This post is in response to two posts by Simon Wren-Lewis on "Zero Lower Bound Denial". First post here and second post here. I'm just collecting my thoughts from my comments on Simon's posts and from my previous posts.
This is what I think:
1. The ZLB is a real wall. Bad stuff happens if the economy wants to cross it but can't.
2. But the ZLB wall is not made of velcro. If we hit it, we don't have to stick to it.
3. An NGDP level path target would reduce the risk of ever hitting the ZLB wall.
4. If the economy is in recession, that does not mean the rate of interest is above the natural rate of interest.
1. The ZLB is a real wall. If the economy wants to cross that wall, but can't, one of two bad things will happen. Either we get an excess demand for money and and excess supply of all other goods (i.e. a demand-deficient recession), or else the balance sheet of the central bank gets very very large as the central bank buys up all the assets that people don't want to hold because they would rather hold currency. As I argued in an earlier post, if the central bank targets too low an inflation rate (and approaches Milton Friedman's Optimum Quantity of Money) we approach an equilibrium in which the central bank owns everything. If the central bank is owned by the state, that means state ownership of the means of production. Let's not go there. (But I can't resist saying that sometimes I wonder if there isn't some sort of commie plot against loosening US monetary policy, and the whole Democrat-Republican thing isn't just a charade.)
2. But the ZLB wall is not made of velcro. If you think that monetary policy really really is the central bank setting nominal interest rates, and that the only way to loosen monetary policy is by lowering nominal interest rates, then the ZLB wall looks like a velcro wall. If you hit it you stick to it, because the only way to get off the wall is to loosen monetary policy by lowering interest rates, which the wall stops you doing. So you have to hope that some non-monetary-policy force, like fiscal policy, or something else that raises the natural rate of interest, will come along and pull you off the wall.
This view that monetary policy really really is setting a nominal rate of interest, and cannot be anything else, is just wrong. This sort of view is what happens when you let institutionalists take over. Especially institutionalists who lack any sort of historical or comparative knowledge. They can't think outside the box of the here and now. The way we do things and think about things here and now isn't the only way of doing things and thinking about things. There are lots of different prices and lots of different assets in any economy. The nominal rate of interest on short safe loans is just one of those millions. For example, it is perfectly possibly to imagine a central bank whose currency is convertible on demand into gold, and which raises or lowers the price at which it will buy-or-sell gold to loosen or tighten monetary policy. And lets all interest rates go to wherever they want to go. No, I'm not a gold bug. Gold was just an example. It could be wheat futures instead of gold. Or even wheat itself, if the central bank has some granaries to store its wheat reserves. Or the stock price index. Or whatever.
The central bank doesn't have to hold the price of gold/wheat/whatever constant. It could raise or lower the price whenever it feels like it. Or it could raise or lower the price in order to hit some other target, like an inflation target, a price level path target, or an NGDP level path target, or whatever.
3. An NGDP level path target would reduce the risk of ever hitting the ZLB wall. It is well-understood that a credible price level path target has a stronger "automatic stabiliser" mechanism than a credible inflation target. Suppose a negative demand shock hits, and the central bank is slow to react, and so inflation falls below target. With a credible price level path target, people will expect temporarily higher inflation as the price level rises back to the original target path. That reduces the real interest rate for any given nominal interest rate. That means the nominal interest rate won't have to fall as much. That reduces the risk it will need to fall below 0%, and so reduces the risk of hitting the ZLB.
A price level path target is better than an inflation target at reducing the risk of hitting the ZLB, but an NGDP level path target would be better still. Because:
3.1. If the Short Run Phillips Curve is fairly flat, the price level will only fall a small amount below the target path, and the automatic stabiliser forces from temporarily higher expected inflation will be small. Temporarily higher than normal expected real GDP growth would also have an automatic stabiliser effect, but people won't know for certain how big an increase in real growth to expect. Under NGDP level path targeting, people will know for certain that if NGDP falls X% below the target path, it will rise again by exactly X% more than normal. If people expect a small rise in the price level they must expect a big rise in real GDP. If people expect a small rise in real GDP they must expect a big rise in the price level. Either way, the strength of the automatic stabiliser is more certain. It's like having a portfolio of two different assets, with perfectly negatively correlated payoffs, creating a much safer total automatic stabiliser effect.
3.2. Sometimes adverse supply shocks hit at the same time as negative demand shocks. Because stuff happens. But that might mean that the price level actually rises above target when the negative demand shock hits. Which means the automatic stabiliser from a price level path target would now become an automatic destabiliser. But NGDP would still fall below target, and so it would still work as an automatic stabiliser.
3.3. The Wicksellian natural rate of interest will probably be positively correlated with the long run natural growth rate of real GDP. This means that periods of low potential output growth would be times of highest risk of hitting the ZLB. A high inflation target provides insurance against hitting the ZLB. An NGDP level path target is like having a higher inflation target for times when potential real growth is low, and a lower inflation target for times when potential real growth is high. In other words, an NGDP level path target gives you the extra insurance against hitting the ZLB at precisely those times when the risk of hitting the ZLB is highest.
4. If the economy is in recession, that does not mean the rate of interest is above the natural rate of interest. Yes, if the central bank sets an interest rate above the natural rate, that will cause the economy to go into recession. But the converse is not true. That's because (very probably, at business cycle frequencies, when you have the output gap rather than output on the horizontal axis) the IS curve slopes up. If a negative demand shock hits, and real output falls relative to potential, the desired saving curve will shift left. But the desired investment curve will shift left too, and probably more than the saving curve. And if it does this, the rate of interest at which desired saving equals desired investment, conditional on that level of output, will fall. So the fact that we observe an economy in recession at the ZLB does not mean that the real interest rate must fall to allow the economy to escape the recession. A loosening of monetary policy sufficient to get the economy out of the recession is perfectly compatible with the real rate of interest rising.
Worthwhile Canadian Initiative 
Not for the first time, you seem to be using the term IS curve to refer to a locus of (r,Y) points traced out by varying the state of long-term expectations. I wish you wouldn’t. It can only sow confusion and we have enough of that. I suppose modesty would inhibit you from calling it the Rowe curve, but can’t you give it some other name? Maybe the ErY locus would do, since it shows what happens to (r,Y) as E[xpected profit] varies.
Kevin: if we ignore G, T, and NX (for simplicity), the IS curve is defined by Y = Cd(Y,r) + Id(Y,r)
Now, what exactly are those Y’s inside the functions? If they are Y’s supplied (what households and firms would like to sell) we have a Walrasian demand curve, which is very different from the Keynesian IS curve. The Keynesian IS curve interprets those Y’s inside the brackets as actual sales, which will be whatever quantity is demanded (in a recession). But does that mean the quantity demanded at this very instant, or the quantity that people and firms expect to be demanded in the near future, based on their actual sales this instant? The whole Keynesian multiplier analysis rests on the (quite reasonable) assumption that people’s and firms’ actual sales affect their expectations of sales in the near future. They don’t expect the recession to end in the very next minute. My IS curve is no different.
Great post.
Clear and simple.
Lays out key Market Monetarist ideas.
Thanks Bill! (But now i need to catch up on my sleep!)
The claim is that the ZLB wall is made of velcro if the CB is targeting inflation.
Whether it’s a strict target or flexible target makes no difference, as long as it’s forward looking. Only a permanent increase in the inflation target unsticks you.
A level target unsticks you at the risk of giving you temporarily high inflation when you don’t want it (after the recession ends).
Nick
Isn’t point 4 better stated as ‘monetary policy has the power to raise (or lower) the natural rate of interest, so that even if all recessions are situations where the market rate is higher than the natural rate, this insight may not be very relevant to the analysis of monetary policy and recessions’.
Max: “The claim is that the ZLB wall is made of velcro if the CB is targeting inflation.”
That’s not my claim. My claim is that the ZLB wall is made of velcro only if the CB is using an interest rate as its policy instrument. Because it needs to lower that rate of interest to loosen monetary policy to get off the wall, but it can’t lower the rate of interest below 0%. And you might not need an increase in the inflation target (either temporary or permanent) to move away from the wall, if the IS curve slopes up.
Ritwik: maybe. But then we have to distinguish between a short run and a long run natural rate of interest. Because the very concept of a “natural rate” implies that it is in some sense invariant to monetary policy. So in a recession, the short run natural rate may be below the ZLB given expected inflation, but the long run natural rate may be above the ZLB given that same expected inflation. But your point is valid nevertheless. And I can’t find a really simple and clear way to express this point.
Nick
It doesn’t have to be short run vs. long run. It’s bad equilibrium vs. good equilibrium. The natural rate is simply an equilibrium ‘solution’ for the cost of capital/ rate of return on capital in the economy.
Think about the central bank – raising gold demand – causing AD crash explanation of the great depression. Call it Hawtreyan. How do you interpret the Hawtreyan explanation? Was the market rate too high? Or was the neutral rate too low?
Of course, my question suffers from a which blade of the scissor problem. But if you had to interpret the Hawtreyan explanation from a market monetarist/ monetary disequilibrium perspective, would you say that the monetary policy error crashed the neutral rate and market rates (real) followed on the way down, or would you say that the error spiked the market rates (real) and deflation kept them up even when nominal rates were falling?
Only one of those explanations is compatible with an upward sloping IS/FF curve.
Nick,
“If the economy is in recession that does not mean the rate of interest is above the natural rate of interest. Yes, if the central bank sets an interest rate above the natural rate that will cause the economy to go into recession. But the converse is not true.”
Is it fair to say that the converse is likelier (generally), at the zero bound?
If nominal rates are at the zero bound with 3 per cent deflation and 3 per cent real, is it likely the IS curve would be monotonically upward sloping in your conception? Wouldn’t there be some sort of upfront, downward kink, as a kick-start?
Conversely, isn’t your upward IS curve conception much likelier to be the case above the zero bound?
Earlier, but different subject (I think):
“For example, it is perfectly possibly to imagine a central bank whose currency is convertible on demand into gold, and which raises or lowers the price at which it will buy-or-sell gold to loosen or tighten monetary policy. And lets all interest rates go to wherever they want to go.”
If the central bank increases the gold price and starts buying what is offered at that price, and what is offered comes to the central bank via the banking system, then the central bank must pay interest on excess reserves or allow the short term risk free interest rate to go to zero. The short term rate will go nowhere else but zero if this is the case. That means the central bank must make a decision on the rate on excess reserves. That means rates can’t go where they want to go, unless you let them go to zero. My impression is that the central bank will choose some non-zero rate to compete against the gold price, such that the CB’s stock of gold will stabilize at some desired level – which again is a central bank choice. And if the central bank only allows gold conversion in exchange for banknotes, that won’t solve the problem because sellers will redeem banknotes, first at the commercial banks and then in turn the commercial banks at the central bank, which will create excess reserves, on which an interest rate decision must be made. So I don’t see how the central bank can let “all interest rates go to wherever they want to go.”
We’ve known each other for four years now, Nick. Unreal.
It seems that the way NGDPT works is as follows:
– A demand shock occurs
– This could in theory be addressed by an increase in the value of money but sticky prices stop this happening
– This will put downward pressure on NGDP that a targeting regime will address by increasing the money supply and preventing the need for the increase in the value of money (that is they avert deflation and prevent sticky prices causing a recovery).
It is pretty clear that it would be possible for an NGDPT to be set and achieved. However this all seems very nominal and appears to assume a direct correlation between nominal and real.
A demand shock may not be just a increase in the demand for money that can be addressed by increasing the supply of money, It may also be a reflection of uncertainty and lack of confidence about the future. The real demand curve (not just the nominal) has shifted to the left as a result of this uncertainty. Unless the increase in the money supply magically fixes the confidence issue then hitting the NGDP target will be achieved via inflation and not a return to the old RGDP level.
Supporters of targeting seem to assume that just setting the target is enough to address the uncertainly issue. But what evidence is there that this is the case? It seems that for the past 3 decades we had a CB that has come quite close (until 2008) to de-facto NGDPT. In 2008 there was a huge demand shock that was not met by a corresponding increase in the money supply and NGDP fell sharply. Since then NGDP has increased at a steady clip for nearly 4 years and we have been close to the 2% inflation target. It seems a stretch to believe that sticky prices are still the issue that hold back RGDP and what we need is more QE to make up for a drop in the NGDP all those years ago. It seems more likely that the lack of confidence and uncertainty that arose 4 years ago is still with us despite NGDP increasing at close to its trend. I do not have much faith that QE3 will address that lack of confidence but rather toy with inflation as a solution, and run the risk of an asset-price driven bubble in the next 2 or 3 years.
Ritwik: desired saving and desired investment depend on the whole time path of expected income. And expected income will depend on expected monetary policy (given SR non-neutrality). Which makes the natural rate of interest a very slippery concept to use when talking about monetary policy. You can define a “long run natural rate” in a model where people might be assumed to know the model’s forecast for the time path for the natural rate of output. But anything other than that becomes a bit too slippery to be useful, I think.
JKH: since desired saving and investment are functions of the real interest rate, and the ZLB is a constraint on the nominal interest rate, there should be no relation between the slope of the IS curve and the ZLB.
“If the central bank increases the gold price and starts buying what is offered at that price, and what is offered comes to the central bank via the banking system, then the central bank must pay interest on excess reserves or allow the short term risk free interest rate to go to zero.”
Sorry, but you lost me there. One possibility is that the central bank could just pay a market rate, minus (say) 25 bp on reserves. So the rate on reserves fluctuates daily with the market.
Don! Yep! I think this means I have been blogging for 4 years now! Seems like I always have been.
Ron: “The real demand curve (not just the nominal) has shifted to the left as a result of this uncertainty.”
What is a “real demand curve”? What do people want to do instead of buying goods? If not hold more money, then what?
Nick, it looks as if I’ve been misreading what you’ve written on upward-sloping IS curves from the start. Sure, we can deduce from Y=C(.)+I(.) that if MPC+MPI>1 then the curve slopes up. I see two problems: (1) with the usual flat-then-rising LM curve we have two intersections, one unstable (I think you did a post about that) and (2) you’ll have a hard time communicating with Simon Wren-Lewis since I don’t think he has that sort of model in mind at all (though no doubt he’s seen the idea; it’s discussed in some textbooks, eg Scarth). Enough already, I clearly need to go back over your older posts.
With BB’s QE3 comments, and your recent posts, do you think the Fed has implicitly decided to choose residential housing as the nominal anchor (rather than gold,wheat,…)? What are your thoughts on choosing an anchor where its value is mostly a function of rate of change in time of credit growth?
Kevin: Yep. On MPC+MPI>1; on the instability problem (which I’ve thought about, and have some ideas, but not sure which is best); and on the difficulty of communicating this to new Keynesians (it’s actually easier with Old Keynesians).
jt: I don’t think so. I sure hope not. I interpreted the housing angle as being part of their transmission mechanism, rather than a target?
‘”Ron: “The real demand curve (not just the nominal) has shifted to the left as a result of this uncertainty.”
What is a “real demand curve”? What do people want to do instead of buying goods? If not hold more money, then what?’
My point is that even in a world of instantly adjusting prices then uncertainty about the future could still lead to reduced RGDP , and increasing the money supply in thos circumstances will just be inflationary. Suppose uncertainty causes business to reduce investment by 50%. This will undoubtedly require large price falls to get back to equilibrium. But even if they take place instantaneously investment and RGDP will still be potentially significantly lower than before.
Nick, I don’t think institutionalism has anything to do with remaining stuck in the here and now, or staying inside of the box. Many of the institutionalists have been associated with the political left, and their emphasis on institutions reflects their belief that the first step in social transformation is social understanding.
So, it’s not about a reluctance to think outside the box. It’s about understanding the exact nature of the current boxes, so we can understand what is actually happening socially and economically, so we can understand what kinds of actions can produce desired effects within the framework of existing institutions, and so we can understand the ways in which achieving certain other kinds of desired outcomes might require altering or abolishing present institutions.
There are different ways in which central bank policies could be innovative. Some innovations might only require doing things that are natural extensions of policies the bank already carries out, or that, while new, are well understood to fall within the scope of the bank’s permitted operations. Other innovations might, while arguably already legal, require the bank to assume a significantly more influential role in the economy and transgress prevailing institutional conventions, and are therefore likely to be met with various kinds of political and institutional pushback. A third type of innovation consists of actions that are presently illegal, and so would require changes to existing law – or else illegal usurpation of power.
And whether or not a central bank should carry out a given innovation is rarely just an economic question. There are questions to be answered about what type of society we want to live in, and how we want wealth and decision-making power to be organized and concentrated.
What kinds of “assets” are you willing to allow a central bank to buy and sell without direct legislative oversight and direction?
As I argued in an earlier post, if the central bank targets too low an inflation rate (and approaches Milton Friedman’s Optimum Quantity of Money) we approach an equilibrium in which the central bank owns everything.
This cannot be. The central bank cannot, for example, purchase labor. Therefore the present value of my labor will not be sold to the central bank. But in the same way, the central bank cannot purchase risky assets. I could incorporate (you would too!) and create “rsj consultancy, inc”, in which I offer equity shares obtained from selling my labor to my employers (or really anything else). The central bank cannot buy shares in the company I create, because everyone would create companies and sell equity to the CB. But in the same way, the central bank cannot buy shares in risky debt, because everyone create risky bonds (not collateralize by anything) and sell shares to the CB, then default, then create more risky bonds.
So the central bank must at least only buy bonds backed by some tangible asset with a real resale value. But now the central bank is getting into the business of conducting fair valuations, so determine collateral values. This is why the government steps in and heavily regulates what the central bank can and cannot buy.
What the central bank can do is to take riskless assets out of the economy with the exception of money, leaving only risky assets + money. But we can (hopefully) see how this will not stimulate investment in risky assets, because you are preventing people from hedging. You are making their overall portfolio more risky and therefore reducing the demand for risky assets. But as the central bank cannot buy those assets either, this policy results in an overall decrease in investment demand.
Thanks for the great posts!
I second Kevin Donoghue | September 15, 2012 at 06:03 AM.
It would be really great if you could lay out your thoughts about this upward sloping IS-curve from the very foundations up. The post that you link to is great, but it seems that there is still plenty of confusion about it since it is brought up in nearly all the comment threads to posts where you mention it. At least I don’t understand what you mean.
Basically I would like to know that what set of points your upward sloping IS-curve is? Given interest rate and output gap, output (gap?) demanded equals output (gap)?
If you would like to illustrate the effect of monetary policy on some variables, wouldn’t it be more suitable to have some measure of monetary policy on one of the axis?
Artturi: Thanks!
“Basically I would like to know that what set of points your upward sloping IS-curve is? Given interest rate and output gap, output (gap?) demanded equals output (gap)?”
Basically, yes. If potential output shifts, that may shift the whole IS curve. So lets just assume potential output is constant, so any changes in actual output reflect changes in the output gap. Then my IS curve is defined in exactly the same way as the standard IS curve from the ISLM model: combinations of r and Y such that quantity of output demanded will equal output. The only difference is my empirical assumption that, in some cases (most likely where the recession is expected to be long-lived rather than very short), and over some range (most likely where the output gap is negative but perhaps not so big that gross investment drops to zero), the marginal propensity to consume plus the marginal propensity to invest together exceed one.
A couple of my earlier posts on this subject here and here
Thanks for the answer Nick. Just to clarify, where does the extra demand to invest and consume come from if together they exceed 1? Peoples money balances?
Arturri: Yes!