Thursday, April 14, 2011

Not like Krugman

A graph like this (from Krugman) can be striking...

The big decline is all I can see. Lucky for you, though (ha-ha), I like to take a second look at these things. (And a third, and so on.) Krugman's graph is the EMRATIO from FRED. Below is the default EMRATIO graph from FRED:

It shows a lot more years, and it is zoomed-in to show all the detail that fits.

The next graph shows the same data, cropped to start at 2006 and end at 2012, just like Krugman's graph:

Looks like Krugman's, except he stretched his up-and-down, turning the default rectangle into a square. By doing that, he made the big decline look even bigger.

In the cropped graph, the lowest value on the Y-axis is 58. On the default graph, the lowest value on the Y-axis is 54. In the next graph, I re-do the cropped graph using the default Y-axis minimum of 54:


Well, no, I don't see how to do that in the FRED. Okay, I'll grab the data and do the graph in Google Docs.

Jeeze... Ever since I installed the Microsoft Office Excel Viewer on this computer, Neither Google Docs nor Microsoft Works will open Excel-format downloads from FRED.

Okay, I opened the file with the Viewer, used CTRL-A, CTRL-C to copy the whole thing, then jumped to Google Docs, opened a spreadsheet and pasted it all in. Here's the graph I made from it. Looks like Krugman's:

This has a minimum Y-axis value of 57, the default used by Google Docs. And now I can change that minimum value to zero and show you how it really looks:

So much for the big decline.

Now, here's the debt graph from my earlier post this morning, which always had the Y-axis starting at zero:

The increase shown on my debt graph is a real increase, not magnified by chart effects.

Update for Jazzbumpa

From How to Lie with Statistics by Darrell Huff:

Jazz, the ratio of private-to-public debt doubled in the decade after 1993. Nothing doubled or fell by half on Krugman's graph. Statistically, the change that I show is far more significant than the change Krugman shows. That is not to say it is far more significant in people's minds or in their lives, as far as they can tell.

Update #2

Well, gee, nobody on the graph makes less than thirty dollars, so we don't really have to show much less than that, I guess.

Wow! Looks like we make four times what those other guys make!


Jazzbumpa said...

Art -

You've put an awful lot of effort into this post, but I fail to see the point of any of it.

The span of the entire data range. going back to the beginning of the record, is from 55 to 65%. Putting that data on a scale that spans that range (or most of it) is not playing a visual trick. And note that PK is specifically talking about current events, not history, so the 5-yr time frame is relevant and valid.

OTOH, the fact that percentage numbers theoretically go from 1 to 100 is pretty much irrelevant in the context of these data.

Showing the "Not Krugman" chart as you did would make sense only if the data range spanned something close to the entire percentage range. What you've done is trivialize a huge drop, by making it look artificially small.

The reality is that this ratio has been dropping for a decade, then fell off a cliff - like everything else - in '07, and had shown no signs of picking up again.

All the gains of the previous 20 years have been wiped away.

There is a serious story behind these numbers. Please do not disguise it as something irrelevant.


The Arthurian said...

Added an update to the post for ya

Jazzbumpa said...

Art -

You've got an apple and a rhododendron.

Each has it's own worth, but asking why the rhododendron doesn't taste like the apple is misunderstanding both.

Nothing doubled or fell by half on Krugman's graph.

A fall from 63 to 58, when the known range is from 65 to 55 is in fact exactly half.

Consider the range of temperatures in a living human body - something like 95 to 105 deg F. Place data points on individuals with life threatening hypothermia and fevers on a sale that spans the freezing and boiling points of water, since we're water based life forms. Nothing there to see, right?

As I said originally, there is a story behind those numbers. I challenge you to give it some consideration.


The Arthurian said...

Ah, the Known Ranger rides again!

What do you suppose will happen to the "known range" of employment numbers as we transition into that neo-feudal society you so often write of?

I am using arithmetic. You are trying to use compassion. Compassion is fine for setting goals, but it is useless when it comes to understanding and evaluation.