Every token is free to define its own economic model… the most important characteristic of which is a supply. Unless you define tokens within the native currency of a blockchain, you can’t know the total amount of economic volume & activity that is within your particular business model, as opposed to that blockchain as a whole.
Without tokens, transactions and smart contracts will only see the amount of the blockchain’s native asset, without being able to see the purpose to which any asset may have been dedicated. When a user converts a native asset into a token, they are dedicating it to that purpose. If your business model doesn’t have or doesn’t require such a purpose, then you have no need to create a token.
An extreme example would be tokens that represent discrete ownership: like 1 token per object in the “real world” (or maybe online, but off the blockchain) with a limited supply. In an even more extreme example you might have only 1 of something, with its own unique token to represent that thing.
In either case these discrete objects could be bought, sold, and their ownership reassigned by smart contracts as well as ledger transactions using their representative token. Without that token capability, the smart contract (only having the native asset to measure) wouldn’t know about that ownership. You might get around the ownership issue by requiring “proofs” of off-blockchain ownership records but then you would be defeating the purpose of doing these transactions on a blockchain that supports tokens which can already handle the ownership proof for you.
Now generalise that idea of discrete ownership into a pool of assets representing the economy of your own business or means of commerce. All those tokens will be indistinguishable from each other, and will be unique from the native asset, and they will entitle the owner to participate in the system according to 1) the smart contracts that operate on your token, 2) the ledger for that token itself (supply & ownership), and 3) whatever smart contract creates the asset (in the simplest case, determining its buy/sell price in the native asset).
Without that token being unique from the native asset, you cannot manage value and most DeFi applications go out the window. For instance, to create a “stablecoin” as you have no doubt heard of, you would have to create and destroy tokens so that your tokens’ total value will roughly match the amount of collateral value that’s confirmed deposited into your financial application. How would a smart contract do that if the ada held by such a smart contract were no different than the 45 billion other ada out there?
Regarding the malicious uses of tokens, a truly decentralised system has to leave these issues to general ethics as well as the governing laws of the jurisdiction in which you operate. By default that means every blockchain user themselves has total responsibility for which tokens they buy and which smart contracts they participate in.
It’s sad that there’s no Wikipedia article on this subject but I’ve noticed that few writers yet are committing to these ideas: I spent a while looking for a “Why create a token?” article on Ethereum / ERC-20, the most popular token platform, and found nothing at all. Your question is a good one & I’ll bet other forum readers would be happy to supply more real world examples & maybe more educational links. 