Economic Sophisms

Now that GDP level targeting is getting a public showing, long awaited and much relished skepticism is at hand. Critique is better than irrelevance. Critics say that GDP level targeting will increase output volatility because the sometimes big revisions to GDP will cause the Fed to unwittingly under or over crank the printing press. This of course begs the question of whether a noisy reading of a key variable (AD) is more useful than a precise reading of an incidental variable (consumer prices).

Still, measurement error is the biggest drawback of level targeting. Drawbacks are relative though and GDP level targeting is still better than current policy. The Fed was able to implicitly level target for the 20odd years before the Great Recession, so I am unmoved by the critique. Level targeting doesn’t mean the Fed need hit 5% GDP growth every quarter, it means they set policy so that the expected level of GDP three or six quarters ahead lies on the 4% trend line. None of the supporters of greater money printing (Market Monetarists, Mankiw, Rogoff, Krugman) are saying central banks should go mad and make up for deviations in one or two quarters. I think their common point is instead that it would be hard to overdo stimulus; or at least that it would have been so back in 2009, three years of unneeded unemployment have a way of crippling AS.

To the title of the post. The RGDP revisions are pretty monstrous, though I was curious to see how big the nominal GDP measurement errors were.  In order to rig my approach against my bias, I compare the advance GDP release to the final estimate, rather than the more accurate advance GDI or preliminary number. In reality we would want to use GDI, not GDP.

Lets start by looking at the advance and final growth rates:

As Professor Sumner pointed out, the big errors happened in 2008, otherwise the two series look pretty similar. Next let’s look at the difference in the growth rates:

My keen time series eye tells me that we are dealing with separate “generating processes” here.  The advance release is an unbiased predictor in the early aughts, becoming autocorrelated in about 2002.  Still, the average overshoot is only .2 percentage points with a behemothian standard deviation of 1.8 percentage points i.e. the advance estimate is unbiased.  This suggests that if the Fed used the advance series to make small adjustments in the policy stance in concert with other data (I suspect that a model mixing in higher frequency data could improve on the advance estimate), that the first round measurement error won’t lead to systematically bad policy. The goal isn’t to hit the target perfectly, its to have a simple way of convincing markets that even if GDP falls 12% below trend, that stimulus is on the way.  You don’t get that kick to expectations with a memoryless growth rate target.

I think histograms are underappreciated in macro.  Look at this fine plot of the above series. Like most everything in economic life, it is not normally distributed.

Finally, assuming the Final estimate were true GDP-it is probably close,black markets excluded-how ignorant would we be if we only had the advance estimate?

I say not terribly ignorant, but still good to have the revisions. Assuming policy doesn’t affect the measurement error,  adopting GDP level targeting at any random point in this series would have tended to over stimulate in the mid aughts and under stimulate* during the Great Recession. I put that * in there because a GDP level target during the Great Recession, even with the measurement error, would have been far more stimulative than actual policy….and that’s why the measurement error argument is obtuse.

Disclosure: I got the advance growth rates by downloading the old BEA press releases and hand entering values. I can’t promise that I didn’t make an error on one or two.  There were a lot of quarters between 2000 and 2011.

I have a feeling that Market Monetarists will look back on the new paper by Goldman Sachs economists Jan Hatzius and Sven Jahri Stehn as a turning point.  Professor Sumner has already succeed about as much as is realistic in persuading the quasi-Austrian camp (he sort of got Arnold Kling and George Selgin and certainly others).  Now great progress is being made changing the minds of-forgive me-politically influential economists.

The Goldman Paper is a big step because it finally brings some econometrics to bear on the issues-and survived the internal critique of a prestigious organization unlike other efforts. We can debate the merits of models. But when we do, the Pro Model camp will carry the day because People like models. We live in a world in which we are constantly reminded of how mighty computers have become and people expect lots of Greek letters and terms they don’t understand. So from a practical standpoint, the econometric simulation of GDP level targeting is a tremendous boon for Sumner’s argument. This is how people who go on Bloomberg will be convinced.

I think the biggest challenge now is to clarify what is meant by level targeting.  Today I heard an economist essentially dismiss the Goldman Paper as no different from current policy-because they were stuck in the rate of GDP growth paradigm.  Part of this effort will be to address the issue of revisions.  Revision are (of course) almost immaterial as nominal GDI (sic) is revised only slightly, but then most economists think that GDP is subject to the same revision volatility as RGDP. No one looks at (N)GDP explicitly.

Anyway, better load up on copper mining shares now.  If Goldman really gets behind this proposal 2012 could be a great year.  The sad thing is that if we really did finally get enough NGDP to allow a recovery, all the analysts and newsmen would attribute it to a mysterious boom in housing and jobs. Oh well, I’d take it.

I am awfully tired and have no insights tonight.  I will leave you with a puzzle though.  Why are entry doors to apartments in America are so flimsy? Locks too.  My apartment doors in China and Sweden (the actual flats, not student housing) would have literally taken a police battering ram to breach.  However, in more crime ridden America one can break down my door (in a respectable apartment complex) with a swift kick, and this is a wide spread across most midrange rental housing.  Perhaps because Swedes and Chinese aren’t allowed to own guns and seek a feeling of protection from thicker doors.  I don’t say that to get political about guns or imply that I own guns (I do by the way), just honest speculation.   I would rather have a foreboding door and not have to take my chances in a shoot out.  Next apartment is going to have an iron gate.

I just found out.  He was one of the greatest capitalists in American history.  Unlike most of us he really mattered too.

I will add him to the list of heroes who fell to cancer the next time I think about eating non paleo food.   I know this sounds like a low thing to say when a Great Man has died but the truth is, carbs kill.  It is highly unlikely Jobs would have developed his cancer had he eaten Paleo, of course there was no way for him to know this sad fact.

The U.S. will escape disinflation and boom in 2013-14 when Mitt Romney is president.  Hold on, let me explain.

My new working model of the Fed is that they hate Obama and want him to fail.  The alternative-that the Fed is full of macroeconomic illiterates who don’t understand that they have nearly perfect control of any given nominal  aggregate-is too horrifying to acknowledge.

Under this assumption I foresee the Fed allowing a mild recession  in order to get a Republican elected, who I assume will be Romney.   Then when Romney is elected, the Fed will announce a price or wage level target, cite Mankiew and with that tame the Hawks, who will suddenly start pointing to Sumner and Woodford….and me.

Who knows though.  Most leer at me with scorn when I hold that the Fed sets the nominal economic aggregates and can move any single statistic (average wages, GDP, a given linear combo of prices) to any level it likes-with a low margin of error.   They get downright rude when I tell them that nominal interest rates are meaningless in and of themselves, that long rates should rise if a policy is expansionary and that the increased oil price from QEII were welcome.  This makes me feel like I am really smart and that I will make a lot of money in the next two years going long copper and oil shares.  After all, a good forecaster is a good trader.

No need to argue, we will find out if I am right.

I will admit a certain satisfaction with this whole Solyndra business.  There are few feelings better than being proven right, part of the reason I enjoy trading so much. With anecdotal evidence piling in form Solyndra type scandals-that is the vast and embarrassing literature of “shovel ready” projects-those who opposed the stimulus look better and better.

The number one economic sophism of our time is be the idea that monetary and fiscal stimulus work though different channels. As Sumner keeps reminding us, if fiscal stimulus works, it is because it raises AD.  If monetary debasement works, it is because it raises AD.

The government could be cut dramatically without any systemically negative consequences so long as the CB made its commitment to offsetting the spending shortfall.  To believe otherwise it to say that cutting government workers reduces AS.  Maybe it does, but I need details.