There are no compound interest benefits to learning to read early. Children who learn to read at 5 and at 7 have indistinguishable reading skills at 11. That's why reading instruction in Finland and Germany begins in grade school, not kindergarten.
I regret I've lost the study (observational, not experimental) on students in North Rhine Westphalia on long term effects of early reading instruction but it showed children who learned to read later doing better at 16 or 17 than the ones who learned to read earlier. I think it was published in an economic of education journal but that's all I remember.
Edit:
The below study shows tiny differences between performance/school readiness of French and German children on entering primary school despite the French system being much more school like and placing a much greater emphasis on reading. Italics in abstract not in original.
French nursery schools and German kindergartens:
effects of individual and contextual variables
on early learning
Eur J Psychol Educ (2011) 26:199–213
DOI 10.1007/s10212-010-0043-4
The present article investigates the effects of individual and contextual variables on children’s early learning in French nursery schools and German kindergartens. Our study
of 552 children at preschools in France (299 children from French nursery schools) and Germany (253 children from German kindergartens) measured skills that facilitate the
learning of reading, writing and arithmetic at primary school. We also evaluated educational family practices and parents’ expectations of their children’s pre-school education. In order to take into account the hierarchical structure of our data, multilevel models were used in the analysis, which was carried out using MLwiN software version 2.02 (Rasbash et al.2005). Although French nursery schools emphasise academic learning, we did not find any significant differences in overall performances between the French and German samples. However, significant differences were obtained for some subscale results. In addition, our results indicate that individual and contextual variables have an impact on the differences observed between children from the two countries.
There aren't enough data points to claim it's a universal pattern, but still, there do seem to be a lot of data points that suggest that cramming stuff into a child very early has all evened out by the time they are an adult.
It's easy to forget, but 0-5 is, in terms of these advanced skills like "reading" (as opposed to "understanding what my body feels like"), not even 1/4th of a childhood. There's no way to stick a concrete number on it, but it's probably less than 10%.
We have a lot of evidence that depriving a child of the basics of life can affect them all their life, but I am not sure I've ever seen a study that has ever supported the idea that it is beneficial to accelerate education at the toddler phase. It's like trying to win a marathon by running really fast in the first 300 meters; unless you are going to be able to maintain that pace for the entire rest of the marathon, you probably haven't accomplished much.
"A sample of 250 kids isn't that accurate... And anecdotally..."
A sample of 250 could be wrong but at least it's a study, and so much better than an anecdote.
It's also much better than the original article, which says, as justification for teaching children to read early, "I will wait for the results of empirical studies before insisting on definitive claims, but my guesses are as follows.... Again, I know this is just speculation, but it seems very reasonable to me, and in the absence of better evidence, I feel justified in acting on what seems very reasonable."
So, to sum up: the best evidence we have says this does not work.
I am unaware of any country that makes a significant effort to tailor instruction to very high ability students in any systematic way. This certainly doesn't occur during primary school age anywhere in the Anglosphere.
I'm also not bullish on the positive effects of primary school on cognitive skills full stop. Unschooled children are only a year behind their counterparts in public schools on average. That's a really small effect for the difference between more than five hours a day of purposeful instruction versus none whatsoever. The same study does show children in structured homeschool doing one to five grade levels better depending on what's tested so there almost certainly are students who benefit from earlier more advanced reading instruction but the implications for mass schooling are unclear. If we can't even have children in different classes by ability level instead of age group I don't hold out much hope for sane educational policy.
The Impact of Schooling on Academic Achievement:
Evidence From Homeschooled and Traditionally Schooled Students
To start the "compound interest" benefits as early as possible?