The purpose of this structure is to render SC liabilities more attractive as a payment instrument. Pegging to the USD is attractive to people living in the U.S. because the USD is the unit of account. Non-U.S. holders may be attracted to the product because the USD is the world's reserve currency. This structure serves to increase the demand for a SC pegged to the USD.
To a macroeconomist, an SC looks like a unilateral fixed exchange rate regime or a currency board. The structure also resembles a money market fund that pegs the price of its liabilities with the USD at par (presently, government money funds in the United States). It also looks like a bank without deposit insurance (bank deposit liabilities are pegged at par value against cash).
The history of unilateral fixed exchange rate regimes is mixed. Hong Kong has successfully pegged its currency to the USD for decades. But the experience for many countries seems closer to that of Argentina. Unless a USD-based SC is backed fully by USD reserves (it needs an account at the Fed for this) or by USD bills (maximum denomination is $100, so unlikely), it may be prone to a bank run. Any other security (including USTs, as the events of March 2020 demonstrated) is subject to liquidity risk -- i.e., a risk that the market for the security suddenly freezes, or demand for the security vanishes as investors seek safer havens. If a SC cannot dispose of its assets at "fair" or "normal" prices, it will fail to raise the money it needs to meet its par redemption promise. The SC will turn out to be not so stable.
The theory of bank runs suggests that SCs might be rendered run-proof if their liabilities are properly designed. The famous Diamond and Dybvig (JPE 1983) model of bank runs is, in fact, a paper that demonstrates how banks can be rendered run-proof. In the first part of their paper, the explain how a credible promise to suspend redemptions when redemption activity is abnormally high can serve to discourage runs (redemptions based on belief of failure, rather than a need for liquidity) altogether. There is no need for deposit insurance (the second part of their paper is devoted to explaining why deposit insurance may nevertheless be needed, but their argument is not entirely satisfactory).
In reality, we do see attempts to render run-prone structures less prone to runs. The Dodd-Frank Act, for example, prevented institutional money funds from pricing their liabilities at par with the USD (only government funds can now do this). In addition, the Act required that fund managers implement liquidity fees and redemption gates in the event of heavy redemption activity. These provisions have not been entirely successful. Even banks that suspended redemptions in the old days did not manage to prevent mass redemption events. The theory suggests that what is needed is a *credible* policy. Evidently, when push comes to shove, people cannot always be expected to follow through on their promises.
Back in my teaching days, I used a "crowded movie theatre" as a metaphor to explain the phenomenon. Imagine a movie theatre that seats 500 people. If someone was to yell "fire!" (for legitimate or illegitimate reasons), people can be expected to rush for the exits. Invariably, some people are likely to be trampled and even killed. If people would instead react to the alarm by rising calmly from their seats and proceeding sequentially to the exits, then only the very last few people in the queue are destined not to make it. Losing one or two people relative to (say) is terrible, but it's preferable to losing 50 people in a mad rush for the exits.
An economist might detect a missing market here. Why not sell tickets with queue positions (in the event of fire)? The tickets with the last few queue positions are likely to sell at a discount (that would depend on the likelihood of the event). This way, if someone yells "fire!", customers will simply show their assigned queue position to the ushers and proceed calmly out of the exits. If the fire does exist, the few people know they are doomed and accept their fate with stoic resignation, knowing that they are dying so that many others may live. (And if it turns out there is no fire, they are saved from the fire and the prospect of death that would have been present had they joined the rush to the exits).
Except we know that's not what is likely to happen. People cannot be expected to commit in this manner. And there's no obvious way to enforce such contractual stipulations.
But this is where SCs may have an advantage over conventional institutional structures. In particular, their use of "smart contracts" means that commitment is not an issue. The terms of such contracts are executed under the specified contingencies whether you like it or not. You may not like it ex post, but such commitment can be valuable ex ante. In the context of SCs, the credible threat of suspending redemptions in the event of abnormal redemption activity may actually prevent any runs from occurring in the first place.
There are, of course, limits to what smart contracts can achieve. They wouldn't, for example, solve the movie theatre problem I just described. This is because people do not live "on-chain." (See also my blog post Smart Contracts and Asset Tokenization.) To some extent, the same issue exists for USD SCs because USDs and USTs exist "off-chain." Nevertheless, money accounts are different than people, so I think the principle described above can apply to financial products.
Related work: Preventing Bank Runs (w/ Nosal and Sultanum, TE 2017).