Wednesday, March 24, 2010

New Varshney/Tootelian statement has new errors

There is a saying that when you are in a hole, stop digging. That's my advice for Dean Varshney and Dennis Tootelian after reading the statement they released last night. It makes many odd arguments, and a few statements about their own work which are demonstrably false. Here are two examples.

First, some excerpts about the cost of regulation study.

Furthermore, our report on the cost of regulations for small business was based on the 2005 federal study by Mark Crain which uses ordinal rankings...Rather than suggesting that regression methodology should not be used at all, our critics should recognize that the methodology comes with inherent issues of multicollinearity that can result in sign reversal. However, this methodology is useful and is no different than the federal study that Varshney and Tootelian modeled that also used ordinal rankings.
It is true that their study is modeled after Crain's federal study, on-line here. Professionally, I have some quibbles with Crain's methodology, but the parts of Varshney's report people are criticizing is when he did not follow Crain. And these are not minor quibbles, but huge, inexplicable errors.

1. He says Crain used ordinal rankings of regulations like them. Not true, Crain used a cardinal index of the intensity of regulations for countries produced by the OECD. See page 68 of Crain's report.

2. Crain uses per capita GDP (GDP/pop) as the dependent variable in his regression, not GDP as Varshney does. This error has been the focus of LAO and others critique. See this post for more.

3. Crain did not enter the projected change in GDP from his regression into an input-output model (IMPLAN) as Varshney/Tootelian do. This is a major error, that almost tripled their already inflated estimate of costs (from $177 billion to $493 billion).

Second, I am really confused by this part of their statement.

4)Varshney and Tootelian have been criticized before for their research
findings, only to be vindicated by the facts. For example the Sacramento
Business Review – a product of Sacramento State – predicted a 10% unemployment
rate for Sacramento last year. Critics took issue with the economic projections
and instead projected an unemployment rate of no more than 8%. Sacramento region was at 12.3% unemployment at the close of last year and is currently at 13.1%
based on January 2010 numbers.

This comment has absolutely nothing to do with the current issue, which is about their AB32 and cost of regulation studies. However, as someone who does a fairly prominent forecast of the Sacramento economy, I feel compelled to correct a few errors of fact in this statement.

I don't recall any critics "taking issue" with the inaugural Sacramento Business Review forecast in January 2009. I also don't recall anyone who was forecasting that Sacramento unemployment would be under 8% at that time, and certainly no one issued a new forecast saying this after the Sacramento State release. I checked the forecast we released a month earlier in December 2008, and we were forecasting Sacramento unemployment would average over 9% for the year in 2009, and our release two months later (March 2009) said Sacramento unemployment would reach 12% by the end of 2009.

Our December 2008 turned out to be too low, but it was the same as Varshney's in January 2009. He is misrepresenting his own forecast. If you look at the unemployment graph from this forecast he is boasting about (see figure 1 on page 8 of this report), you will see the Sacramento unemployment line briefly slips above 10% for the seasonal peaks, but is below 10% for almost the entire year! I don't get this statement of "being vindicated by the facts" and "critics taking issue."

They sent a news release on the very first monthly employment report after their very first forecast (January 2009 employment report released in March 2009) bragging about how they had acurately predicted that seasonally unadjusted unemployment spikes in the month of January, and he is doing the same thing today. I do recall privately criticizing that news release to a few people . Predicting an unemployment spike in January is like predicting that gasoline sales increase in the summer, retail sales increase during the holidays, the #1 seed will beat the #16 in the NCAA tournament, and the sun will rise tomorrow. Varshney keeps comparing his not seasonally adjusted peak values to others seasonally adjusted forecasts, but that isn't valid.

He needs to stop misrepresenting the facts and others' work. Stop complaining about politically motivated criticism of your politically motivated study. Put the shovel down.

Update: Matt Kahn and James Sweeney published a harsh and entirely correct op-ed in the LA Times today on the AB 32 study, which I didn't even mention above. It keeps getting deeper.

No comments:

Post a Comment