Keep pounding on Sterling?

Over at Macro Man, Macro Clown wonders if it might be time to go long GBP or at least close out shorts. Though he points readers away from GBPUSD because of dollar risk, I think it’s OK to look at how the major has evolved to see where Cable might be going. Quickly, as far as the PPP theory of exchange rates is concerned I’ll start and finish with a quote from Rudy Dornbusch “…strict versions (of PPP theory) are demonstrably wrong while soft versions deprive it of any useful content.”

For context, after its post crisis peak of about 1.71 in July of 2014 GBPUSD has been a steady decline.

Looking at rate differentials going back four years, the divergence that opened in late 2014 has persisted and indeed is at its widest at its present rate of 1.65%.

Equities tell a similar story. I indexed the S&P 500 and FTSE 100 to 100 at March 14, 2013.

Here is a table of Sharpe ratios for three time periods:

FTSE 100 S&P 500
4 Year 0.28 0.88
1 Year 1.38 1.62
Since US Election 2.64 4.60

The forward p/e of the S&P 500 is a little higher than that of the FTSE–18.27 to 13.37–but the difference doesn’t indicate a revaluation would deliver convergence. The punchline is that further portfolio flows should drive further weakness.

Last but certainly not least, the UK current account deficit is at its lowest level since data begin in 1955, and it’s not showing signs of rebounding as quickly as it did at previous local minimums. The first order effect is persistent weakness in Cable.

Finally, market positioning is net short (via the CFTC CoT reports) but at a rate that suggests there’s room for more shorts to pile on before we’re in danger of a big squeeze on a rally.

So where do we go from here? Take a simple rate differential model using the 5s back to July 2014. Here’s the scatter plot with the r-squared:

The current value of 1.65% implies about 1.11 (rounding so as not to convey a false sense of precision). Combine this with the equity return differential and the current account deficit and it’s reasonable to think we’d see a test of the 1985 low of 1.05; maybe we could even see parity. Technically, the pair has been making lower lows and lower highs in the trading range from October 2016 to present, with the recent rally looking to make a new lower high, and a following push below 1.20 could precipitate the move I’m expecting.

Rates: The Long and Short of it

A great post over at globalmacrotrading on long term global rates. We indeed have seen a large move down in bonds over the entire curve and it appears consensus that “rates are going to rise” and the market is “rotating into risk assets”. And when there’s consensus there’s usually little reason to hop on board. Here’s a longer term chart of the ten year with a back-of-the-envelope fundamental model – rolling 10 year average nominal growth plus the term premium (here I’m using Kim and Wright, 2005):

(Source: FRED/Author’s calculations)

The r-squared is about 0.92 and the current reading of about 2% suggests the ten year is a little high. And indeed it had worked it’s way down from the December 15th high of 2.6% down to about 2.3% before shooting up again yesterday on comments San Francisco Fed’s Williams, and we’re sitting close to 2.5% as I write this.

Thinking about the long term value of the ten year, we have to think about long term nominal GDP and the term premium. Decomposing nominal GDP into the real and inflation components, we see that if we assume that monetary policy has reached a state of technology that enables the Fed to control inflation, then the degree of freedom is real GDP. What does that look like long term?

I built a simple Cobb-Douglass-inspired model of real GDP growth as a function of productivity growth, growth in the capital stock, and growth in the labor force. The data go back to 1951 and the r-squared is 0.73. Here’s a table of the data for the whole period and the last 10 and 20 years:

RDGP TFP K L
1951-2014 3.18% 2.13% 3.53% 1.47%
Last 20 2.44% 2.08% 2.39% 0.88%
Last 10 1.51% 1.41% 1.76% 0.48%
Forward? 1.80% 1.43% 1.19% 0.50%

For the forward estimate of TFP and K I projected forward the recent recovery. For labor growth I used the BLS’s projection to 2022. Then if we assume an inflation rate of 2% going forward we come to a nominal GDP estimate of 3.8%. Plugging this back into our model, assuming the small term premium of today and we get about 2.6%. So most of the move might already be in. It all depends, of course, on assumptions about future growth and the term premium. I’m frankly not certain how to think about long term forecasts of the term premium, but it seems to correlate with the rate itself. For growth and inflation, if you think we’ll be stuck at 1.5% real GDP and 1.5% inflation then the 10 year moves below 2%. Think we’ll return to the nominal growth rates of the late 1990s, printing 6% per year? Then the 10 year goes above 4%. But with productivity growth averaging about 2% long term, and seeing as how we’re not likely to get a surge in immigration, skilled or otherwise, over the next eight years (yeah, I said it), and seeing as how the BLS is pretty good at what they do, any additional growth has to come from a huge increase in K. That’s another post, but the spoiler is that it’s not likely.

The Deal with US Wage Growth

Nick Bunker over at equitable growth wonders what’s the deal with US wage growth, citing the absence of “expected” growth given the unemployment rate. That is, he’s wondering why the “Wage Phillip’s Curve” isn’t holding. Here’s an illustration using the most commonly cited wage figure, Average Hourly Earnings (AHE):

(Source: FRED/BLS/Author's calculations)
(Source: FRED/BLS/Author’s calculations)

I don’t think this is too surprising. Here’s wage growth regressed against unemployment rate going back as far as the BLS data go:

(Source: FRED/BLS/Author’s calculations)

Not impressed? Me neither. However, over at the Atlanta Fed’s macroblog, there’s an interesting post about their so-called wage growth tracker (WGT). Rather than looking at average hourly wages in aggregate and tracking the growth in that figure, they use the Fed’s Current Population Survey to track wages of people who had a job the last time they were surveyed. The data go back to 1997. Here’s that regression:

(Source: BLS/Atlanta Fed/Author’s calculations)

Not bad. And just to make sure we’re not cherry-picking the data, here’s a regression for the 1997-2016 period using the data from the first regression:

(Source: FRED/BLS/Author’s calculations)

So, it’s a little better than the first regression, but not nearly as good of a fit as the WGT data. Let’s now replace the AHE data in the first chart with the WGT and see what happens:

(Source: FRED/BLS/Atlanta Fed/Author's calculations)
(Source: FRED/BLS/Atlanta Fed/Author’s calculations)

Look at that! Finally, I ran Granger causality on the WGT and AHE series and there’s evidence below the 5% level that WGT Granger-causes AHE but not that AHE Granger-causes WGT.

So, what’s the deal with wages? We might induce that if unemployment keeps falling then wages of those already employed will keep rising and this will raise the average rate of wages as popularly measured. The next question is how the Fed will react to the coming wage growth, but that’s another post.