estimate_mrr_mod: Estimates the model of Madhavan, Richardson & Roomans (1997)...
In simonsays1980/tim: Estimation of Trade Indicator Models

Description Usage Arguments Details Value References

View source: R/MRR.R

Estimates the model of Madhavan, Richardson & Roomans (1997) using the approach by Theissen & Zehnder (2015)

1 2	estimate_mrr_mod(price_diff, price, indicator, indicator_lag, indicator_lag2, midquote, midquote_lag)

`price_diff`	a numeric vector containing the series of first price differences.
`price`	a numeric vector containing the price series.
`indicator`	an integer vector containing the trade direction with a buy as 1 and a sell as -1.
`indicator_lag`	an integer vector containing the first lag of the trade indicator series.
`indicator_lag2`	an integer vector containing the second lag of the trade indicator series.
`midquote`	a numeric vector with the midquote price series.
`midquote_lag`	a numeric vector containing the first lag of the midquote series.

The function estimates for given data the trade indicator model of Madhavan, Richardson & Roomans (1997) using the modified estimation approach by Theissen & Zehnder (2015). The latter authors use a three-step estimation approach to overcome the negative bias in the estimation of the effective spread. The major part of this bias appears to come from the adverse selection component θ in the trade indicator model of Madhavan, Richardson & Roomans (MRR).

The authors propose a three-step estimation procedure that results in an unbiased estimation of both effective spread and also the adverse selection component as they explicitly account for endogeneity in their model, namely a correlation between the trade innovation and the public information arrival. The estimation procedure involves a VARMA part that is estimated using the dse-package.

Standard deviations for the VARMA model (step 3) are estimated using the delta method.

A data.frame with the following values:

n: the number of observation used in estimation.
rho: the estimated first order autocorrelation, ρ, of the trade indicator series, i.e. the parameter of the AR(1) model in step 1.
rho_std: the standard deviation of the estimated first order autocorrelation, ρ of step 1.
a1: the first parameter estimate of the OLS model in step 2, α_1.
a1_std: the standard deviation of the first parameter estimate of the OLS model in step 2, α_1.
a2: the second parameter estimate of the OLS model in step 2, α_2.
a2_std: the standard deviation of the second parameter estimate of the OLS model in step 2, α_2.
psi1: the parameter (1,2) of the VARMA model in step 3.
psi1_std: the standard deviation of the parameter (1,2) of the VARMA model in step 3.
psi2: the parameter (2,2) of the VARMA model in step 3.
psi2_std: the standard deviation of the parameter (2,2) of the VARMA model in step 3.
varv: the variance of the trade innovation, ν from step 1.
varu: the variance of the public information arrival, u from step 2.
covuv: the covariance of the trade innovation, ν from step 1, and the public information arrival, u from step 2.
spread_eff_emp: the empirical effective spread measured from the data.
spread_eff_est1: the estimation of the effective spread using the traditional estimation approach.
spread_eff_est2: the estimation of the effective spread using the modified estimation approach.
stable: the stability of the estimated polynomials, i.e. if all roots are inside the unit circle.

Madhavan, Richardson & Roomans (1997), "Why Do Security Prices Change? A Transaction-Level Analysis of NYSE Stocks," Review of Financial Studies, Vol. 10, No. 4, pp. 1035-1064.

Zehnder & Theissen (2015), "Estimation of Trading Costs: Trade Indicator Models Revisited," unpublished.

simonsays1980/tim documentation built on July 19, 2019, 7:35 a.m.