Immediately after I pressed the submit button on my last comment that thought occurred to me, but instead of correcting myself I waited to see what others thought. Except for the eternal skeptics, I think most people that comment on this blog think the same. The level of rates don’t represent monetary policy stance! ]]>

You said:

“Tyler does not oppose NGDP targeting, he just thinks its time has come and passed. It would have been effective in 2009, not now. Just ask him, don’t take my word for it.”

You don’t even know what “effective” means. He probably said that right now it wouldn’t have much impact on unemployment, unlike 2009. But I believe that! If he supports it then he supports it all all times. NGDP targeting is not something that you just do in certain years and not others.

Jeff, Thanks, but your attempt to educate Ray reminds me of this link:

http://www.metmuseum.org/collection/the-collection-online/search/484972

]]>Suppose you observe two series: (i) Q, the quantity of X bought and sold and (ii) P, the price of X. Econ101 tells us that in a plot with P on the vertical axis and Q on the horizontal axis, there is a demand curve for X that slopes down and a supply curve that slopes up. Where the curves intersect determines price and quantity simultaneously. If this equilibrium always holds, then it follows that the only way the P or Q ever change is if one or both of the curves move or change somehow.

For simplicity, imagine both “curves” are actually straight lines that are completely characterized (as is any straight line) by their slopes and intercepts. Furthermore, we will simplify things even more by assuming the intercepts can change over time, but the slopes never do. When the intercept on the demand curve increases, we call this a rightward shift of demand, and if the intercept of the supply curve increases, that’s a leftward shift of supply.

We don’t observe the curves, all we see are observations of P and Q. The identification problem is this: From observations alone, we can make only limited inferences about those slopes and intercepts.

For example, if we observe a period in wherein the P went up and Q went down, we can infer that the supply curve shifted left, but since there may also have been a demand curve shift, we can’t say much more than that. It could be that the demand curve is steeply sloped and it also shifted left, but by a bit less than the supply curve did, or it could be that the demand curve shifted right and is not very steep. If we want to use the data to estimate the slope of the demand curve, we have to make some further assumptions.

One simple assumption we could make is that the demand curve never shifts or changes in any way, i.e., all the changes in P and Q are due to changes in the intercept of the supply curve. If that’s true, we only need two observations that are not identical to know everything about the demand curve, as two points in the P-Q plane are enough to determine a straight line.

If the intercept of the demand curve is actually a bit random, we can still estimate the slope and intercept of the demand curve by regressing price on quantity, and our estimates should get more accurate with more observations. But we still have to make some kind of identifying assumptions to do so. There isn’t any way around this.

Note also that even if we get a decent estimate of the demand curve, without further assumptions we still can’t say much about the supply curve. For example, suppose we see both P and Q increase. We know that this can only happen if the demand curve shifts right, but whether the supply curve also shifted or not we can’t tell.

]]>*But let’s face it, even after the FED raise rates, monetary policy will still be very accomodative, don’t you all think?*

Interest rates are low, but that’s mainly because people expect low inflation, i.e. they expect money to be tight.

Even allowing for dynamic feedbacks like the above, it’s hard to describe the stance of monetary policy because one can always ask “compared to what?”

Probably the most proper description of the stance of monetary policy is how it relates to the trinity of RGDP growth, employment, and purchasing power. Obviously the government could set a negative 20% inflation target (perhaps even running surpluses to fund the Fed’s buying of dollars) and absolutely crush RGDP and employment while increasing purchasing power; we would all agree that is suboptimally tight. They could also set a positive 20% inflation target and that might help maximize RGDP and employment, but would certainly impair purchasing power.

(I’m extremely skeptical that monetary policy is intrinsically superneutral in the long run, I suspect that’s just usually the practical effect of how CBs react to events.)

So where are we today? Well, the optimal monetary policy seems to be to have NGDP growth of around 4-5%, that seems to best promote the trinity of economic welfare. **If NGDP growth is outside that range, monetary policy is either too accommodative or not accommodative enough**.

How many times has Sumner and his followers on this blog posted charts of NGDP and employment/output and said these correlations are “strong evidence” that the single variable NGDP is the cause?

]]>Sumner: “You say the Fed follows the market 80% of the time, and I want them to follow 100% of the time. So we are just 20% apart.” – I wish. If you could explain how the Fed printing money and buying paper ‘follows the market’ I would love to convert to NGDLT. In a sense, you may be thinking of the ‘real bills doctrine’ which back in the day said any Fed purchase of commercial paper freely offered cannot be bad. There is some logic to this pro-cyclical proposal–is that what you are referring to?

Sumner: “And you seriously think you can talk about Sims VAR studies without understanding the identification problem?” – yes, I cited the Sims paper for the proposition that the Fed has little or no influence over the economy, not for the identification problem. See also this: ‘King and Levine (1993) did not find evidence to support the hypothesized relationship between real interest rate and economic growth in a cross-section of countries. Taylor (1999) found that the link between real interest rates and macroeconomic aggregates such as consumption and investment is tenuous.’ (from a paper by Richard A. Werner). Granted interest rates are not NGDPLT, but go to IS-LM model analysis, but if they have no influence in the economy, with dozens of years of analysis vested in them, why would the unstudied and more complicated proposal of NGDPLT have influence? Ockham’s Razor dictates otherwise. Can’t you see your proposal is bogus? Have you vested so much of your time in promoting NGDPLT that you cannot let go, no matter what the fact say? The power of confirmation bias, as Haidt wrote about.

]]>Dan, You still don’t get it. Saying “GDP” just isn’t enough when you need to distinguish between real and nominal GDP. From the context of your quote you are using NGDP data for an RGDP argument.

Ray, I’ve never seen Tyler say he opposes NGDP targeting, when he does I’ll respond.

You say the Fed follows the market 80% of the time, and I want them to follow 100% of the time. So we are just 20% apart.

And you seriously think you can talk about Sims VAR studies without understanding the identification problem? That’s like talking about the theory of relativity without understand mass and energy.

]]>