Wednesday, 11 December 2013

Guest post - When is "good news" "bad news"?

"Bad news sells because the amygdala is always looking for something to fear." - Peter Diamandis 

Please find below a great guest post from our good friends at Rcube Global Macro Asset Management. In this post our friends go through a quick and easy method to identify time periods during which "good news is bad news.

Since last May and the sudden addition of the word “tapering” to the financial lexicon, we have observed quite a few better‐than‐expected US economic releases resulting in short‐term market corrections, and vice versa.

The underlying reason for this market behavior is quite obvious: nowadays, when an economic release indicates a strong US economy, investors fear that this will incite the Fed to prematurely reduce the dosage of morphine it injects into the economy through QEs and ultra‐low rates.

In this paper, we will present a quick and easy method to identify time periods during which “good news is bad news”. We won’t use the vast mathematical toolbox aimed at this kind of task (regime‐switching models such as HMM). Instead, we’ll keep things simple, visual and intuitive.

We’ll use Citigroup’s US Economic Surprise Index (CESIUSD Index) as an indicator of economic momentum. CESI indicators reflect how recent economic releases have fared compared to expectations. As far as we know, they are never revised (unlike the vast majority of economic indicators) and as a result, they reflect the surprise factor in real‐time. Their main shortcoming resides in their rather short history (around 11 years). For stocks, we’ll simply use the S&P 500.

A casual look at both time series’ levels doesn’t yield many interesting insights:

Even when we attempt to “stationarize” the S&P 500 by applying a momentum transform, it is still
quite difficult to visually distinguish between periods where economic momentum and equities move
in opposite directions from periods where they move together.

One popular solution to identify regime switches is to perform a rolling regression between time series differentials. What follows is a 6‐month exponentially‐weighted regression* between daily changes in economic momentum and daily equity returns:

*By using an exponentially‐weighted regression, we avoid the “cut‐off” effect that is inherent to rolling windows (in which the disappearance of past observations from the rolling window arbitrarily changes our assessment of the current correlation level).

Due to outliers, the regression time series is very noisy and difficult to interpret. Moreover, the regression level is an inherently lagging indicator of regime switches.

As an alternative to rolling regressions, we therefore propose the following time series, which we will call the “Cumulative Surprise Impact” (on the S&P 500, in our case):
Cumulative Surprise Impact: 
Cumulative Sum of [ (CESI daily differential) x sign (S&P daily Change) ]

The Cumulative Surprise Impact increases (by the amount of the CESI daily differential) when the CESI daily differential and the S&P go in the same direction, and decreases when they go in opposite
directions. The goal is not to predict anything, but to analyze the S&P’s short‐term response to positive (or negative) surprises.

The level of the Cumulative Surprise Impact is not directly interpretable (CESI daily differentials represent the intensity of surprises released during each day, but their value do not represent anything). What matters is the direction of the Cumulative Surprise Impact:

‐ The slope is positive when equity investors yearn for positive economic news (“good news is good news”).
‐ The slope is negative when investors fear that the Fed’s proverbial “punch bowl” is going to be removed (“good news is bad news”).

As illustrated by the following graph, there is a clear relationship between the Cumulative Surprise
Impact and the Fed’s monetary policy:

‐ Before mid‐2004, as long as the Greenspan Fed stubbornly kept its rates at 1%, good news was generally good news.
‐ From mid‐2004 to mid‐2007, a period which started with the Fed hiking rates by 25bp at every meeting, good news became bad news.
‐ From mid‐2007 until mid‐2013, as we entered the “Great Recession,” rates were eased and we switched back to a “good news is good news” mode. Although we reached the zero‐bound for
rates in late 2008, the massive QE programs that followed acted as a substitute for rate decreases (the rate equivalent of 1 additional Tn$ in the Fed’s balance sheet is anyone’s guess, here we arbitrarily displayed 1 Tn$ as 1%).
‐ Last May, as soon as Bernanke hinted that the Fed might contemplate reducing the amount of monthly bond purchases (under the right set of circumstances), good news became bad news again.

At the risk of stating what might be obvious, we can formulate the following rule:

‐ When the Fed is in easing mode, good news is generally good news.
‐ When the Fed is in tightening mode, good news is generally bad news.

We now seem to be at the onset of a new “good news is bad news” period. As we have seen, these periods can last for several years. We therefore believe that bulls should be careful what they wish for regarding future US economic releases.

"He who laughs has not yet heard the bad news." - Bertolt Brecht 

Stay tuned!

No comments:

Post a Comment

View My Stats