Unemployment: Playing with Numbers

A little while back, I came across this article published by Mark Perry, a contributor to the American Enterprise Institute.  The headline reads:  “Texas, the ‘great American job machine,’ is largely responsible for the +1.2M net US job increase since 2007“. 

The article doesn’t make any direct ideological claims, but was posted just three days after President Obama proudly touted figures illustrating America’s recovery from the great recession in the 2015 State of the Union Address.  I can’t know Perry’s intentions for certain, but a quick scan of the comments on the site gave a clear indication of what people (specifically, the sort of people who read AEI blogs) took away from it: Obama was claiming credit for the job-creating accomplishments of the conservative bastion of freedom that is the state of Texas. Some comments (including the Facebook post that made me aware of this article) credit Texas’ low taxes and lax regulatory environment for their job creation.

Digging a little deeper, I came across this response piece by Barry Ritholz in the Houston Chronicle, which argues that Perry stacked the deck by starting his analysis in December 2007, before the effects of the financial crisis. According to Ritholz, Texas was not hit as hard by financial crisis of 2008 as other states, and thus managed to lose fewer jobs during the recession.  Ritholz argues that Perry wrongfully attributes Texas’ huge share of 2007-2014 job growth to the oil and natural gas boom, and instead credits regulations in Texas that insulated the state from lending practices that contributed to widespread mortgage collapse.

This piqued my curiosity, so I decided to take a look at the data for myself. A quick note: I told myself I would try not to make any overtly ideological posts here, and I’m going to make an effort to adhere to that while discussing the data.

What AEI Claims: 

“It’s a pretty impressive story of how job creation in just one state – Texas – has made such a significant contribution to the 1.169 million net increase in total US employment (+1,444,290 Texas jobs minus the 275,290 non-Texas job loss) in the seven year period between the start of the Great Recession in December 2007 and December 2014. The other 49 states and the District of Columbia together employ about 275,000 fewer Americans than at the start of the recession seven years ago, while the Lone Star State has added more than 1.25 million payroll jobs and more than 190,000 non-payroll jobs (primarily self-employed and farm workers).”

So According to Perry, from December 2007 to December 2014, Texas gained 1.169 million jobs, and the rest of the nation lost 275k.  Perry used overall employment data that includes farm employment, but I used non-farm data from the Bureau of Labor statistics because I couldn’t find overall employment data going back to 2007.  Because we’re using different measures of employment, my numbers will look a little different, but I don’t think it changes the overall point.  (You can find the API calls I used to obtain the data, and SQL code I used to clean it in this Github Repository.

The first order of business was to recreate Perry’s findings using BLS data. From December 2007 to December 2014:

Texas: 1.23 million jobs added (12% increase)

Rest of nation: 493k jobs added (0.4% increase)

So using different measure of employment, the conclusion is still essentially the same: when you look at the data with 2007 as a baseline, Texas appears to dominate any job growth that occurred across that period.

But as Ritholtz points out, starting from December 2007 combines two different events: the financial crisis and the subsequent recovery.  The trough of the Great Recession, by most metrics, occurred around September of 2008.  In order to avoid capturing any seasonal effects, I start my analysis at December of 2008.  From December 2008-2014:

Texas: 1.18 million jobs added (11% increase)

Rest of the nation: 3.70 million jobs added (3% increase)

That’s quite a difference.  Texas still outpaces the rest of the nation, but the rest of the country created over 3 times as many jobs as Texas did.

Part of the reason is that in the additional year that Perry included, from December 2007 to December 2008, Texas actually created 52k jobs, while the rest of the nation lost 3.2 million jobs.  So Texas has done well to create jobs, but the employment picture after this initial loss of 3 million+ jobs doesn’t support Perry’s claim that Texas was almost entirely responsible for recent American job creation.

To Perry’s credit, he never explicitly claims to be looking at recovery and never directly states that he means to contradict Obama’s claims about recovery.  Still, and especially given the proximity of Perry’s article to the State of the Union address, it seems disingenuous to lump a period of heavy losses and a period of heavy gains all into one category, and call the overall effect “job creation.”  

It’s worth briefly discussing Ritholtz’ claim that Texas was spared the wrath of financial crisis by their strict regulations on mortgage lending, and not by their lax corporate regulatory and taxation policy.

The reason – surprisingly for a state that likes to think of itself as a model of free-market economics – are regulations enshrined in the Texas Constitution and other legislation. Texas is the only state that limits home-equity borrowing, capping total mortgage debt at 80 percent of a home’s fair market value. That helps prevent using one’s own house as a piggyback for engaging in reckless speculation, as we saw in the rest of the country in the mid-2000s.

This post article from 2010 explains the phenomenon in more detail.  Beyond the restriction on home-equity loans, homeowners in Texas were unable to use “cash-out” refinancing – increasing the total balance of their mortgage in exchange for short term cash – because of the constitutional cap on mortgage debt.  These restrictions, the author argues, prevented sub-prime borrowers from taking out exorbitant debt against the value of their homes, leaving them less vulnerable during the nationwide economic downturn.

On face, it may appear that Texas owes its resilience during the recession to these borrowing restrictions, but Texas also benefited from the fact that it never developed a large housing bubble in the first place.

On average, the home-resale prices of the 20 metro areas in the Case-Shiller Home Price Index peaked in 2006 after more than doubling since 2000. In Dallas, one of the 20 areas, they rose just 25 percent, gradually, and have barely declined.

Texas’ unique lending restrictions only applied to loans on existing mortgages, so it’s doubtful that they alone were enough to prevent a bubble from forming.

So it seems Perry and Ritholz each tell part of the real story. Ritholz is probably right that Texas home-owners have less unmanageable debt thanks to the state’s unique lending restrictions, which helped them weather the recession, but these alone can’t explain the lack of housing bubble in the first place.  Perry’s claim that Texas has outperformed the nation seems to hold up even when we look at post-recession numbers, though the difference is far smaller than he claims.

Let’s take a step back.  This map, based on the same BLS non-farm employment data, shows the percentage change in employment for each state between December 2008 and December 2014 – it cuts out the first year that AEI used.  Screen Shot 2015-02-03 at 11.38.37 PM

You may not see the point I want to make with this map, and that’s because I’m not really making one.  This patchwork map is a reflection of raw numbers, stripped of the narratives we may be tempted to weave around them.  I could mention the strength of Texas and North Dakota to talk about the Shale boom, while ignoring the counter-example of Wyoming.  I can take a slice of the country and show how in that region, blue states outperform red states or vice versa. I could single out the home states of presidential hopefuls and allege mismanagement or gubernatorial incompetence.

But the numbers don’t do any of that themselves, numbers don’t do anything.  There are many ways to manipulate statistics to tell your story, and I don’t mean to disparage anyone for doing it as it’s considered essential for both politicians and pundits.  Maybe it was wrong for Obama to claim credit for the strong jobs numbers and maybe it was equally wrong for the GOP to use weak job numbers as a stick to beat him for so many years.

Fortunately, as I don’t have a defined ideological readership to please, I don’t have to take a stance and boil a complex issue down to one and only one brand of policy.

Statistics don’t reveal truths, they only simplify portions of the truth. I personally believe we would all be better served if we made the effort to question numbers and dig deeper, even if those numbers align with narratives that we like.  Wait, that last sentence sounds a bit like an ideology… oops.


Github Repository