Shadow education is the collective term for paid private tutoring, exam preparation, or private lessons outside regular school. Worldwide, it has grown into a multibillion-dollar industry. Parents invest enormous sums, often believing that extra lessons will automatically lead to better results. A new international study by Cao and Huo casts serious doubt on that assumption by examining the effectiveness.

The researchers analysed TIMSS 2019 data from more than 222,000 students across 38 countries. They examined whether students who received paid tutoring performed better in mathematics and science while controlling for a wide range of background characteristics, including socioeconomic status, parental education, and other relevant factors. Their conclusion is remarkably straightforward: overall, shadow education produces little, if any, learning gain.

After accounting for those background characteristics, students who received tutoring did not outperform comparable students who did not. The picture becomes even less encouraging. The association was generally slightly negative. That does not, of course, mean tutoring makes students worse. Many students receive tutoring precisely because they are already struggling. Yet even after the researchers applied various statistical techniques to compare similar groups of students, the overall pattern persisted.

You might think that students who receive tutoring over a longer period would eventually pull ahead. The data did not support that idea either. Students who had received tutoring for more than eight months still performed no better, on average, than students who had not received tutoring.

Does this mean shadow education is pointless? Probably not. There were exceptions. In a handful of countries, including South Korea, Japan, and to a lesser extent, Taiwan and Turkey, the researchers found small positive associations. Interestingly, these are countries where shadow education is highly organised and often closely aligned with the official curriculum and examinations. Even there, however, the effects remained relatively modest.

To me, this study primarily shows that shadow education is no miracle cure. I also suspect that the quality of tutoring varies enormously, as do the reasons why students seek it. Some are looking for an extra challenge, while others are simply trying to catch up after falling behind. The TIMSS data allow only limited differentiation between these groups. One tentative implication is that shadow education may contribute less to educational inequality than is often assumed, although this study was not designed to answer that question directly.

I do think the study has one important limitation. The data come from 2019, before the COVID-19 pandemic. Since then, the global market for shadow education has expanded rapidly, fuelled by lockdowns, learning loss, and the rise of online tutoring. At the same time, China dramatically reduced its tutoring sector through its well-known Double Reduction policy. Those developments make today’s landscape much more complex than the one captured in TIMSS 2019. The study, therefore, tells us little about the current size of the industry. It does, however, address the more important question: Does paid tutoring improve learning outcomes on average? Based on these data, the answer appears to be: not by very much.