@SamHLevey Not true in the Sims model.

@mean_field_zane Link?

@SamHLevey Sims invented FTPL.

@mean_field_zane I know that, but I'm wondering if you have a specific model in mind here, because all the versions of FTPL that I can recall reading have made this same mistake, IMO.

@SamHLevey What’s the mistake?

@mean_field_zane The basic premise is that for a non-"passive" fiscal policy, the price level adjusts today to make the real value of the debt equal to expected future surpluses. The mistake is in treating future surpluses as a policy variable exogenous to private actions. Specifically:

@mean_field_zane if private actors can refuse to sell goods to the gov (which of course in real life they can), then that makes private actors the "last movers" in determining future gov surpluses in most circumstances. So, instead of the price level adjusting, expected future surpluses adjust.

@SamHLevey If you’re talking about a Ramsey problem here, this is studied.

@mean_field_zane No, it's simpler than that. The gov's surplus is always modeled as a discretionary policy variable, but it's not, because private actors have the power to increase the gov's surplus, by refusing to sell to the gov. For most contexts this does not matter, but for FTPL it does.

@SamHLevey What do you mean by “refuse to sell to the government”? The government collects taxes. That can’t be refused.

@mean_field_zane I'm referring to government purchases. The government purchases goods and services from suppliers who always have the option not to sell to the government if they don't want to.

@SamHLevey You don’t need to have government spending in this model at all.

@mean_field_zane Yes, that's true, and in that case, I'd be inclined to agree with the model. But of course in the real world we do have government spending.

@SamHLevey The idea is just to show how the failure of Ricardian equivalence matters. You could fairly simply add a government that employs workers and purchases some of the final good for a price as well I think if you wanted the path of real surpluses to be endogenized.

@mean_field_zane I *think* that what I'm saying is still true regardless of whether Ricardian equivalence holds or not.

@SamHLevey Sure I just don’t see the importance tbh. The important channel is the anticipation of the tax and the HtM households it doesn’t affect.

@mean_field_zane The significance is that FTPL says private actor choices cause today's price level reacts to exogenous future surpluses. If future surpluses aren't exogenous to those private choices, then the model fails - it cannot determine the price level.

@SamHLevey I don’t see how the supply of goods to the government matters on the margin, or why the government isn’t able to buy at market price.

@mean_field_zane Ok let's do an example. Consider a 1-period model, that begins with $100 in private hands. The gov buys some goods, let's call them hammers, and these hammers arbitrarily begin with a price of $20. The gov intends to buy 3 hammers, and collect 5 hammers worth of taxes, leaving a

@mean_field_zane real surplus of 2 hammers. With a real surplus of 2 hammers and $100 that have to be collected before the game ends, FTPL says that the price level must be $50 per hammer. If the hammers only cost $20, then there would be 60 worthless dollars left in private hands at the end of

@mean_field_zane the game. Anticipating this, private actors bid up the price of hammers to $50 before the gov runs the surplus. That's standard FTPL. What I'm saying is, private actors have another choice: they can just not sell hammers to the gov. If they choose not to do this, then the gov

@mean_field_zane will buy 0 hammers but still collect 5 hammers worth of taxes, for a surplus of 5 hammers. At the original arbitrary price of $20, gov will collect all $100 in taxes, so private actors have no need to bid up prices. So, no price level has been determined. FTPL has no content.

@mean_field_zane To summarize more succinctly: both stories are about people not wanting money, but the standard FTPL story forgets that people have the option to not accept more money that they don't want.

@SamHLevey You need to write this in the standard Sims model for me to accept it.

@mean_field_zane Here's a quick but extremely verbose model. Actually most of the first 2/3rds are ultimately worthless, the key is just the usual intertemporal budget equation, but I just wanted to prove that there's no tricks up my sleeve here.

@SamHLevey You should do this in infinite time and then endogenize g, as well as write out a commitment/expectation/belief structure. Then p is a function of agents expectation of real surpluses that will actually be levied which is a story I am willing to believe.

@mean_field_zane I'm sure you could do that, I just don't see why it matters. As I tried to show above, the basic logic of the FTPL story goes wrong before you even get to that stuff. It shows up in even the simplest possible versions of FTPL, which is why I like demoing it that way.

@SamHLevey I just do not see how this is any different than regular FTPL because either agents can anticipate or have an expectation of g OR g being below is an unanticipated shock which we already model. Like either the goods are sold to the government or they’re not. Then the extra stuff

@mean_field_zane I'm sorry, I didn't understand what you are saying here, maybe rephrase it?

@mean_field_zane As short/clear as I can get it: FTPL is entirely based on pointing to 1 equation (the IGBC) and saying that there's only 1 free variable (P), so it's determinate. I'm saying there's 1 equation (the IGBC), and 2 free variables (g and P), so it is not determinate.

@SamHLevey I’m willing to buy this only in a richer model. If the government declares it wants gbar but agents only sell it g, then g is a fixed variable as before unless you put p somehow into the decision for what determines g. If you just exogenously declare that g is g, then gbar

@mean_field_zane My intuition here is that g is not a function of gbar at all except in edge cases. For example, if gbar is already not binding in a period, then increasing it further does absolutely nothing. The edge case is that private actors do need enough money to pay their taxes in every

@SamHLevey Do you think the government has no role in determining how much it’s able to purchase? Stop using it as a budget constraint (because it’s not a budget it’s a preference) and start realizing it’s a bliss point.

@mean_field_zane Haha in the model or in real life? In real life of course the gov determines this. In the model, if you endogenize gov spending, then the amount they spend is determined by taxes + intertemporal preferences, except when you collide with gbar. That's how my other model works.

@SamHLevey What about intertemporal preferences for how much the government spends or buys? You can’t just have the government blindly try to buy some gbar and have agents only care about how much they sell to it because that is essentially saying G in the real world is only a function of