@ Josep-Lluís Carrion-i-Silvestre and María Dolores Gadea
November, 2022 @
This README file tries to clarify the MATLAB codes used to elaborate the paper
Carrion-i-Silvestre, J.L. and Gadea, M.D. 2023. Testing for multiple level shifts with an integrated or
stationary noise component, forthcoming at the Journal of Applied Econometrics.
The main purpose is the detection and estimation of multiple level shifts regardless of the order of integration of the time series.
The codes replicate the two empirical applications presented in the online material of the paper.
-------------------------------------------------------------------------------------------------
The folder functions_empirical contains several auxiliary functions for the analysis.
We use two databases:
1.- The first one collects data for 17 countries from the end of XIXth century to present provided
by Jordà, Schularick and Taylor (2018) -hereafter, JST database.
2.-The second database that is used is the PennWorld Tables (PWT) extension by Feen-
stra, Inklaar and Timmer (2015) -hereafter, PWT database.
REMEMBER CHANGING THE PATH ACCORDINGLY
-------------------------------------------------------------------------------------------------
The codes should be run in the following order for each database:
1) JST database
1.0 The data are in excel format in Tab_cpi_updated.xls and Tab_xrusd_updated.xls for the consumer price index and exchange rate, respectively.
1.1.The code main_JST_Ftest_0.m computes the Ftest_0 with both the one-at-a-time break dates estimating strategy and the the joint break dates estimating strategy
1.2 The code main_JST_joint.m computes the number and position of structural breaks with the joint break dates estimating strategy; it also saves the results in the structure RDOS_joint.mat.
1.3 The code main_JST_one_at_time.m computes the number and position of structural breaks with the one-at-a-time break dates estimating strategy; it also saves the results in the structure RDOS_one_at_time.mat.
1.4 The code main_JST_ur.m tests the PPP and calculates persistence measures.
2) PWT database
2.0 The data are in pwt100.xlsx
2.1 The codes main_PWT_Ftest_0_cons.m, main_PWT_Ftest_0_da.m and main_PWT_Ftest_0_gdpo.m compute the Ftest_0 for each of the price definitions.
2.2 The codes main_PWT_cons_joint.m, main_PWT_da_joint.m and main_PWT_gdpo_joint.m compute the number and position of structural breaks with the joint break dates estimating strategy for each price definition; they also save the results in the structures RDOS_cons_joint, RDOS_da_joint and RDOS_gdpo_joint.mat.
2.3 The codes main_PWT_cons_one-at-a-time.m, main_PWT_da_one-at-a-time.m and main_PWT_gdpo_one-at-a-time.m compute the number and position of structural breaks with the one-at-a-time break dates estimating strategy for each price definition; they also save the results in the structures RDOS_cons_one-at-a-time.mat, RDOS_da_one-at-a-time.mat and RDOS_gdpo_one-at-a-time.mat.
2.4 The codes main_PWT_cons_ur.m, main_PWT_da_ur.m and main_PWT_gdpo_ur.m test the PPP and calculate persistence measures.
All the figures are saved in formats .pdf, .png, .eps.
The codes also generate the latex tables presented in the paper.
____________________________________________
If you use this programme for a paper, please cite the authors and the web page.
If you have any problem, please do not hesitate to contact to us.
Good luck,
Josep Lluís Carrion-i-Silvestre & Lola Gadea
PS: Remember, any code works at the first try.