Goldman, as well as everyone else, was dead wrong when it predicted that the tumbling gas prices are “unambiguously good” and would unleash a wave of spending unseen since the Lehman collapse – because all those “gas savings” had to go somewhere, right? Instead, what happened was that retail sales disappointed 5 months in a row, and personal spending data that has consistently missed expectations.

So now that the plunge in oil, and gas prices, is history, what does Goldman do? Instead of admitting that like every other economist, it was wrong in expecting a surge in spending, it take a deep look at the “consumption puzzle” and concludes that the spending boost is merely deferred.

But before we dig into the humor, we should note that Goldman isn’t entirely wrong. It does observe that “the puzzling lack of impact so far is likely explained in part by consumers’ initial skepticism that low gasoline prices would last. As Exhibit 2 shows, only over the last couple of months have consumers come to believe that lower prices are here to stay.”

Then Goldman does the “economist” thing: it goes full retard.

How puzzled should we be by the absence of a boost from cheap gas prices so far? To answer this question, we use a range of models to estimate the usual time pattern of the consumption response to changes in energy prices. We find that low gas prices should have boosted Q1 consumption growth by 0.5-1pp, reinforcing the Q1 disappointment. However, the models also imply that only one-fourth to one-half of the eventual consumption impact should have been felt by Q1, suggesting plenty of remaining upside in coming quarters.

And even provides a pretty chart that should somehow justify it being right when it was already so massively wrong:

But while we can’t confirm if Goldman is merely wrong about consumer expectaions of future gas prices, as it has been wrong about everything else, or if indeed consumers are merely confused, one can easily ignore all of the above because there is one simple problem with all of Goldman’s assumptions shown above.