Variables selection

Estimation ML/Bayesian estimation
1

I have a DSGE model, and I am trying to find the right variables for estimation, since my topic is innovative, and I have no idea what they should be. So I am looking to find the right variables from Fred’s data for US data for the following: More, I would kindly like to help me understand which ones should be. Most of them are so-called commonly used, but I did not understand which ones they should be or their definition.

I have in my model Technology Shocks.
These shocks represent changes in productivity or technological advancements.

Financial Shocks
These shocks represent disruptions in financial markets or changes in financial conditions. They are modeled as changes in the cost of borrowing.

Financial frictions
Common definition

Supply Chain Shocks: These shocks capture disruptions in the supply chain that affect production and prices. They are modeled as changes in production costs:

Fiscal Policy Shocks
These shocks capture changes in government spending (G_t) or taxation policies (T_t).

Also, I have capital, and alpha is the output elasticity of labor.

Total factor productivity

My question is: which variables are commonly used for these? Mostly for technology shocks, financial, shocks, financial frictions, and supply chain shocks as well as with the remaining ones

thank you for any help you could provide

2

In that case, you would probably use the standard series like GDP, investment, consumption, and government spending. Given the financial frictions, you may want to use things like spreads or credit aggregates. That depends on the particular model. In any case, it makes sense to see what other people have done.

1 Like
3

Thank you so much for the reply. My big doubt here is which variable should be used to check for technology shocks. My dsge model has to do with policy implications and social aspect assumptions. I’d need the classical variables, I guess. But which one will fit better for technology shocks?

Thank you!

4

When working with US data, you can rely on Fernald’s TFP series. However, most studies identify technology shocks from the joint dynamics of national account and hours worked data. There is usually no 1:1 mapping from shocks to particular data series.

1 Like