I Trample On Jim Geraghty's Thanksgiving

Jim Geraghty of NRO points out that the latest Federal Reserve forecast of continued high unemployment bodes poorly for Obama's re-election chances in 2012.

To which I gloomily reply, not so fast.  Prof. Ray Fair's model, which we have kicked around for years, tries to predict Presidential vote share based on inflation and GDP growth.

Unemployment is not a variable in this model (nor are unpopular wars), so the gist is this: if the economy manages to average a paltry 1% growth through the first three quarters of 2012, the model forecasts an Obama win with a razor-thin 50.06% of the popular vote (I put inflation at 1.5% and gave Obama the two quarters of growth above 3.2% that were booked in 2009).

I suspect Ray Fair would be the second to say that his model might not apply in quite that scenario (I am hereby the first), since no prior elections in the sample are based on four years of high unemployment, low growth and a grim war.  However, Prof. Fair no doubt included unemployment as a predictive variable and (we presume) concluded that it is not as important as one might guess.

I am more suspicious of the model than of Mr. Geraghty's political instincts, but still...

ERRATA:  Fed minutes and forecasts here.

 

Comments

[Rick Ballard]

Dr. Fair's model may be a little wobbly (with a tendency to veer left) but I find this bit from the FOMC:

The central tendency of participants’ projections of real GDP growth in 2010 was a narrow band from 2.4 to
2.5 percent, down from 3.0 to 3.5 percent in June.

to be much more worthy of derision. If my memory serves, June was around five months ago and is generally acknowledged to be the midpoint of most years. That means that the highly trained and well credentialed economists of the Fed blew their prediction by 20 - 31% with half the data in hand.

That doesn't qualify for a Marksman badge, let alone an Expert bar. They do note that consumers are devoting income to debt reduction but they don't provide even a hint of a rationale as to why that will stop short of household balance sheets being brought into actual balance. That's going to take another two years (if home prices stop dropping) and I don't see why consumption should be expected to increase until it occurs.

[Jack is Back!]

As an economics Phd. once told me regarding forecasting: "I am a highly educated and trained economist whose guess is as good as yours". Plus we had an election in November and Obama lost by an almost unprecedented scale. I believe his nomination and election will ride heavily on the tension or lack of it between the executive and congress for the next 18 months.

[Pagar]

Rick B,
How many times do you think they willget the Jobless claims figures correct before they forecast those 2012 quarterly figures?

"In what is now 38 times in 2010 (out of 46), the BLS has revised its prior weekly initial jobless claims number higher,"

[Ignatz]

Models of chaotic systems are highly useful until they're not, which is invariably the next time you use one.

[JB]

For kicks I looked at Obama's Indian astrology chart, and assuming his birth time's right it does not look good for him.

It looks like he's entering the same kind of period in the summer of 2012 that he briefly did when he lost his race against Bobby Rush.

Of course the opponent's chart could look even worse, but I doubt it. Simply put Obama's been riding Jupiter since the summer of 1996 (roughly when his legislative career was about to start), and that will come to an end as the Saturn-ruled period begins.

In other words, his lucky streak will end.

[Jane (get off the couch - come save the country)]

JB - give us a link so we can revel in it. It seems to me his lucky streak has been over for a while now. How fun that it gets even worse for him.

[JorgXMcKie]

I've been following Fair's model for several presidential elections. It's general margin of error is around 2%. Since a great many presidential elections are settled by less than a +/- 2% I don't pay him much heed.

I suspect that, overall, the model is right. There is a decided Democratic vote bias [I'd be willing to guess that his current C of 48+% Dem is wrong, however, and it almost certainly is wrong with a decent 3rd party challenger] to the model, but if you have a couple of years of a decent economy, it's pretty hard to beat the incumbent, or the incumbent party.

Having said that, it is way less accurate in a choppy economy or with a 'hot' war or in many other circumstances, and hasn't been tested in an economy with a prolonged 8+% unemployment rate [I'm guessing Obama will be looking at a 4 yr such stretch even if the economy is running at 3+%].

While I do believe that Obama's re-election depend some on whether he gets a strong challenge from within the Dem party and who the Republican candidate is, it is certainly the case that he's unlikely to have an easy path to re-election.

I suspect he'll need to run an entirely different type of campaign next time, and I'm not sure he can.

Also, I'm not totally convinced he'll run again and risk getting beat.

[centralcal]

"... and that will come to an end as the Saturn-ruled period begins."

Woohoo - does that mean his chickens turkeys will come home to roost?

I think I like astrology.

[RichatUF]

Jorg-

Fair includes a few scenarios for 2012 in his paper: The first set assumes a modest economic recovery, with no more
good news quarters from the one that has already happened and a value of G of 1
percent. The inflation forecast is left the same. The second set assumes a double
dip recession, with no more good news quarters and a value of G of -3.0 percent.
In the modest recovery, Obama gets 49.12 percent, and in the double dip recession,
he gets 46.44 percent. The corresponding House forecasts are 46.06 percent and
44.51 percent. [link, 12-13]

He also worked on the equation after the 92 election because that was the only one that he got wrong (the paper is 96a). He also includes a war variable that is part of C that captures elections a few elections around WW1 and WW2 (I'd have to look around to see if changing the war variable would have made the error for the 04 election less, because Bush significantly underperformed the model). However, I think that the US forecast is too optimistic,though, it is useful to think about the possibility that Obama would be re-elected.

Happy Thanksgiving everyone.

centralcal-

Astrology? That explains it!

[RichatUF]

Sorry, WAR is a non-economic variable and is not in C. Have too many things going on to even be looking at this.

[clarice]

I vote for astrology, too. It's been around longer than economics.

[RichatUF]

Better predictions too.

[wobblit]

clssify all that foreign aid loan write offs? Its not like would do that considering his past.Valand Joe just got jobs at cia for their kods like their daddies did before them. agents protect their clients i mean informants and their fsmlies helping with work schooling and careers like Os dad, families were tjeir jobs sd so is the work.

[Charlie (Colorado)]

Okay, JB, how'd you get connected to Vedic astrology?

Tom, the thing is that unemployment and rate of GDP growth are highly correlated; thus adding unemployment to the model wouldn't add much new information. The other side of that is that a time when GDP growth was relatively high but unemployment stayed high -- which might happen, for example, when added long-term costs or the probability of those costs discourage hiring, say via upcoming tax increases and expensive health care bills -- they may be decoupled, giving unemployment more influcence than the model has allowed for.

[Melinda Romanoff]

Chaco-

Think resonant frequencies when considering impending tax bills and the rates of change to structural unemployment.

It's more precise to say they're harmonic, not just correlative.

But that's me and the way I view the relationship.

[PD]

I think I like astrology.

We got yer astrology right here.

[J]

Through the management of obama and company, they will soon have a large portion of our citizens on their knees.
People on their knees tend to gravitate towards big daddy government that will give them food and medicine.

[cathyf]

If, as chaco points out, unemployment and economic growth are highly correlated, it may be that elections are decided by employment levels and growth is simply the wrong variable. So growth worked well enough in previous applications of the model, but only because it's a proxy for employment.

I would also point out that our "non recessionary" quarters have a significant level of book-cooking. They include counting stimulus spending, cash-for-clunkers, etc., while not accounting for huge hits on the economy from housing deflation. Not to mention declaring victory in the first announcement and then quietly going back and revising later. How much of the economic growth that is happening is being double counted?

[glasater]

Ah but there was no job growth during President Bush--or so the libs say.
How they get away with that canard is beyond me when back just a few years ago unemployment was under five per cent.

[MarkJ]

I fearlessly predict the following conversation will occur Election Night 2012:

David Axelrod: "I have some good news and bad news, Mr. President. Which do you want to hear first?"

Obama: "I'll go with the good news."

Axelrod: "The good news is that you managed to squeak by and get reelected!"

Obama: "That's great! What's the bad news?"

Axelrod: "Umm, well, the bad news is that the Republicans now have veto-proof majorities in the House and Senate, they now control 3/4 of the governorships and state houses, and the Democratic Party doesn't exist anymore outside of the big cities."

Obama: "So what? I WON...and that's ALL that matters."

[I R A Darth Aggie]

This all assumes that the real estate market doesn't do anything hinky.

I'm thinking we haven't seen hinky yet.