase until there was no longer any shortage.

      One reason why I prefer talking in terms of prices (i.e. interest rates) rather than quantities is that our language seems to become really imprecise when we’re discussing quantities. Nick and others talk about “excess demand for money”, but to me it seems like their story would be better described as “excess supply of money”—in particular, that even when 0% is above the market-clearing nominal interest rate, it is always possible to obtain more of an asset that pays 0%. Nick points out that this is because money is the medium of exchange (otherwise it would just be rationed at the above-market interest rate and nothing would happen at the margin), which is a good point, but if the Fed was committed to meeting all demand for base money at its target rate, we’d get the same effect—and that seems like “excess supply” to me.

  15. Lee, Nick, and David:

    All right, I’m sorry for distracting the conversation by mentioning base money, then. I was trying to understand where our differences really were—and since “money” construed more broadly includes the “liquidity” I’m discussing in this post (though I admittedly wasn’t very explicit about what I meant), I figured you were referring to something more specific.

    But I don’t think that money being a medium of exchange is really the key here. It is tough to come up with a counterexample, because conventionally the medium of exchange is nominally riskless, and therefore will have the lowest yield among all assets, but here is my attempt…

    Imagine that everyone is mandated by law to use an asset called “scrip” in exchange. Scrip is not the unit of account, and the price path of scrip (including its expected nominal appreciation, which we call the “nominal yield” on scrip) can vary depending on the marginal transactions services it provides. In theory, the nominal yield can become arbitrarily negative, as the marginal transactions services provided by scrip go to infinity as the supply of scrip approaches zero. The price of scrip is completely flexible.

    Meanwhile, there is a unit of account called “bling” (thanks to Nick for this term). All prices, wages, and contracts are (stickily) quoted in bling, so that it is (nominally) riskless.

    Bling has a nominal yield of zero. The central bank’s policy tool is the federal funds rate—the yield on overnight debt in terms of bling. The central bank is obligated by law to buy and sell bling in a narrow corridor around its target rate.

    Suppose that the trend inflation rate is 3%. Some shock hits the economy and dramatically lowers the equilibrium real interest rate, to -4%. The central bank lowers the federal funds rate to 0%, but this is still not enough to achieve the equilibrium real interest rate, and the central bank is unable to generate higher inflation expectations. The artificially high real interest rate induces a recession.

    Now, I slipped in a key sentence there: “the central bank is obligated by law to buy and sell bling in a narrow corridor around its target rate”. Without this sentence, I don’t think my model would work: as Nick points out, bling would just be rationed, and nominal interest rates on other assets could fall to an arbitrary extent (at the margin, getting the guaranteed 0% by adding more bling to your portfolio is not an option). But I hope this shows that conceptually, the medium of exchange is not a fundamental part of a “general glut”; instead, what’s important is a specific property of the medium of exchange as we usually understand it (that it always pays a certain nominal yield *and* that it can never be rationed).

    If the unit of account is never rationed, then we have exactly the same phenomenon.

    • Matt: this post and discussion are very good. We are all trying to come to understand our differences (and our own positions) better.

      My brain isn’t working very well this week. Still coming back up to speed after a fortnight’s holiday on another planet. I tried to get my head around your thought-experiment above, and failed. My fault, not yours. But I want to try a simpler one.

      At root of the difference is your belief that, even if the monetary policy instrument is not an interest rate, the monetary policy transmission mechanism always works through interest rates. I can see that way of looking at the world; it’s the way I have been trained too, and how I sometimes think. But is it right?

      Here’s a simpler thought-experiment. A representative-agent economy in a stationary equilibrium. Because all agents are identical, no borrowing or lending takes places in equilibrium. Suppose the equilibrium interest rate is 5%. Assume the stock of money is fixed. Now assume that a crazed political party, the “Pro-Usury Party”, takes power. It passes a law saying that all loans below 10% interest are illegal. (And everybody obeys the law). What happens?

      I say nothing happens. There is an excess supply of loans at 10% interest, but since nobody is borrowing or lending anyway (being identical agents), that has no effect on anything. Since the stock of money stays fixed (by assumption), there is no change in the price level. But if you changed the stock of money, that would affect the economy in exactly the same way it would effect the economy if the Pro Usury Party had never gained power.

  16. Scott Sumner

    Matt, Thanks for the replies. I misunderstood your ATM comment–I should have known you understood the point I made.

    Regarding your other reply, I suppose I can’t really disagree with anything too strongly. When you say:

    “Since the effect of monetary policy inevitably must work through interest rates, investors enforcing an NGDP targeting rule will be observing interest rates themselves when they decide whether to buy or sell.”

    I assume you mean observe “state contingent interest rates.” Thus in principle if investors see nominal rates shoot up from 3% to 10%, it could be from tight money, or it could be from money so easy that it creates fear of double-digit inflation. If that’s what state-contingent interest rates means, then I have no objection to your claim.

  17. brit

    Hi Matt,

    I thought this was a great post, and got right to the heart of the matter, thanks a lot. One thing I would add is that you say the economics profession hasn’t worked this out yet, but I think Kiyotaki and Moore have in their new paper. Check it out, I think it’s awesome and formally models the stuff we’re discussing above:

    http://www.princeton.edu/~kiyotaki/papers/ChiKM6-1.pdf

    Thanks for the great post.
    B

    • Jeff Hallman

      @brit:
      I started to read the linked Kiyotaki and Moore paper and stopped when I realized that there is no medium of exchange in their paper at all. What they call “money” is just the more liquid of the two tradable assets. So what they’re really talking about is T-bills, not money. They haven’t modeled any of what Nick Rowe talks about.

      They do, in their introduction, mention why they don’t like a cash-in-advance constraint as a way to bring money into a model.

      Although the … cash-in-advance model … has proved important to monetary economics and policy analysis, it is not well-suited to answering larger questions to do with liquidity. By endowing money with a special function in the otherwise frictionless economy with complete Arrow-Debreu security market, one is
      imposing rather than explaining the use of money, which precludes the possibility that other assets or media of denomination may substitute for money.

      Of course, there is no such thing as a frictionless economy with a complete Arrow-Debrue security market, so what they’re really doing is assuming away the reasons for the existence of money and modeling the demand for something that isn’t the medium of exchange, i.e., that isn’t money.

      • Exactly. From Keynes to Friedman to the Austrians, economists have confused money as asset with money as medium of exchange. Friedman, for instance, reiterates everywhere that money as he considers it is an asset. And then he tries to relate it with things like GDP for which what matters is money as a medium of exchange. The same with Keynes and the same with the Austrians.

        The day economists realise that any sensible economic model must incorporate banks and that money can be created only when banks make loans (or when people move money from their savings deposits into their demand deposits, which is just a loan to themselves) they will have understood what money as. Hyman Minsky is the only one to have approached anywhere close to this model.

        Banks figure even in the example that Matt cites. When I use a credit card the bank is giving me a loan. When I cannot use credit cards the money supply in the economy contracts.

  18. John

    Rognile doesn’t understand, “why was demand for businesses’ products expected to fall so much in the first place?”

    China, which will not accept US imports, and the closing of 59,000 factories +/- in Bush’s 8 years

    What is it that economists have against thinking? Any reader of James Clavell knows that business people to their bes to climb up mountains and look out into the Sea, searching for Flying Cloud.

    Real economists, list Robert Drugger, even come up with names for such activities, like Fiscal Adjustment Cost (FAC) discounting (how company executives and investment fund managers assess the effects of fiscal adjustments on company balance sheets and earnings).

    If fool who observed the tax cuts and other plans of George Bush stopped investing in this Country.

    I mean, seriously people, have you ever been to Plato, Missouri, now the center of country, and asked yourself, “My God, what madness made us think that these people can make a living if we open the doors to competition from China?”

  19. Jovf

    You’ve produced some incredibly good posts recently. This is one of the few blogs I still read with fascination. Truly, thank you.

  20. Pingback: Why do safe, liquid assets become so expensive in a financial crisis? | Matt Rognlie

  21. Pingback: Why the zero lower bound matters | Matt Rognlie

  22. Pingback: What’s needed in a macro model? | Matt Rognlie

  23. Pingback: The dark, dark era of money demand | Matt Rognlie

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s