It looks like you're new here. If you want to get involved, click one of these buttons!
Imagine for a moment a world in which it was impossible to translate games into other languages. Every game that is made could be made in exactly one language, and that's it. Some gamers understand multiple languages pretty well, and some don't--but playing a game in a language that they don't understand at all wouldn't be very viable.
What languages do you think online games would be made in? There would still be a lot of games made in English, because it's a language that a lot of online gamers speak. There would still be a lot in Korean, and a lot in Chinese. Without the possibility of translation, there would perhaps be fewer of each, but there would still be a lot of them.
But how many games do you think would be made in Dutch? How about Swedish? Swahili? Tagalog? Languages that aren't spoken by that many people would see few games made in them, and no really big budget games. If you spend $100 million to make the greatest game ever, but it's only in Basque, even if everyone who speaks Basque (under 1 million worldwide) and plays online games (probably a small fraction of the total who speak the language) loves your game, you still lose money.
So what does this have to do with sandboxes? Make a list of some theme park games. Let's say WoW, LotRO, Champions Online, EverQuest II, Vanguard, Guild Wars, Runes of Magic, and throw some more games on the list of you like. Whatever games you stick on your theme park list will probably have a lot in common. There are important differences that could mean you like one game and not another, but there are a lot of similarities.
Now make a list of some sandbox games. Let's say EVE, Darkfall, Fallen Earth, Uncharted Waters Online, A Tale in the Desert, Puzzle Pirates, Xyson, and throw whatever other games on the list that you want. (But please don't be one of those silly people who wants to put Vanguard and EQ2 on lists of sandboxes.) Now what do they have in common? A lot less than the list of theme parks did. We range from free for all full loot PVP to no combat at all. The list of what a sandbox is "supposed" to have is far less standardized than for theme parks.
So why does this matter? A single theme park game can appeal to a large fraction of the players who like theme parks. This isn't just hypothetical, either; WoW actually does this in the real world. There's no reason why another theme park couldn't do so.
But with sandboxes, that's not the case. If you make a really good sandbox game, then no matter what design decisions you put into it, a large fraction of sandbox fans will refuse to play it. If it has free for all full loot pvp, then a lot of sandbox fans will refuse to play it on that basis. If it doesn't have PVP, then a lot of sandbox fans will refuse to pay it because of that. If it has a heavy emphasis on the economy, more so than on combat even, then a lot of sandbox fans won't play that. If the economy is only a minor side thing, as in most theme parks, then a lot of sandbox fans won't play that. No matter what design decisions you make, if you have a sandbox MMORPG that is the best game ever, probably a majority of sandbox fans won't even consider it. That's not the case with theme parks.
So a sandbox game has a much lower revenue ceiling than a theme park game. If you want to make good money on the game, you'd better spend a lot less to create the game than what you can plausibly make by selling it. So you end up with smaller budget niche games, just like in the hypothetical case of small budget games in obscure languages.