Suppose an increased supply of government bonds to finance deficit spending were met with an increased demand for government bonds from the private sector, so that interest rates stayed the same, and the bond issue were fully subscribed. That is exactly what would happen for a bond-financed tax cut under full Ricardian Equivalence, for example. I would interpret that "very successful" bond auction as very bad news for fiscal policy. It means fiscal policy won't work.
A failed bond auction however is good news.
Let's take an extreme example (because it's simpler): the "nightmare scenario". Suppose people are willing to hold the existing stock of government bonds, but refuse point blank to buy any more, at any interest rate. What would that mean, for fiscal policy, and for aggregate demand?
It would not mean that the government could not run deficits; it would mean that deficits would have to be 100% money-financed. Government deficits that are 100% money-financed are the most powerful form of demand stimulus. Bond-financed deficits increase future tax liabilities, which may cause some crowding out of private consumption or investment demand. Money-financed deficits do not increase future tax liabilities, unless they cause inflation and thereby increase the inflation tax. But they won't cause inflation unless they do increase aggregate demand. And if they do cause inflation, and expected inflation, that reduces real interest rates and causes an even bigger effect on aggregate demand.
So, the "nightmare scenario" would not neuter fiscal policy; it would make fiscal policy more powerful. That's good news.
If people refused to buy more government bonds, this would mean that all private savings would henceforth have to be held in either money or private debt. Some people would continue to save, and choose to hold money. Some people would continue to save, and choose to hold private debt. And some people would stop saving, and would spend instead.
The first group, who switch from bonds to money, are not a problem. The central bank will accommodate their desire through an open market purchase of bonds. That's neither good nor bad news, providing the central bank responds properly.
The second group, by switching from government bonds to private debt, will reduce the interest rates on private debt, and so encourage investment demand. That's good news.
The third group, who stop saving, will increase consumption demand. That's good news.
With the honourable exception of Leigh Caldwell, everyone else I read seems to think failed bond auctions are bad news. Their reasoning seems to be that failed bond sales mean higher interest rates on government bonds, which will cause higher interest rates generally, and high interest rates mean low demand for goods. This is a mistake, for three reasons:
First, because if people refuse to hold more government bonds, and interest rates rise when the government tries to sell more bonds, the central bank can simply substitute money for bonds and bring interest rates back down again.
Second, because the refusal to buy more government bonds means that government bonds are seen as less attractive relative to other assets, so the interest rates on government bonds should rise relative to interest rates on other assets, rather than pushing up all interest rates.
Third, because high interest rates do not (always) mean low demand for goods. This confuses shifts in the curve with movements along the curve. Take an IS curve, add in expected inflation, risk-premia, and everything else that creates a wedge between any given interest rate and the implicit marginal rates of substitution and transformation between present and future spending. The job of the central bank, if it wishes to increase demand, is to lower the rate of interest and move the economy down along the IS curve. The job of the fiscal authority, if it wishes to increase demand, is to shift the IS curve upwards, and to raise the equilibrium interest rate for any given level of income, so the central bank can lower it again and move the economy down along the IS curve.
I would be much more worried if government bond sales did not put upward pressure on interest rates and force the central bank to step in and lower them again. If an increased supply of bonds from the government were met with an increased demand for bonds from the private sector (as it would be for a bond-financed tax cut under full Ricardian Equivalence), that could mean that the IS curve did not shift, and fiscal policy would fail.
If bonds and money were perfect substitutes, of course, fiscal policy could shift the IS curve and not put upward pressure on interest rates. The "failure" of recent bond auctions suggests they are not perfect substitutes. This implies that an open market purchase of bonds by the central bank will also work. And that's good news too.
Morning Nick,
First: "First, because if people refuse to hold more government bonds, and interest rates rise when the government tries to sell more bonds, the central bank can simply substitute money for bonds and bring interest rates back down again."
This I think is just false, if the central bank could bring down long rates by printing money it wouldn't need to buy the bonds directly. The fact is, short term real rates are already low and we still need more stimulus. Thus bringing down long term real rates is the next logical step. However...
Second: "because the refusal to buy more government bonds means that government bonds are seen as less attractive relative to other assets, so the interest rates on government bonds should rise relative to interest rates on other assets, rather than pushing up all interest rates."
Well, I think the original idea was to bring down government real yields, essentially getting people into real assets by taking away the supply of risk-free assets. To the extent that required returns are the sum of the government yield and a risk premium then you'd be lowering the required return by lowering the govie yield. On the other hand, if you get people into real assets by making the gov't yield look riskier and thus lowering the yield premium that people require to hold the real assets that could work just as well and the equity rally could be an indication of just that. So here I basically agree, but with the exception of a huge caveat for which I need to think about how to express succinctly...
Third: yes, I basically agree with your reasoning in the last two paragraphs. But, high real risk-free rates mean high consumption growth. This goes with a high demand for present consumption/investment goods if consumption is expected to be even higher in the future. Is that the case here? It could be, I agree that what we are seeing is consistent with a general belief that we are basically near the bottom in the real economy. On the other hand, what if the government debt is just aquiring a real risk premium? Then central banks can't step in and lower rates, we basically just don't have a risk free asset. Then fiscal policy fails by the mechanism you describe
Posted by: Adam P | March 28, 2009 at 06:54 AM
Nike, under what conditions would a lack of interest in UK government bonds be a bad thing?
Posted by: pointbite | March 28, 2009 at 08:12 AM
Nick: "Suppose an increased supply of government bonds to finance deficit spending were met with an increased demand for government bonds from the private sector, so that interest rates stayed the same, and the bond issue were fully subscribed. That is exactly what would happen for a bond-financed tax cut under full Ricardian Equivalence, for example. I would interpret that "very successful" bond auction as very bad news for fiscal policy. It means fiscal policy won't work."
The thing is, even withoug the government fooling anyone, this sort of thing needn't mean fiscal policy completely fails. After all, if demand is temporarily low then the government takes some of the idle cash, spends it and amortizes the debt over the next 30 years. Since all of the spending is today but paying it down is spread over many years when, hopefully, real income will be much higher, you still have an effect today. As far as I see though, this only gets around full Ricardian equivalence if the real interest rate on government debt is below that of the private sector (debt or equity) and I guess, maybe below the growth rate of income over the long term (or something like that). If the government can't borrow at a low enough real interest rate to grow away the debt the what? People don't want the bonds but we're still back to full Ricardian equivalence.
Posted by: Adam P | March 28, 2009 at 08:23 AM
I'm with pointbite on this, this just isn't likely to be good.
Posted by: Adam P | March 28, 2009 at 08:25 AM
pointbite: A lack of interest in government bonds would be a bad thing if we were trying to reduce demand (or even not trying to increase it). Especially bad if we needed to have big temporary government expenditure for some other reason (like a war or similar emergency), and couldn't raise taxes (or didn't want to), and didn't want demand to increase. In fact, under almost any conditions except for right now, it would be bad news.
Oh, and if we didn't have our own central bank it could be bad news.
Morning Adam! And you were up early too (or in a different timezone)!
I sort of ignored the term structure of government bonds. If the government wanted to sell 20 year bonds, and nobody would buy them, the central bank would need to do an open market purchase of 20 year bonds, to prevent interest rates on 20 year bonds from rising. (I also ignored the idea that fiscal policy and quantitative easing might possibly be complementary policies, rather than substitutes. I'm not sure if that's right, but it seems to be an idea worth thinking through.)
"On the other hand, if you get people into real assets by making the gov't yield look riskier and thus lowering the yield premium that people require to hold the real assets that could work just as well and the equity rally could be an indication of just that. So here I basically agree, but with the exception of a huge caveat for which I need to think about how to express succinctly..." Yes, that idea was at the back of my mind. I started to write a post on the economic consequences of people believing you caught cooties from owning government bonds. But it got too difficult to figure out a metaphor where the risk of catching cooties increased with the national debt. So I switched to assuming that people were happy to hold a certain amount of bonds, but refused to hold any more at any price. An exogenous cooties-premium would still be neat to work out, but probably not so closely-related to current experience.
" On the other hand, what if the government debt is just aquiring a real risk premium? Then central banks can't step in and lower rates, we basically just don't have a risk free asset. Then fiscal policy fails by the mechanism you describe" I think it depends on whether money also has a risk-premium (which it would if it were an inflation risk rather than a default risk), and on whether the risk premium on bonds were increasing in the quantity of bonds outstanding (as in my cooties discussion above).
8.23: there are many reasons Ricardian Equivalence could fail. But I can't see how Ricardian Equivalence could be true if the supply of government bonds did not create its own demand. If, for some reason people wanted to save all their tax cut, but use it to buy private bonds rather than government bonds, private borrowing for investment should increase, I think?
I'm still trying to get my head around all the possible reasons why a government bond auction might fail, what people would do instead of buying bonds, and how this would affect aggregate demand.
Posted by: Nick Rowe | March 28, 2009 at 09:24 AM
I like this post. I have a comment and a question:
1. Wouldn't there be data from the secondary bond market that resolved some of this. I know very little about bond auctions, but I assume that if the auction failed, then long term yields on government bonds must have jumped higher the day of the auction. Otherwise, why would the auction fail? If so, couldn't you resolve one of the issues you raise by comparing what happened that day to long term yields on gilts, with long term yields on corporate bonds. My hunch would be that both would rise, but the corporate bond yields would rise less.
2. My initial reaction was that a failure of the bond auction would be due to higher nominal growth expectations, and that would be a good thing. There is a fairly strong correlation between nominal GDP growth expectations and nominal yields. (Both incorporate expected inflation, and the real expected GDP growth rate is positively correlated with the real interest rate.) So a fiscal policy expected to produce higher nominal growth, should raise nominal interest rates, and this might cause the bond auction to fail. Corporate yields would rise less, as economic growth would reduce default risk. Perhaps I am just restating your point in different language.
Posted by: Scott Sumner | March 28, 2009 at 11:29 AM
I think pointbite had a good point in the last thread (one that he has been making for a while): that the introduction of risk to the most risk-free savings instrument may not necessarily lead to spending or private investment, but rather to second-best forms of saving. Gold & Oil I see as somewhat unlikely candidates for a variety of reasons.
In an open economy, a shift towards buying the sovereign debt or forex of other countries is what I would see as being most likely. I guess the usual candidate for this would be the Swiss, but they have already undertaken QE. I think that this was probably coordinated with UK & US partially to minimize capital flight. I'm not sure how many other places there are to save relatively risk-free (including QE/forex risk) these days. maybe Germany? I dunno, I don't really know a whole lot about bond markets, but I could see a flight towards non-QE countries.
In a situation where the investment flows out of the country, I don't see devaluation necessarily boosting real growth. If import prices just go up, but there is no growth in domestic investment, and no growth in wages, and exports don't improve quickly enough, you can get cost-push inflation & inflation expectations without growth. I guess what I'm saying is that there is potential for a temporary stagflation (USA, Spring 08) that is bound to slide back into outright deflation (USA, August 08). You need some sort of wage inflation to really anchor general inflation IMO.
Posted by: bob | March 28, 2009 at 12:14 PM
Thanks Scott!
1. The second Bloomberg story I linked in my previous post said that 40-year gilt yields rose 10 basis points, with smaller rises on shorter gilts. I don't know what happened to UK corporate bond yields on the day. My hunch too would be that the spread decreased.
2. I hadn't thought about that expected real growth story. I'm not sure if it's a different channel, or the same equilibrium effect looked at from a different perspective. I think the former.
By the way, though I haven't thought this through clearly, I'm wondering (as I said to Adam above) whether we (you, me, others) should start thinking about quantitative easing and fiscal policy as complements, rather than alternate ways of trying to increase demand. I can dream up a theory in which one would fail, and I can dream up a theory in which the other would fail, but I find it hard to dream up a theory in which both together would fail. So the effect on confidence of the joint policy could be much greater than the sum of the two independent policies. Sell 40-year (or whatever) gilts to the central bank, and spend the proceeds, and tell everyone that's exactly what you are doing.
Posted by: Nick Rowe | March 28, 2009 at 12:21 PM
"So the effect on confidence of the joint policy could be much greater than the sum of the two independent policies. Sell 40-year (or whatever) gilts to the central bank, and spend the proceeds, and tell everyone that's exactly what you are doing."
Bingo. That's my favored policy. In a general sense I still see QE on its own as pushing on a string. Fiscal stimulus I view as being better because it is like pulling on a string. But the best would be pushing and pulling the string at the same time.
I think that a coordinated policy between a number of countries would be ideal. The basic case for the US would be:
1) Congress announces somewhere in the neighborhood of 3-5 trillion in fiscal stimulus
2) The Treasury issues 3-5 trillion in long-term treasuries
3) The Fed prints 3-5 trillion to buy them
If the US pursued this course in isolation, the dollar would get slaughtered, but if it works according to plan, that would be a good thing. The important part is that fiscal stimulus will support wages and domestic demand during a shift towards more balanced trade (you don't have to count on private investment/spending from QE alone), giving the export & domestic industries time to build on their new-found competitiveness. The degree to which the forex swing takes place would be dependent on how many other countries take part in this stimulus plan, and in particular how Asian nations respond.
Posted by: bob | March 28, 2009 at 12:40 PM
this is the kind of expectations trap that I think monetary policy can't really break on its own, which fiscal stimulus would help to break:
Downturn Hurts Expectations in Portland
A Downturn Wraps a City in Hesitance
http://www.nytimes.com/2009/03/27/business/economy/27portland.html
Posted by: bob | March 28, 2009 at 02:14 PM
bob: an individual can save by buying gold, but if all people try to save by buying gold, they fail (because they can only buy it from each other), but the price of gold rises (I'm ignoring the effect of a higher gold price on new production). Either the rise in gold price satisfies their desire to accumulate wealth, and so savings stops, and they start spending on consumption goods, which is fine; or when gold gets expensive they decide to continue saving but in something other than gold. Not sure if it would be the same with oil.
Yes, I realised after I had posted that I was implicitly assuming a closed economy. I expect really I had the whole world at the back of my mind, rather than a specific country. Otherwise I would have had to analyse the exchange rate, and that gets tricky. If the exchange rate depreciated that might cause a one-shot rise in the price level. I agree it would be simpler if all countries did it at the same time.
But $3-$5 trillion would be big for the US. If I'm right -- that money financed deficits would be powerful -- I wouldn't think you would need that much.
Posted by: Nick Rowe | March 28, 2009 at 02:37 PM
Scott: "....couldn't you resolve one of the issues you raise by comparing what happened that day to long term yields on gilts, with long term yields on corporate bonds. My hunch would be that both would rise, but the corporate bond yields would rise less."
I just checked the FTSE100 http://ca.finance.yahoo.com/q/bc?s=^FTSE&t=5d It fell on Wednesday, around the time of the auction, but then recovered it all in the afternoon. But then Paris and Frankfurt(?) did about the same thing.
http://ca.finance.yahoo.com/q/bc?s=^FCHI&t=5d
http://ca.finance.yahoo.com/q/bc?s=^GDAXI&t=5d
So if stock prices tell the same underlying story as corporate bond prices, it's not inconsistent with private yields staying the same.
Posted by: Nick Rowe | March 28, 2009 at 03:49 PM