Use of currency to pay taxes = use in transactions. It is because I expect that in future someone will give me real goods in exchange for the currency that it has value to me now. (In the case you refer to, the real goods are avoiding whatever consequences non payment of tax has. So, yes, "coercion" if you want.)

But the government is not the only body who might be prepared to exchange real goods for currency for you in future. In fact, it's readily easy to write down a model in which a fiat currency such as bitcoin has a value because everyone believes that everyone else will accept it for payment in future. The problem with these models is that they always have at least two equilibria. One in which people coordinate on believing that everyone else will accept the currency for real goods, and so they too are happy to accept the currency for real goods, and another in which no one believes this. Both are fully rational. (They are sub-game perfect Nash equilibria of the dynamic game.)

Macroeconomists have been interested in the role of nominal tax payments not because they're necessarily the fundamental source of a currency's value, but because they provide a way of ruling out the bad equilibrium of the model, in much the same way that gold's use as jewellery ruled out the bad equilibrium in societies in which gold was used as currency. It is interesting to note that there's nothing ruling out the bad equilibrium for bitcoin. However, this is not the case for e.g. namecoin, for which the (apparently) fixed price of a domain name provides a lower bound on value. The bad equilibrium should also be ruled out for coins such as primecoin for which the mining process is productive, since if I value prime chains sufficiently highly I should be prepared to accept primecoins for real goods in order to incentivise others to mine primecoins.