Dividends are a way for a company to tell investors “We don’t think we can use this money to develop our future earnings better than you would do it at other places so we are returning it to you”.
Companies that don’t distribute dividends do so on the assumption that using that money to further their growth is the best use of investors’ money at that time. Some are right to think so, some will never be profitable and are just burning money quicker the more of it they get.
Ideally, an educated investor should be able to trim the bad weed out and, in that model, reinvesting dividends globally would mean taking money from the companies who don’t think they can turn enough of a profit out of them toward companies who anticipate high growth, thus it would be an efficient use of that money.
Finding mostly good stocks, or weeding out the bad ones, takes a formidable amount of time, education and skill, so it’s mostly out of reach of individual investors. In the index fund version of this, we would trust market participants to identify the good and bad companies for us and adjust their price accordingly and we would tag along for the ride.
In a 100% efficient market, reinvesting dividends in the broad index should yield the most efficient outcome. Companies that distribute dividends should be properly valued to account for the future cashflows they are likely to generate, and companies that don’t should also be properly valued to account for their potential future returns, or lack thereof.
Assuming some level of inefficiencies, it probably isn’t too far a stretch to assume reinvesting dividends broadly compounds the small missevaluations that can happen in the market. This probably would affect smaller cap companies more than the others, since they’re probably harder to value (their prospects are either small - meaning few investors would take the time to value them properly - or further into the future - meaning investors would need to make a lot of assumptions to value them properly).
If we believe in the efficient market hypothesis enough to invest in broadly diversified index funds, then it wouldn’t seem too much of a strech for me to trust it some more and reinvest the dividends in those same funds, according to our allocation. If we don’t believe in it and provided we are individual investors with limited time, resources and knowledge, we should either hire a wealth manager (go for actively managed funds) who’s got the time, resources and skills for it, or adapt our allocation to minimize the inefficiencies we would anticipate. Focusing on large/mid caps instead of total market would be a way to do it in this situation, in my opinion. Then again, small caps have only a small effect on the global fund so the distortions coming from them could be considered minor enough for us to deal with them.
That being said, I’m not a professional fund/wealth manager/economist like some Bogleheads are, and just a guy on the internet. Make of this what you want.