Archive for the Category Forecasting


Lars Christensen’s new market monetarist newsletter

Lars Christensen has a new newsletter called the Global Monetary Conditions Monitor, which I highly recommend for people interested in international monetary policy.  It is by subscription at this link, but Lars is allowing me to quote from the newsletter.  (There is a discount for academic users and think tanks.)

Lars has constructed a monetary conditions index for a wide range of currencies. This basically measures whether the current stance of monetary policy is too easy or too tight to hit the target.  (A value of zero means right on target.)

On pages 8 and 9 of the May issue there is a discussion of policy credibility:

The approach here is to evaluate a central bank’s credibility based on our monetary conditions indicators.

We consider a central bank to be credible if it succeeds over time in keeping the monetary indicator close to zero. This can be measured by how long each central bank keeps the indicator within a range between -0.25 and 0.25 over a rolling five year period. This also means a central bank’s credibility can and will change over time.

By this criterion, the central bank of New Zealand has the highest credibility:

This can be illustrated by looking at developments in New Zealand over the past five years.

If monetary policy is (highly) credible, we would expect monetary conditions to be ‘mean-reverting’ – meaning that if the monetary conditions indicator is above (below) zero, we should expect it to decline (increase) in the subsequent period.

This is precisely the case for New Zealand. The graph below shows monetary conditions in New Zealand six months ago and how they changed over the following six months.

The line should go through zero, with most of the points being in the upper left and lower right quadrants.  To give you a sense of what a lack of credibility looks like—consider Turkey, one of the least credible central banks:

Maybe Lars will eventually incorporate the Hypermind NGDP forecast into his analysis.

Hard data beats sentiment in Q1

For the past 3 months there’s been a raging battle between the Atlanta Fed and the New York Fed.  The Atlanta Fed relies mostly on hard data when predicting GDP growth, whereas the New York Fed puts relatively more weight on consumer sentiment.  Republicans became much more optimistic after Trump was elected, so the sentiment indicators pointed to much stronger growth than the hard data indicators.  The following graph shows the huge divergence that developed in recent weeks:

Screen Shot 2017-04-28 at 10.14.15 AMThe actual growth was only 0.7%, which was much closer to the bearish Atlanta Fed’s 0.2%, than the New York’s Fed’s bullish 2.8%.  This reminds me a bit of the post-Brexit vote growth in the UK.

I don’t like either hard data or sentiment; I like market forecasts.  Unfortunately we lack a NGDP futures market (I’m working on setting one up again, and will have an announcement soon), but we do have some market indicators.  The preceding graph and the following quotation were from an April 12 article:

Tying in with the earlier point, the rally in the ten-year bond is consistent with the Atlanta Fed’s forecast for low growth in Q1.

So the bond market seemed to sense that growth was weakening.

Is it too early to attribute any of this to Trump?  I’d say so.  But the Trumpistas all crowed in early February when the strong January jobs report came in.  This was attributed to the magic powers of Trump, despite the fact that he had not even taken office when the January survey was conducted.  I don’t know how they’ll reconcile this GDP report with their dreamy predictions of 4% growth as far as the eye can see, but I’m sure they’ll think of something.

There’s likely to be some bounce back in Q2 (poorly measured seasonality depressed Q1), but I’m sticking with my view that America’s new trend RGDP growth rate is 1.2%, or 1.5% if Trump succeeds in getting his supply-side reforms passed.

PS.  Core PCE is up 2% over the past year, so the Fed is hitting both its price and employment targets.  For the moment, they are fulfilling their dual mandate. That’s a problem for Trump, who needs some Arthur Burns-style recklessness to paper over his personal incompetence when it comes to developing supply-side policy reforms.

Update:  I got the core PCE inflation data from the FT.  Ant1900 points out the true figure is 1.7%, still below target.

PPS.  I have a new post at Econlog explaining job shortages.

Print the legend

As I get older, I become increasingly interested in the mythological folktales that are believed by most economists.  For instance the idea that LBJ refused to pay for his guns and butter program, ran big deficits, and kicked off the Great Inflation.  All you need to do is spend 2 minutes checking deficit data on FRED to know that this is a complete myth, but apparently most economists just can’t be bothered.

Screen Shot 2016-05-04 at 10.23.16 AMDuring the 1960s, the budget deficit exceeded 1.2% of GDP only once, in fiscal 1968 (mid-1967 to mid-1968.  LBJ responded with sharp tax increases in 1968, and the deficit immediately went away.

The LBJ guns, butter and deficits story is too good to drop now, it’s in all the textbooks. It would be like admitting that the textbooks were wrong when they tell students that the classical economists believed that money was neutral and that wages and prices were flexible.  We can’t do that, it’s too confusing.

Another one I love is that monetary policy impacts the economy with “long and variable lags”.

I’ve talked about this before, but today I have a bit more evidence.  The idea that monetary policy affects RGDP with long and variable lags has three components, one or more of which must be true for the theory to hold:

1.  Monetary policy affects NGDP expectations with a long and variable lag.

2.  Changes in NGDP expectations affect actual NGDP with a long and variable lag.

3.  Changes in actual NGDP affect actual RGDP with a long and variable lag.

All three are false.  The third claim is obviously false; NGDP and RGDP tend to move together over the business cycle.  So the entire theory of long and variable lags boils down to the relationship between monetary policy and NGDP.

The first claim is also obviously false, as it would imply a gross failure of the EMH. Now the EMH is clearly not precisely true, but it’s also obvious that market expectations respond immediately to important news events.  Even EMH critics like Robert Shiller don’t claim that an earnings shock hits stocks two week later; it hits stock prices within milliseconds of the announcement. That part of the EMH is rock solid.  There is no lag between policy shocks and changes in expectations of future NGDP growth.

So the entire long and variable lags theory rests on the second claim, that NGDP responds with a lag to changes in future expected NGDP.  Unlike the first and third claim, that’s possible.  But it’s also highly, highly unlikely.  While we don’t have an NGDP futures market, the markets we do have strongly suggest that markets (and hence expectations) move with the business cycle, not ahead of the cycle.

Perhaps the best period to test this theory is the 1930s.  That decade saw massive RGDP and NGDP instability, which was clearly linked to asset price changes.  Put simply, the Great Depression devastated the stock market.  Here’s the correlation between stock prices and industrial production, from my new book:

Screen Shot 2016-05-04 at 10.42.33 AMThe stock market is clearly not a leading or lagging indicator; it’s a coincident indicator.  And that’s not just true in the 1930s; it’s also true today:

Screen Shot 2016-05-04 at 10.48.03 AMThe onset of the recession lines up, as does the steep part of the recession.  The stock recovery in 2009 did lead by a few months, but the recent slump in IP led stocks by a few months.  In any case, there are no long and variable lags; it’s basically a roughly coincident indicator when there are massive changes in NGDP.

If there actually were long and variable lags between changes in expected NGDP and changes in actual NGDP (and RGDP), then forecasters would be able to at least occasionally forecast the business cycle.  But they cannot.  A recent study showed that the IMF failed to predict 220 of the past 220 periods of negative growth in its members.  That sounds horrible, but in a strange way it’s sort of reassuring.

Suppose that the business cycle is random, unforecastable, as I claim.  And suppose that declines in GDP occurred one out of every five years, on average.  In that case, the rational forecast would always be growth.  As an analogy, if I were asked to forecast a “green outcome” in roulette, I never would.  Each spin of the wheel I’d forecast red or black.  I’d end up forecasting 220 consecutive “non-greens” outcomes.  And yet, there would probably end up being about 11 or 12 green outcomes during that period, and I’d miss them all.  A 100% failure to predict greens.  Because I’m smart.

Of course if there really were long and variable lags, say 6 to 18 months, then there would be occasions where the IMF would notice extremely contractionary monetary policy, and accurately predict recessions a year later.  I’m not saying they’d always be accurate. The lags are “variable” (a cop-out to cover up the dirty little secret that there are no lags, just as astrologers cover their failures with the excuse that their model is complicated, and doesn’t always work.)  No, they would not always be successful, but they’d nail at least some of those 220 recessions.  But they predicted none of them. And that’s because there are no lags.  Because recessions begin immediately after the thing that causes recessions happens.

That’s the message the markets are sending loud and clear.  But economists can’t be bothered; they have their comforting stories. Who can forget this line from The Man Who Shot Liberty Valance:

Ranson Stoddard: You’re not going to use the story, Mr. Scott?

Maxwell Scott: No, sir. This is the West, sir. When the legend becomes fact, print the legend.

As expected, low rates for as far as the eye can see

It wasn’t expected by everyone, but MMs have been predicting this for quite some time (although they are even lower than I expected):

World markets may have recovered their poise from a torrid start to the year, but their outlook for global growth and inflation is now so bleak they are betting on developed world interest rates remaining near zero for up to another decade.

Even though the U.S. Federal Reserve has already started what it expects will be a series of interest rate rises, markets appear to have bought into a “secular stagnation” thesis floated by former U.S. Treasury Secretary Larry Summers.

Of course Summers was a latecomer to this idea; I’ve been talking about slower trend growth and low interest rates as the new normal for many years.  Tyler Cowen’s book entitled “The Great Stagnation” also preceded Summers.

If you believe the press (and many economists), this period of low interest rates represents “easy money”,  That’s right, the implied claim is that the “liquidity effect” (normally very transitory) has now lasted for a decade. And there’s more to come:

Take overnight interest rate swaps. They imply European Central Bank policy rates won’t get back above 0.5 percent for around 13 years and aren’t even expected to be much above 1 percent for at least 60 years.

Japan‘s main interest rate won’t reach 0.5 percent for at least 30 years, they suggest, and even U.S. and UK rates are set to remain low for years. It will be six years before U.S. rates return to 1 percent, and a decade until UK rates reach that level.

“Although interest rates are low, they’re not accommodative,” said Harvinder Sian, global rates strategist at Citi in London. “The era of zero rates will be with us for years and years, it wouldn’t surprise me if we’re looking at another five to 10 years.”

Sixty more years!?!?!  At least there is one guy dissenting from the view that low rates mean easy money.  It will be interesting to see how long it takes the others to figure this out.  Let’s hope it’s not 60 years.

Update:  Commenter BC pointed out that the US interest rate data looks fishy, given that 5 year T-notes currently yield well over 1%.  So the article I cited may not be accurate.

And as predicted by MMs, the Swiss decision to revalue the franc has backfired. Now that markets understand that the Swiss are willing to let the SF appreciate over time, Swiss interest rates are forced below the already very low eurozone rates (due to the interest parity condition.)  In contrast, the Danes fought the speculators off, and are now reaping the (admittedly small) benefits, of having slightly higher bond yields than Germany:

The five countries or economic blocs currently with negative deposit rates have yields below zero on all their bonds from a minimum of five years’ maturity (Denmark) to a maximum of 20 years (Switzerland).

The pro-revaluation crowd thought that the SF would no longer be expected to appreciate, if speculators could be placated with a revaluation upwards.  But that was like feeding meat to sharks, it just increased their appetites.

PS.  Here’s what Tyler Cowen said 13 months ago, about the ability of the Danes to fight off speculators:

I would bet against them [the Danes], in any case this will be a neat test case for our judgments of Switzerland.

PPS.  Lars Christensen says the US may be about to enter a recession.  Possibly, but I’m less confident than he is.  BTW, here’s the track record of IMF economists in predicting downturns:

As The Economist noted, between 1999 and 2014, the International Monetary Fund, in its April forecasts, failed to predict every one of the 220 instances in which one of its members suffered negative annual growth in the next year.

Ouch!  I wonder if they ever predicted anyone will have negative growth?  Kudos to Lars for going out on a limb.

PPPS.  Tyler Watts sent me a music video on “Tall Paul” (Volcker), which might be fun to use in undergrad money/macro classes.

Update:  Timothy Lee has a great post on Draghi’s screw-up today.


Who predicted what, when and why

Let’s go back to March 3, 2009.  Here’s Paul Krugman:

As Brad DeLong says, sigh. Greg Mankiw challenges the administration’s prediction of relatively fast growth a few years from now on the basis that real GDP may have a unit root — that is, there’s no tendency for bad years to be offset by good years later.

I always thought the unit root thing involved a bit of deliberate obtuseness — it involved pretending that you didn’t know the difference between, say, low GDP growth due to a productivity slowdown like the one that happened from 1973 to 1995, on one side, and low GDP growth due to a severe recession. For one thing is very clear: variables that measure the use of resources, like unemployment or capacity utilization, do NOT have unit roots: when unemployment is high, it tends to fall. And together with Okun’s law, this says that yes, it is right to expect high growth in future if the economy is depressed now.

But to invoke the unit root thing to disparage growth forecasts now involves more than a bit of deliberate obtuseness.

And here is Greg Mankiw’s reply:

Paul Krugman suggests that my skepticism about the administration’s growth forecast over the next few years is somehow “evil.” Well, Paul, if you are so confident in this forecast, would you like to place a wager on it and take advantage of my wickedness?

Team Obama says that real GDP in 2013 will be 15.6 percent above real GDP in 2008. (That number comes from compounding their predicted growth rates for these five years.) So, Paul, are you willing to wager that the economy will meet or exceed this benchmark?

And here’s what I wrote, 5 years later:

Krugman wisely decided to avoid this bet, which suggests he’s smarter than he appears when he is at his most political. In any case, the actual 5 year RGDP growth just came in at slightly under 6.3%. That’s not even close. Mankiw won by a landslide.

In January 2011, Tyler Cowen wrote a book entitled “The Great Stagnation.”  So far Tyler’s hypothesis has proven correct. (Oddly, the media often refer to Larry Summer’s stagnation hypothesis, which (AFAIK) came much later.)

In 2013 Tyler made a bet with Bryan Caplan, that unemployment would not fall quickly back to 5%:

Tyler just bet me at 10:1 that U.S. unemployment will never fall below 5% during the next twenty years.  If the rate falls below 5% before September 1, 2033, he immediately owes me $10.  Otherwise, I owe him $1 on September 1, 2033.

Readers of my blog know that I would have agreed with Bryan.  Tyler Cowen responded by pointing to reasons why these bets are not a good idea:

Bryan Caplan is pleased that he has won his bet with me, about whether unemployment will fall under five percent.  I readily admit a mistake in stressing unemployment figures at the expense of other labor market indicators; in essence I didn’t listen enough to the Krugman of 2012.  This shows there were features of the problem I did not understand and indeed still do not understand.  I am surprised that we have such an unusual mix of recovery in some labor market variables but not others.  The Benthamite side of me will pay Bryan gladly, as I don’t think I’ve ever had a ten dollar expenditure of mine produce such a boost in the utility of another person.

That said, I think this episode is a good example of what is wrong with betting on ideas.  Betting tends to lock people into positions, gets them rooting for one outcome over another, it makes the denouement of the bet about the relative status of the people in question, and it produces a celebratory mindset in the victor.  That lowers the quality of dialogue and also introspection, just as political campaigns lower the quality of various ideas — too much emphasis on the candidates and the competition.  Bryan, in his post, reaffirms his core intuition that labor markets usually return to normal pretty quickly, at least in the United States.  But if you scrutinize the above diagram, as well as the lackluster wage data, that is exactly the premise he should be questioning.

As I’m the only one in this exchange fessing up to what I got wrong, and what I still don’t understand, and what the complexities are, in a funny way…I feel I’m the one who won the bet.

I agree with Tyler’s skepticism regarding the utility of public bets; they oversimplify a very complex set of issues.  They also subtly imply that greatness is a function of not being “wrong” about particular questions.  I’d argue that one doesn’t become a truly great scientist until one’s views have been partially discredited.  That means people are taking your ideas seriously, and pushing them to the point where they are no longer intellectually progressive.  (Think Copernicus, Newton, Einstein, Darwin, etc.)

However, as someone who agreed with Caplan, I don’t entirely accept the implication of Tyler’s final sentence.  I can’t speak for Bryan, but here are the views I’ve expressed:

1. Cycles in unemployment are largely caused by nominal wage stickiness, and unemployment will usually revert back to the natural rate, which tends to be fairly stable in the US (but not completely stable).

2. The US is entering a Great Stagnation, where 3% NGDP growth will be the new norm, measured RGDP growth will also slow sharply, but of course it’s not clear what RGDP actually is, because it’s not clear what economists mean by the term “price level.”

3. The Labor Force participation rate has historically been unstable, unlike the natural rate of unemployment, responding to demographics, welfare reform, disability insurance, prison incarceration, etc., etc., etc.)  Wage stickiness doesn’t explain this.  But Tyler was also skeptical of how far Bryan and I pushed the wage stickiness concept.  Since our view is that wage stickiness explains changes in the unemployment rate, but not the LFPR, Bryan winning his bet is at least as small point in favor of the sticky wage model.

4. Fiscal austerity would not slow growth in 2013, a claim Paul Krugman contested.  I was right and Krugman was wrong.

5. Repealing the extended unemployment benefits in early 2014 should have modestly increased new job creation, by boosting the supply of labor.  This would be true even if NGDP growth (i.e. AD) did not accelerate.  Paul Krugman also contested this claim.  Again, I was right.  Job growth in 2014 was substantially above the 2010-13 rate, despite very modest growth in NGDP. Of course Krugman has been right about many things, especially when he agreed with market monetarists.  Thus he has criticized the mainstream conservative prediction that “easy money” would lead to high inflation.

6. I’ve consistently predicted that unemployment would fall faster than the Fed thought, and that NGDP growth and inflation would be less than the Fed thought.  That’s actually sort of threading the needle, as faster falling unemployment would normally be associated with faster than normal NGDP growth.

Despite the fact that I’ve recently ended up being right more often than wrong, I think the importance of specific predictions is overrated.  If I had been blogging in 2006-09 I would have been wrong about lots of things, because the market was wrong about lots of things.  The economy is very hard to predict, and hence I’ve been lucky.  More important is the reasoning process used.  Here is how I’ve approached this:

1. For low NGDP growth and inflation predictions I’ve relied on market forecasts, which generally seemed more bearish than Fed forecasts.

2. For the Great Stagnation, early on I noticed a “job-filled non-recovery”, which many people oddly called a jobless recovery.  The unemployment rate fell sharply despite anemic RGDP and NGDP growth (for a recovery period).  So I reasoned that if RGDP growth was only 2.0% to 2.5% during a period of fast falling unemployment, then the trend rate of growth must be really slow.  So far I’ve been correct (and I hope my prediction is soon refuted, for the sake of the economy.)

3. For the unemployment compensation issue I relied on basic theory, and on previous studies of the effect of UI on jobless rates.  Back in 2008, Brad DeLong predicted higher unemployment as a result of a very small increase in benefits under President Bush.  It seemed to me that people like Krugman were abandoning mainstream economics for ideological reasons.

4. Standard theory, pre-2008, also implied that the monetary authority drove AD, and that fiscal policy would only impact growth by shifting the AS curve.  I saw no reason to abandon the standard view.

5. As far as wage stickiness and unemployment, my views were shaped by many factors, including my study of the Great Depression, and the fact of wage stickiness documented by many studies.  I also relied upon the strong theoretical implication that if nominal wages are sticky then nominal GDP shocks will lead to volatility in hours worked.  As far as the unemployment rate recovery predicted by Caplan, I relied on both theory (Friedman/Phelps) and evidence—the fact that the US natural rate seems fairly stable at around 5%. That’s the best I could do, and in this case I was right.  But if I’d lived in Western Europe in the late 1970s and early 1980s, I would have been wrong and Tyler would have been right, as the unemployment rate jumped sharply, and never went back down (except in a few countries like Germany, and even then only much, much later.)

Market forecasts are the best we can do.  I suggest that readers pay less attention to who predicted what, and more on the reasoning process behind their predictions.  Occasionally people will get lucky and nail a prediction that the markets missed (think Roubini, John Paulson, Shiller, etc.)  But when you look at their overall track record it’s clear that luck was involved; no one can consistently predict the macroeconomy. Nor should we focus on who has the most impressive mathematical model. Instead we should focus on who has a coherent explanation for what is occurring, an explanation that is consistent with well established theory, and that can be applied to a wide range of cases.  I hope market monetarism is one of those coherent explanations.

PS.  I just returned from the Warwick Economics Summit, and was very impressed by the Warwick students.  I would especially like to thank Ibraheem Kasujee, who helped arrange the visit.  It was good to get out of the Puritan States of America for a few days, and attend a student ball where drinks are served to 18 year olds.  At Bentley even the faculty can’t drink a glass of wine at the Holiday Party.  And Bentley just banned smoking everywhere—basically telling the smokers (who used to huddle outside in the cold) to go away, we don’t want you here.  This university is only for PC puritan paternalists.  (On the plus side, our students do get good jobs.)

I had planned on being home for dinner on Monday, but instead (due to snow at the Boston airport) I was 30 miles north of Reykjavik (at 11 pm Iceland time), in the middle of nowhere, standing in a cold wind with no hat on, looking straight up at a zillion stars in the sky–and a few northern lights as well.  But now I’m back and have a huge amount of catching up to do. (Here’s my earlier unexpected layover in Iceland.)

For readers who didn’t get their fill of Krugman bashing here, I also have a new Econlog post.