Stock studies extended

Markets have gone into lock-down mode ahead of today’s FOMC announcement (and, to a lesser degree, the US trade data.) Most trades appear to be closing rather than opening risk, although yesterday’s bullish candlestick patterns in the likes of EUR/USD and cable have moderated (though not eliminated) some of the concerns arising from Friday’s mysterious dollar rally. The relief rally was partially spurred by none other than Easy Al Greenspan, who has gone all Sakakibara on us by commenting on currencies after his time has past. Regardless, the major source of Macro Man’s misery at the moment is the SPX/DAX spread, which probably requires a full-fledged reversal in stock prices to come back into the black. For what it’s worth, Macro Man expects the Fed to acknowledge an economic soft patch, to note that housing is seeing some signs of stabilization, but to maintain the bias to tighten. Jeffrey Lacker clearly shouldn’t vote for a rate hike, so Macro Man favours the end of his dissension. However, it is probably a close call.

Following yesterday’s post, Macro Man had a couple of requests to extend the study further back, if possible. Luckily, he managed to lay hands on P/E data going back to 1968. The results of the extended study were interesting, particularly as the 1970’s were a period of extremely high inflation and extremely low P/E ratios. The results of the study appear to confirm that the 2% threshold of Treasury yield over earnings yield is a consistently effective sell signal, with improved return/risk ratios in the 1980’s 1990’s, and 2000’s.

The model never had a chance to ‘sell’ in the 1970’s or late 1960’s however, as the S&P 500 earnings yield traded at a healthy premium to Treasury yields for much of that period. Equity multiples got crushed in the 70’s due to high inflation, while bonds traded much too rich until Paul Volker resolved to stamp out inflation.
Source: Datastream,

As a result, there are a couple of ‘premium buckets’ where equity market performance is mediocre. This largely reflects equity market weakness in the late 1960’s, and then again during the first oil crisis in 1973-74. While Macro Man could probably add some sort of inflation filter onto the study to explain the weakness of stocks despite their apparent cheapness 30+ years ago, that smacks of data mining. He is therefore left to conclude the potency of ‘buy signals’ may be less than that of ‘sell signals’ (i.e., when Treasuries yield >2% more than the S&P 500 earnings. Fortunately, the ‘value filter’ equity beta model is long by default, with only the sell signal in place as a market-timing device.

Source: Datastream,

Equity weakness in the 1970’s notwithstanding, extending the time frame of the study by an additional 18 years has done little to alter the underlying conclusion that there appears to be a strong correlation between the Treasury ‘yield premium’ and subsequent stock market returns. Applied since 1968, this value filter strategy would have doubled the returns of a simple buy and hold strategy, excluding transaction costs and dividends.

Source: Datastream,
While there were no applicable signals in the 60’s or 70’s, the strategy has consistently improved passive performance over the past 25 years. Notably, the strategy would have kept one out of the market during the 1987 crash, as well as during much (though not all) of the aftermath of the dot-com bubble a few years ago. Notably, the performance this decade with the filter moves from slightly negative to comfortably positive.
Source: Datastream,

Is it the answer to guaranteed returns? No. But as a ‘beta plus’ strategy on which to overlay portable alpha, it doesn’t look half bad.

Next Post »