You Can’t Make This Up, But You Can Apparently Just Make Up The Data

Back in 2014, the Federal Reserve was convinced that the labor market was better than it appeared to be in various data accounts. Though it was called the “best jobs market in decades”, researchers at the central bank were keen on showing it. Primarily lacking in wages and incomes, the labor segment was suspected of missing the very elements of sustainable economic growth.

They came up with the Labor Market Conditions Index (LMCI), a factor model purported to give less weight to any of the 19 data points embedded within it that might be outliers. I assume they really thought the weaker points would be those outliers, and therefore the LMCI would overall gravitate toward suggesting the very robust jobs market they were sure was there.

The LMCI, of course, behaved in the opposite fashion. It suggested instead that the labor market was weak and getting weaker, not strong and getting stronger. Worse, after suggesting something like recession, which even GDP revisions have subsequently shown as the correct position, the LMCI failed to indicate a robust rebound befitting the by-then ultra-low unemployment rate.

In a fit of “data dependence”, the LMCI was unceremoniously scrapped.

The current mystery in terms of contradictory economic data is one step further along in the economic process from jobs. The Fed’s economists and policymakers don’t understand inflation. They openly proclaim their ignorance now.

So in that spirit, back in May, the Fed made up its own inflation index. The Underlying Inflation Gauge wasn’t worthy of much notice until the latest publicized reading (h/t ZeroHedge). Like the first few months of the LMCI, it purports to show what the Fed expects it to show (weird how models derived this way start out behaving so well).

Should we expect some time in 2020 the Fed to disavow once again its own work? I wouldn’t bet against it. They are, after all, data dependent in the (opposite) extreme.

Leave a Reply

Your email address will not be published.

four × one =