Sentiment Indicators - The Singapore PMI and ST Sector Indices - Student's Post

Armed with sector information from the FTSE, I have followed up on the prior post looking at the PMI and the STI. The PMI is a sentiment index of purchasers in the manufacturing industry, so I have identified the relevant sector indices to be the FSTBM, FSTIN, and FSTCG, corresponding to the Basic Materials, Industrials, and Consumer Goods indices. I then ran a simple regression on their returns, much like the prior post:

We can see that the PMI doesn’t do that great as a coincident indicator on any of the relevant sectors; the best we get is 34 basis points, and as mentioned in the prior post you can get 44 bps just wagering that this month’s returns are the same as last month’s. In other words, even if you somehow managed to get insider info on the PMI or forecasted the PMI accurately (also suspect; I have recorded PMI survey data on my data subsite for your perusal), you would not be likely at all to make much alpha.
Nothing to write home (or here) about.
However, great news is about “man bites dog”, and a little messing around with the lags show this:
This shows the same regression run on the PMI lagged by 1 month. The R-squared on all three sectors leaps by 7x, 4x, and 9x respectively – in particular, the previous month’s PMI seems to explain almost 3% of the FSTCG Index’s movements over this past decade (N=135). This is a magic number – some quants say that “People have gotten rich off of a 2% r-squared” – however an adult user of statistics would not get overly excited about this if only because the sample size is still not that impressive.
I know some of this talk about R-squared puts some people off, so here’s a simple trading rule conclusion you can make from this study:
When the PMI minus 50 was lower than -0.4 (purchasing managers were bearish), the FSTCG moved up 2.59% in the next month on average. When the PMI minus 50 was above 2.1 (bullish), Consumer Goods stocks dropped on average. The current PMI is at 50.7.