# Kaiji Motegi's Website

Tenure-Track Associate Professor at the Graduate School of Economics, Kobe University
Doctor of Philosophy in Economics (University of North Carolina at Chapel Hill)

## Matlab Codes

### Time Series Analysis

1. MFVAR Toolbox: This zip file contains m-files for estimating Ghysels' (2012) mixed frequency vector autoregression and implementing related analysis like impulse response functions, forecast error variance decomposition, and Ghysels, Hill, and Motegi's (2016, JoE) mixed frequency Granger causality tests. MFVAR_example.m is the main m-file illustrating all these tools via an artificial trivariate example.

2. Max-Causality Toolbox: This zip file contains all Matlab codes for "Testing a Large Set of Zero Restrictions in Regression Models, with an Application to Mixed Frequency Granger Causality" (with E. Ghysels and J. B. Hill). They serve as the toolbox of max tests for mixed frequency Granger causality. They cover local power analysis, Monte Carlo simulations, and empirical applications. The main codes for simulations generate latex tables of rejection frequencies automatically.

3. Max-Correlation Toolbox: This zip file contains all Matlab codes for the simulation section of "A Max-Correlation White Noise Test for Weakly Dependent Time Series" (with J. B. Hill). They serve as a toolbox of white noise tests (in particular the max-correlation test). The main codes generate latex tables of rejection frequencies automatically.

### Fuzzy Cluster Analysis

1. opt_level: Find optimal levels of a partition tree based on fuzzy decision with max, power mean, and arithmetic mean.

2. maxminTM: Compute max-min transition matrix of (asymmetric) fuzzy graph F.

3. connectivity: Compute the degree of weak connectivity, unilateral connectivity, and strong connectivity of a fuzzy graph F. We also judge whether each individual is a weekening point, neutral point, or strengthening point.
This code requires another function maxminTM as a sub-routine.

4. MIDAS_Rsq: Compute a similarity matrix containing (adusted) R^2. Both mixed data sampling (MIDAS) regression and usual single-frequency regression can be used. This is a main function file for my working paper "Fuzzy Cluster Analysis with Mixed Frequency Data". 