The Markit Services PMI flash reading for March rebounded from the sub-50 reading in February, but only slightly. The calculation continues to suggest that the “services economy” is following the manufacturing and “goods economy” even if with some lag. The internals of the survey were not any better, with the new orders component falling to the lowest level of the relatively short series. From Markit last week:

U.S. service providers indicated a return to business activity growth during March, following a decline that was partly driven by east coast snow disruptions in February. However, the rebound in services output was only marginal, suggesting an underlying slowdown in growth momentum so far this year. At the same time, latest data indicated that new business levels expanded at the weakest pace since the survey began in October 2009.

With the ISM Non-Manufacturing version also decelerating sharply since last fall, the anecdotal suggestions appear to be continued slowing if to some as-yet unknown extent. There should be some solid data to confirm or deny the thesis, but the Commerce Dept.’s PCE data has been so volatile of late that it’s reliability is in question. In some ways that isn’t surprising given that far too much of what counts as “spending” on especially services is either completely made up (as in purely phantom activity) or imputed, thus interjecting even more potential bias.

The data variability also infests the national income accounts, as they have been equally subject to continuous revisions (almost always downward). The net result is the personal savings rate that when plotted with each revision looks like the branch of a tree rather than a stable data series conducive to honest analysis.

The latest updated version shows the savings rate still higher than in 2013 and 2014 even if we hold no confidence as to its actual range. It is at least consistent with the idea that overall consumer spending isnot robust or even improving, and nowhere near what would be harmonious with either the unemployment rate or the predicate potential for “overheating.”