You'd think so, but banks aren't "all or nothing" businesses. It's not like you can either get a loan for any amount you want, or you can't borrow at all. There's a bunch of intricacies that make is much more interesting.
Say you lend $200 to a guy who also has a house worth $100, you're the only creditor. In the event of a default, you can take that $100 house. That means you only lose $100, not he $200 you're actually out. This is called the Loss Given Default, or LGD. That number encapsulates the you first point, and half of point two.
The other half of point two, and the entirety of point three is covered in what we call the Probability of Default, or PD. The chance that a given debtor is going to default.
I hope you can see how these two number interact, especially when you then also take into account the upside of giving the loan. Providing loans might entice a large counterparty to do more business with you, or it might provide you with access to a new network. It might make sense to make a risky loan if the downside is very small, or conversely it might not make sense to make a pretty safe loan, if the downside is huge. In practice you can multiply these two numbers together to get an "expected cost" of proving the loan.
Now when someone comes and asks you why you aren't going to lend him $300, you then have to be able to trace all that data back to the source.
Say you lend $200 to a guy who also has a house worth $100, you're the only creditor. In the event of a default, you can take that $100 house. That means you only lose $100, not he $200 you're actually out. This is called the Loss Given Default, or LGD. That number encapsulates the you first point, and half of point two.
The other half of point two, and the entirety of point three is covered in what we call the Probability of Default, or PD. The chance that a given debtor is going to default.
I hope you can see how these two number interact, especially when you then also take into account the upside of giving the loan. Providing loans might entice a large counterparty to do more business with you, or it might provide you with access to a new network. It might make sense to make a risky loan if the downside is very small, or conversely it might not make sense to make a pretty safe loan, if the downside is huge. In practice you can multiply these two numbers together to get an "expected cost" of proving the loan.
Now when someone comes and asks you why you aren't going to lend him $300, you then have to be able to trace all that data back to the source.