They could theoretically keep their $$$$ in reserves, but that really wouldn't do anything. If the dumped their Treasury holdings (dollar deposits in securities accounts), which is ultimately what the doomsday crowd goes on and on about, the only way to shed those dollar holdings would be to buy real goods and services from us. We'd get the $$$$$, they would get the medical technology, software, clothing, pharmaceuticals, automobiles, etc. Do you not see the increase in Chinese demand for our real goods and services resulting in dollar appreciation?
At the end of the day, when all is said and done, all we owe China is bank statement from the FED. This is what people don't understand, it can never become a financial strain in any capacity, unless credits and debits between reserve accounts and securities accounts entails more than accounting entries.
In order to solve our broader macro problem, in my opinion, the right fiscal policy will result in us having enough spending power to buy employment output domestically and real goods and services from the foreign sector. We should structure it so we can reach full employment, increase our standard of living, and increases output irrespective of the any type of trade gap. If we can do this, and do it as actual public policy, what China does would then become essentially inconsequential in a broader, international aspect with regards to the US economy on the global stage.
There is a well-known principle in linear programming that a LP problem is solvable only if the number of independent variables (policy tools) is equal to or more than the number of variables in the objective function (policy goals). So if we have more goals than tools there is a pretty good chance the solution set will be empty. In practice, this dovetails with the fact that many of the equations used in macroeconomics are accounting identities, not functions. So if you try to explain everything, you have an overdetermined system.
For example, the Monetary Identity is MV=PY. If we have a theory of what determines M, P, and Y (i.e. an exogenous money supply, a theory of prices or inflation, and a theory of output), there is no need for an independent theory to explain V. If you use a theory to explain monetary velocity functionally, you have to drop one of the other theories. People are forever trying to posit a theory for each of the four and then are surprised that their system is overdetermined.
So some things just have to come out in the wash. The current balance in trade theory is one of those things that usually has to come out in the wash. It's the price we pay for having floating exchange rates. If we concentrate on domestic employment, production, price stability and growth, with variable exchange rates and comparatively free trade, there is no room for a separate theory of trade balances. Unless of course you prefer to give up a production function (theory of growth), theory of prices, or impose new trade restrictions.
The same reasoning explains why the federal deficit is another remainder term. There is no policy that can target the deficit without major impact on prices, employment, and growth. You get to pick which you want to achieve, but you can't choose them all.