Postlude: CNN Money Kingsport Small Business Award

[Update 4:45pm: The City of Kingsport representative has confirmed, in an email, the accuracy of the original analysis, adding: "I will be talking to Census and CNNMoney.com on Monday about correcting the data sets." See this subsequent Bjournal.com article for more details. Issue resolved?] 


Postlude (20 November 2009 10:45am): Two days ago, on a call between a reporter and a representative of the City of Kingsport, two claims surfaced regarding CNN Money’s ranking Kingsport as the best place to launch a small business: first, that our analysis of the error was itself errant; second, that another error existed in the Census data used to rank Kingsport number one.

Unfortunately, the representative had not read the original blog post, prior to making these same claims to city and community leadership,  nor had the representative been able to pinpoint the second error. The reason is quite simple: there is no other error.

A website the representative pointed to as proof that a second error existed is the same website we used to analyze the error in the first place.  Often referred to as CenStats, the site allows one to choose an MSA and a year and then generates detailed info about an MSA - the exact detail CNN Money uses to populate the snapshot section of its rankings. 

• Choosing the Bristol VA MSA for 2004 yields the 1779 businesses that were pointed out in the original blog post as the businesses missing from the baseline. 
• Choosing the Kingsport Bristol TN VA MSA for 2004 yields the original errant baseline number of 4235 businesses.

Notice a pattern? In two clicks on the CenStats website, the error emerges, consistent with the original analysis. One more click (choosing the Kingsport Bristol TN VA MSA for 2007) yields the correct 6087 businesses at the end of 2007.  

Of even more interest, an extra two clicks of CenStats obliterates an objection mentioned in the first post, namely that the Bristol VA and Kingsport MSAs were combined together all the way back to 2000. CenStats clearly shows the Bristol VA and Kingsport MSAs as separate all the way up to 2006, combining them together in 2007. If one tries to choose the Bristol VA MSA for 2007, the end result is a screen that says “No record found.” 

Year	Kingsport MSA (24700)	Bristol MSA (14980)
2004	4235	1779
2005	4222	1813
2006	4210	1801
2007	6087	0

The full chart shows the Kingsport MSA’s small business numbers declined in 2004 - 2006, while the Bristol VA MSA’s small businesses actually grew.  The original post intentionally avoided pointing this out, but the requirement to “show our work” (like in long division in elementary school) now openly reveals the small business decline in the Kingsport MSA. 

How will Kingsport address this Census Bureau administrative/CNN quality control issue? 



Finally, what about jobs?  Did we gain anything?

Yes, we did. The combined MSAs saw a total of 8,085 jobs added for the four year period.

The CenStats data doesn’t tell us how many of those were small business jobs, though, as it only provides business size and aggregate employees, per industry, without breaking them down to aggregate employees per industry per business size.

I suspect most of these jobs were gained from large businesses, for four reasons:
First, small business declined through most of the timeframe, as we only added 73 small businesses over the four years.

Second, even if we assumed all these 73 small businesses were made up of the maximum number of employees (49 employees) that would only bring us to 3,577 jobs created, less than half of the total 8,085 jobs created in that same timeframe.

Third, a maximum possible 3,577 small business jobs number can't possibly be true, because the number of 20-49 person businesses only grew by 15 total for the timeframe, yielding a maximum possible of 735 jobs.

Fourth, even if we run a formula based on a consistent weighted average, where the average is the percentage of maximum employees per business size (100% would mean that 1-4 person companies would have four people in them, while 50% would mean of 1-4 person companies have 2 people in them) we end up with a decline in the number of small business jobs.

At a 50% consistent weighted average, we actually lose 2,147 small business jobs.
So, while there is no doubt we can say the combined MSA gained small businesses and gained jobs, we can't say for certain we gained jobs in small businesses.

Jobs	Kingsport MSA	Bristol MSA	Aggregate
2004	76,694	29297	105,991
2005	77,040	29289	106,329
2006	81,214	30420	111,634
2007	114,076	0	114,076
			Total Jobs Added 2004-07	8,085

Food for thought for someone like me who is always touting small business as the growth engine -- maybe Kingsport is unique after all, appearing to have had significant mid- to large-sized business growth.

Theaters, theaters, everywhere, and not a one being bought by a city

v

Embarrassment in Entrepreneurship?

A great quote from the founder of LinkedIn:

"If you are not embarrassed by the first version of your product, you’ve launched too late."

http://www.openforum.com/idea-hub/topics/technology/article/the-iterate-fast-and-release-often-philosophy-of-entrepreneurship-ben-parr




iPhoned

Kingsport's Number One? CNN Money award revisited

[Updates: Bjournal editorial, "neighbor to the south" numbers, other CNN Money miscalculations. Phone call with City representative. Additional details appear at end of post.] 


Unlike the accolades I've enjoyed writing about Kingsport, both here and at LaunchPin.com, this is a painful article to write. Kingsport has won numerous awards, but the community must decide how to address the recent receipt of an erroneous award, finding our voice before others speak for (or, more likely, against) us.

Significant interest has been generated in Kingsport's status as a small business hotbed, based on CNN Money's designation of Kingsport as the USA's fastest growth area for small businesses between 2004 and 2007.

Once CNNMoney.com posted the list, the news quickly spread, including a mention in the widely read Kingsport! blog and several articles in the Kingsport Times-News (October 13 and October 15). I blogged the news, via DevelopedEconomy.com, which I heard in London, where I and another Tri-Citian were speaking at a new media/marketing convention. (Brits at the conference coined the term "Digital Hollow" in our honor, which we Tri-Citians morphed into "Digital Holler".)

The number-one ranking was welcome confirmation of northeast Tennessee's potential as a whole, and of Kingsport's role specifically, especially for those of us who love Kingsport and trumpet it as a great place to live and work.

Yet something didn't seem right with the numbers.

A hunch may be the basis for questioning a statement, but it means nothing without confirming data. Since returning home to Kingsport, I've spent several hours combing CNN and local data to understand whether the "gut feel" was unfounded or accurate.

Unfortunately, the answer is that CNN Money's calculation, boosting Kingsport to the number one spot, is highly flawed. The proper calculation yields a growth rate of 1.21%, not 43.7%, meaning that CNN Money overstated the Kingsport area's true business growth by over 3600%.

To understand how this error occurred, here's a look behind the numbers.

Improbable growth rate. According to CNN Money, 5.0% was the national average in small business growth for all US mid-sized metro areas from 2004-2007. Yet Kingsport's metropolitan statistical area (MSA) grew by 43.7%, to a total of 6,087 small businesses in early 2008?

Considering that the growth rate of St. George, Utah, number 2 on the CNN Money list, was 29.8%, the 43.7% growth rate that CNN Money cites for the Kingsport metro area implies a superheated business climate.

CNN Money's other business statistics for Kingsport yield an economic snapshot that isn't so flattering as the purported small business growth. During the years in question, the Kingsport MSA lagged behind the national average for all mid-sized metro areas in vital economic indicators:

• Population growth (2% versus 5.73% nationally)
• Percentage of 25-34-year-olds with bachelor's degrees (18.9% versus 26.9% nationally)
• Per capita income ($29,730 versus $35,547 nationally)
• GDP growth (17% versus 32% nationally)

In fact, the only two categories in which Kingsport outperformed the national average during this time were property crime (3,706.1 versus 3,562.2, per 100K inhabitants) and small business growth (43.7% for Kingsport versus 5.0% nationally).

To put some perspective on those increases, note that the increase in Kingsport's property crime—not something the area would be proud to claim—is approximately 4% over the national average.

Kingsport's small business growth, on the other hand, is reported to outstrip the national average by 880% (43.7% growth versus 5% growth) for 2004-2007.

Improbable number of startups. CNN Money correctly reports that the Kingsport MSA had 6,087 businesses by the end of 2007.

A growth rate of 43.7% in four years means that the Kingsport would begin 2004 with only 4,235 small businesses (6087 ÷ 1.437) and would add 1852 businesses in four years' time.

Does it seem plausible for an MSA with only 4,235 small businesses to add almost half as many companies again to its rosters in a four-year period?

If so, where did those entrepreneurs come from? Rough estimates from the Kingsport Office of Small Business and Entrepreneurship (KOSBE) show that KOSBE helped approximately 350 businesses during the 2004-2007 timeframe, about half of which were start-ups.  This means KOSBE can account for about 175 new small businesses, or less than 10% of Kingsport's growth as reported by CNN Money.

The local incubator, Holston Business Development Center (HBDC), can account for around another 20 startups in that timeframe, leaving 1,647 business startups unaccounted for.

The HBDC and KOSBE numbers assume that no existing small businesses shuttered during that same four-year period, so a more accurate reflection might be to say that 100 small businesses were added during that time, leaving an approximate 1,750 small business gap.

Flawed methodology. What about the other 1,750 new businesses?

While 1,750 new businesses don't exist, based on the CNN Money error in deriving growth numbers for the Kingsport MSA, they are real: prior to 2004, at least 1,750 additional small businesses existed in the same geographic region, but were not counted in CNN's 2004 baseline.

In its methodology section, CNN Money explained how it compared MSAs. "We began our search with the U.S. Census Bureau's 363 Metropolitan Statistical Areas. Each MSA includes at least one urbanized area of 50,000 or more inhabitants." It then gathered information from standard government and industry sources. For data about small business populations, it cited U.S. Census Bureau 2007 County Business Patterns. "This tally is for the number of businesses in the MSA with 1-49 employees. We measured three-year [sic] small business growth from 2004-2007."

CNN Money missed the fact the Kingsport MSA's composition changed between 2004-2007. It's easy to see why they missed this: no other MSA in the top 25 list changed in this time period.

In the census data used to by CNN Money create the Kingsport MSA baseline, the MSA contained only three counties. For those three counties, a total of 4,235 small businesses existed. Data from the Census Bureau website (published on June 10, 2003) shows what CNN Money used as its baseline for the MSA 28700, the Kingsport MSA:

Kingsport-Bristol, TN-VA Metropolitan Statistical Area

47073         Hawkins County, TN

47163         Sullivan County, TN

51169         Scott County, VA

Beyond the mid-2003 baseline numbers, though, the Kingsport MSA composition changed. At least according to the Census Bureau website. Data for Kingsport's MSA (28700), in early 2008:

Kingsport-Bristol-Bristol, TN-VA Metropolitan Statistical Area

47073         Hawkins County, TN

47163         Sullivan County, TN

51169         Scott County, VA

51191         Washington County, VA

51520         Bristol City, VA

The "new and improved" MSA "added" the city of Bristol, VA, and Washington County, VA. (Bristol, TN is included in the Sullivan County portion of the MSA, but Virginia recognizes cities as separate from counties for both state and federal census models.)

For those four counties and Bristol, VA, a total of 6,087 small businesses existed.

The math that CNN used took the correct final number (6,087) for all four counties and Bristol, VA, and subtracted the faulty baseline number (4,235) from the first three counties in the Kingsport MSA.

This change in composition means that CNN Money derived inaccurate small business growth (6,087 - 4,235 = 1,852 new businesses) with an errant growth rate (43.7%) that made Kingsport the number-one growth area in the nation.

Actual Results. In reality, in the mid-2003 Census Bureau data, Bristol, VA, already had 647 small businesses and Washington County, VA, already had 1132 small businesses.

This means a total of 1,779 businesses need to added to the 2004 baseline to derive an accurate growth rate. The true growth calculation, then, would be to compare 6,014 businesses in 2004 (4,235 + 1,779) to 6,087 businesses in 2007.

The corrected MSA numbers would show that the Kingsport MSA gained 73 total small businesses for the 2004-2007 timeframe, a 1.21% growth rate.

CNN Money's calculation would be correct if Bristol, VA, and Washington County, VA, had zero small businesses in mid 2003, but added 1,779 small businesses in four years. 

If that were true, the supergrowth in Bristol and Washington County, VA, would be huge news. Yet is as inaccurate as the errant 43.7% growth rate for the Kingsport MSA.

Interestingly, the corrected model shows Bristol, VA, moved into the negative column, losing small businesses between 2004 and 2007. Small business gains were mainly in Hawkins County, TN, perhaps thanks to the HBDC's joint effort with the Tennessee Small Business Development Center, which started offering courses at HBDC in 2007 on how to start a small business.

Sullivan County gained a total of 15 businesses for a .45% growth rate. That's 0.45%, not 45.0%.

What if CNN Money had chosen a different set of Census Bureau data for its baseline? Data from Census website for late 2003 (published to the web on Feb 25, 2004) yield proper MSA composition for MSA 28700:

Kingsport-Bristol-Bristol, TN-VA Metropolitan Statistical Area

47073         Hawkins County, TN

47163         Sullivan County, TN

51169         Scott County, VA

51191         Washington County, VA

51520         Bristol City, VA

Why CNN Money chose the mid-2003 data rather than the more recent late-2003 data is unknown, but it led directly to the calculation error of 43.7% growth.

Unfortunately, for those of us who want to champion Kingsport's small business potential, correcting the CNN Money calculation mistake will destroy Kingsport's standing in the growth rankings, pushing us far out of the top 10, 25, and even 100.

Does this incorrect calculation mean that Kingsport itself is not growing? Of course not. Look at downtown's "overnight success" which was planned long before new business and property owners came on to the scene. Going as far back as the 1999 economic development summit, the city has held a proactive approach to small business growth and diversity of business types. And our MSA growth rate appears to be higher than our direct neighbor to the south.

Action. Now we come to the hard part: what action to take. We didn't ask for or seek the award; nothing we could do as a community would have affected the outcome; and the error is certainly not "our fault" in the sense that Kingsport had nothing to do with the skewed results.


While "do nothing" may seem an appropriate action, my professional experience in crisis management PR suggests that it is a short-term, self-deluding action, as in "waiting for the other shoe to drop." The shoe is poised; I am comfortable others have also discovered this error, and will probably publish it as well.

Acknowledging the error to CNN, then, appears the best way to avoid a potential crisis.  Why?

First, if Kingsport business leaders let the accolade stand, knowing that we didn't earn it, we can't broadcast our ranking with integrity and self-belief. It's like knowing your grade-school teacher transposed a test score from a 79 on your paper to a 97 in the grade book. Is the A deserved?

Second, anyone really considering Kingsport on the basis of the ranking will see the disconnect when reading the less-favorable snapshot numbers.

Third, if someone else points out the error to CNN—perhaps #2, Provo, UT, or would-be #10, Las Vegas—the ensuing publicity will lead to shame and humiliation for Kingsport.

Fourth, shaping a message to market ourselves as the number one "most improved" spot in the nation is extremely difficult; marketing ourselves as an emerging spot to watch is more effective.

Finally, if CNN opts to change the rankings, there's the possibility they will publicize the change. The publicity we will gain from being seen as an "honest" small city is much better than the publicity that any "most improved" award could ever generate. Probably not, as it becomes CNN's crisis to manage, but we've got a better chance at it if we're the ones who bring it to their attention.

[Update, 3 Nov 2009 4:30pm: Apparently CNN Money miscalculations have happened before, with communities in a neighboring state proactively addressing the issue.]

[Update, 3 Nov 2009 5:15pm: One of confirming data sources sent new data correcting my  "neighbor to the south" statement. MSA 27740 grew at 4.0%, versus national average of 4.1%.]


[Update, 4 Nov 2009 8:15am: Asked why I didn't include this Times-News article; only comment is that the commenting seemed to be more sniping than constructive, and jokes about adding meth lab businesses seem to be a bit out of place when a real issue is at hand.]


[Update, 4 Nov 2009 11:10am: As mentioned in the main article, and as can be seen on blog post comments in multiple locations other than DevelopedEconomy.com, several others also had that "huh?" moment when hearing about the win. It's taken a bit longer for me to figure out the true error (the process of which I'll address in a subsequent post) and even longer to choose to publish it. 


I did so once I confirmed that an editorial in next week's Business Journal will note that at least one economic development entity had this information shortly after the initial ranking announcement. 


The editorial will state the entity chose to ignore the error, going forward with plans to broadcast the errant win. Whether they felt they were doing their community a service by withholding information, or whether it was a move of self-preservation, the facts are clear. 


We have less than a week to address this before the loudest voices heard come from outside the region.


Update, 20 Nov 2009 10:24am: Had a rather unbelievable call with a City of Kingsport representative a few days ago. Ugh.]