Advertisements
Advertisements

There is no bond market bubble

Bubbles, bubbles, everywhere bubbles. There is a lot of talk about bubbles among commentators and central bankers. One of the most common bubble fears is a fear of a bubble in the US bond market (just take a look at this recent “bond bubble”-story). Generally I am very skeptical about all kinds of bubble fears and that also goes for the fear of a bubble in the US bond market.

The general bond bubble story more or less assumes that quantitative easing from the Federal Reserve and other central banks has pushed down bond yields to artificially low levels and once QE is over the bond bubble will burst, bond yield will spike dramatically and send the US economy back into a major recession.

Should we fear this? Not really in my mind and I will try to show that in this blog post.

Inspired by Krugman and Mankiw

I very often disagree with Paul Krugman, but no one can dispute that he is a great communicator. Krugman is able to present complicated economic stories in a few sentences. This is exactly what he did in one of my favourite Krugman-blog posts back in 2010. The topic of the post was exactly the “bond bubble”. This is Krugman:

Here’s a thought for all those insisting that there’s a bond bubble: how unreasonable are current long-term interest rates given current macroeconomic forecasts? I mean, at this point almost everyone expects unemployment to stay high for years to come, and there’s every reason to expect low or even negative inflation for a long time too. Shouldn’t that imply that the Fed will keep short-term rates near zero for a long time? And shouldn’t that, in turn, mean that a low long-term rate is justified too?

So I decided to do a little exercise: what 10-year interest rate would make sense given the CBO projection of unemployment and inflation over the next decade?

… I decided to use the simplified Mankiw rule, which puts the same coefficient on core CPI inflation and unemployment. That is, it says that the Fed funds rate is a linear function of core CPI inflation minus the unemployment rate.

Krugman is basically using the Mankiw rule to forecast the Fed funds rate 10 years ahead and then he compared this forecast with the 10-year US government bond yield. It turned out that the 10-year yield was pretty well in line with the forecasted path for Fed Funds rates. I will now show that that is still the case.

Using the Mankiw rule to predict US monetary policy 10-years ahead

I have recently been playing a bit with the Mankiw rule (see here and here) so it is only natural to re-do Krugman’s small “experiment”.

Krugman is using CBO’s projections for core PCE inflation and unemployment. I will do the same (see the latest CBO forecast here) thing, but I will also use the FOMC’s recent projections (see here) for the same variables. I plug these projections into the Mankiw rule that I recently estimated. This gives us two forecasts for the Fed funds future rate for the coming 10-years. The graph below shows the two “forecasts”.

Mankiw rule FOMC CBO

Both forecasts (or maybe we should say simulations) point to interest rate hikes from the Fed in coming years. The forecast based on FOMC projections for unemployment and core inflation is a bit more “aggressive” in the rate hiking cycle than the Mankiw rule based on the CBO forecasts for the same variables.

The reason for this is primarily that the FOMC members expect unemployment to drop faster than forecasted by the CBO. Both the FOMC and CBO expects inflation to gradually increase to 2% over the coming 4 years.

The rule based on the FOMC projections indicates that the Fed funds target rate should be close to 3% in the “long run” (after 2018), while the CBO based rule is indicating a Fed funds rate around 2.6% i the long run. This difference is due to the FOMC expects unemployment at 5.0% in the long run, while the CBO expects unemployment at 5.5% in the long run.

I should stress that this is not my forecasts for the Fed funds rate as such, but rather an illustration of how we should expect the Fed’s policy rate to development over the coming 10 years if the Mankiw rule in general holds and we use the FOMC and CBO’s macroeconomic forecasts as input in this rule.

Drawing a (simplified) yield curve

We can now use these “predictions” to construct a (quasi) yield curve. Not to make things overly complicated (and spending to much time calculating the stuff…) I have simply constructed the “yield curve” by saying that “forecast” for for example the 2-year yield is simply the average of the predicted of the Fed funds rate in 2014 and 2015. Similarly the 5-yield is the average of the forecasted policy rate for 2014-2019. Hence, I disregard compounded interest and coupon payments.

The graph below shows the actual US yield curve compared with the two quasi-yield curve based on the two Mankiw rule based predictions for the Fed funds rate in the coming 10 years.

yield curve Fed Mankiw

Looking at the graph we imitatively spot two things:

First of all we see that the FOMC curve and CBO curve are considerably “higher” than the actual yield curve for the next couple of years. This should not be a surprise given the fact that we already know that forecasts based on the Mankiw rule is too “hawkish” compared to the actual Fed policy in 2014. Hence, the “predicted” rate for 2014 is 75-100bp too high. The reason for this is among other things that the simple Mankiw rule does not take into account “discouraged worker”-effects on the labour market, which seems to have been a a major problem in the past 5-6 years. Furthermore, the rule ignores that the Fed over the past 5-6 years more or less consistently has undershot it’s 2% inflation target. I have discussed these factors in my previous post.

These factors mean that we should probably pushed down the “rate path” in the next couple of years and that means that the yield curve does not look to be too “low” for 2-year or 5-years (very broadly speaking).

Second, we see that if we look at the 10-year yield we see that it is more or less exactly where the FOMC curve “predict” it to be (around 2.6%). We can of course not directly compare the two as I have not taken compounded interest and coupon payments into account (which would push the FOMC curve up), but on the other hand we should also remember that the Mankiw rule is too “hawkish” for the “early period” (which should push the FOMC curve down along the curve).

There is no “bond bubble”

I believe that the discussion above shows that US bond yields pretty well reflect realistic expectations to Federal Reserve policy over the coming decade given the FOMC’s and the CBO’s expectations for US unemployment and core inflation and it is therefore hard in my view to justify the claim that there is a bubble in the US bond market. That of course does not mean that yields cannot go up. They very likely will if FOMC’s and CBO’s expectations particularly for the US labour market are correct.

And the bond market might of course also be 50bp wrong is one or the other direction, but I find it very hard to see why US bond yields should suddenly spike 200 or 300bp as some of doomsayers are claiming.

And finally I should stress that this is not investment advice and I am not making any recommendations to sell or buy US Treasury bonds and the market might go in whatever direction.

Instead my point here is to argue that policy makers – the Fed – should not be overly concerned that quantitative easing has caused a bond bubble. It has not. If anything bond yields are this low because the Fed has not eased monetary policy enough rather than too much.

Related posts:

There is no bubble in the US stock market

The stock market has reached “a permanently high plateau” (if the Fed does not mess up again…)

If there is a ‘bond bubble’ – it is a result of excessive monetary TIGHTENING

Advertisements

The Mankiw-Darda rule tells the Fed to wait a bit with hikes

Greg Mankiw has a blog post commenting on my previous post on the so-called Mankiw rule.

I show in my post that according to both the original and a re-estimated version of the Mankiw rule the Federal Reserve should be hiking rates right now. I should stress again that I don’t think the Fed should hike interest rates – I am only using the Mankiw rule to illustrate why we likely are moving closer to a rate hike from the Fed (that is the is a difference between thinking about what the Fed will do rather than what it should do).

Greg makes some good points why the Fed should not hike rates yet:

Taken at face value, the rule suggests that it is time for the Fed to start raising the federal funds rate.  If you believe this rule was reasonably good during the period of the Great Moderation, does this mean the Fed should start tightening now, as the economy gets back to normal?

Maybe, but not necessarily. There are two problems with interpreting such rules today.

The first and most obvious problem is that odd things have been happening in the labor market for the past several years. The unemployment rate (one of the right hand side variables in this rule) may not be a reliable indicator of slack.

The second and more subtle problem is the nagging issue of the zero lower bound.  For several years, the rule suggested a target federal funds rate deeply in the negative territory.  We are out of that range now, but should the past “errors” influence our target today?  An argument can be made that because the Fed kept the target rate “too high” for so long (that is, at zero rather than negative), it should commit itself now to keeping the target “too low” as compensation (that is, at zero for longer than the rule recommends).  By systematically doing so, the Fed encourages long rates to fall by more whenever the economy hits the zero lower bound. Such a policy might lead to greater stability than strict adherence to the rule as soon as we leave negative territory.

I agree with both points. It seems particularly problematic for the original Mankiw rule that there seems to be a major problem on the US labour market with a discouraged worker effect – as a result the actual unemployment data tend to underestimate just how bad the situation has been (and still is?) on the US labour market. Many have simply given up looking for a job and left the labour market.

My good friend Michael Darda (MKM Partners) has suggested to deal with the discourage worker effect and other demographic problems by use the prime age employment ratio rather than the unemployment rate when estimating the Mankiw rule. Michael also uses core CPI rather than the core PCE deflator as a measure of inflation. The graph below shows the Mankiw-Darda rule (Michael’s estimate):

Mankiw Darda rule

 

We see that while the Mankiw-Darda rule has become increasingly “hawkish” since early 2011 it is still not “recommending” a rate hike – the predicted Fed fund rate is still negative (around -1.5%). Hence, it seems like the Mankiw-Darda rule is better at actually describing what the Fed is doing than the original Mankiw rule. This is not totally surprising.

Inflation targeting or price level targeting?

In an update to his post Greg makes some highly relevant comments, which will appeal to any Market Monetarist including myself:

There is another (related) argument for not raising rates now to offset shortfalls in the past. It is not about the interest rate. It is about the price level, the ultimate goal of monetary policy and measure of its performance.

If you plot the PCE deflator, there is a clear shortfall relative to a 2% price-level target. A 2% price level target fits very well during Greenspan’s time.  By the end of 2008, we were exactly on the 1992-target. But when I look at that plot starting in 2009 until the most recent data I see a gap.

A price-level target rule is optimal in normal times (Ball, Mankiw, and Reis) but is also an optimal policy in response to the dangers of the zero lower bound (Woodford). We have to catch up for the shortfall in the price level right now. And if you look at inflation expectations from surveys or markets, there seems to be no catch up expected, indicating that policy is still too tight.

Obviously I would prefer a nominal GDP level target to a price level target, but I think a price level target as here suggested by Greg is much preferable to an inflation target.

As I noted in my own previous post “inflation has drifted lower” since 2008. This is exactly the point Greg makes (based on comments from Ricardo Reis). Since 2008 the Fed has failed to keep the PCE price level on a 2% path trend. The graph below illustrates this:

Price gap US

 

The graph shows that it looks as if the Fed prior to 2008 had a 2% price level target and the actual price level closely followed a 2% trend line. However, since then inflation has consistently been below 2% and as a result the “price gap” has become increasingly negative.

This is paradoxical as the Fed now officially has 2% inflation target, while it prior to 2008 did not have such a target.

This is obviously another argument for why the original Mankiw rule at the moment is too “hawkish”. On the other hand one can certainly also discuss whether the Fed should close the “price gap” or not. Here I have the relatively pragmatic view that the Fed should let bygones-be-bygones as we over the past five years have seen some supply side adjustments. Furthermore, the decline in the price gap does not only reflect demand-side factors but likely also reflects a positive supply shock. In that regard it should be noted that longer-term inflation expectations in the US still remain above 2% (5-year/5-year US breakeven inflation this morning is 2.4%).

Therefore, if I was on the FOMC I would not favour a one-off money injection to close the price gap, but on the other hand the Mankiw-Darda rule and the consideration about the price gap also shows that the FOMC should not be in a hurry to tighten monetary conditions either. The Fed’s gradual and fairly well-communicated policy to continue “tapering” and then sometime next year gradually start increasing the fed funds rate therefore is consistent with a policy of ensuring nominal stability and it is also reducing the risk of a 1937-style premature tightening of monetary conditions, which would send the US economy back in recession. Said in another way I find it hard to be very critical about how the Fed at the moment is balancing risks both to the upside and to the downside.

A NGDP level target rule solve our problems

The discussion about the Mankiw rule illustrates two problems in common monetary policy thinking. First there remains a major focus on the US labour market. The problem of course is that we really don’t know the level of structural unemployment and this is particularly the case right now after we in 2008 got out of whack. Second, while inflation clearly has remained below 2% since 2008 we don’t know whether this is due to supply side factors or demand side factors.

There is of course a way around these problems – nominal GDP level targeting – and as I have argued in a recent post it in fact looks as if the Fed has followed a 4% NGDP level target rule since July 2009.

That would not have been my preferred policy in 2009, 2010 or 2011 as I would have argued that the Fed should have done a lot more to bring the NGDP level back to the pre-crisis trend-level. However, as time has gone by and we have had numerous supply shocks and some supply adjustments have gone on for nearly six years I have come to the conclusion that it is time to let bygones be bygones. We can do little to change the mistakes of six years ago today. But what we – or rather what the Fed can do – is to announce a policy for the future, which significantly reduces the likelihood of repeating the mistakes of 2008-9.

Therefore, the Fed should obviously announce a NGDP level target policy. Whether the Fed would target a 4% or a 5% path for the NGDP level is less important to me. It would be the right policy, but it would also be a pragmatic way around dealing with uncertainties regarding the state of the US labour markets and to avoid having supply side shock distorting monetary policy decisions.

PS My friend Marcus Nunes also comments on the Mankiw rule. Marcus seems to think that I am advocating that the Fed should tighten monetary policy. I am not doing that. All I have been saying it that the original Mankiw rule indicates that the Fed should tighten monetary conditions and that this is an indication of the direction we are moving in.

PPS My “playing around” with the Mankiw rule should be seen in the perspective that I am currently thinking quite a bit – in my day-job – about when the Fed will actually hike relative to what the markets are thinking the Fed will do. If the Fed moves earlier than expected by markets then it obviously is going to have clear implications for the global financial markets.

Mankiw rule tells the Fed to tighten

The most famous monetary policy rule undoubtedly is the so-called Taylor rule, which basically tells monetary policy makers to set the key monetary policy interest rates as a function of on the one hand the inflation rate relative to the inflation target and on the other hand the output gap.

The Taylor rule is rather simple and seems to at least historically have been a pretty good indicator of the actual policy followed by particularly the Federal Reserve. Often the Taylor rule is taken to be the “optimal” monetary policy rule. That of course is not necessarily the case. Rather one should see the Taylor rule as a empirical representation of actual historical Fed policy.

A similar rule which has gotten much less attention than the Taylor rule, but which essentially is the same thing is the so-called Mankiw rule. Greg Mankiw originally spelled out his rule in a paper on US monetary policy in the 1990s.

The beauty of the Mankiw rule is that it is extremely simple as it simply says that the Fed is setting the fed funds rate as function of the difference between core inflation (PCE) and the US unemployment rate (this of course is also the Fed’s “dual mandate”). Here is the original rule from Mankiw’s paper:

Federal funds rate = 8.5 + 1.4 (Core inflation – Unemployment)

The graph below shows the original Mankiw rule versus actual Fed policy.

Mankiw rule

I have also added a Mankiw rule estimated on the period 2000-2007: Federal funds rate = 9.9 + 2.1 (Core inflation – Unemployment)

We see the Mankiw rule more or less precisely captures the actual movements up and down in the Fed funds rate from 2000 to 2008. Then in 2008 we of course hit the Zero Lower Bound. From the Autumn of 2008 the Mankiw rule told us that interest rates should have been cut to somewhere between -4% (the original rule) and -8% (the re-estimated rule). This is of course is what essentially have justified quantitative easing.

Mankiw rule is telling the Fed to hike rates 

Since early 2011 the Mankiw rule – both versions – has been saying that interest rates should become gradually less negative (mostly because the US unemployment rate has been declining) and maybe most interestingly both the original and the re-estimated Mankiw rule is now saying that the Fed should hike interest rates. In fact the re-estimated rule has just within the past couple of months has turned positive for the first time since 2008 and this is really why I am writing the this post.

Maybe we can use the Mankiw rule to understand why the Fed now seems to be moving in a more hawkish direction – we will know more about that later this week at the much anticipated FOMC meeting.

BUT the Mankiw rule is not an optimal rule

I have to admit I like the Mankiw rule for its extreme simplicity and because it is useful in understanding historical Fed policy actions. However, I do certainly not think of the Mankiw rule as an optimal monetary policy rule. Rather my regular readers will of course know that I would prefer that the Fed was targeting the nominal GDP level (something by the way Greg Mankiw also used to advocate) and I would like the Fed to use the money base rather than the Fed funds rate as its primary monetary policy instrument, but that is another story. The purpose here is simply to use the Mankiw rule to understand why the Fed – rightly or wrongly – might move in a move hawkish direction soon.

PS One could argue that the Mankiw rule needs to be adjusted for changes in the natural rate of unemployment, for discourage worker effects and for the apparent “drift” downward in the US core inflation rate since 2008. Those are all valid arguments, but again the purpose here is not to say what is “optimal” – just to use the simple Mankiw rule to maybe understand why the Fed is moving closer to rate hikes.

PPS One could also think of the Mankiw rule a simplistic description of the Evans rule, which the Fed basically announced in September 2012.

%d bloggers like this: