Tuesday, May 4, 2010

5.08.2010 On the Rorscharch test of pretty charts...

Recently, the blogosphere has been a flutter about this post by Mike Mandel. It contains this chart:

He observes:

This one fact explains much of the fiscal stress at the state and local level—why states such as New York, New Jersey, and California are in such a mess. State and local governments pay more than $1 trillion in compensation annually (actually, that’s an astounding number–I had no idea it was that high). If compensation is 5% higher than it should be, that’s $50 billion in excess pay costs for the state.

And lo and behold, that $50 billion would roughly cover the total size of the state budget gaps. For example, in February a survey found that the combined budget gap of all 50 states was $55 billion for the 2011 budget year and $62 billion for the 2012 budget year . (The survey was done by the National Governors Association and the National Association of State Budget Officers)

Of course, this chart doesn't actually tell us this. It does show that somewhere around 2006Q01, something happened where public and private wages started to diverge (or perhaps nothing happened, and this is a common event in the business cycle). Whether private wages were held low to where they should be or public wages were inflated higher than where they should be is unanswered by the chart. Indeed, the chart merely confirms a belief already held by the viewer, as we see in the comments where LP2 states:

Yes, the arithmetic IS very clear. Just eyeballing that graph, virtually all of the difference comes from stagnation in private sector pay between 2004 and Q1 2008. The recession started in December 2007 by most measures.

This graph doesn’t show higher wages for public workers “while the private sector struggles;” it shows the private sector sticking it to private sector workers.

Same chart, diametrically opposed conclusions.

So both sides partake in a thesis in search of data, and nothing looks more like data than a nice chart with some squiggly lines on it.

This is, of course, the opposite of what we should do. Upon observation of a divergence from either a historical relationship or a seeming identity, we should avoid the temptation to immediately shoehorn it into our existing world view. Rather, we can and should attempt to build a model of items in question, and from that we can see if, at the inflection point, what, if anything, changed.

Much of this is informed by my professional career in retail asset management. When faced with a market that swings every day and with clients calling wanting to know why their portfolios are having intraday swings greater than your Adjusted Gross Income, the temptation to declare something over or under priced, to declare the collective wisdom of everybody else in the market as wrong, is great. Indeed, the entirety of the asset management business is based off of being smarter than the market. But I felt it was always hubris to, in my small office in Los Angeles, declare myself smarter than every blowhard on TV. What we did better than they did, in my opinion, was to be more humble, to know what our clients wanted, and to know the limitations of our experience. We levered the information that we had and where our knowledge failed us, we had the humility to trust the collective knowledge (market).

The creation of useful knowledge - which is the primary unit of investing and economics - requires a willingness to embrace not knowing an answer. This requires work, and lots of it. If you are going to accuse the collective wisdom of a multitude of actors of making a mistake, your reasons should be better than your reading of Atlas Shrugged or A People's History...

Which is to say I'd love to look at Mr. Mandel's data and method backing this up, and see what this relation looks like going back further than 2001Q01.

No comments:

Post a Comment