I am impressed with your ability to include graphs with your posts, Eugene.
Rich hit the nail on the head. Ultimately, when you buy US stocks you are buying a share in the aggregate future earnings of US companies, which itself is a slice of the US GDP. When price/GDP ratios are high, you are paying a lot, and when price/GDP ratios are low you are paying a little.
Rich’s method is sound. Divide that historical (Stock price)/GDP analysis up into two pieces:
(Stock prices) / GDP = A x B, where
A = (Stock prices) / (Smoothed earnings), and
B = (Smoothed earnings) / GDP.
Sure, A and B move over time, but the further they get from historical values, viewed over several generations, the less likely they are to stick. When both ratios are high compared to historical values, the warning lights start to flash twice as fast.
Smoothing earnings is best done using an approach like Shiller’s (that Rich also uses), employing a long term moving average. Over the long run, this kind of smoothed result is guaranteed to converge to the actual results, making it an unbiased estimate. Your approach to smoothing is interesting, but it is far too subjective to be an unbiased or otherwise reliable estimate of underlying earnings.
