The Bond Market for Dummies

The government wants money. One of the ways it can get money is by borrowing it. One of the ways it can borrow money is through bonds, which, when the government issues them, are sometimes called treasury bills, sometimes called treasury notes, and sometimes called treasury bonds. When the government issues a bond, it basically says: “if you buy this bond for $100, then in X years (printed on the bond) I will give you back $100 + Y” – where Y is some extra percentage. The government offers many of its bonds at auction, where Y is not directly set by the government, but they offer to pay the lowest Y that bidders at the auction are willing to buy them for.

In Siggy's OP, he observes that the "lowest Y" at the most recent government bond auction rose considerably. According to CNBC, yields rose on every single bond offering of the US government, with the highest rise being in the US 20 year bond, which rose from 4.605% to 4.752%. That's a rise of 0.148% in one day, which is well within normal range for something like bitcoin, but not for US bonds. They are supposed to be very steady.

A rise in yields such as this is usually treated by the media as an indicator that something went wrong in the bond markets. But whenever I read headlines about yields rising and how that's a bad indicator, they seem confusing to me at first because rising yields sound positive: when something's price rises, that usually means lots of people want that thing, signaling confidence.

But remember, the government auctions treasury bills for the lowest price bidders offer. When their price rises like it did today, that means no one wanted these treasury bills at a lower return, i.e. no one bid to accept a lower yield than 4.752%. i.e. "I won't take your terrible bonds unless, when they mature, you pay me almost 5% more than I'm giving you now." Not wanting the government’s bonds might mean inflation expectations are high. I.e. buyers might think the dollar will be worth less when the bond matures than it’s worth now, perhaps by an amount close enough to the treasury bond’s yield percentage (e.g. 4.752%), so they want more than that to break even.

Importantly, every time the government sells a bond, it takes out a loan, which means they owe more than they got. Where do they get the extra money they need to pay back the loan? They have four options:

1. get it from taxes
2. print money
3. take out more loans
4. just don't pay off some old loans

Usually they use option 3, but these loans keep getting more and more expensive, adding to the federal deficit, so that every year it gets harder and harder to keep doing that. In theory they can keep taking out more and more loans forever and keep paying them off, but that only works if people keep buying their bonds. If buyers will only do so while simultaneously asking for higher and higher yields (due to higher and higher inflation expectations), the government will end up paying ridiculous amounts of money to bond buyers, indicating that #2 is occurring. Eventually this goes the way of #1, #2, or #4, none of which are good: high taxes, high inflation, and theft are all very bad.

This criticism of Robin's bridge model seems to incorrectly state that verifiers can burn user funds in step 8. That is not my understanding of Robin's model, I believe that if the prover does not process a withdrawal, the verifiers can burn some funds that the prover himself put up as a kind of stake to dissuade dishonest behavior, but user funds do not get burned, they simply move to a new m-of-m multisig controlled by the remaining verifiers.

One of them becomes the new prover and must process the withdrawals now otherwise they lose their own stake and the cycle repeats until one of two things happens: either some verifier processed the withdrawals properly when it was their turn, or all user funds become the sole property of the last verifier, locked to a 1-of-1 where he or she is the only remaining keyholder.

At that point, this final verifier can simply send the users their money without hindrance. But if he or she is dishonest, users are out of luck, the last verifier can run off with all user funds if no previous verifier could or would perform the withdrawals honestly. (This is largely why you must always trust at least one verifier to be honest, in Robin's model.)

Also worth pointing point: my unisob bridge model is unaffected by this criticism. The prover sources the funds to perform a withdrawal when the user first deposits money to the bridge, and every depositor gets to withdraw up to the amount in their own individual contract with the prover. If the prover is unable to source sufficient funds, the would-be depositor simply aborts and never makes their deposit.