I have two types of households who consume different goods including durables. We often say that non-ricardian can not optimize inter temporally because they do not have access to financial markets. But if they consume, and accumulate, durable goods, the household choose in period t the stock of durables for the period t+1. Isn’t that a kind of inter temporal optimization for the non ricardian ? Plus, the ricardian can always sell its durable good at an ulterior date.
Yes, buying durables provide some consumption smoothing over time. How strong that actually is depends on your utility function. Usually durables are not perfect substitutes for non-durables, so being able to partially smooth utility via durables is not a perfect substitute for access to financial markets.
Thank you @jpfeifer . I have a utility function with an unitary elasticity between non durables and services from durables. SO they are imperfect substitutes.So that means that the consumption smoothing thanks to durables is not that strong?
Usually, yes. But it’s hard to judge how strong the effect will be.
Thank you @jpfeifer. I have another question regarding lump-sum transfers to households.
I have two types of households: a ricardian one who accumulates K and gets dividend from a firm, and a non-ricardian households. Both receive lump sum taxes. In the model, I assume Tr_ricardian = Tr_nonricardian. Transfers depend on the budget constraint of my government who raises taxes and spends. Tr is calculated in a way that It allows the budget constraint of the government to hold. My issue is that when dividends and revenues from capital taxation increases for the ricardian household, it mechanically raises the Transfers to the ricardian AND non-ricardian household by the same amount.
Is that increase of Tr considered as a positive income shock for both agents ?
Yes, it is a positive income shock for both agents. That’s why the transfer rules are crucial for the results in most settings.
Thank you @jpfeifer. Would you have any paper recommandations considering transfer rules? I’m in a perfect foresight model
Is there any way to have a rule for the government which does not apply such a strong hypothesis ? Because I’m afraid that transfers as a way to balance government’s budget is making the results a little bit wild.
Unfortunately, no. In older papers like Gali/Lopez-Salido/Valles (2007) “Understanding the effects of government spending on consumption” the role of transfers is actually hidden. A similar issue pops up in newer HANK models where you have to distribute firm profits.
Can it be a good idea to do a robstuness check with government spendings G doing the adjustment and not Tr ? So Ill have a model with a redistributive effect and the other one no redistributive effect?
Thank you @jpfeifer
There are various ways to check robustness. You can give transfers to just one of the two agents, you can use expenditures instead of a rebate or simply throw the resources away as a deadweight loss.
Thank you @jpfeifer. How do I throw the ressources away ? I mean I do I formalise that in the model ? Thank you
The tax revenues are simply not rebated to anyone. Essentially, the government never receives the tax revenue.
Thank you @jpfeifer. If the government never receives the tax revenues, how do I close the model? This DWL is going to appear in the ressource constraint of the economy, isn’t it ?
Plus, in the case of a perfect foresight model with anticipated shocks, do you agree that income changes are viewed as permanent ?
And in the case of PF model with unanticipated shocks at each date, do agents view the income changes as transitory?
I’m following the extended path for my (deterministic) unanticipated shocks. So agents think that after t shocks are 0. Even in that case the shock is viewed as permanent ?
Only if you specify them to be. You as the model builder are setting the information structure and the persistence of shocks.
But how do you specify the persistence in case of a deterministic shock ? My first guess is that they are actually viewed as persistent because the agents know they are jumping to a new SS at (t+1).
And regarding the deadweight loss, if G does not appear in the utility function and I assume that G is doing the adjustments, isn’t this considered as a deadweight loss as it does not profit to households ?