Sunday, September 29, 2013

Chop Shops

For me, the goal is always to get as much data as I can. The more years, the better. Even so, I generally ignore the first ten years.

There's nothing scientific about that. It's just a habit I developed while looking at graphs of economic data. It's tempting to think you can imagine what the numbers were doing in the years just before the start of the graph, but I've been wrong often enough to know better. And since I cannot see what happened just before the first years of a graph, I don't trust the stories about what I do see in those first years...

I trust that what's there in those first years is valid. I don't trust the interpretations of it or the explanations of what brought it about. But that's just me.

Marcus Nunes (7 September 2013) quotes Ed Lambert:

Volcker put the breaks on the liquidity coming from the Fed rate in the early 80′s. And the babies born from 1960 on graduated from college into a world of tight liquidity due to tight Fed policy. They became disadvantaged. They did not find job openings. As a generation, they fell behind. I was one of those babies that graduated into the heart of the 1982 recession.

A little too personal for me, Ed, but I take your point. (BTW it's brakes, like brakes on a car: for slowing or stopping the car or (in this case) the liquidity.)

Marcus responds:

Actually it appears he had greater luck than those who graduated a few years earlier! The charts show how lucky Edward was!

Here's the second of Marcus's graphs, from that same post:

Graph #1 from Marcus Nunes
Marcus's graph makes it look like employment was stuck at a low level until Ed Lambert needed a job, and then things suddenly picked up. That's not really what happened. The graph creates a false impression because it only goes back to 1979 and only shows 10 years in all. Marcus uses the false impression to build his story.

Here's how FRED shows Employment for the 1979-1989 period, just like Marcus:

Graph #2: Employment, 1979-1989
Looks just like Marcus's graph. But wow, my numbers are way different from Marcus's numbers. Marcus, você pode explicar isso? Maybe we used two different data series? And yet, the patterns appear to be identical.

Here's the same FRED data for the full period, with Marcus's ten years in red:

Graph #3: Employment since 1939
You have to tip your head to the left a bit, so that the employment line since 1960 looks like a flat line. That's the "trend". (Employment was slowing before 1960, and also after 2000, relative to this trend.)

When you look at that trend since 1960, you can see a flat line with high spots and low spots in it. A low spot near every recession bar. One of those low spots, the big red one, occurred just when Ed Lambert graduated.

The low spots are not good times to go looking for work. Lambert went looking for work at one of the low points of employment. I don't see how Marcus can call that luck, unless he means bad luck.

Marcus chops his graph down to just a few years, making it look like the employment numbers support the story he tells. But what looks like a flat spot on Marcus's short graph is really a low point in a generally rising trend.

Ed Dolan (23 September 2013) does a pretty decent evaluation of the Money Multiplier, contrasting old and new views. He opens with a look at the older view:

The multiplier posits that there is a stable ratio between M2 ... and the monetary base

Dolan writes:
There is just one problem. As the following chart shows, something has gone badly wrong with the money multiplier in recent years. For most of the 1990s and 2000s, it was steady as a rock. From 1994 to 2007, the 12-month moving average of the multiplier stayed in a narrow range, between 8.0 and 8.4. Then it fell off a cliff. By July of this year, it had reached a record low of 3.24.

Yep. Dolan could have stopped the graph in 2007, if he wanted to defend the "stable ratio" argument. He didn't do that, maybe because he he's not defending the stable ratio argument. Or maybe because he wants to look at the graph and see what actually happened before he picks an argument to defend. That would be good.

Here's the FRED graph that corresponds to Dolan's graph:

Graph #5

Here's the full FRED dataset:

Graph #6
Who in his right mind would call that "stable"? And why does Ed Dolan not show the years before 1993?

Warren Mosler (24 September 2013) writes:
Interesting chart- inventory of existing homes for sale vs the labor force participation rate…
new home sales track closely as well…

I couldn't find data for inventory of existing homes for sale but FRED has new home sales, so I looked at that. Since 2005, like Mosler's graph:

Graph #8 (since 2005)

Next, the full dataset:

Graph #9 (all years)

"New home sales track closely," Mosler says. That wouldn't be my argument. My argument would be: You can take any little short period of time when two numbers are moving in the same direction, and zoom in on just that period of time, and pretend to show evidence of whatever you want to claim.

EDIT: Concluding sentence removed.


Jazzbumpa said...

My comment just disappeared into the ether. Blogger apologizes for the inconvenience. MEH!

I read the first part of your posr, re: Nunes v Lambert, will get back to the 2nd part.

Maybe two posts for two topics would be better.

I went through the same FRED exercise as you last night, with the same thoughts.

Nunes is cherry picking - that's dishonest, but it seems to be SOP for the market monetarists. I've called out David Beckworth for doing that on a few occasions.

More disturbing, What is Nunes point? He seems to have a case of the ass for AB. But he's been gracious to me off-line, which I find very confusing.

Is he just looking for a way to make Lambert look bad? I seriously do not get it.


Jazzbumpa said...

I'm going to defend Dolens. He specifically identifies, more than once, that he's talking about a specific period. Plus, he hints pretty strongly that the plateau was an aberration.

He's also suggesting that the multiplier might be a dead duck, and goes into a bunch of contributing factors, with an emphasis on the post-plateau decline.

My assessment is that his approach is valid, and he is not cherry picking.

Also, unlike Nunes, he is trying to transmit real, useful knowledge, rather than reach too far in a vain attempt to make some questionable point.


The Arthurian said...

It *did* turn into a long post (but I already split off Marcus's first graph from the same post...)

Marcus's short graph strengthens his argument.
Ed Dolan's short graph weakens his argument.
Warren Mosler's short graph gets him into the Labor Force Participation Rate discussion.

Marcus's was still festering in my brain when I found the other two posts, on consecutive days. There seemed to be a common theme: The Short Graph.

João Marcus said...

Art & JzB
I´m not "ungracious" on or offline! And I don´t think of myself as dishonest.
And I was not 'cherry-picking'. EL talks about a very specific period of time. Why should I bring into the argument the distant past?
I thought I was clear when I showed that he "went into the market" at the very beginning of a long boom.
That´s surely better than having graduated a few years earlier.
And Art knows I have posts that show data for 50 or more years, when that´s relevant for the story.

Jazzbumpa said...

Marcus -

Anything looks like a boom if you start counting from the bottom. Art, in a follow up post, shows exactly why the "distant past" is relevant. [Otherwise the distant future isn't either.] The new trend has a lower slope than the old trend. And Lambert entered the job market at the worst time between WW II and the GR.

You are absolutely cherry picking, and it is absolutely dishonest, whether that is your intention or not.

I entered the job market in 1968. Since then, unemployment has never been lower. My entire career occurred during an employment bust. By your reasoning, Ed had the good luck, and I had severe misfortune.

There is no way I can make sense of that.

Really - what is your point?