Accounting is a complex subject, I do recommend pursuing some kind of formal training if the subject is of interest to at least get the basics of common scenarios, like ammortization, accrual.

The issue of liquidity I am not sure how it plays out, but I'm assuming there would be many systems by which it can be consistently accounted for, maybe there is an actual loss of value that is tallied up as a part of the transaction cost, (of which there are many other subcomponents), but it is expected to be profitable in the long term, so the loss in value by liquidity reduction can be ammortized over time maybe?

I also think there's legitimate reasons by which internal accounting methods can differ from accounting methods and books used for tax purposes. Of course the most common difference between such books is tax evasion and fraud, but one can imagine an entity that pays tax according to the tax books, but holds some accounting methodology (even if implicit) by which it values assets differently, and this allows it to find underpriced assets to buy (or overpriced assets to sell), otherwise we would all value assets identically and a whole class of incentive to engage in trade would dissapear.

Whether the case of building a business model around welfare systems is in general fraud, or whether it depends on the details and if so does this case meet those elements, is a complex question that I'm definitely not trained to answer, but to my layman perception, it seems like something that, if not already fraudulent, I'd bet that a discovery process would find accessory fraudulent elements, and even if it doesn't, and everything is legal, it sounds one of those very problems that congress was designed to crack down on.