Ok - that could well explain it. The EIA C&C line is the most negative of the four I used, so that would tend to give a higher base decline. I'll check quantitatively tonight.

I've redone the calculation using the O&GJ dataset:

The regression is:

5.2963e+3 - 2.64 * Year
R2=0.8552

The 95% confidence interval for the slope is [-2.64 -2.62]. Still higher than yours, whereas using the EIA data I got:

5.7767e+3 - 2.88 * Year
R2=0.8657

The fit quality is slightly better with the EIA.