The best traditional studies show that GPA accounts for about 3% of job performance, as measured by income and performance reviews among other things. (IIRC Alfie Kohn cites this study in his book What Does It Mean To Be Well Educated?) It will be interesting to see whether or not running the study backwards yields similar results.
There was an interesting study a few years back that looked at the "outcomes" of a cohort of people who had all been admitted to an Ivy League school. Part of the study group attended and graduated from the Ivy League school they were admitted to. The control group ended up matriculating elsewhere for whatever reason.
Past studies showed at a significant improvement in earning potential for Ivy League grads, but they didn't have a distinguished control group. What this recent study found was that for most students of similar qualification (all judged to be Ivy League material by the admissions boards at Ivy Leage schools), attending an Ivy conferred no obvious advantage compared to attending a non-Ivy.
The only difference was for students who came from more challenging backgrounds, those whose families may have flirted with poverty and/or didn't have significant educational attainment. In those cases, attending an Ivy conferred some advantage over going elsewhere.
There are many confounding factors: People who are more intelligent, have parent(s) who pursued higher education, have parents with higher income, or are white are both more likely to receive a bachelor's degree AND on average earn more than other people with the same amount of education. (In the past, being male was a predictor for both higher education and higher earnings, but this is not a significant factor for people born after 1980.)
In addition, the statistic ignores the fact that pursuing higher education shortens one's career -- people who spend 4-5 years in college on average work for 2-3 years less than those who started their careers immediately after high school. As a result, comparing "40-year careers" is bogus -- the only reasonable comparison is between the total careers of individuals, including the time when one person is at college while the other is working.
Finally, even if the statistic were computed correctly and reflected causation, it would be misleading: Different careers are not influenced equally by education. Generally speaking, people don't finish their education before they start thinking about what career the will pursue; someone who was planning on becoming a plumber isn't likely to suddenly decide that he wants to become a brain surgeon just because he finds that he has a bachelor's degree.
Decide what you want to do (either as a single career, or more likely thinking of a number of careers which seem interesting), and look at what qualifications are needed for those. And never forget what Benjamin Disraeli said: "There are three kinds of lies: lies, damned lies, and statistics".