Dynamic Simulation and Testing for Single-Equation Cointegrating and Stationary Autoregressive Distributed Lag Models

Abstract

While autoregressive distributed lag models allow for extremely flexible dynamics, interpreting the substantive significance of complex lag structures remains difficult. In this paper we discuss dynamac (dynamic autoregressive and cointegrating models), an R package designed to assist users in estimating, dynamically simulating, and plotting the results of a variety of autoregressive distributed lag models. It also contains a number of post-estimation diagnostics, including a test for cointegration for when researchers are estimating the error-correction variant of the autoregressive distributed lag model.

Exploring meaningful visual effects and quantities of interest from dynamic models through dynamac

“…dynamac (2020) implements a framework of estimating and interpreting autoregressive distributed lag (ARDL) models in R. dynamac uses stochastic simulation techniques (Jordan & Philips, 2018a, 2018b to easily recover traditional quantities of interest, such as short-and long-run effects, even from complex dynamic specifications, including models with cointegration. These simulation techniques bring a wider set of inferences from complex models to users in fields as diverse as environmental science (Danish, 2020) and economics (Sharma, 2019).…”

Asymmetric and Symmetric Link between Quality of Institutions and Sectorial Foreign Direct Investment Inflow in India: A Fresh Insight Using Simulated Dynamic ARDL Approach

“…Dynamic Autoregressive Distributed Lag Simulation model (SARDL) is an advanced form of orthodox ARDL, developed by [47]. There are several advantages of SARDL over the simple ARDL approach: (i) SARDL is used to overcome the issues in simple ARDL estimator for estimation in the long and short run.…”

dLagM: An R package for distributed lag models and ARDL bounds testing

“…Since these CVs are based on large samples, Narayan [46] provides small-sample CVs for Cases II-V. ardlBound() function calls pssbounds() function from nardl package [23] to get the CVs. This function is originally included in pss package and currently available in dynamac package [26]. For Cases II-V, pssbounds() function produces small-sample CVs in the brackets of 5 observations from 30 to 80 and produces asymptotic CVs for the sample sizes greater than 80.…”

“…The package dynlm [25] has just one main function to fit dynamic linear models by preserving time series characteristics. A recently published R package, dynamac [26] is specifically designed to simulate the effect of some independent series on dependent by dynamic simulations and run the ARDL bounds test of Pesaran et al [12]. dynamac has some functions for fitting the ARDL models and plotting and simulating results.…”