ExternalInstrument

ExternalInstrument <: AbstractIdentificationMethod

Identification method using external instruments for structural shock identification in time series models.

External instruments serve as proxies for structural shocks and are not included in the estimation model (e.g., a VAR), but are used in instrumental variable estimators such as two-stage least squares (2SLS).

This method identifies relative structural impulse response functions (IRFs), meaning that the scale of the identified shock is normalised. Specifically, the IRF is scaled such that the response of the treatment variable equals one at a given normalising horizon.

This approach also generalises to the Local Projections Instrumental Variables (LP-IV) framework and supports the use of multiple instruments.

The method is based on:

Stock, J. H., & Watson, M. W. (2018). Identification and Estimation of Dynamic Causal Effects in Macroeconomics Using External Instruments. The Economic Journal, 128(610), 917–948. https://doi.org/10.1111/ecoj.12593

Fields

  • treatment::Union{Symbol, Int}: the variable used to normalise the IRF
  • instruments::AbstractMatrix: matrix of external instruments
  • normalising_horizon::Int: the horizon at which the IRF of the treatment variable is set to one
ExternalInstrument(treatment::Union{Symbol, Int},
                   instruments::Union{AbstractMatrix, DataFrame};
                   normalising_horizon::Int=0)

Constructs an ExternalInstrument identification method using a specified treatment variable, one or more external instruments, and an optional normalisation horizon (default is 0).

Arguments

  • treatment::Union{Symbol, Int}: treatment variable in the model data specified either as the column index in the data or by variable name.
  • instruments::Union{AbstractMatrix, DataFrame}: observations of external instruments. Instruments must be observed over the same period as the model data.

Keyword Arguments

  • normalising_horizon::Int: horizon of the treatment variable with respect to which all IRFs are normalised.