Career Development and Employee Retention: The Evidence

Evidence grade: Mixed. Credible causal studies exist and disagree, or the effect holds only under conditions the studies can name. This grade describes the published evidence. It is not a prediction about your workforce and not a result RetainScore has produced.

Two different claims wear the same name

The first claim is that people leave for lack of career development. The second is that providing career development keeps them. They get cited interchangeably, usually with the same statistic attached, and they rest on completely different kinds of evidence. This page grades the second, because the second is the one a budget gets attached to.

What the number one reason actually measures

The figure that anchors most of this conversation is Work Institute's. In its words, "Career Development, or the lack of it rather, is the number one reason listed as to why employees quit their jobs", and it has held that rank for more than ten consecutive years, accounting for about one in five moves to another employer in 2019. Those numbers come from the firm's own exit interviews, "sentiments from nearly 100,000 exit interviews across diverse industries" in the current report.

That is a large and consistent body of stated reasons, and it deserves the weight of a stated reason. Our grade on exit interviews lays out why departing employees manage that conversation: they protect references, avoid naming managers, and reach for explanations that are safe to say out loud. A better opportunity elsewhere is the safest explanation there is. The stated reason is real. It is not a measurement of what a development program does.

What happens when employers actually pay for development

Four studies measure the thing itself, and they are more interesting than the survey.

Tuition reimbursement

Manchester studied a non-profit education employer that introduced tuition reimbursement in 1999 and tracked staff through 2005. Among people hired after the program began, using it "reduced their probability of leaving within five years by over 50 percentage points". A large effect, with two things attached to it. Only 4.5% of staff participated, which is a self-selected group, and the author is direct that "A shortcoming of the case study analysis is whether the findings can be generalized to other employers and programs".

The earlier study Manchester reviews, by Benson, Finegold and Mohrman at a large manufacturer, adds the twist that matters most for anyone designing such a program. Reimbursement kept people while they were studying. Once they finished a graduate degree, "these individuals have a greater probability of leaving than non-participants", unless the company moved them: "promotion decreases the probability of leaving for employees who obtain a graduate degree". Pay for the skills and then leave the person in the same job, and you have funded their exit.

Training

Dietz and Zwick used a German panel of 149 establishments from 2006 to 2010, and solved the usual selection problem neatly: they compared people who took training with colleagues who had been booked onto the same course but could not attend for reasons outside their control. Training raised the chance of still being there a year later. In the simplest model "training participation increases the retention probability in the training establishment in the next calendar year on average by 8%", and 9% to 11% in the models that handle individual differences and endogeneity. The effect shrinks as the training becomes more useful to other employers: "portable training reduces the retention effect", and certificates and external providers reduce it further. It never turns negative. "The total retention effect even of credible portable and visible training is positive".

The randomized trial

The one controlled experiment is from Indian garment factories, where Adhvaryu, Kala and Nyshadham randomized roughly 2,700 workers into on-the-job soft-skills training. Productivity rose 20%. Retention did not move: "Wages rose only modestly with treatment (by 0.5 percent), with no differential turnover". The timing is the useful part. "Retention was actually higher in the treatment group relative to control during the program period; this effect diminished after program completion".

The pattern

Read together, the studies say something more specific than development improves retention. Development retains people while it is happening, and afterwards only to the extent that the employer is the place the new skills get used. Skills the market prices better than you do, or a degree with no promotion behind it, tilt the other way. The stated-reason data fits this too: people who say they left for career development are describing an employer that stopped being where their growth could happen.

Why the grade is mixed

Mixed is the grade for credible studies that disagree, or for an effect that holds only under conditions the studies can name. The randomized trial found no retention effect; the two quasi-experimental studies found large ones. The conditions that reconcile them are named in the papers: whether the program is still running, whether completion is followed by promotion, and how portable the skills are.

This page replaced one of ours

Until today this URL held a page titled Lack of Career Growth as a Cause of Turnover, built on the Work Institute figure and treating it as a settled structural cause, on a par with pay and management quality. The figure stays on this page with its proper weight. The claim that a decade of exit-interview rankings proves a cause was more than the evidence supports, so the page was replaced rather than left up.

The honest boundaries

• The big effect is one employer. The tuition-reimbursement result comes from a single non-profit with 4.5% participation. Treat its size as an upper bound, not a forecast.
• The German result is observational. The accidental-non-participant comparison is a good design, and it is still not random assignment.
• The experiment is soft skills, not careers. Frontline garment workers on a training program are not an analyst waiting on a promotion. The null result says skill-building alone does not retain; it does not test career paths.
• One study is read second-hand. The Benson, Finegold and Mohrman paper sits behind a publisher paywall with its abstract withheld; the claims here are the ones Manchester's review states.
• The ranking is a vendor's exit interviews. Work Institute sells retention services and the data are its clients' leavers. That does not make the ranking wrong. It makes it a stated reason from a particular population.

If you are building a development program

• Plan for the finish, not the start. The retention risk sits at completion. Decide before anyone enrols what changes in role, scope or pay when they finish, because the graduates who got nothing were the ones who left.
• Expect retention during, not after. Both the trial and the tuition data show the effect concentrated while the program runs. If the goal is keeping people beyond it, the program is the on-ramp, not the mechanism.
• Measure against people you did not develop. Participants are usually stayers to begin with. Compare against similar hires who did not enrol, or the same roles before the program existed.
• Treat the stated reason as a lead, not a diagnosis. When leavers say growth, find out what stopped: the manager, the promotion cycle, the scope of the job. The retention diagnostic is built to separate those.

Read the companion grade on whether mentorship programs work, the intervention most often bundled with development, which has the same shape of evidence.