Thursday, February 6, 2014

Indiana Rep Turner Mischaracterizes Economic Data in Support of Legislation

In Indiana there is a current political battle over whether the legislature, which currently has a GOP supermajority in both houses, and a GOP governor, will pass a Constitutional amendment defining "marriage" as solely between a man and a woman, and includes a ban on any legal protections that are similar to marriage (such as civil unions). While this would seem to be a moral/religious issue, advocates for the ban are trying to make an economic case for why the ban would be beneficial. Large Indiana businesses, like Eli Lilly, and Cummins, along with various Chambers of Commerce, have argued in hearings that such legislation makes it difficult for them to recruit qualified candidates to Indiana. In response to these business claims, advocates of the amendment produced testimony such as the following:
Eric Turner, R-Cicero, an author of the amendment, cited statistics he said showed business officials were wrong in their analyses. “Eight of the top nine states with the highest rate of job growth in the private sector have an amendment to define marriage,” Turner said. According to Turner, the four states with the highest gross domestic product per capita – as well as four of the top five fasting-growing per capita income states – all have an amendment to protect their states’ definitions of marriage as one man and one woman. Micah Clark, executive director of the American Family Association of Indiana, echoed Turner’s sentiments and called the claims of economic harm a make-believe “boogeyman.” He said North Carolina – the most recent state to pass a marriage amendment in 2012 – has experienced economic gains in the calendar year since the 2012 passage of its amendment. Clark cited the North Carolina Chamber of Commerce saying, “business investment increased more than 4 percent, unemployment dropped, and the state added more than 42,000 new jobs after voting to protect marriage.” Franklin College Journalism
As a sociologist, I tend not to accept "anecdotal" evidence, such as the unsystematic, poorly documented narratives from Lilly and Cummins about their experiences, since the experiences of a handful of businesses may not represent the majority of experiences, and without systematic documentation, their perceptions may not even match reality. Further, I also tend not to accept decontextualized "snapshot" data, such as the statistics quoted by Turner and Clark. While they may be accurately reporting statistics, they neither demonstrate that a) these statistics accurately characterize the broader patterns of these states, nor b) do they provide any kind of causal mechanism for the relationship between them. As Nuvo news editor, Rebecca Townsend states, "Correlation does not causation make."

After hearing the testimony of the corporations, as well as the claims by Turner and Clark, I perused the publicly available data from the Census, and the Bureau of Labor and Statistics--a fuller look at the exact same variables that Turner and Clark use, tell a completely different story than the narrative they attempt to generate. In an unsolicited e-mail correspondence with Ms. Townsend, specifically in response to the statistics presented by the pro-amendment advocates, I presented an alternative reading of the data, one which takes the view of a social scientist. In our case, our job is not to find any data that supports our political or business position, the presumed approach of a campaign manager or marketer. The scientific process is based on an objective rendering of data, with a goal of creating the most accurate picture of the data that is possible, not just to support one's particular cause. Below is an abbreviated copy of the e-mail I sent to Ms. Townsend, which represents my look at the variables that Turner and Clark cited--but placed into the broader context of what the economic data is actually conveying about these states. Full disclosure, Ms. Townsend subsequently highlighted, in a Nuvo article, a recent analysis I produced of the economic impact of same-sex marriage legislation on Great Lakes states, which I also posted on my blog, where I will post updated pdf's of my research on this topic.

I wanted to specifically address Turner’s peculiar statements regarding job growth and income. There are 3 approaches I would take when trying to understand whether Turner's narrative is accurate, in terms of the story he wants to generate by quoting these statistics. The first is whether his statistics are cherry-picked. For example, why "8 of 9" states for job growth, and why "4 of 5" for per capita income? Those seem like odd numbers. I played around with the data (which I downloaded from the BLS and BEA web sites after reading his statistics), and I can use the exact same data to tell the opposite story. For example, regarding the real GPD/capital growth, I can say that "6 of the bottom 7 growth state have constitutional amendments or legislative bans on same-sex relationships." Regarding the jobs growth numbers, I can say, "13 of the bottom 15 states have constitutional amendments or legislative band on same sex relationships." I'm using the same data, but just using the bottom instead of the top numbers.

From a more scientifically accurate perspective, if you create 3 groups of all 50 states--those that have constitutional amendment bans (CB), those that have legislative bans (LB), and those with no bans (NB), then you can average the jobs and real GDP/capita growth for each of those groups. In that case, for jobs growth, you get 1.8% (CB), 1.4% (LB), and 1.7% (NB)--states with constitutional bans and without any bans have statistically equivalent jobs growth. Similarly, when you look at real GDP/cap growth, the numbers you get are: 1.2% (CB), 1.1% (LB), and 1.4 (NB)%-- the states that have no bans at all, actually have the highest rate of GDP/capita growth!. A very different picture from the narrative Turner is trying to generate.

However, even this analysis is a relatively poor way to really understand these relationships, because it’s a 1-year static approach, and it’s a look at all 50 states, when CA, HI, NY and FL will clearly have very different socio-economic dynamics than Indiana. So the second approach I would take is to recognize that the data he's using to support his case, are literally economic snapshots of one point in time over too broad an area. The study I attached looks at a 10-year period over just the Midwestern/Great Lakes states. A great example of why Turner’s approach is not very useful, is North Dakota--economically, it is in a unique position, because it is in the middle of an oil boom. Looking at the data that Turner chose to use, ND has GDP/cap growth rates of 13.2%, and a jobs growth of 10.4%. Contextualize that in the broader US perspective: the average US GDP/Cap growth rate was 1.2%, with the 2nd highest rate at 3.3%; the average jobs growth rate of 1.7%, and the 2nd highest growth rate of 3.9% --both the ND GDP/cap and jobs growth are about 3x as high as the next highest state, and about 10x the national average!!! That has nothing to do with their constitutional amendment banning same sex marriage--they are literally in the middle of an oil boom. But it gets "counted" in the statistics that support the anti-gay marriage constitutional amendment folks.

So focusing just on Great Lakes states, in my opinion, is a far more reasonable approach. If you limit the question to "what is the likely economic impact of a same-sex ban on Indiana?" comparing us just to other Great Lakes/Midwestern states, you come up with a very different narrative from what Turner describes. Using the same measures that Turner chose, and doing the same calculation as above, but just from 12 Midwestern states, I got the following, both of which show that states with no bans perform economically superior to states with no bans or limited bans:
Real GDP/Capita growth: complete ban=1.0% growth; limited ban=2.1% growth; no ban=2.2% growth.
Jobs growth: complete ban=1.4% growth; limited ban=1.2% growth; no ban=1.5% growth.

The third approach, is that I would question what definitions he is using when he talks about states that have “amendments defining marriage”. Some states have constitutional bans "only" on same-sex marriage, but their bans don't mention other types of legal protections, like domestic partnerships and civil unions. Some of those states, like Wisconsin, have legislatively allowed civil unions, even though their constitution bans same sex marriage. Other states with constitutional bans allow for local protections, like Ohio, where their domestic partner registry has been upheld by the courts, and many cities have adopted these kinds of protections. Finally, other states ban everything and no exceptions are allowed. Which of these definitions is Turner using? If you create 3 new groups, going back to using all 50 states since that was Turner’s original method—1) states with the compete bans (CB), 2) states with "limited bans" (LB), 3) and states with no bans (NB) --the real GDP/cap numbers are 1.0% (CB), 1.3% (LB), and 1.6% (NB)! The jobs growth numbers are statistically similar for each group, but the states with no marriage restrictions are still slightly ahead--1.7% (NB), 1.6% (LB & CB). Again, a very different picture than what Turner is telling, since there is a clear linear progression, with the most restrictive states doing the poorest, and states with no bans doing the best, even using Turner’s own measure.

So no matter how you analyze these numbers, as long as you put them back into a broader context, whether using all 50 states, or focusing just on Midwestern states, it all points to the fact that same-sex marriage bans seem to be economically bad for states. Turner is either listening to people who are intentionally manipulating facts, or who simply don't know how to do such analyses, and should be banned from using Excel.

1 comment:

  1. Excellent .. and this needs to be IN the media