2. Road Map for Prices and Markets: Tools
Demand and Supply Analysis:
Demand fundamentals; Supply fundamentals; Market equilibrium
Gains from Trade: Consumer surplus, Producer surplus
Competitive markets are Pareto efficient
Taxes as a source of inefficiency; Many reasons for taxes
Today: Demand and Revenue
Phenomenon/Example
Data to estimate demand functions; Elasticity of demand
Information content of estimates
Costs
Costs determined by production decisions and optimal input mix
Different kinds of costs and cost curves
Costs to ignore and costs to include
4. Coca Cola: b. 1886 in Atlanta GA.
Brad’s drink: b. 1893 in a New Bern NC.
• quickly rechristened Pepsi-Cola.
1931: Coke dominant; Pepsi near bankruptcy.
Pepsi president Guth buys 12oz recycled beer bottles (doubles volume)
• Tries to sell at 10c, twice the price of 6oz Coke (doubles price). Fails!
1933. Brilliant idea: Why not sell 12oz Pepsi @ 5c = price of 6oz Coke
• Jingle: Twice as Much for a Nickel Too
Sales explode! Pepsi comes out of bankruptcy.
• Profits: $2m in 1936; $4m in 1938.
“... brilliant marketing ploy ... saved the company” Really?
Cola Wars Episode I: The Pepsi Menace?
5. A 100% increase in price leads to a 40% decline in demand
Price Elasticity = −
% change in demand
% change in price
= −
−40
100
= 0.4
Uber Demand Curve
Sub-Segment Elasticity
New York 0.61
Los Angeles 0.33
Evening Rush 0.50
Morning Rush 0.52
Weekend Day 0.66
Weekend Evening 0.54
Surge 2.4 – 3.0 1.01
Number of rides 47.5 million
Using Big Data to Estimate Consumer Surplus: The Case of Uber
Peter Cohen, Robert Hahn, Jonathan Hall, Steven Levitt, Robert Metcalfe
NBER Working Paper No. 22627, Sep 2016
Great for users
Bad for Uber
Consumer surplus =
$6.8bn
Uber profit = -$4.5bn
Drivers earn $21.07
per hour; $10 net
Men earn 7% more
6. • When surge was first launched, uber learned that going
from no surge to 1.2x resulted in a drop of 27%.
• They also learned that going from 1.9x to 2.0x resulted in
a 6 times larger drop in demand than in going from 1.8x
to 1.9x surge….simply because 2.0x felt viscerally larger
and unfair.
• Surprisingly, they learned that going from 2.0x to 2.1x
surge induced more rides, because consumers assumed
an intelligent algorithm must be at play, better able than
humans to set a fair price.
Uber Surge
9. 1973: 2.75 % of global production withheld; Prices quadrupled
1979: 5.68 % of global production withheld; Prices doubled
“When demand is inelastic, small withdrawal of capacity can lead to
disproportionate increase in prices”
28
53.39
A Puzzle: Oil Price Shocks in 1973 and 1979
10. Procedure:
1. Write down model (equation) for demand with unknown coefficients
2. Collect data
3. Fit line or curve to data points using statistical techniques
(regression)
Sources of Data:
1. Consumer focus groups; Surveys
2. A/B testing; Randomized controlled trials
3. Historical (real) data: cross-section, time-series, or both (panel)
Scanner data: Walmart has 500TB of data
Consumer Expenditure Survey (U.S., Bureau of Labor Statistics)
Family Expenditure Survey (U.K., National Statistics)
Budget des Menages (France, INSEE)
Euro Monitor (INSEAD library)
Data as competitive advantage (Uber, Amazon, Google…)
Know Your Demand: Regression Analysis
13. Can only give us information of realized price changes.
1. We observe only some prices (more extreme prices, try an
alternative)
2. We observe customers` past behavior. What if this has changed
over time? (for changes, try an alternative)
3. Alternatives are usually consumer surveys, consumer focus groups
or market experiments.
Some Limitations Regression Analysis
16. What is the demand curve for oil in December 2013? Substitute values
of everything else
Income: $36,865
Price of new cars: 144.36
Price of old cars: 148.183
Year: 2013
Demand function: ln (G) = 3.649 – 0.051*ln (PG)
We can plot this!
From Demand Functions to Demand Curves
ln 𝐺 = 22.562 − 0.051 ln 𝑃𝐺 + 0.905 ln 𝑌
−0.101 ln 𝑃𝑁𝐶 − 0.115 ln 𝑃𝑈𝐶 − 0.014𝑌𝑒𝑎𝑟
17. From Demand Functions to Demand Curves
1
1.5
2
2.5
3
3.5
4
4.5
5
35 35.5 36 36.5 37 37.5 38 38.5 39 39.5 40
Gasoline
Price
Gasoline Demand
Demand for Gasoline
An increase in income?
18. From Demand Functions to Demand Curves
1
1.5
2
2.5
3
3.5
4
4.5
5
35 35.5 36 36.5 37 37.5 38 38.5 39 39.5 40
Gasoline
Price
Gasoline Demand
Demand for Gasoline
A decrease in car prices?
19. Strong growth raises incomes in US. What would happen to the demand
curve for gasoline?
Demand curve would shift right/up. Movement of the demand
curve
What if the price of gasoline fell by $0.50? Do you need to write a new
demand equation?
No! Own price changes are movements along the demand curve
Trump tariffs increase price of all cars, what would happen to the
demand relationship?
Demand curve would shift left/down. Movement of the demand
curve
Movement Along and Movement of Demand Curve
20. When an
increase in…
causes demand to…
increase decrease
Income Normal Inferior
Other Price Substitute Complement
Price and Income Effects: A Taxonomy
21. Measures responsiveness of demand to changes in prices/income
Price elasticity
Income Elasticity
Q
P
dP
dQ
P
Q
E
in
change
%
in
change
%
Q
I
dI
dQ
I
Q
EY
in
change
%
in
change
%
Price Elasticity and Income Elasticity
x
y
y
x
y
x
cross
Q
P
dP
dQ
P
Q
E
in
change
%
in
change
%
Cross Price elasticity
22. Bookseller
Own Price
Elasticity
Cross Price
Elasticity
0.6 0.2
4.0 3.5
Amazon vs. Barnes & Noble.Com
Summary
Amazon demand is price inelastic; BN is price elastic
A 1% decline in BN.com’s price reduces quantity at Amazon by 0.2%
A 1% decline in Amazon’s price reduces quantity at BN by 3.5%
Source: Measuring Prices and Price Competition Online by Chevalier, J and Goolsbeee, A.
24. Short-Run Long-Run
Price 0.009 0.009/(1 - 0.968) = 0.281
Income 0.017 0.017/(1 - 0.968) = 0.531
Oil Demand: Long Run vs. Short Run
Define Long Run:
State of Rest Today’s consumption = Yesterday’s consumption
G = G(-1) LG = LG(-1)
Coeff.
Standard
Error
t Stat P-value
LG(-1) 0.968 0.008 119 0
LPG -0.009 0.001 -7.9 0
LY 0.017 0.010 1.66 0.098
LPNC -0.043 0.006 -7.03 0
LPUC 0.011 0.003 3.81 0
YEAR 0.0001 0.0002 0.42 0.676
25. Oil Price Elasticities
Country Short-Run Long-Run
Australia 0.034 0.068
Austria 0.059 0.092
Canada 0.041 0.352
China -0.001 -0.005
Denmark 0.026 0.191
Finland 0.016 0.033
France 0.069 0.568
Germany 0.024 0.279
Greece 0.055 0.126
Iceland 0.109 0.452
Ireland 0.082 0.196
Italy 0.035 0.208
Country Short-Run Long-Run
Japan 0.071 0.357
Korea 0.094 0.178
Netherlands 0.057 0.244
New Zealand 0.054 0.326
Norway 0.026 0.036
Portugal 0.023 0.038
Spain 0.087 0.146
Sweden 0.043 0.289
Switzerland 0.030 0.056
UK 0.068 0.182
USA 0.061 0.253
Source: Annual data from OPEC Review
26. 1. The more close substitutes a good has, the ________ elastic is
demand.
2. Demand for a particular brand (Viewsonic) or type (17” flat panel) is
_________ elastic than demand for the entire category (computer
displays).
3. The more differentiated the brand, the ________ elastic is demand.
4. Advertising usually both increases demand and makes it _________
elastic.
5. Demand is __________ elastic in long run (after consumers have
time to adjust).
6. But short-run demand is _________ elastic if consumers expect price
change to be temporary.
Elasticity Sudoku
27. Demand estimation
• Use statistics to estimate demand based on limited data
• Importance of modeling step
Demand functions and demand curves
• A change in a good’s price causes a movement along the
demand curve
• A change in some other variable causes a shift in the demand
curve
Price (Income) elasticity is a measure of responsiveness of demand to
its price (income)
• Short run vs. long run elasticities
Wrap Up