This is an article I wrote several weeks ago that the ProJo chose not to publish in its printed edition.
.......................
I've always admired weather forecasters. Whenever they want
to know precisely what the current conditions are, all they have to do is look
out the window. Things aren't quite that simple for economists. A great deal of
the data we use is survey based. And, predictably, survey data are often
revised, occasionally in ways that tell a very different story than what the
originally released data showed.
This is the case Rhode Island right now. After my March Current
Conditions Index report release, which showed that based on the existing data
Rhode Island was flirting with the double-dip recession, I was informed by the
Rhode Island Department of Labor and Training (DLT) that the likely upcoming
revisions to their data will tell a strikingly different story. Instead of
seven or eight months of consecutively declining employment, the upcoming data revisions
apparently show that employment actually rose throughout that time period. What
they did not say, but that is every bit as important, is that if employment has
actually been rising, a number of other key indicators will also be affected,
not the least of which is our state's unemployment rate.
Some of this was apparently discussed at the recent Revenue Estimating
Conference and reported by the local media. However, with the release of the
April labor market data, we only heard about the existing labor market data,
which we now know is faulty. Whenever anyone turned on their television or read
the local newspapers, they were told that Rhode Island's unemployment rate rose
to 11.2 percent as employment fell yet again.
What an extraordinary time! I honestly can't remember ever
being informed this close to the most recent rebenchmarking (data revisions) that
such dramatic changes were coming. This placed the local media in quite a
predicament, as they chose to report the April data as released by the DLT even
though, as I pointed out to a number of them, we shouldn't put very much confidence
in that data or the obvious conclusions that emerge from analyzing it.
So, at this point it is appropriate to quote the character Emily
Litella of Saturday Night Live fame concerning Rhode Island’s large number of
employment declines and the increase in our unemployment rate above 11 percent:
Never Mind!
The origin of the situation we now find ourselves in is the
result of cost cutting at the US Bureau of Labor Statistics (BLS). Soon, it
will be taking over the task of producing the monthly employment numbers that
was historically done by our DLT (this is also true for all other states).
While this may well lower costs, its greatest cost to the people of Rhode
Island will be the loss of all the experience and expertise of our DLT
possesses. Furthermore, the way the BLS will produce their estimated labor
market values will not incorporate as much known data as the DLT has in the
past. Instead of beginning projections after the third quarter of the prior
year, the BLS will start after the second quarter. Furthermore, and more
troubling, Rhode Island will apparently be homogenized. By this I mean that
exceptional circumstances or events that would routinely be analyzed and
meaningfully incorporated into the labor market data by our DLT will now often be
ignored by the BLS. As Rhode Island has an extremely idiosyncratic economy,
this homogenization will make our state’s labor market performance appear to be
very different from what it actually is at times. Ironically, the most obvious
impact of this homogenization will be to make Rhode Island appear to more closely
resemble overall US economy. If you don’t believe that, take one look at what
the BLS has done with their estimation of
our state’s manufacturing wage (especially look at the charts after Read More ...)!
Because of these extraordinary circumstances, I found it
necessary to build a small econometric model of Rhode Island’s labor market in
order to estimate and simulate various labor market indicators. According to my
model, payroll employment has not been consecutively declining, as the existing
data show, but is in a mild uptrend. At the Revenue Estimating Conference, the
DLT offered tentative projections of where they believed payroll employment
might be as of March. My model produces slightly more optimistic numbers. As of
April, my estimate of payroll employment is slightly above 462,000, which is
higher than the official April value of around 458,000. Instead of having eight
consecutive employment declines in the last nine months, as indicated by the
current DLT data, my model shows consecutive increases for seven of the past
eight months, although not necessarily by large amounts. Along with this, my
estimate of the April unemployment rate shows it declining to 10.7 percent, not
rising to 11.2 percent. Even though my estimated jobless rate may appear to be
“better” than the DLT’s official value, its foundations are less than
flattering -- it is accompanied by declines in both our labor force and
resident employment.
The current divergences in labor market data are not the
fault of our state’s DLT, but the result of something forced upon them by the
federal government -- a different methodology. While I continue to have the
utmost faith in our state’s DLT, I am very irritated by the apparent attempt to
politicize our state’s jobless numbers by the DLT’s spokesperson, Laura Hart. She
recently offered an utterly ridiculous explanation as to why our state’s jobless
rate is so high -- Rhode Island doesn’t have the economies of scale that states
like Massachusetts have. If her hypothesis were correct, Delaware, another
small state, would have a very high jobless rate, while California, an
extremely large state, would have a very low jobless rate. For April, Delaware
had a 6.8 percent jobless rate while that for California was 10.9 percent.
It’s bad enough that the diverging data exists. Having DLT’s
spokesperson offer such ad hoc rationalizations only makes things worse.
