Consumer Sentiment and the Totality of Data

If you were still worried about the weak December retail sales numbers, the rebound in consumer sentiment in February should help ease your mind.

Recall that last week I questioned the relevancy of the retail sales decline given the magnitude of the decline is recessionary in nature and the economy was clearly not in recession in December. If in recession, the jobs numbers would have fallen off a cliff.

What about data besides the employment report? Consumer spending had declined sharply in January before partially rebounding in February. Is there a story here?

First off, the January decline was greater than one standard deviation, but that is not an uncommon event even outside of recessions:

There is still a well known relationship between consumer sentiment and spending that we should acknowledge:

On quick visual inspection, the decline in January, especially coupled with the February rebound, don’t suggest much reason for concern about consumer spending. That said, the two-axis graph is anathema to respectable economic and financial analysis as it can be manipulated to suggest stronger relationships than actually exist. I find it preferable to at a bare minimum convert the graph to a scatter plot and add the results of a liner regression:

The implication in this chart is that the surge in confidence after Trump’s election did not translate into a surge of consumer spending; spending had persistently fallen short of what would have been expected given the confidence numbers. Spending should be higher if the boost in confidence reflected real improvement in household financial conditions associated with Trump’s election. Consequently, the January decline by itself was also likely not meaningful for the underlying pace of spending.

This regression, however, isn’t really my favorite; the two-axis, two-variable format limits the analysis. At a minimum, we can add a lag of spending to the regression (the resulting three-variable equation though doesn’t fit neatly on a scatter plot, hence I dropped the fitted line):

Now the predicted values dragged up modestly (to the top of the pink dots) by the high levels of confidence but still hold fairly constant due to the persistent impact of the lagged spending variable.

In reality, the December spending number is likely to be depressed below these forecasts, an artifact of the weak retail sales report. Still, the fact that consumer sentiment has not fallen off a cliff in December or thereafter suggests that this is a short term impact. At the same time, note that consumer sentiment is probably overstating the strength of actually spending, so even a decline in sentiment would have to be quite significant to be consistent with deteriorating household spending. That kind of situation is typically associated with substantial deterioration in the jobs market, which so far has not occurred.

Bottom Line: Consumer sentiment is another variable indicating that near term recession concerns are misplaced. The decline in January’s sentiment itself wasn’t reason for concern, and the rebound in February backs this up. Also, if your favorite pundit has a penchant for two-axis charts, request at a¬†minimum conversion to a scatterplot with a regression.¬†