Skip to contents

Prono (2014) Heteroskedasticity-Based Identification with GARCH

Source: R/prono.R
generate_prono_data.Rd

This file implements the Prono (2014) procedure for using conditional heteroskedasticity (GARCH) to generate instruments for identification in triangular systems with endogenous regressors.

Usage

generate_prono_data(
  n = .hetid_const("N_SMALL"),
  beta1 = c(0.05, 0.01),
  beta2 = c(0.097, -0.005),
  gamma1 = 1,
  k = 1,
  garch_params = list(omega = .hetid_const("GARCH$OMEGA_DEFAULT"), alpha =
    .hetid_const("GARCH$ALPHA_DEFAULT"), beta = .hetid_const("GARCH$BETA_DEFAULT")),
  sigma1 = 1.5,
  rho = 0.3,
  seed = NULL
)

Arguments

n

Sample size

beta1

Coefficient vector for X in first equation (portfolio return equation). Default c(0.05, 0.01) gives realistic portfolio returns in percent.

beta2

Coefficient vector for X in second equation (market return equation). Default c(0.097, -0.005) gives mean market excess return of 0.097% matching Prono.

gamma1

Coefficient on Y2 in first equation (the "beta" in asset pricing)

k

Number of exogenous variables (excluding constant)

garch_params

List with GARCH parameters: omega, alpha, beta. Default values give realistic volatility clustering for weekly returns.

sigma1

Standard deviation of epsilon1 in percent (portfolio idiosyncratic risk)

rho

Correlation between epsilon1 and epsilon2 (endogeneity)

seed

Random seed

Value

Data frame with generated variables (Y1 and Y2 are in percent)

References

Prono, T. (2014). "The Role of Conditional Heteroskedasticity in Identifying and Estimating Linear Triangular Systems, with Applications to Asset Pricing Models That Include a Mismeasured Factor." Journal of Applied Econometrics, 29(5), 800-824. Generate time series data for Prono's triangular model

Generates data matching Prono (2014) asset pricing application with returns in percent (like the original paper).

Prono, T. (2014). The Role of Conditional Heteroskedasticity in Identifying and Estimating Linear Triangular Systems, with Applications to Asset Pricing Models That Include a Mismeasured Factor. Journal of Applied Econometrics, 29(5), 800-824. doi:10.1002/jae.2387

See also

run_single_prono_simulation for running a single simulation create_prono_config for default configuration run_prono_monte_carlo for Monte Carlo analysis