“Language Testing” just published an article about using the TOEFL and IELTS tests to predict academic success at university. The authors’ meta-analysis of 32 studies indicates that there is a weak positive correlation with academic achievement and no significant difference between the predictive power of each test.
Note the following conclusion, though:
Although significant, the overall correlation was low; thus, practitioners are cautioned from using standardized English-language proficiency test scores in isolation in lieu of a holistic application review during the admissions process.
Perhaps the test creators would agree that the tests should be used only to assess a test-taker’s ability to use academic English, and not be used to predict their academic success. Indeed, there are other tests which can be used for the latter purpose.