DP 0253 - Are Fiscal VAR’s Non-Fundamentalness Easily Reversible Through the Addition of Informative Variables?

The VAR/SVAR (Vector Autoregressive and Structural Vector Autoregressive) models are the cornerstone of the contemporaneous empirical macroeconomic research, in particular for being able to measure the impact of fiscal policy shocks. They may be employed as atheoretical models, as well as a mean to support the estimation and testing of DSGE (Dynamic Stochastic General Equilibrium) models – the main theoretical tool for modern macroeconomics. Nevertheless, VAR models may be subject to pathologies, such as the non-fundamentalness. It is capable of biasing the estimates in any direction or intensity, and it consists of the non-invertibility of the MA (Moving Average) representation on the positive powers of the lag operator. This is associated with the insufficiency of the econometrician’s data to estimate the model’s correct parameters or with model misspecification. This study is the first to employ the latest and most efficient tests for non-fundamentalness on fiscal data for the USA: the Forni and Gambetti’s (2014) and Canova and Sahneh (2018) tests. The data and model were found to be non-fundamental.

Limit theorems for locally stationary processes

We present limit theorems for locally stationary processes that have a one sided time-varying moving average representation. In particular, we prove a central limit theorem (CLT), a weak and a strong law of large numbers (WLLN, SLLN) and a law of the iterated logarithm (LIL) under mild assumptions using a time-varying Beveridge–Nelson decomposition.

Assessing the Relationship between Presidential Rhetorical Simplicity and Unilateral Action

Research from Shogan (2007) and Lim (2008) on the executive branch proposes that the American presidency has adopted an anti-intellectual approach to leadership, such that there is a concerted rejection of thoughtful political discourse from the president. This has been reflected by what appears to be a relative decline in both the linguistic and substantive complexity of presidential rhetoric. Shogan’s (2007) work, while focused on examining whether Republicans are more apt to employ anti-intellectual leadership than Democrats, raises an additional topic worthy of empirical examination: the potential relationship between anti-intellectual leadership and unilateral action from the president. If anti-intellectual leadership is a defiant form of leadership that opts to publicly demonstrate the rejection of external expertise, the usage of anti-intellectual rhetoric from the president might be able to predict the usage of unilateral action. On the other hand, anti-intellectual rhetoric might be used as a straightforward and quick means to explain unilateral action, such that change in the level of unilateral action can predict the usage of simplistic rhetoric. Unfortunately, no one has yet to empirically test whether rhetorical simplicity predicts unilateral action, unilateral action predicts rhetorical simplicity, or there is a multi-directional relationship present. This project makes an initial attempt to remedy this gap in the literature. The project contrasts the monthly average simplicity level of the presidential weekly public address with the monthly number of executive orders emanating from the executive branch, using information spanning between February 1993 and May 2015. The initial findings from the vector autoregression and moving average representation analyses suggest that prior change in rhetorical simplicity predicts the usage of executive orders, and that an increase in rhetorical simplicity helps produce an increase in the number of executive orders offered by the president.

A UNIFYING APPROACH TO FRACTIONAL LÉVY PROCESSES

Starting from the moving average representation of fractional Brownian motion, there are two different approaches to constructing fractional Lévy processes in the literature. Applying L2-integration theory, one can keep the same moving average kernel and replace the driving Brownian motion by a pure jump Lévy process with finite second moments. Alternatively, in the framework of alpha-stable random measures, the Brownian motion is replaced by an alpha-stable Lévy process and the exponent in the kernel is reparametrized by H - 1/α. We now provide a unified approach taking kernels of the form [Formula: see text], where γ can be chosen according to the existing moments and the Blumenthal–Getoor index of the underlying Lévy process. These processes may exhibit both long and short range dependence. In addition we will examine further properties of the processes, e.g., regularity of the sample paths and the semimartingale property.

Measuring Tax Multipliers: The Narrative Method in Fiscal VARs

This paper argues in favor of empirical models built by including in fiscal VAR models structural shocks identified via the narrative method. We first show that “narrative” shocks are orthogonal to the relevant information set a fiscal VAR. We then derive impulse responses to these shocks. The use of narrative shocks does not require the inversion of the moving-average representation of a VAR for the identification of the relevant shocks. Therefore, within this framework, fiscal multipliers can be identified and estimated even when, in the presence of “fiscal foresight,” the MA representation of the VARs is not invertible. (JEL C32, E62, H20, H62, H63)

A REPRESENTATION THEORY FOR POLYNOMIAL COFRACTIONALITY IN VECTOR AUTOREGRESSIVE MODELS

We extend the representation theory of the autoregressive model in the fractional lag operator of Johansen (2008, Econometric Theory 24, 651–676). A recursive algorithm for the characterization of cofractional relations and the corresponding adjustment coefficients is given, and it is shown under which condition the solution of the model is fractional of order d and displays cofractional relations of order d − b and polynomial cofractional relations of order d − 2b,…, d − cb ≥ 0 for integer c; the cofractional relations and the corresponding moving average representation are characterized in terms of the autoregressive coefficients by the same algorithm. For c = 1 and c = 2 we find the results of Johansen (2008).