The hidden seasonal layers of PMI

The PMI index released by S&P has a large effect on financial markets as it gives an early indication of the direction of economic activity. Recently, the correlation to growth has declined, and we analyse whether this is due to the seasonal adjustment method used by the S&P. Further, we show what market participants can learn from the seasonal adjustment method in predicting PMIs and thus markets.

S&P use both the Census Bureau X12-ARIMA model and a discretionary adjustment to seasonally adjust PMI data (see appendix). The discretionary adjustments tend to be positive in the first half of the year (thereby increasing PMIs) and be negative in the second half for both eurozone and US manufacturing, while eurozone service PMIs are adjusted up in Q2 and Q3 and down in Q1 and Q4.

In the US, we find that PMIs adjusted by solely an X12-ARIMA model outperforms the combined S&P inhouse discretionary and X12-ARIMA model (the official data) in predicting GDP growth. The difference is largest in the post-pandemic years, where the X12 adjusted PMIs have a 11-percentage point higher correlation to growth.

Using solely the X12 model (i.e. excluding the discretionary factors) for adjustments makes recent service PMIs look weaker and manufacturing PMIs stronger compared to the official data.

Discretionary adjustments does not improve GDP correlation

We first examine whether the PMIs adjusted by both the in-house discretionary model and the X12-ARIMA (i.e. the official data) are better predictors of GDP growth compared to the PMIs adjusted solely by an X12-ARIMA model, in the same quarter. We take quarterly averages of the monthly PMIs and compute the correlation coefficient to quarterly GDP growth. Results are shown in table 1.

Download The Full Research Global