BE103/512

THE UNIVERSITY OF HONG KONG

FACULTY OF BUSINESS AND ECONOMICS

MECON [6089] [Modern Econometrics for Business Strategy]1

GENERAL INFORMATION

Instructor: Dean He
Email: deanhe@hku.hk
Office: 8/F, The Jockey Club Tower, Centennial Campus
Phone: +852-5398-6187
Consultation times: Mondays 5pm to 6pm or by appointment at Cyberport Campus

TF: Karen Mai

Email: maixt@hku.hk

Course Website: HKU Moodle

COURSE DESCRIPTION

This course introduces the science and art of building and using econometric models in the analysis of business
strategy. It aims to equip business and economics students to understand and appreciate econometric analysis in
economic and business strategy analysis, and to provide students with an understanding of three widely used
techniques in modern econometrics: randomized control trials, regression discontinuity, and differences-in-differences.

Data analysis in economics, or “econometrics” as it is called by practitioners, has moved away from mathematical
complexity and towards simpler tools that are accessible to businesses and can be applied easily to big data. After
learning how these tools provide superior analytic results than traditional regression techniques in making inferences
about the real world, students will gain the practical knowledge to wield them successfully and make better business
strategy decisions with data.

COURSE OBJECTIVES

1. Prepare students become skeptical about claims of cause & effect

2. Prepare students realize when there are flaws in others’ causal claims and communicate reasons to be doubtful
3. Prepare students understand why experimentation is the best method for solving the causal-inference problem in
business strategy making and why randomization is key to experimentation
4. Prepare students understand when “natural” experiments are a good substitute
5. Prepare students be able to describe examples of successful experiments, why we feel confident in the results
and the limits to what we learn from any experiment
6. Prepare students appreciate how to design data systems and websites to make experimentation easy
7. Prepare students understand statistical concepts more intuitively
8. Prepare students understand how to quantify uncertainty, using confidence intervals and statistical power
calculations and be able to analyze experimental data & interpret results
9. Prepare students able to design, pilot, and pitch your own experimental analysis
10. Prepare students appreciate the common things that can go wrong with an experimental analysis and what to do
about it: a) Non-compliance: implementation is incomplete; b) Placebo: just being in an experiment affects
outcomes; c) Attrition: outcomes can’t be measured for all subjects
11. Prepare students understand how to find natural experiments and the most reliable methods for analyzing them

1
This course is adapted from a course taught by Professor Jonah Rockoff at Columbia Business School in 2019/2020. This is to acknowledge
and thank Professor Rockoff for inspiring me and allowing me to gain access to his course materials and case studies.
1
PROGRAMME LEARNING OUTCOMES

PLO1. Understanding of fundamental theories and new development in economics

PLO2. Mastering of skills in analyzing economic data
PLO3. Demonstration of ability to apply economic knowledge and analytical skills to address policy and business problems
PLO4. Awareness of ethical concerns in economic issues
PLO5. Mastering of communication skills

COURSE LEARNING OUTCOMES

Aligned Programme Learning
Course Learning Outcomes
Outcomes

CLO1 Students understand and appreciate econometric analysis in economic and PLO 1 and PLP 2
business strategy analysis, and to provide students with an understanding of three
widely used techniques in modern econometrics: randomized control trials,
regression discontinuity, and differences-in-differences.

CLO2 Students know how to use the business econometric tools to provide superior PLO 3 and PLO 4
analytic results than traditional regression techniques in making inferences about the
real world

CLO3 students gain the practical knowledge to wield them successfully through PLO 5
effective business communication, in order to make better business strategy
decisions with data.

COURSE TEACHING AND LEARNING ACTIVITIES

Expected contact Study Load
Course Teaching and Learning Activities
hour (% of study)

T&L1. Lectures every week

36 36%
T&L2. Problem Sets (2 of them) 36 36%

T&L3. Consultation and Self Study 12 13%

15 15%
T&L4 Group Project Presentation

Total 100%

Aligned Course Learning

Assessment Methods
A1. Class Participation
Brief Description (Optional) Weight
Outcomes
In-class discussions are an integral part
20% CLO 1
of the course, and students are expected
to contribute to the learning experience
of the class by asking relevant
questions, offering insights into the topic
at hand, and generally behaving in a
professional manner. Quality of
contribution matters more than quantity.
Class participation scores will also
account for attendance, lateness, and
completion of pre-class and in-class
surveys

There will be TWO problem sets for the

course. the first one covering CLO 2 and CLO 3
experimental methods and the second 40%
A2. Problem Sets
one covering analysis of natural
experiments. They must be written up
and turned in individually. Students can
discuss concepts related to the
problems, but not their own individual
2
 submissions or solutions, with members
of their learning team and/or other
classmates.

Students will work in teams (of 4-5

A3. Group Project
people) to develop a causal question
related to a real business strategy,
design a randomized experiment to test
their question, run a pilot study to inform
their design, and “pitch” a proposed 40% CLO 2 and CLO 3
design and implementation strategy
(directed to the key decision maker in
the company) during the final exam
period. Each student must submit an
experimental idea (10% of project grade)
which is graded individually. Then each
group must meet with course instructors
to present an initial design plan and a
revised version (10% of the project
grade), pitch the idea at a full-class
session during Final Exam period (20%
of the project grade) and submit a
written version of this pitch (60% of the
project grade); these elements are
graded at the group level.
TOTAL 100%

STANDARDS FOR ASSESSMENT

Course Grade Descriptors

A+, A, A- Strong evidence of superb ability to fulfill the intended learning outcomes of the course at all
levels of learning: describe, apply, evaluate and synthesis.

Strong evidence of ability to fulfill the intended learning outcomes of the course at all levels of
B+, B, B- learning: describe, apply, evaluate and synthesis.

Evidence of adequate ability to fulfill the intended learning outcomes of the course at low levels
C+, C, C- of learning, such as describe and apply, but not at high levels of learning such as evaluate and
synthesis.

D+, D
Evidence of basic familiarity with the subject.

F Little evidence of basic familiarity with the subject

Assessment Rubrics for Each Assessment:

The rubrics will be discussed with the class in specific details with the dissemination of the problem sets .

3
 COURSE CONTENT AND TEACHING SCHEDULE (as of 17/03/2023)

Session Date Time Topic Required Readings

Section 1: Experimental Design

Class #1 • Introduction 1. "Is Your Digital Advertising Campaign
20/03 6:15pm to 9:45pm • Average Treatment Effects & Working?"
Selection Bias 2. "Why Businesses Don't Experiment"
3. "Do We Really Know What Makes Us
Healthy?"
4. MM pp 3-15; RRE pp 57-58
Class #2 • Precision and Power 1. Testing for Statistical Significance
23/03 6:15pm to 9:45pm (Managerial Stats)
• Randomization and Balance 2. MM pp. 4-5,16,19-22
3. Distribution of Problem Set 1 (work in
team of 2)
Section 2: Econometric Analysis
Class #3 1. HBS case: Advertising Experiments at
• Regression with Covariates Restaurant Grades
27/03 6:15pm to 9:45pm 2. FE pp. 102-105
3. MM pp. 215-217

Class #4 1. FE pp. 296-298

30/03 6:15pm to 9:45pm • Heterogeneous Effects 2. Background on Bangladesh SC Experiment
3. Discussion of Group Project Idea with the
course instructor
Class #5 03/04 6:15pm to 9:45pm • Non-Compliance 1. MM pp. 116 – 121
2. Problem Set 1 DUE before class
3. Distribution of Problem Set 2 (work in
team of 2)
Class #6 06/04 6:15pm to 9:45pm • Placebo and Hawthorn Effects 1. FE pp. 161-163
2. "Virgin Atlantic Tested 3 Ways to Change
Employee Behavior"
Class #7 13/04 6:15pm to 9:45pm • Sample Attrition 1. "Does Working From Home Work?" (pp 176
- 183)
Section 3: Natural Experiments
1. MM pp. 150-152
Class #8 17/04 6:15pm to 9:45pm • Regression Discontinuity (RD) 2. "Using Big Data to Estimate Consumer
and its application: Uber Surplus: The Case of Uber" pp. 6-10
3. Problem Set 2 DUE before class
4. Return of Problem Set 1
Class #9 20/04 6:15pm to 9:45pm • Difference-in-Difference (DID) 1. Bargaining at Ebay
and its application: Bargaining 2. Group Project Draft Due (OPTIONAL)
at Ebay
Class #10 24/04 6:15pm to 9:45pm • Project Presentation 1. Group Project DUE

REQUIRED/RECOMMENDED READINGS & ONLINE MATERIALS (e.g. journals, textbooks, website addresses etc.)

There is no required textbook, but readings from sources including textbooks will be posted on Moodle or distributed
by email and some readings are required. Case studies are available via Canvas.

For those who would like to have a textbook on hand, excellent books covering much of the course material are:

• Gerber and Green’s “Field Experiments: Design, Analysis, and Interpretation”

• Angrist and Pischke’s “Mastering ‘Metrics.”
• Angrist and Pischke’s “Mostly Harmless Econometrics” (also great book but is more formal/mathematical).

These can be relied on for questions about classwork, homework, and your final project.

4
MEANS/PROCESSES FOR STUDENT FEEDBACK ON COURSE

o conducting mid-term survey in additional to SFTL around the end of the semester

o Online response via Moodle site

✓ Others: ___SFTL____________ (please specify)

COURSE POLICY (e.g. plagiarism, academic honesty, attendance, etc.)

Attendance: while attendance is not counted towards the final grading, it is taken into consideration for final grading.

The course expects you to be:

Present: Your success depends on being on time and present for the entire class every session. Attendance is an
integral of class participation.

Prepared: Bring your nameplate, clicker, and complete any pre-work needed for class discussion. Expect the
professor to cold call-in class.

Participating: Active participation calls for no electronic devices such as laptops, smartphones, or tablet computers
during class, except if you are told to use them as part of in-class work.

ADDITIONAL COURSE INFORMATION (e.g., e-learning platforms & materials, penalty for late assignments, etc.)

Course Roadmap (tentative)

Experimental Design Econometric Analysis Natural Experiment

Classes 1 to 2 Classes 2 to 7 Classes 8 to 10

• Introduction • Regression w/ Covariates

• Average Treatment Effects and • Heterogeneous Effects • Regression Discontinuity
Selection Bias • Non-Compliance • Differences-in-Differences
• Precision and Power • Placebo/Hawthorne Effects
• Randomization and Balance • Sample Attrition