Labor Department’s disgraceful discrimination case against Oracle fails

Labor Department’s disgraceful discrimination case against Oracle fails

In 2017, on its way out the door, the Obama Labor Department filed an action against Oracle for alleged pay discrimination against women and minority group members. The case was based on a flawed — indeed an indefensible — statistical analysis. I explained why in this post.


The Trump administration should have dropped the case straight away. Instead, Secretary of Labor Alex Acosta pursued it. So did Gene Scalia after he became the Secretary in 2019. Both lacked the spine to buck liberal career employees determined to press forward with the Obama Labor Department’s parting gift to the left.


Yesterday, an administrative law judge ruled in favor of Oracle. The judge’s opinion, nearly 300 pages long, is here.


The essence of the ruling comes in this statement, which is similar to my critique of the DOL’s case:



The statistical evidence does not support an inference that Oracle is engaged in the alleged intentional compensation discrimination. Dr. Madden’s analysis [she was the DOL’s expert on statistics] is highly aggregated and not attuned to potentially important differences between groups within job functions.


Dr. Madden’s analysis does not similarly situate employees with respect to the work performed. The jobs at issue require particular skills, experience, and expertise. They involve work on particular products or types of product. Organization matters for compensation and for making relevant comparisons between employees.


Dr. Madden’s measures of experience and education are very rough estimates and poorly capture the sort of education and experience that matters for compensation at Oracle.


Dr. Madden’s analysis relies largely on assumption about aggregation and the view that it is unnecessary to control for variances between employees at a group level, but this assumes away the important question about potential explanations for the raw disparities and thus undermines the inferential power of the model.



This is basic stuff. Even a cursory review of the file should have persuaded Acosta and Scalia that the DOL had no case. Maybe it did persuade them, but they failed to act accordingly.


When the government has no case, it has a duty not to proceed. The government’s duty is to seek justice. It is never to try to nail a private party by prosecuting a meritless case.


Yet, that’s what the Trump Department of Labor did here. Oracle was required to spend millions of dollars to defend itself from a ludicrous case concocted by left-wing ideologues. The DOL, for its part, expended a huge amount taxpayer dollars pushing a laughable theory of discrimination. The ALJ’s opinion lists 13 DOL attorneys as appearing in the case, which gives some idea, though not a complete picture, of the resources the government poured into the matter.


The ALJ’s opinion should finally put this litigation to rest. It’s sad that it took three and a half years, and a judge, not the Trump administration, to drive a stake through the case.


Technically, the case isn’t over. The Department could appeal the ALJ’s decision to its Administrative Review Board. That’s the body that Alex Acosta wouldn’t fill with fresh appointees, allowing Obama-era holdovers to remain in place.


It’s full now with members appointed in 2019 and 2020. I consider it highly unlikely that the board would reverse the ALJ’s sound decision.


But it shouldn’t come to that. The ALJ’s decision should clear the way for the Trump administration finally to let go of this awful case. If the Department appeals, there should be an investigation.


Here is the bottom line on the DOL’s action against Oracle for $400 million: It netted zero dollars recovered, flushed millions of taxpayer dollars down the toilet, cost Oracle millions of dollars to defend, injured Oracle’s image, and damaged the Labor Department’s reputation and integrity.


On the plus side, it satisfied career lawyers in the Labor Department Solicitor’s office, at least for a few years.