When tomorrow comes - Latest thought piece by Brett Gillespie
November 2017 | BY Brett Gillespie
When tomorrow comes
There are two possible outcomes: If the result confirms the hypothesis, then you’ve made a measurement. If the result is contrary to the hypothesis, then you’ve made a discovery. Enrico Fermi
First off, let’s be happy. At least in an economic sense, the world is in a very good place.
US, UK and Japanese unemployment rates are lower than the best levels achieved last business cycle, just prior to the financial crisis. Europe will likely be there within 2 years.
Yet the news cycle is always doom and gloom. No longer do we hear about secular stagnation and that the world will never recover from its GFC growth funk. Now we hear debt is too high, equities are overvalued, and we are on the cusp of a recession.
The truth will likely be far more boring.
That’s not to say there is not some serious work ahead of central banks as they normalise interest rates. After all look where they are now compared to before the crisis.
But it is “good” work.
Equities certainly think so. US equity indices are up 15-25% YTD. Europe 10-15%. Asia typically 15-30%. Alas, Australia and UK 4-5%
So why are there always scary headlines about impending disaster?
Simple. It sells.
But wait, I hear you say, am I not also forecasting a scary disaster? Well, possibly. I have been giving the chance of a scary bond sell-off about 30% in recent pieces. That is still 70% chance of a more benign outcome. So not totally sensational…
How good is my forecast? Well, if I may, I will come back to that.
But let me first digress to Deputy Governor Guy Debelle, who opined on forecasting last month, in a speech titled “Uncertainty”. In it he references Tetlock’s “Superforecasting” book. It came out a couple of years ago, and explains what makes a good forecaster. Tetlock is a University Professor at University of Pennsylvania who co-lead the “Good Judgement Project” (GJP), a forecasting competition, for anyone who cared to enter, asking over 100 questions per year for 4 years. And very random questions. They were asked to forecast anything from “Will Serbia be officially granted European Union candidacy by 31 December 2011?” Or “Will the London Gold Market Fixing price of gold (USD per ounce) exceed $1850 on 30th September 2011?” The best forecasters weren’t professionals. They weren’t even academic experts or IQ geniuses. They simply had above average intelligence and above average math’s skills, and they “Fermi-ised”.
Wonderful. With just a little bit of Fermi-ising, most of us can be super-forecasters. How exciting!
I was even more excited to learn that I already Fermi-ise! Most of us do when we seriously try and make a forecast. So what is it?
It is named after Enrico Fermi, the creator of the world’s first nuclear reactor in 1942. Post WW2 he was Professor of Physics at Chicago University, and he had a favourite question for his students;
How many piano tuners are there in Chicago?
Obviously without google, let alone the yellow pages (now half the people reading this don’t know what I am talking about) most students had no idea. But they had to guess. He would encourage students to break it down, and then guess/estimate. Let’s say one assumed the following;
- There are approximately 9,000,000 people living in Chicago.
- On average, there are two persons in each household in Chicago.
- Roughly one household in twenty has a piano that is tuned regularly.
- Pianos that are tuned regularly are tuned on average about once per year.
- It takes a piano tuner about two hours to tune a piano, including travel time.
- Each piano tuner works eight hours in a day, five days in a week, and 50 weeks in a year.
Based on the above, the answer is 225. The actual answer was 290. Not a bad guess.
By breaking down a forecast into components, and then determining the confidence of an estimate of each of those components, one can build a much more robust and objective forecast. We do that all the time, from our GDP forecasts, to the prospect of a central bank hiking, or even war on the Korean peninsula.
Another interesting observation by Tetlock is most “superforecasters” tend to have relatively boring forecasts. And that’s because they are concerned with being right, not getting attention.
We don’t have a boring forecast at the moment… The last few months we have written extensively about how US rates will rise significantly over the coming year. Are we being sensationalists? Should I be concerned?
Well, first off we have Fermi-ised. We break our forecast down into components. We model growth. We model inflation. We model wages. We review Fed reaction functions. We review asset market forecasts under various macro environments. We do a lot of Fermi-ising. We score each component, and we monitor and adjust those scores with new developments.
At the end, we have some factors we are very confident about, and some we are relatively confident about.
So let’s start with the very confident.
a) G7 economic data is surprisingly strong. The chart below measures the strength of economic data relative to consensus. It has rarely been stronger.
b) Our now-casting models collate all the data to estimate GDP in real time. They show G7 growth is very strong and accelerating.
c) Further our models based around financial conditions say this is going to continue. Indeed conditions have never been easier since the crisis!
So global growth is tracking very strongly, and it will continue. Normally this would automatically mean higher interest rates.
The general consensus agrees. 87% of investment professionals think US yields will be higher next year (that in itself is a concern).
Nonetheless, we are very confident central banks will continue to normalise. We take account of debt, debt servicing, demographics and productivity and estimate that the long run neutral cash rate will be lower. But there is still a lot of “normalising” to do.
How much? The chart below shows our estimate of how much (in orange). For most countries, it is around 150 basis points. And in blue we show what markets expect. On average about half that. We are very confident of most these countries getting back to the “new long run neutral” rate over the next 2-3 years. Possibly they will have to go higher, and sooner…
The journey to neutral is as we know well underway in the US. Canada fell in line in July, and now the UK. We expect Australia and NZ to join next year, and Europe and Japan the year after. They are all hoping for an orderly procession.
So should you be concerned if you hold equities, credit or even your house? Not yet. As I said at the outset, normalising interest rates is “good” work. Asset markets returns are generally fine, albeit not exciting, in these periods.
So when do you get concerned? I guess if you have read our forecasts for US inflation and wages you might be concerned. But will we be right?
Models work. Most of the time. Or we learn. At the end of 2015, the price of oil fell from $110 to $45. My models said this would be a big boost to US consumption (lower petrol prices freeing income). Yet growth slowed. Why? Because there had been a boom in shale investment in the US when oil was $100, and that investment slumped to near zero. The world changes, and sometimes we only realise when the models don’t work. As Enrico Fermi said, “If the result is contrary to the hypothesis, then you’ve made a discovery.”
I think Tim does amongst the best econometric work I have come across in my career. So I am excited by his forecasts. But inflation and wage modelling in the US? That has been the most difficult to forecast post the financial crisis. Indeed, many have simply thrown their hands in the air. So how confident is Tim Toohey? Well the forecasts are classic Fermi work. Break it down into forecastable baskets, and work on each forecast. We are “quietly” confident. After all, we are trying to tackle the hardest forecast gig in town. That and the impact of central banks winding down their balance sheets- quantitative tightening, which Tim has also modelled.
What we do now is monitor the incoming information relative to our models. What data is important? What are the big calls?
Well right now we have two big calls.
- The weak patch in US inflation over the last 6 months will be followed by a strong patch over the next 6 months.
- Global quantitative tightening will push US 10 year yields some 90 basis points higher over the next 12 months.
The first call – inflation – is updated monthly in the US. We will know over the next few months.
The second we will only know when it happens. The effect of quantitative easing/tightening is notoriously difficult and controversial to forecast. We think we have a more robust methodology. Even if we are half right, it is a big deal. And a big deal is a lot of alpha for us.
We also have what I think is one easy call. Regardless of inflation and quantitative tightening, all central banks are going to normalise as the global economy continues to strengthen and unemployment falls. This is the bread and butter of macro funds. Pass the jam.
 Yes the neutral rate is lower. And yes, higher debt levels mean less rate hikes. But debt servicing levels are still very low. So on average assume half the hikes we might have assumed pre-crisis.
 Superforecasting – The Art and Science of Prediction. Philip E. Tetlock and Dan Gardner
 If the index falls, the financial environment is more supportive of growth. A one point drop, all else equal, suggests growth will be 1% higher over the next 12 months.