In one important sense, there is a "bubble" in US government bonds, and we should be worried about it.
There is always something normative in saying that the price of something is above the fundamental value, so that people are paying "too high" a price, more than they "should", more than its "real value". I want to make that normative element explicit.
If the US and world economy were operating as it should be, with expected and actual inflation higher than it is, expected and actual real growth higher than it is, and the expected and actual return to investment in real assets higher and safer than it is, the price of government bonds would be lower than it is. And the price of real assets (like houses) would be higher than it is.
I approach the question of "fundamental value" as a macroeconomist, not a microeconomist — from a general equilibrium, not partial equilibrium perspective. If we define the "fundamental" value of an asset as the price that asset would have if all markets, not just the market for that asset, were in long-run equilibrium (and with inflation at target), then bond prices are above their fundamental values. And if we define "bubble" as a price above that fundamental value, then bond prices are a bubble.
A bubble in house prices is a bad thing. It will cause over-investment in building new houses, under-investment in other things, and under-consumption. A bubble in bond prices is a much worse thing. It will cause under-investment in everything, and under-consumption in everything, because it causes under-employment of everything. That's because bonds and money are close substitutes. A bubble in bonds causes a bubble in money. And a bubble in money can cause a bubble in bonds. Or perhaps they are just different aspects of the same bubble, in both money and bonds.
It's not the bubble in bonds per se that is the big problem. If there were only a bubble in bonds, and no bubble in money, it would be no worse than a bubble in houses. It might lead to the wrong mix of real investment and consumption (presumably too little real investment and too much consumption, due to a wealth effect). It's when a bubble in bonds spills over into a bubble in money, the medium of exchange, that we get a big problem. An excess demand for the medium of exchange is what causes, and is the only thing that can cause, a general glut of all goods. And that causes employment and output to fall, and both consumption and investment to fall.
That's why we should be worried about the bond bubble.
If the US and world economy returned immediately to long-run equilibrium, and expected and actual inflation increased to target, the price of US government bonds would immediately fall. And people who held bonds would suffer a large loss when the bubble burst. But perhaps it won't return to long-run equilibrium for a long time. That is what holders of bonds must be forecasting, because if they are right in this forecast, their decisions to hold bonds at current prices are rational.
And maybe they are right. Who am I to know better? But, like all bubbles, the beliefs that sustain the bond bubble are, at least partly, self-fulfilling. The bond bubble, and the associated money bubble, create the general glut, and prevent the economy returning to long-run equilibrium. And the belief that the economy will not return soon to long-run equilibrium is what sustains the bond bubble.
We need to burst the bond bubble. Bursting the bond bubble will help the economy recover more quickly.
Worthwhile Canadian Initiative 
Nick, let me approach this situation as a composite of many situations like this: Someone has been living a bit high off the hog for a few years – not crazy in-debt spending but spending their entire salary. The economy goes south and things are uncertain so this person cuts back a bit on spending and starts saving a little money. Because the economy is uncertain and they are afraid they might lose their job they put the money they save in safe investments: gov’t bonds.
So there is over-investment in bonds as a collective. But surely for each individual person like the one I described, it is the right thing to do.
Therefore in the situation a government policy has to stimulate us out of the mess. Either much lower interest rates or spending. How does the government burst the bond bubble given that it is right for my hypothetical person to save some money now in very safe investments?
More generally – surely government policy is hardest when government must take actions which run counter to the actions individuals are taking.
Nick: interesting post.
I don’t agree with the mechanism you posit – a massive demand for bonds causing zero interest rates and this translating into a massive demand for cash because the two are pretty much similar at those rates.
If Microsoft 10 year debentures fall to 6%, and the yield payed by Microsoft 10 year senior bonds happens to be 6%, you can’t say individuals will now be indifferent to the two bonds because they pay the same interest rate.
If Microsoft 10 year senior bonds fall to 5%, the same level at which Intel 10 year senior bonds trade, that doesn’t mean individuals will be indifferent between the two.
As in the above examples (and I could provide plenty more), money and bonds (the actual stuff, not the idealized stuff in economist’s models) have many different features built into their fine print, of which interest rates are only one dimension. To top it off they are also issued by entirely different institutions. It seems to me that there should be far more to the story you tell than just rates, although if that’s all you wish to focus on, then cash and bonds will indeed by substitutes at 0%.
David,
“A good chunk of these assets are “hard” assets, like mortgages (ultimately backed by property, in the event of default)”
But mortgages are not hard assets; they are soft assets that happen to be secured by hard assets. Even if fiat money becomes worthless, the mortgagor can still discharge his obligation by paying in fiat money. A bank may have some hard assets, because of repossessions and such, but it usually tries to get rid of them as soon as possible. Moreover, if a bank defaults, its depositors are only entitled to claim their hard assets up to the value of their claims, which are denominated in the unit of fiat money and therefore only as valuable as fiat money.
Suppose a borrower were to have issued paper denominated in tulips during the Dutch tulip bubble. And suppose this paper were secured with something more substantial, such as senior mortgages. When the tulip bubble collapsed, the paper would still become worthless (or nearly so), no matter how solid the securing assets and the other remaining assets of the debtor. Tulip certificates have no more intrinsic value than actual tulips (which is admittedly not zero, but during the peak of the tulip bubble, it was a small fraction of what the exchange value would have been). Bank money is to fiat money as tulip certificates are to tulips.
Andy H. @ 7:04,
“Bank money is a claim on the bank’s net assets. Virtually all a bank’s assets are claims payable in a fixed value of fiat money. If fiat money has no intrinsic value, then neither does bank money. We can argue about whether fiat money has intrinsic value in its own right, but it makes no sense to start from the premise that bank money has intrinsic value.”
That occurred to me.
A bank is a collection of financial assets, liabilities (typically, mostly deposit money), and capital.
As you say, bank money is effectively a (senior) claim on bank assets.
And bank assets are themselves financial claims payable in fiat money.
Assume fiat money has no intrinsic value.
And assume the line of reasoning that suggests that because bank money is effectively an ultimate claim on fiat money, it also has no intrinsic value.
But surely this must mean that financial assets held by the bank also have no intrinsic value, because they are also claims payable in fiat money. If bank money has no intrinsic value because it is an ultimate claim on fiat money, how can bank assets have intrinsic value when they are also a (more immediate) claim on fiat money?
Moreover, banks can hold just about any type of financial asset.
So this in turn must mean that no financial asset has intrinsic value, because all financial assets are claims payable in fiat money.
Unless that conclusion is acceptable – that no financial asset has intrinsic value – there would seem to be a contradiction in the assumption that bank money has no intrinsic value – at least using this line of reasoning.
@Nick: I don’t understand your response to my point at all; the math is not at issue, and Gordon growth model is fine (I agree with your math). V, as you put it, has no bubble component, and no expectation of capital gains is needed to get the value (as with the treasuries) – you get the value from the cash flows.
The bubble part is where you do the estimate of the cash flows, the valuation doesn’t work (P is much greater than V, even when plugging in new expectations of growth in cash flows), and instead ‘the investor’ (speculator) breaks the model into cash flow for some period plus a terminal value that is arbitrarily higher (with no explanation of how that future sale price is determined). The point is, the ‘bubble price’ requires capital gains to make sense, requires someone else in future to pay an absurdly high price (the underpants gnomes theory of finance, if you will).
Or, with somewhat more sophisticated buyers/sellers, Gordon Growth valuation is ‘fixed’ by inserting high growth rates/inappropriately low interest rates.
For your T-bill example, yes, there are values implicit in the discount, but this is not a capital gain, just interest (cashflow) paid out. In your example, ‘bubble’ pricing for a T-bill would be at or above par, because investors would seem to be hoping that someone is going to magically pay them more for the T-bill than cash at some point before redemption – i.e. counting on capital gains rather than cashflow. Anything less is just low interest being paid.
JKH:
“Unless that conclusion is acceptable – that no financial asset has intrinsic value…”
It’s acceptable to me (although I would note that there are quasi-financial assets, such as TIPS, that clearly do have intrinsic value).
Of course, it does lead to a paradox: if the only value of money is liquidity value, then how can illiquid financial assets have any value at all? My resolution of the paradox is to argue that all financial assets are essentially bubble assets, which have value only because they are expected to have value to others, for no particularly good reason (although I think the bubble is a Nash equilibrium, but maybe not subgame perfect).
Then again, money does actually have a small intrinsic value, in that it can be used to pay existing debts (including, but not limited to, outstanding taxes). If the “intrinsic” demand for money is sufficiently inelastic (i.e., if debtors have a sufficiently strong aversion to default), then the government can, by controlling the supply of money, make the intrinsic value arbitrarily high. Then as long as people are expected to continue contracting debts in monetary terms (which the government can assure to be the case by contracting such debts itself, via tax liability, and also encourage by providing debt enforcement mechanisms that only work for monetary debts — i.e. making money legal tender), the government can promise to control the ultimate value of money in such a way as to validate its current value. So perhaps financial assets are not all a bubble.
It seems to me that there is an element of circularity in the argument that money is valuable for its liquidity. What is liquidity? To me, liquidity is confidence in the value that something will have at some ultra-short horizon starting at some arbitrary point in the future. An artwork is illiquid because, even if it is quite valuable at a long horizon, it risks having a value of zero at an ultra-short horizon if I can’t find a buyer. Money is liquid because, no matter how short the horizon, I’m confident that it will have value. So liquidity is (a certain sort of) confidence in the value of something. But if the value only exists because of the liquidity, doesn’t that put me in kind of a Tinkerbell situation? Why should I be confident in the value of money if the only reason it has value is that people are confident in its value? That sounds remarkably like a bubble: I’m enough of a fool to hold money specifically because I know that the world is populated with a very large and widespread number of greater fools.
Andy and JKH, setting aside whether or not federal reserve debt has any intrinsic value I think you can certainly argue that bank debt does have intrinsic value.
Here’s JKH: “And bank assets are themselves financial claims payable in fiat money”
Let’s imagine a world without a government issued currency, does that mean bank debt wouldn’t exist? No, of course not. Bank assets are often claims to real resources.
If a bank loans you the money to buy your house then the house secures the loan, if you don’t pay the bank back it can take the house and hand it over to its creditors. If the bank can also transfer the claim on the house over to its creditors as payment of its debt.
Now, would anything be fundamentally different if the bank was just a big cooperative institution where you go when you need to build a house. The bank supplies you with all the materials and labour to build the house and you agree to pay the bank back by supplying a certain amount of your labour (or the materials if you have them) in building the houses (or whatever else) of others that the bank loans to. So here a bank loan is the loan of actual resource and you pay back with actual resource, is that really different from what we actually do?
So, put succinctly and expanding beyond houses, the bank supplies you with the labour and materials to build whatever it is you want and you pay it back with some combination of labour and materials supplied by you to the banks other customers.
Of course then when multiple banks appear and start transferring the claims to resource amongst themselves you might end up supplying your labour to someone from another coop or receiving services from them but what’s the difference. As long as everyone lives up to their obligation to whomever is currently holding the claim on their resource it works.
The bank’s money are just the transferable claims to the resources, the fact that the management of this system of claims is facilitated with outside money is not relevant once we’ve established where the outside money derives its value (and I’d tell a similar story about the government using resource and issuing transferable debt in return, then taxing you in real resource, where you can return the debt to government in order to extinguish the tax liability).
JKH and Andy I got myself off on a rambling tangent in the last comment, sorry about that. Let me try this again.
The point I want to make, which I guess is what JKH has been saying, is that most bank assets are secured by claims to things, like houses, that have intrinsic value. Thus, bank debt, which is secured by those assets, acquires the same intrinsic value.
When Andy says that bank assets are just claims to fiat money he refers to the legal tender aspect of currency, if you give the bank enough outside money then they can’t take your house from you. But the legal tender property works for the bank on its liabilities as well and works for the creditors of the bank, thus I’m claiming it’s a wash.
So, while the legal tender property of government issued currency aids its liquidity value it doesn’t change the intrinsic value of the bank’s assets. And since the bank debt is secured by those assets it aquires the same intrinsic value.
So I think JKH is right here.
But if the value only exists because of the liquidity, doesn’t that put me in kind of a Tinkerbell situation? Why should I be confident in the value of money if the only reason it has value is that people are confident in its value? That sounds remarkably like a bubble: I’m enough of a fool to hold money specifically because I know that the world is populated with a very large and widespread number of greater fools.
Andy: yes, money/liquidity is a bubble (in the sense that the object in question is valued beyond its underlying fundamental value; the premium being, at least in part, generated by a self-fulfilling expectation that others will believe in the liquidity properties of the object in question).
Not sure that I agree with your “greater fools” interpretation. It can be shown theoretically that when there is an “asset shortage” (e.g., a la Ricardo Caballero), a bubble can make everyone better off. Collectively, the construction of a bubble can be rational. The widespread emergence of assets whose values are based largely on their ability to facilitate payments suggests that there is some empirical relevance to this theory. (There are downsides to bubbles too, but I wish to emphasize the other side of the argument here).
David, I think I am assuming that values of assets equal their fundamental value. A $1 federal reserve note (or dollar bill) has a fundamental value of $1 doesn’t it?
Scott: I don’t think there necessarily is any right or wrong way to define “fundamental value.” And it appears that we are defining the term differently. Imagine that you were stranded on Gilligan’s island. What is the intrinsic value of your USD? It may serve as toilet paper, perhaps. This is my definition of “fundamental value.”
Andy H.,
“Money is liquid because, no matter how short the horizon, I’m confident that it will have value. So liquidity is (a certain sort of) confidence in the value of something. But if the value only exists because of the liquidity, doesn’t that put me in kind of a Tinkerbelle situation? Why should I be confident in the value of money if the only reason it has value is that people are confident in its value?”
I usually think of liquidity value as something that is reflected in yield rather than capital value per se.
The yield of elementary fiat money is zero. The holder of fiat money is paying for the benefit of liquidity by accepting a yield as low as it is.
Yield eventually amounts to changes in accumulated principal value over time. So liquidity value/cost is reflected in expected changes in value rather than present value per se.
Liquidity is part of a broader risk spectrum. Other risk components include credit risk and interest rate risk, also reflected as value through yield components. There is arguably no credit or interest rate risk in the most elementary form of fiat money (currency I think).
Fiat money is arguably risk free. It is at the top of the monetary/financial capital structure.
Bank money is one level down, with the help of bank capital and FDIC insurance.
Other financial assets range down below that.
So my starting point in characterizing elementary fiat money would be that it is more or less risk free, in the sense of liquidity risk, credit risk, and interest rate risk.
The absence of these risks is reflected in a zero yield, as well as in the absence of changes in value over time.
So rather than characterize fiat money as comprised of either intrinsic value or liquidity value, I would simply define it as risk free capital.
And I’d say its risk free because the state can issue fiat money without limit, and therefore the state can never be forced (involuntarily) to default on its obligations.
Bank reserves are also a form of fiat money. I think it’s debatable whether currency or bank reserves are the more elementary form of fiat money. Bank reserves are really a mechanism required in the creation of a banking system structure whereby not everybody needs to bank directly with a single central bank.
But bank reserves are (no longer) precisely risk free, due to interest on reserves. The principal value may be certain, but commercial banks actually face a minor interest rate risk dilemma in that the Fed can always change the interest rate on reserves. That’s an interest rate risk that has no effect on the principal value of reserves, but it is still a well defined risk in the sense of its potential impact on future commercial bank net interest margins.
David, I’m not sure I follow the desert island example. My air conditioner has value to me. If I was stranded in Nome, Alaska, it would not have value to me. It seems to me that “fundamental value” can be contingent upon the environment in which you live.
Gold is supposed to have “intrinisic value.” But seriously, would you rather have a gold bar or a can of beans if stranded on a desert island?
JKH, Cash earns a real yield in liquidity services. In economics there is no distinction between an asset that earns a cash return, and one that gives you a return in a good or service. Some transactions are easier to make with cash, and hence they save people time. That is a service. (Of course the dirty secret with cash is that most is held to evade taxes, or to hide illegal transactions. So in that case the “service” is the taxes you don’t have to pay.)
Scott,
I see no conflict between a real yield in services and a zero monetary yield.
In fact they’re quite compatible.
This is like watching a combined episode of Entourage and the Sopranos. My entire fantasy economist blogging team on one thread. But I’m starting to think I should have drafted Leland Yeager.
I don’t find Nick’s definition of bubble useful because, as Scott and Adam P. said, it requires conditional speculation about knowing the equilibrium real interest rate and inflation target at full production. The most important hole here is that while natural rates may be expected rise when an output gap is filled, we have no idea where they should rise to. There are of course multiple full production equilibria with different structures of real rates, based on changing variables in the euler equations — including risk aversion. Real risk free rates were declining even during the boom. Arguably we had a bubble in falsely identified risk free assets. Maybe the real problem is that we have a bond puddle (caused ultimately by the zero bound) and with current preferences (not just beliefs!) and demographics we need higher bond prices (even if the natural rate would rise a bit with full production). You can’t just say bubble and assume that question away.
Money is normally defined as a bubble because its value is ultimately dependent on other people valuing it, with no end user. Alas, the effectiveness of this Ponzi scheme actually creates end user value (“Liquidity”) and throws us into a semantic frenzy. Ebay can be called a bubble in the same way (end user value contingent on mass use of something that would be worthless without mass use). Yes, this makes all nominal assets “bubbles” in some fairly uninteresting way. This terminology doesn’t mix well with the other definition of bubble — violation of EMH or “fundamental value,” because whether we call these Ponzi-turned-instrinsic values fundamental or not is a matter of preference. But Nick wants to distinguish here between a bubble that is sustainable at full output (fiat money) and one that may not be (possibly current bond prices). But unfortunately we don’t know enough about the world to really do that, I don’t think.
David, I’m not sure I follow the desert island example. My air conditioner has value to me. If I was stranded in Nome, Alaska, it would not have value to me. It seems to me that “fundamental value” can be contingent upon the environment in which you live.
Scott,
Yes, “fundamental value” (the way I defined it) can be contingent in the way you describe. Indeed, in an Arrow-Debreu securities market, different assets will be valued according to their time and state-contingent properties.
Let me consider another island economy – that of Robinson Crusoe (prior to Friday). Can we agree that everything on that island will be valued according to its “fundamentals” by Crusoe? To put things another way, none of the assets on that island would be valued for their ability to facilitate payments. There can be no liquidity premium. The “fundamental value” of a dollar bill would be equivalent to that of toilet paper.
Adam,
“bank debt, which is secured by those assets, acquires the same intrinsic value”
Y’all keep saying stuff like that, but it makes no sense to me. Suppose I believed that fiat money were fundamentally worthless and would soon revert to its fundamental value. How would I speculate to profit from that expected reversion? I would buy a house with a mortgage.
Suppose I then turned out to be right. What would happen? There would be a hyperinflation. I would pay off my mortgage with worthless fiat money, and I would keep the house. The bank’s depositors would be screwed. All their silly bank money would be shown to have no intrinsic value. They couldn’t get their hands on my house or anything else with intrinsic value.
Bank money (along with all other financial assets) has only a derived value, derived from the value of fiat money. It is all but irrelevant that there may be conditional claims on real assets involved. Those conditional claims exist only to secure the derived value — to make sure that the actual derived value is as high as the stated derived value.
If I write an option and secure it with a margin deposit, the margin deposit adds no intrinsic value to the option over and above its derived value; it merely serves to guarantee that derived value. If the option expires out of the money, the margin deposit becomes completely irrelevant. It’s the same way with mortgages and bank money: bank money is a derivative (like an option), and the conditional claim created by a mortgage is just like the conditional claim created by a margin deposit. Generally, when we talk about valuing derivatives (especially if they’re guaranteed by a clearinghouse) we ignore the value of the margin deposit and simply assume that it is adequate to guarantee the derivative. Why should it be different with bank money?
Andy,
I think you make an excellent point, but then carry it too far.
Let me see if I understand your basic point correctly. First, assume a legal tender regime (all debts dischargeable in the legal tender). Second, assume a fixed exchange rate regime (all bank money is legally stipulated to exchange at par with the legal tender). Third, assume that the legal tender is fiat (in the sense of being intrinsically worthless). Then it follows that bank money must be intrinsically worthless.
To demonstrate this claim, imagine that I take out a $100K bank loan (banknotes or electronic digits created by the bank and credited to my account) to purchase a $100K home. I transfer my bankmoney to the account of the original homeowner; the latter is now a “depositor.” On the surface, it appears that the bank loan is collateralized by the $100K home. But now, imagine that hyperinflation makes fiat worthless. I can discharge my bank loan at zero cost. The depositor discovers that his “deposits” are also worthless — there is no hard asset backing up his bank money.
This argument seems right to me. But does it imply that bank money is intrinsically worthless? I do not think so. You have only described one contingency in which it becomes worthless. Let me now describe another contingency.
Imagine that the price-level remains constant over time. Repeat the home purchase scenario described above. Now, imagine that I refuse to go to work to acquire the money credits I need to service my mortgage. Is the depositor screwed in this case? No, the depositor (via the bank) has recourse to my property. The bank liability in this event is evidently NOT intrinsically worthless.
If what I just said makes sense, then your statement that “Bank money (along with all other financial assets) has only a derived value, derived from the value of fiat money” goes a little too far. It might be more accurate to say that some of the value is derived from the value of fiat money, but not all its value.
Thoughts?
You can define intrinsic value as uniformly zero because of risk (Andy H.?).
Or you can define intrinsic value as risky but not zero until it is zero.*
And it depends whether you are defining value in real or nominal terms.
The case of hyperinflation actually gives bank money enormous protection in nominal terms, and in a way that still distinguishes it from fiat money in its ultimate connection to real assets.
*For example, this is consistent with the options definition of intrinsic value.
Seems to me that Andy, in the back of his mind, is thinking that intrinsic value is not state-dependent. Perhaps that’s what makes it “intrinsic”?
Personally I agree with what David A. said above, the fact that nominal debt holders lose in inflation is no different from anybody holding an asset whose real value is falling.
As JKH says, if the claim allows you to seize an intrinsically valuable asset in some states then, before the state is revealed, the claim has some of that intrinsic value. But the claim’s value is conditional on the state so perhaps Andy is saying that it isn’t “intrinsic” to the claim.
Andy?
Andy: “Bank money (along with all other financial assets) has only a derived value, derived from the value of fiat money.”
I agree with you in part. Your observation applies to a modern western banks that haven’t faced inflation threats for decades. But I don’t think it is a universal rule for bank money always-and-everywhere. As economies tip towards hyperinflation, bank’s will begin to build into the debt they create the sorts of features that will protect them when fiat becomes worthless.
Andy: “all financial assets are essentially bubble assets”… Here I’d disagree. Your points seem to apply to bonds and other securities that promise future fixed cash redemption, but most definitely not stocks. Stocks are claims to underlying assets – even when fiat becomes worthless, a stock promises its holder accessibility to some portion of a factory, warehouse, or whatnot. Who cares what the unit of account is.
David,
What I’m saying is that, if fiat money is a bubble asset, then so is bank money, because the value of bank money is dependent on the value of fiat money. If you give me an example where fiat money retains its value, then the bubble is still going on, so naturally the derivative asset retains its value.
If I understand the law correctly, even if I default intentionally, the bank doesn’t have the right to take my house for its own use. It only has the right to recover the amount that I owe, and the amount that I owe is defined by its fiat money equivalent.
My debt is a promise to pay in fiat money (although bank money will usually be accepted as a substitute, but only because my mortgagee trusts other banks to pay it in fiat money if necessary). If it were unsecured, its value in terms of fiat money would be lower, because of the greater risk that the fiat money value might not be recovered. So part of the fiat money value comes from collateral. The collateral happens to be intrinsically valuable, but its intrinsic value is irrelevant; all that matters is its fiat money value.
Suppose that instead of mortgaging my house, I pledged a Treasury bond as collateral. Would the resulting bank money really have less intrinsic value than if it were secured by a mortgage? (Or suppose that a bank had its entire loan portfolio secured by T-bonds. Would its deposits really have less intrinsic value than those of a bank whose portfolio were secured by mortgages?) Only in a very unusual scenario (which becomes impossible as the maturity of the collateral approaches zero) would the bank recover more from mortgage-secured loan than from the bond-secured loan. I submit that the collateral is only a substitute for fiat money, and its intrinsic value is irrelevant except to the extent that it makes it a better substitute for that (putatively) intrinsically valueless asset.
I think that there is a lot of confusion over what the job of the government is. JKH says that a bond bubble is better than a money bubble, I agree, but only if the government is prepared to do its job. The bond bubble should be the first step in a process to stabilize the economy.
The governments job is to collect and distribute information and “money” into the economy in such a way as to get the “optimal” usage of our resources. Optimal is defined as to capacity (whatever that may be, but employment and growth are probably good indicators)
If the government is doing its job it the bond bubble lets it borrow cheaply and invest in medium term projects (let’s say research and education) to create jobs that have a chance to induce growth. There is no draw back since that money is going to be spent of EI, Social assistance, etc anyway. This investment will be taken into account and the “bond bubble” will deflate naturally because expectations of inflation and growth arise.
The real problem is that the persistence of a long term bond bubble (10 year bonds yields are falling still) is an indicator that people believe (with good reason) that the government is failing to do its job and what we are going to see is not a self fulfilling prophecy of deflation and an economy operating under capacity (high unemployment) due to a bond bubble, but the exact opposite.
People are giving money to the government because they are supposed to be the ones who know what’s best to do with it. It’s backed by their faith that the economy will grow and that those in charge won’t mess it up and default on payments the way the private sector would. The government is supposed to act unselfishly, that’s the only safe investment there is. Unfortunately that’s not happening.
If the bond market bursts without government stimulus its because we are imploding and people have lost faith in our ability to run an economy at all. God help us if it happens.
Suppose a large depositor who is not FDIC insured loses 50 per cent of his money in a bankrupt bank.
Then the fiat money value of the deposit liability is irrelevant.
The depositor has lost money because the intrinsic value of his bank money has declined, according to the apportioned breakup value of the bank.
Fiat money is risk free. Bank money is not risk free (in this example).
So bank money doesn’t derive its value in all states from fiat money.
Andy,
Well, you certainly have an intriguing way of thinking about the issue. I am not completely convinced, but I see where you’re coming from. Let me mull it over…
Thanks,
DA
I’m just back from vacation and I see y’all have been having fun.
David: I’d like to understand your alternative interpretation (from early in the thread) better. I don’t yet see it is plausible, but I’d bet that you’ve thought about the possibilities harder than I have.
“The “excess” demand for bonds … reflects a rational “flight to relative safety” in the face of some fundamentally depressing news concerning the likely future return to capital investment.”
This statement confuses me. I associate “flight to safety” with a story about a change in level of risk. Depressing news about the returns to investment sounds to me like a “first-moment” sort of story. Both sound very different to me from the sort of interpretations I was hearing in 2009, when some said that the price of risk had suddenly increased. How do you see this?
I have a skeptical but not very informed prior on interpretations that rely on changes in the future return to capital investment. This prior reflects my impression that
1) our ability to forecast variations in the private return to capital investment is poor at both short and long horizons.
2) given the positive surprises in US productivity growth over the past 2 years, our forecasting models do not make us revise long-term US productivity growth downwards.
3) We’ve seen broadly similar falls in nominal and real interest rates in Canada and the US, despite quite different fiscal and productivity performance.
How do you think that I should update my prior?
JKH:
Couldn’t you apply an analogous argument to any secured derivative (such as a futures contract secured by a margin deposit)? We don’t normally say that pork belly futures are derivative of both pork bellies and margin. We just say they’re derivative of pork bellies.
My main point is that the additional security doesn’t make bank money any less of a bubble than fiat money. If anything, perhaps it prevents bank money from being more of a bubble than fiat money.
By analogy, if pork bellies are in a bubble, and you’re long in the futures market, your position is just as precarious as if you were long in the spot market, regardless of whether your counterparty has posted adequate margin. Perhaps if your counterparty didn’t post adequate margin, then you’re doubly speculating on the pork bellies and on the financial health of your counterparty. But if your counterparty has posted adequate margin, that doesn’t add any fundamental value to the overvalued pork belly position you’re holding.
“If I understand the law correctly, even if I default intentionally, the bank doesn’t have the right to take my house for its own use. It only has the right to recover the amount that I owe, and the amount that I owe is defined by its fiat money equivalent.”
Andy, here you’ve gone too far. The bank certainly gets possession of the failing debtor’s property. Haven’t you heard of seizures of collateral upon bankruptcy? Foreclosures? Repossessions? The repo man?
JP,
When a bank forecloses, it auctions the property and pays off its associated expenses and, in almost every case, ends up recovering less than it was owed. In the very few cases where it recovers more than it was owed, I believe it is required to return the difference to the debtor.
Simon,
Welcome back. Yep, the fireworks have been blazing. It’s that trouble-maker Nick at it again!
You are right to be confused by the “flight to safety” phrase. In the model I had in my head, it is a simple portfolio substitution that rebalances E[MPK | n] = R, when n falls (a bad news event). See here for details: http://andolfatto.blogspot.com/2010_07_01_archive.html
You are right, this is a simple first-order moment story. I could add risk-aversion and higher-ordered moment shifts, but this seemed like plausible first-pass. Your remaining comments question the plausibility of this interpretation. I certainly do not believe that this was the only thing going on. But in any case, let me address your questions:
1) our ability to forecast variations in the private return to capital investment is poor at both short and long horizons.
OK, no problem with this. However poor these expectations are, all I need is form them to move around in response to information relating to future fundamentals.
2) given the positive surprises in US productivity growth over the past 2 years, our forecasting models do not make us revise long-term US productivity growth downwards.
I don’t really trust aggregate measures of TFP (composition bias; the weak have been cleansed, so measured productivity increases). And I trust the predictions of forecasting models even less! Are you really so sure that it was unreasonable to revise long-term US productivity growth downward during the depths of the crisis?
3) We’ve seen broadly similar falls in nominal and real interest rates in Canada and the US, despite quite different fiscal and productivity performance.
I’m not sure why you would use this as evidence against a “news over future capital return” interpretation. Complicated interactions between monetary and fiscal policy can generate all sorts of outcomes. And mix this in with surprise shocks to realized productivity (relative to earlier expectations) and I’m not sure what to expect. (Why do you ask — are there other interpretations out there consistent with these facts?)
How do you think that I should update my prior?
It’s really up to you to decide that! I do think that this interpretation should be explored in greater depth however (there are people doing this). The classical writers, including Keynes, stressed the importance of expectations in generating fluctuations in investment. Declines in what they called the “marginal efficiency of investment” were responsible for leading agents to substitute in money and bonds. I was just trying to propose this as an alternative to the interpretation offered by Nick.
Andy H.,
Let me try and respond to your last point at Aug 29, 8:56 by regrouping and leading up to it:
Let’s assume it’s agreed that fiat money has zero intrinsic value. I’ve never thought about it in quite those terms, but I can live with it OK, and it seems to be more widely accepted.
Fiat money in its elementary form – currency – is risk free. There is no risk to the nominal value of currency. It may be viewed as the risk free asset.
Assume a simple bank balance sheet with bank assets, bank deposits, and bank capital.
Bank money itself is not entirely risk free. As per my earlier example, a large depositor can end up taking a loss in a bankrupt bank.
But most of the time, bank money is risk free. You might say that on average it’s very nearly risk free. Bank money effectively has a senior claim on the value of bank assets, and is protected from first loss asset price erosion by bank capital. The residual exposure to depositors is effectively an out of the money short put option position on bank asset value. Accordingly, bank capital protects bank depositors from most of the downside risk to bank asset value. In addition, small depositors enjoy a long put option to the FDIC so that they are fully protected in bankruptcy.
My point here is that in order for bank money to enjoy a risk free attribute nearly equivalent to risk free fiat money, specific capital protection (private and FDIC) is required.
That type of protection is not required for the pure risk free asset itself.
Where does the protection for the pure risk free asset come from?
It comes from its fiat characteristic – not from a risk management overlay on a risky asset.
You might say it’s sui generis.
Thus, the nature of the (gross) risk structure of fiat money and bank money is very different.
Why is this relevant?
It’s relevant because it might affect one’s perception/identification of intrinsic value in each case.
We are agreed there is no intrinsic value in the case of the pure risk free asset – fiat money.
The risk free attribute of fiat money is “sui generis”. But the (near) risk free attribute of bank money is the result of elaborate financial engineering.
Basically, due to the capital/option structure, bank money is (near) risk free because most of the risk in the assets on which bank money has a senior claim is “put” to capital providers (private and FDIC).
The risk absorption effect of both types of capital is such that bank money has a senior claim on an equal nominal value of assets whose aggregate risk has mostly been stripped away. This happens when bank asset value erodes to the point where capital is gone and bank asset value equals bank deposit value (in the assumed simplified balance sheet).
Therefore, bank money effectively has a senior claim on financially engineered (near) risk free assets of equivalent nominal value.
That senior risk free claim is payable in fiat money.
But that’s very different than the idea that bank money is a claim on fiat money.
Rather, bank money is a claim on risky assets whose “risk adjusted” nominal (near) risk free value (as per above) is payable in fiat money.
Bank money can be exchanged for fiat money.
But bank money is not a claim on fiat money, due to the residual risk that can affect the nominal value of bank money (as in the bankruptcy example).
It’s on that basis that I would continue to argue that the intrinsic value characteristics of fiat money and bank money are very different. Fiat money has no intrinsic value. But bank money has an intrinsic value that is a function of its risk structure as described.
Moreover, that intrinsic value of bank money is itself is subject to risk, which is what really differentiates it from the property of fiat money that makes fiat money purely risk free.
I’m not quite sure how to interpret the “bubble” issue at this point or how to interpret your analogy as per your Aug 29, 8:56 p.m. I see what you’re saying about the redundant value of collateral or margin posted when the market in question is in a bubble, but I’m having great difficulty trying to fit this in precisely with the discussion here. I’m not saying it’s not applicable; I’m just not seeing it clearly yet. The closest I might come is to suggest that when the call option value of bank capital is very high – i.e. when asset values are increasing – then any purported differentiation of intrinsic value as between fiat money and bank money is less binding in interpreting what’s going on. But I don’t think the fact that it becomes less relevant in the call option zone means that it is not relevant in the put option zone. That said, it’s very possible I’m still missing your point, because I’m having difficulty getting comfortable with it.
I am in the camp that does not like your bubble definition. Yes I know – it’s your blog, and you can define what you want to. But if your purpose is to communicate with the outside world, arbitrary definitions are counterproductive.
The distinctive feature of bubbles as commonly understood is that the positive feedback element is as follows: “I am buying because I think others will buy.” That is not what is happening with bonds. People are buying bonds because they don’t want to buy anything else, not because they are counting on greater fools. There remains the usual paradox-of-thrift feedback loop, but that is not the same thing at all.
Now, you could argue that “I am not buying non-bond assets because I think others will sell.” That is kinda-sorta self-reinforcing in a bubble-like way, taking figure for ground. But 1) that is not in fact what you are arguing, and 2) it isn’t true anyway.
Andy: “In the very few cases where it recovers more than it was owed, I believe it is required to return the difference to the debtor.”
Ok, I see your point now. Such debts are denominated and redeemable in fixed nominal money amounts (even when repossessed), so their value depends on the given fiat unit of account. Makes sense now.
It is not universal that all debt contracts derive their value from fiat money. In the old days lenders protected themselves with gold clauses.
Your observation that the value of all financial assets is derived from fiat money’s value goes too far. Seems to me it only applies to certain these types of assets, the ones paying fixed nominal money amounts… most definitely not equities.
David:
Thanks for the reply! Didn’t get a chance to check yesterday, so I just saw it today.
“However poor these expectations are, all I need is form them to move around in response to information relating to future fundamentals.” No problem there. We both understand that, with rational forecasts, models with poor explanatory power have forecasts that are not sufficiently volatile. Just show me your forecasting model. I honestly do not know of any forecasting model that (a) gives rational forecasts and (b) gives such volatile forecasts now.
“Are you really so sure that it was unreasonable to revise long-term US productivity growth downward during the depths of the crisis?” I’ve never seen any forecasting model that fits the data and would support such a forecast. If you’ve got one, many of us would love to see it. If you’ve got one that investors also seem to be using, it would shift my prior even more.
Regarding point 3, what I had in mind was some commentators (Plosser comes to mind as just one example) arguing that the steep rise in US fiscal deficits has created a sudden and large revision of future expected marginal tax rates which in turn is responsible for the collapse in US investment and interest rates. I haven’t seen compelling evidence for this.
But reading your comments here, I’m wondering whether we substantially agree in the following sense. (a) we think that Nick is most likely correct, but (b) we recognize that there is a degree of ambiguity that keeps us from being completely certain. Research on alternative explanations (such as those you put forward) might even do good! (as part of a balanced research agenda.) Shifting my priors, however, would be much more effective if you could refer me to the sorts of forecasting models I mentioned above.
The bubble in bond prices exists only in crazy minds of fixed income traders. Shut them down and get to something more productive and worth talking about. Any bond of US government will pay 100 at maturity with coupons which were fixed at issue date.
Simon,
Looks like I am about to learn something (again).
You appear to hang much on this notion of a “rational forecasting model.” Can you please explain what you mean by this? In my mind, I have an image of an econometrician working with some limited set of historical data X, imposing some restriction (e.g., linearity) on this information set, then deriving some forecast (over some variable of interest) by some method (like minimizing SSE). The econometrician then discovers that, based on this estimation procedure (and the limited information at his/her disposal), it would have been “irrational” for business sector participants to forecast an unusually low return to planned capital expenditure.
If I have that just about right (I realize that I may have missed the boat entirely), then I’m not sure what to make of it.
Expectations are notoriously difficult to observe directly. The actors in the economy almost surely have more information than the econometrician. There is indirect evidence that expectations are tremendously volatile; evidence in the form of asset price movements, for example. The preciptuous drop in capital spending is another (indeed, it is precisely these phenomena that led Keynes, Pigou, and other before them to hypothesize the capricious nature of business sector expectations).
I try to put myself in the place of someone, in the depths of the financial crisis, contemplating a planned capital expenditure at that time. I don’t think I would have paid much attention to the forecasts generated by your forecasting model. I would have likely thought that it was a terrible time to invest, stemming from what I would have considered a justifiably bearish outlook.
Now, trying to distentangle whether my bearish outlook was in fact based on “fundamentals” or whether it might have been the product of some “animal spirit” is an interesting question. But either way, it is easy to see how a depressed economic outlook could lead investors to flock to government bonds, increasing their price, and leading to the “bond bubble” that concerns Nick here. Basically, I am reversing the causality put forth by Nick. Who is right or wrong, I have no idea. But I do think my interpretation is one that should be considered.
Does this make any sense?
David;
Thanks for your reply….I’ll answer as best I can.
I would have thought that we mean the same thing by a rational forecasting model; a projection (not necessarily linear) on an information set the provides “efficient” conditional forecasts (by efficient here I just mean that the conditional expectation of their forecast errors is zero and that their conditional variance is minimized. I’m pretty sure this is just what you were thinking of.
However, you go on to say “it would have been “irrational” for business sector participants to forecast an unusually low return to planned capital expenditure. ” I don’t think either of us said or implied that. I mean, we could say that a certain forecast is inconsistent with the forecast from a given model, but we both understand the importance of the conditioning information, no?
But your next statement surprises me. “Expectations are notoriously difficult to observe directly.” How would we know? It seems to me that this conclusion is as hard to reach as the one you wrote about someone behaving “irrationally.” And for similar reasons. Both require important auxillary assumptions that I try (and sometimes fail) to keep in mind. In the case of measuring expectations, some measures of expectations (I’m thinking of surveys, for example) don’t line up well with what they would have to be to satisfy rationality tests, or finanancial market volatility, or other things. I would have thought that this implies a choice: expectations are poorly measured or people and markets don’t quite behave the way our simple models predict (or both, of course.) My business card claims that I’m a finance professor; I teach my students that there are serious, smart, well-trained people I respect in both camps. But it seems that you have something to teach me about microeconomics and the measurement of expectations.
I also want to respond to the question of what someone in the depths of the financial crisis should do in terms of investment. The little that I know about the physical investment literature (you’re the macro guy, so feel free to correct me here) is that cashflow plays a key role in determining investment levels. There’s plenty of empirical evidence for the US (and Canada I think) to support this and plenty of theory to explain why (such as asymmetric information and the differing net costs of financing from internal cashflows, external debt and external equity.) I also think there’s quite a literature drawing on this that argued that bank lending or other forms of credit played a special role in some recessions and that systemic financial problems which impaired companies’ access to external finance at times when cashflow was low would have unusually severe effects. My intuition is that in such models, the expected profitability of investment need not vary to explain flucuations in investment. Rather, the cost of external finance to the firm could be where much of the action is.
I put that kind of mechanism up against what I know about forecasts of output and financial returns. On the output side, recessions look like short-lived, transitory phenomena. Professional forecasters don’t seem to be able to predict recessions more than about 2Q into the future. And if there is predictability in financial market returns, it is pretty small. If I’m therefore contemplating physical investment projects (buying a fleet of jetliners, building an oil pipeline, etc.), my growth forecasts don’t move much for when those things come online. I therefore have trouble reconciling volatile expected returns with both the historical evidence on output growth and professional private-sector forecasts.
So I see two models in this case; one emphasizes access to finance and its real cost while the other emphasizes expected profitability. Both have microfoundations. But (so far), when I ask about the expected profitability story, I have yet to get the same level of empirical support that I thought exists for the other side.
But, as a macroeconomist, I’m sure that you’ve followed this more closely than I have.
Simon,
Wrt your first and second paragraphs, OK.
Wrt your third paragraph. I am delighted to report that I am just as surprised by your reply!
I don’t want to put words in your mouth, but are you suggesting that expectations are easy to observe? How would we know whether expectations are easy to observe or not??? I don’t get it. Personally, I have no idea what people are truly expecting…do you? My technique is to observe behavior (e.g., plummeting investment and asset prices) and then try to interpret observed behavior in the context of some economic model embedded in my brain. One (certainly not the only) interpretation is that expectations (over future profitability of investment) have fallen. I can’t tell whether we are agreeing or disagreeing here. Please straigten me out here (I ask for some leniency, as I am presently jet lagged).
Wrt your fourth paragraph, and in particular, your italicized comment. I agree that desired investment may theoretically contract even in the face of relatively stable profit expectations. I did not mean to suggest that my interpretation is the only plausible one (I simply offered it as an hypthesis worthy of discussion). Debt constraints that bind more tightly is another hypothesis, and one that I have some sympathy for. However, I am led to ask what caused these debt constraints to bind more tightly (or the cost of external finance to increase)? I think it may have had something to do with lower profit expectations, perhaps on the part of the creditors. I also wonder how this interpretation squares with the evidence. My understanding is (I could be wrong) that the business sector is flush with cash. If this is true, then how does one square reduced investment with bullish (is this true?) surveys of professional forecasters?
Wrt to your fifth paragraph. Just because recessions have been short in the recent past, does not mean that they should be forecasted to be short in the future. I have in mind here “growth slowdowns.” Japan, perhaps?
Thanks for bearing with me!
Simon: “We both understand that, with rational forecasts, models with poor explanatory power have forecasts that are not sufficiently volatile.”
If the variable we are trying to forecast is I(0), this seems clear. But what if it’s I(1), or I(2)?
For example, a rational forecast of white noise never changes at all. A rational forecast of a random walk is as volatile as the walk itself. But if the growth rate were a random walk, wouldn’t that give you more volatility?
Or are you just saying that, empirically, we just can’t seem to make them volatile enough?
David;
I’m sympathetic about your jet lag. Been there. Done that. I’m sure that, after a good night’s (day’s?) sleep, you’ll reread my third paragraph and clearly see the difference between what I wrote in my 3rd paragraph and your reply. (You’ll also see that I’ve already answered your question.)
Asking what caused the debt constraints to bind tightly is a good question. It also has many good answers that do not depend on lower profit expectations, but rather depend on the ability of the financial sector to extend credit.
I also did not think that the business sector was flush with cash. I’d be interested to see recent data on business cashflows and investment. (That’s the kind of thing that can move my priors.) I did not think that the latest SPF surveys were bullish although I’ve mentioned that they still do not put high odds on a “double-dip”.) But see for yourself:
http://www.philadelphiafed.org/research-and-data/real-time-center/survey-of-professional-forecasters/2010/survq310.cfm
As I’ve said, their contraction odds are not really useful beyond 2Q (which gives me an excuse to shamelessly plug a forthcoming paper: http://svannorden.files.wordpress.com/2010/07/galbraith-and-van-norden-ijf.pdf )
As for the “Japan scenario”, I’m just wondering what model you have in mind that produces this as a “rational” expectation? Even Krugman and DeLong are more optimistic than that!
test
test