Thursday, May 15, 2008

Coming up with monetary SoL by state without dropping $95

I'd like to do a follow-up on Steve Sailer's monetary standard-of-living by state post, but using a more valid cost-of-living measure than he did. The Missouri Economic Research and Information Center (MERIC) quarterly puts out a nationwide cost-of-living by state table. That's where Steve's CoL data come from.

MERIC gets its numbers from ACCRA's (now called C2ER) cost-of-living index, which measures CoL by city across six cost categories, which are then used to compute each city's total CoL index score. MERIC comes up with a simple average for each state using all of the cities within that state that participate in the survey. It makes no adjustment for population size. So in California, the cities of Palm Springs and Los Angeles are given equal weight in determining the CoL for the state as a whole.

Conveniently, this method makes Missouri appear more affordable relative to other states than it should. ACCRA uses 100 as its index average, so a city's score is always relative to the other city's participating during the same time period. MERIC boasts that Missouri has the fifth lowest cost-of-living in the country, at 90.3. If Missouri's composite score is adjusted for population size, it rises to 92.2. Only two of the seven participating cities are in the more expensive northwestern section of the state (Kansas City and St. Joseph), even though they represent nearly half of the participating population.

But for most states, the MERIC simple average actually makes them appear more costly than a weighted average would. The MERIC method yields a simple state average of 104.3 and a population-weighted state average of 106.3 for the US as a whole.

Since ACCRA's average is actually 100, the MERIC method is inflating most states' CoL, while deflating Missouri's. Again, Missouri's score of 90.3 using the MERIC method is lower than the 92.2 it receives when population is taken into account, something ACCRA presumably does. But the nationwide average is 104.3 or 106.3 using the MERIC method--higher than the ACCRA average of 100. Sneaky devils!

With access to ACCRA data for all participating cities across the country (MERIC's site shows the numbers for each of Missouri's particpating states), it would be possible to come up with a more accurate state CoL comparison than what MERIC provides. However, in Excel format, it runs at $95 for a single quarter.

Setting myself back like that is psychologically tough to do. I love blogging and wouldn't want (okay, wouldn't be able) to make a penny from it, but incurring accounting costs* in doing so is a hard pill to swallow, even a trivial amount like this. So if anyone happens to have access to the data and would be willing to send it to me, it'd be greatly appreciated. I wouldn't make the ACCRA data public, of course.

* As active bloggers know, work is an infernal enemy always nipping at the ankles. Economic costs certainly do exist.


Anonymous said...

I find two main issues with the MERIC data.

First (and probably lesser known), St. Louis Metro Area prices are not collected by ACCRA, which then deflate the numbers.

Second, as you mentioned, the contributing MSAs are weighted equally. A Connecticut research group has been working on this for a few years (and recently presented at a conference I attended on it). They use the geometric average of total wages * population to give a weight for each contributing MSA.

For example, because about 70% of metro population & wages are in NYC, when aggregating to a state index, the city weighs much more heavily. I feel their methods give a more accurate picture.

I would recommend you to their site "State Cost of Living Index" for more information, but it seems to not be working at the moment.

Audacious Epigone said...


I'm not getting anything with that address. But what you describe sounds much better than MERIC, which is too problematic to use as a basis of gauging the monetary standard of living by state in the US. Please let me know where I might find the site of this research group.