When President Barack Obama singled out St. Louis’ own LaunchCode as a national model for job training in early 2015, corporate donors and politicians couldn’t get on the bandwagon fast enough.
After reporting $241,000 in revenue for 2014, LaunchCode’s revenue rose to $3.7 million by the end of 2015, according to federal tax filings. Then in 2016, Vice President Joe Biden came to the tech jobs nonprofit’s office on Delmar Boulevard, once again putting them in the national spotlight.
“I came for a simple reason, to try to amplify what you’re doing here,” Biden said Sept. 9. By the end of 2016, LaunchCode’s annual revenue topped $5.4 million.
LaunchCode, which rapidly expanded to six locations nationwide after it was founded in 2013, sees itself as an innovative approach to education and job placement in a domestic tech industry expected to add 500,000 jobs by 2024. It says it places nontraditional candidates into computer coding jobs, and creates educational partnerships to train people who aren’t ready for apprenticeships, all for free.
However, how much LaunchCode has accomplished and its leaders’ methods now face increased scrutiny. Its Rhode Island hub was quietly closed in May after falling well short of expectations; LaunchCode has publicly feuded with Washington state officials over bureaucratic hurdles there; and the organization has had three executive directors in just the last year.
Documents and interviews also showed LaunchCode takes credit for educating more students than it has graduated, falls short of diversity goals, exaggerates its impact according to economists, and has, occasionally, resorted to aggressive tactics to gain support.
But LaunchCode founder Jim McKelvey looks at any setbacks or tumult with optimism, saying any young nonprofit goes through growing pains. After opening with just a few employees in 2013, LaunchCode now has 39 staff members combined nationwide.
“Frankly, I would be more worried if there wasn’t some chaos,” McKelvey said in an emailed response to questions.
LaunchCode’s main focus is putting people in apprenticeships, which pay $15 an hour and include coder training in an office environment working alongside a mentor at the business. In 2014, McKelvey said he expected LaunchCode in two years to be putting 2,000 computer engineers in jobs annually, available to anyone willing to put the time and effort into learning code.
“We say to people who want these jobs, they’re open to you,” McKelvey said last September at the public roundtable discussion with Biden.
LaunchCode since its founding in 2013 says it has placed more than 700 people in apprenticeships nationwide and about 81 percent of their apprentices get full-time jobs. But a few thousand more who went to LaunchCode for education or job placement have either been turned away or eventually dropped out of its programs.
McKelvey said only one-third of people who turn to LaunchCode will get a programming job, but he claims demographics have no bearing on who succeeds.
“The dirty little secret about LaunchCode is we can only do that with about a third of the population,” McKelvey told attendees at a business conference Wednesday hosted by Washington University. “The good news about LaunchCode is one-third of the population seems to be completely unaffected by age, gender, race or any other demographic.”
Fourteen percent of LaunchCode’s apprentices are black or Latino, according to numbers provided by LaunchCode. Twenty-four percent of LaunchCode’s apprentices are women.
LaunchCode leaders tout CoderGirl, a weekly class supporting women learning computer programming, and say that 40 percent of students in 2016 who spent at least two weeks in a LaunchCode education program were female. But they admit that reaching minorities has proven more difficult.
“For people of color, there’s larger questions about lifetime educational disparities and educational attainment,” said Haley Shoaf, LaunchCode director of education operations.
Fading industries and outsourcing have threatened middle-class jobs while the costs of higher education in the United States rise. In response, politicians and civic boosters have latched on to job training programs as a win-win partnership.
McKelvey, an area artist and tech entrepreneur who co-founded payments processor Square Inc., gave the nonprofit immediate clout in regional business and politics.
But not everyone was sold on LaunchCode’s promise.
The St. Louis Agency on Training and Employment, a federally funded city agency, spent 2015 delaying a funding partnership with LaunchCode, skeptical of what the nonprofit promised. By November, LaunchCode leaders were losing patience with Michael Holmes, who was SLATE’s executive director at the time.
Jeff Mazur, then LaunchCode’s point man on government partnerships, told then-executive director Brendan Lind and board member and former Missouri Secretary of State Robin Carnahan he had a plan “in service of putting a little pressure on our buddy Mike.” Mazur, a former labor leader in Jefferson City and former aide to Gov. Jay Nixon, forwarded the message to more LaunchCode employees.
“FYI only, but once Brendan gives the green light we’ll try to get some leverage on Holmes,” Mazur wrote.
The email was disclosed by former employees, who said they did not still have a copy of the plan attached to the email. Mazur did not respond to a request to see a copy of the document attached in the emails. Lind, who left LaunchCode in August, would not comment. Carnahan did not acknowledge interview requests.
Now LaunchCode’s executive director, Mazur said in an email that LaunchCode and SLATE “worked through some philosophical differences and today we enjoy a positive working relationship.”
Former LaunchCode employees say it was a whisper campaign to discredit a public employee for not moving forward on a job training and placement contract. Following the email, LaunchCode called a meeting with Mayor Francis Slay, and in January 2016 SLATE and LaunchCode signed a memorandum of understanding.
Holmes, who was let go by Mayor Lyda Krewson earlier in August, was shown the emails and would not comment for print. To date, SLATE has not cut any checks to LaunchCode.
A plan to pressure Holmes wasn’t insidious, Mazur said, but a routine way of finding “access to partners and resources that could help us pursue our mission.”
The vast majority of LaunchCode’s apprentices come from St. Louis. More than 100 placements have been made in Miami; 55 in Kansas City; and 20 in Rhode Island. In Portland, Ore., and in Seattle, where LaunchCode only recently began operations through a federal grant, the nonprofit claims two apprentices so far, both in Seattle.
About the same time as Obama’s shout-out, the U.S. Department of Labor was reviewing LaunchCode’s apprenticeship program, including its affirmative action plan. A review required by the Equal Employment Opportunity Commission found the nonprofit was underutilizing the St. Louis area’s ethnic minority and female populations in its job candidate pool.
Out of 166 people placed in apprenticeships by early 2015, about 18 percent were women and 15 percent were black or Latino. The review led LaunchCode to sign an agreement with the department “to make a good faith effort” to get each category above 20 percent.
The number of female apprentices improved to 24 percent as of August, but the number of blacks and Latinos in apprenticeships receded to 14 percent in the two years since that review.
Shoaf said LaunchCode was working on programs that reach potential candidates with lower computer aptitude, who tend to come from poor and ethnic backgrounds.
“And that’s where our Pre-LC 101 program and other elements of our admissions process and other support mechanisms are still being developed,” she said.
LC 101 is LaunchCode’s standard computer coder training course, which combines online coursework and in-person mentoring. Pre-LC 101, which is still being developed, would give students a base of computer competency needed before learning how to code.
More than 3,400 people have enrolled in courses organized by LaunchCode, according to the nonprofit’s 2016 annual report. LaunchCode counts them all as “students educated,” although that number includes everyone who spent at least two weeks in programs that can last up to 20 weeks and doesn’t reflect completion rates, which are about half that number, Shoaf said.
A 2016 report by the federal Equal Employment Opportunity Commission challenged tech leaders’ assertions that a talent pipeline is all that stands in the way of diversifying the tech industry. The report notes that the tech industry employs higher percentages of white people, Asian-Americans and men than other private industries overall.
“Despite rapid transformation in the field, the overwhelming dominance of white men in the industries and occupations associated with technology has remained. This tendency includes occupations requiring less education than a four-year bachelor’s degree,” the report concludes.
In an emailed response, McKelvey said LaunchCode would have to “lower standards” to accept more applicants for apprenticeships.
“LaunchCode provides opportunity to everyone. But, that really means everyone, we do not prioritize any demographic over any other,” McKelvey said.
But other LaunchCode leaders acknowledged they needed to account for people from different backgrounds who lack basic computer skills necessary to learn coding. To help, LaunchCode brought in help from a Chicago nonprofit, BLUE1647, that mostly trains people from poor and ethnic groups in computer programming.
That arrangement lasted for one year, BLUE1647 founder and CEO Emile Cambry said. During that period, BLUE offered 12-week computer training classes to 60 to 80 people under the age of 18 and to 40 to 60 adults.
“About a dozen” went on to be apprentices or take further coursework through LaunchCode, Cambry said.
“I think boot camps across the country have had a challenging time,” Cambry said. “At the same time, you want to be able to make the average person successful as opposed to just the top 10 or 5 percent.”
For 2015 and 2016 combined, LaunchCode accepted apprenticeship applications from nearly 27 percent of people applying. Nineteen percent were interviewed for apprenticeships, and about 11 percent were placed.
Ultimately, less than 10 percent were hired to full-time jobs, according to LaunchCode’s own numbers.
LaunchCode earns placement fees of about $5,000 per candidate from employers, its most reliable annual revenue source. LaunchCode received $1.1 million in fees in 2015, about one-third of its total revenue, according to the most recently available federal tax filings.
LaunchCode reported 97 individual donors in its 2016 annual report. Many are current or former LaunchCode employees, members of its board of directors, and family of those individuals. In 2016, a board member made a $300,000 loan to LaunchCode to be paid back without interest by May 2018, according to an audited financial statement that does not identify the board member.
If government and foundation grants eventually dry up, LaunchCode’s failure to create a base of individual donors means the nonprofit could rely on placement fees even more in the future. McKelvey said employers’ needs should be driving LaunchCode’s work anyway, but former employees say that pressure makes it even harder for LaunchCode to focus on education and workforce diversity.
“If we ever lose sight of what businesses need, we will fail everyone,” McKelvey said.
In New Jersey, LaunchCode currently partners with Express Scripts, the St. Louis County-based pharmacy benefit manager, to train potential employees in COBOL, a computer coding language used internally by Express Scripts. COBOL, which traces its start to the end of the 1950s, is widely viewed as an outdated code in the tech industry, which LaunchCode officials acknowledge but aren’t necessarily concerned by.
“We talk to companies and make sure there is a long-term path for those people,” said Emre Sarbak, LaunchCode vice president of technology.
LaunchCode has produced two measurements of its economic impact authored by Sarbak and other LaunchCode leaders. Both have appeared in numerous funding applications and in the nonprofit’s annual report, with the most recent claiming LaunchCode apprentices and students will “reinvest over $266 million back into the economy over the course of their careers.”
“The net impact for each placement was calculated after factoring in the individual’s current earning potential given educational background, estimated years left in labor force, and whether the individual already has a (computer science) major,” the report explains.
Shown those numbers, multiple economists said they probably overstated LaunchCode’s impact.
“They are garbage,” said Lindenwood University economics professor Howard Wall, formerly vice president and regional economics adviser at the Federal Reserve Bank of St. Louis.
Wall said the impact statements left out a number of variables or make faulty assumptions. Wall said he thought it was common for jobs nonprofits, not just LaunchCode, to overstate their impact, even when they call their numbers conservative.
“Each trainee will get some personal benefit above the cost, which is an indisputable economic impact,” Wall said. “It’s a rare occurrence when the external impacts are anything like what they claim so ineptly.”
Mazur said there were different ways to measure economic impact, “but however you tally it, LaunchCode’s programs add many millions of dollars in economic benefit to the communities in which we work.”
Rik Hafer, another Lindenwood University economics professor, was also skeptical, saying LaunchCode’s economic impact estimates don’t account for any number of potential changes in the tech industry, such as whether those jobs will be in as much demand or pay as well in the next 20 to 30 years.
“In the end, their back-of-the-envelope numbers are just that, and should be given just about the same amount of credence,” Hafer said.