Code for "A User's Guide for Inference in Models Defined by Moment Inequalities"

This code is not yet ready for public use.

This repository contains the code for the paper "A User's Guide for Inference in Models Defined by Moment Inequalities" by Canay, Illanes, and Velez available here.

Structure

The code is organized into four folders:

• data contains fake data intended to be similar to the data used in the empirical application in the paper. The data is stored in several csv files. The file data/README.md contains a description of the data.
• matlab contains the code for the Matlab implementation of the algorithms. The file matlab/README.md contains a description of the Matlab code.
• python contains the code for the Python implementation of the algorithms. The file python/README.md contains a description of the Python code.
• r contains the code for the R implementation of the algorithms. The file r/README.md contains a description of the R code.

In each of the three code implementations, there is one script for each table in the paper. This script produces the output for the table. The scripts are named table_1a.m (or .py or .R), table_1b.m, etc. The scripts are self-contained and can be run independently of each other.

Outputs

Each implementation produces outputs in a folder named _results. The results folder is not included in the repository. The results folder is created when the code is run. The results folder contains a language-specific output data file as well as a _results/tables-tex folder containing the tables from the paper. Each table in the paper is a separate file in the tables-tex folder. The tables are named table_1.tex, table_2.tex, etc. The tables are in LaTeX format and can be included in a LaTeX document. You can check the output of the code by comparing the tables in the results folder to the tables in the paper as well as in the Tables section of this README.

Tables

Matlab tables

Table 1

Panel A

	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
$\bar{V}$=500	self-norm	[-16.0 , 23.4]	[-40.0 , 39.3]	45.9
	bootstrap	[-13.9 , 22.4]	[-40.0 , 38.5]	180.0
$\bar{V}$=1000	self-norm	[-40.0 , 29.1]	[-40.0 , 63.1]	54.4
	bootstrap	[-40.0 , 26.8]	[-40.0 , 60.2]	216.3

Panel B

	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
$\bar{V}$=500	self-norm	[-14.3 , 22.6]	[-40.0 , 35.9]	5.1
	bootstrap	[-13.1 , 22.1]	[-40.0 , 34.8]	13.1
$\bar{V}$=1000	self-norm	[-40.0 , 28.3]	[-40.0 , 57.4]	4.2
	bootstrap	[-40.0 , 26.6]	[-40.0 , 54.7]	12.8

Table 2

	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
$\bar{V}$=500	self-norm	[-21.0 , 17.1]	[-40.0 , 39.1]	15.1
	bootstrap	[-16.0 , 14.6]	[-40.0 , 36.6]	41.5
$\bar{V}$=1000	self-norm	[-40.0 , 17.0]	[-40.0 , 62.8]	14.0
	bootstrap	[-40.0 , 13.9]	[-40.0 , 55.1]	41.5

Table 3

Test Stat.	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
CCK	self-norm	$14.2^{\dagger}$	[-40.0 , 12.8]	25.0
RC-CCK	self-norm	[-35.4 , 44.0]	[-40.0 , 13.8]	36.8
RC-CCK	bootstrap	[-35.6 , 43.3]	[-40.0 , 13.0]	42.7
RC-CCK	SPUR1	[-39.2 , 53.2]	[-40.0 , 18.4]	54.4

Table 4

	parameter	linear	quadratic	linear	quadratic
	$\theta_{1,1}$	[ -22.2 , 43.7]	[ -22.4 , 76.7]	[ -40.0 , 49.6]	[ -40.0 , 82.0]
Coca	$\theta_{1,2}$	[ -20.0 , 50.0]	[ -20.0 , 50.0]	[ -20.0 , 50.0]	[ -20.0 , 50.0]
Cola	$\theta_{1,3}$	[ 0.0 , 0.0]	[ -10.0 , 10.0]	[ 0.0 , 0.0]	[ -10.0 , 10.0]
	$\theta_1(\mu)$	[ -79.9 , 133.7]	[ -167.8 , 157.5]	[ -100.0 , 134.4]	[ -190.0 , 195.3]
Energy	$\theta_{2,1}$	[ -40.0 , 53.6]	[ -40.0 , 67.6]	[ -40.0 , 78.2]	[ -40.0 , 91.6]
Brands	$\theta_{2,2}$	[ -20.0 , 50.0]	[ -20.0 , 50.0]	[ -20.0 , 50.0]	[ -20.0 , 50.0]
	$\theta_{2,3}$	[ 0.0 , 0.0]	[ -10.0 , 10.0]	[ 0.0 , 0.0]	[ -10.0 , 10.0]
	$\theta_2(\mu)$	[ -75.1 , 99.0]	[ -105.8 , 119.9]	[ -75.1 , 126.0]	[ -105.8 , 142.7]
Comp. time		11.0	11.9	8.6	8.7

R tables

Table 1

Panel A

$\bar{V}$	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
500	SN2S	[-16.0, 23.0]	[-40.0, 39.0]	6.18
500	EB2S	[-12.0, 22.0]	[-40.0, 38.0]	586.89
1000	SN2S	[-40.0, 29.0]	[-40.0, 63.0]	5.95
1000	EB2S	[-40.0, 26.0]	[-40.0, 60.0]	517.84

Panel B

$\bar{V}$	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
500	SN2S	[-14.3, 22.6]	[-40.0, 35.9]	0.99
500	EB2S	[-11.9, 21.7]	[-40.0, 34.6]	39.88
1000	SN2S	[-40.0, 28.3]	[-40.0, 57.4]	0.94
1000	EB2S	[-40.0, 26.8]	[-40.0, 54.1]	39.13

Table 2

$\bar{V}$	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
500	SN2S	[-21.0, 17.1]	[-40.0, 39.1]	1.75
500	EB2S	[-15.3, 13.9]	[-40.0, 36.1]	167.82
1000	SN2S	[-40.0, 17.0]	[-40.0, 62.8]	1.62
1000	EB2S	[-40.0, 13.1]	[-40.0, 54.9]	161.40

Table 4

	Parameter	Linear	Quadratic	Linear	Quadratic
Coca-Cola	$\theta_{1,1}$	[-22.2, 43.7]	[-21.9, 76.7]	[-40.0, 49.6]	[-40.0, 82.0]
	$\theta_{1,2}$	[-20.0, 50.0]	[-20.0, 50.0]	[-20.0, 50.0]	[-20.0, 50.0]
	$\theta_{1,3}$	[0.0, 0.0]	[-10.0, 10.0]	[0.0, 0.0]	[-10.0, 10.0]
	$\theta_{1}$	[-18.7, -16.3]	[-17.8, 8.6]	[-40.0, 2.3]	[-40.0, 14.2]
Energy Brands	$\theta_{2,1}$	[-40.0, 53.6]	[-40.0, 67.6]	[-40.0, 78.2]	[-40.0, 91.6]
	$\theta_{2,2}$	[-20.0, 50.0]	[-20.0, 50.0]	[-20.0, 50.0]	[-20.0, 50.0]
	$\theta_{2,3}$	[0.0, 0.0]	[-10.0, 10.0]	[0.0, 0.0]	[-10.0, 10.0]
	$\theta_{2}$	[0.0, 0.0]	[0.0, 0.0]	[0.0, 0.0]	[0.0, 0.0]
Comp. Time		1.20	1.37	1.35	1.20

Python tables

Table 1

Panel A

$\bar{V}$	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
500	SN2S	[-16.0, 23.0]	[-40.0, 39.0]	2.346
500	EB2S	[-15.0, 22.0]	[-40.0, 39.0]	416.250
1000	SN2S	[-40.0, 29.0]	[-40.0, 63.0]	2.081
1000	EB2S	[-40.0, 27.0]	[-40.0, 61.0]	419.781

Panel B

$\bar{V}$	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
500	SN2S	[-14.3, 22.6]	[-40.0, 35.9]	1.045
500	EB2S	[-13.7, 22.3]	[-40.0, 34.5]	30.341
1000	SN2S	[-40.0, 28.3]	[-40.0, 57.4]	0.769
1000	EB2S	[-40.0, 27.4]	[-40.0, 54.1]	30.521

Table 2

$\bar{V}$	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
500	SN2S	[-21.0, 17.1]	[-40.0, 39.1]	1.374
500	EB2S	[-17.3, 15.3]	[-40.0, 36.5]	121.747
1000	SN2S	[-40.0, 17.0]	[-40.0, 62.8]	1.148
1000	EB2S	[-40.0, 13.8]	[-40.0, 54.8]	120.689

Table 3

Test Stat.	Crit. Value	$\theta_1$: Coca-Cola	$\theta_2$: Energy Brands	Comp. Time
CCK	SN2S	[nan, 14.2]	[-40.0, 12.8]	0.987
RC-CCK	SN2S	[-35.4, 44.0]	[-40.0, 13.8]	1.193
RC-CCK	EB2S	[-36.5, 43.4]	[-40.0, 12.6]	31.191
RC-CCK	SPUR1	[-40.0, 54.5]	[-40.0, 18.3]	183.142

Table 4

	Parameter	Linear	Quadratic	Linear	Quadratic
Coca-Cola	$\theta_{1,1}$	[-22.2, 43.7]	[-22.4, 76.7]	[-40.0, 49.6]	[-40.0, 82.0]
	$\theta_{1,2}$	[-20.0, 50.0]	[-20.0, 50.0]	[-20.0, 50.0]	[-20.0, 50.0]
	$\theta_{1,3}$	[0.0, 0.0]	[-10.0, 10.0]	[0.0, 0.0]	[-10.0, 10.0]
	$\theta_{1}$	[-18.7, -16.3]	[-17.8, 8.6]	[-40.0, 2.3]	[-40.0, 14.2]
Energy Brands	$\theta_{2,1}$	[-40.0, 53.6]	[-40.0, 67.6]	[-40.0, 78.2]	[-40.0, 91.6]
	$\theta_{2,2}$	[-20.0, 50.0]	[-20.0, 50.0]	[-20.0, 50.0]	[-20.0, 50.0]
	$\theta_{2,3}$	[0.0, 0.0]	[-10.0, 10.0]	[0.0, 0.0]	[-10.0, 10.0]
	$\theta_{2}$	[0.0, 0.0]	[0.0, 0.0]	[0.0, 0.0]	[0.0, 0.0]
Comp. Time		0.518	0.832	0.707	0.736