Tactical Price Management

Question 1

Heuristics

Answer 1

a. Cost-Oriented Pricing – e.g. costs + mark up
i. Not ideal – customers might be willing to pay much more or are not willing to pay that much
ii. Dangerous to ignore the willingness to pay

b. Customer-oriented - e. g., Perceived-Value
i. Could be that you are no longer covering your costs

c. Competition Oriented - e. g., Going-Rate
i. Competition is much cheaper then you will not make a lot of profit

–> strategies only focus on one thing (competition, costs or customer, you want to look at all of them)

Question 2

Optimization based on response functions

Answer 2

• Estimate a response model: Sales = f (Price)
• Determine values of marketing mix variables that maximize profit
• More complex models:
  1. Decision Support Systems (DSS)
  2. Marketing Analytics
• Models cannot replace managers!
• -> If model results do not match up with managerial intuitions, both have to be reviewed

Question 3

Advantages of model-based planning:

Answer 3

1. Assumptions have to be revealed

2. Models yield a better understanding of problems

Question 4

Elasticity

Answer 4

= percentage change in a dependent variable (e. g., unit sales) divided by the percentage change in an explanatory (independent) variable (e. g., price)

Question 5

Direct price elasticity

Answer 5

a. Rule: |e| > 1 resp. e < -1
b. Exceptions (-1 < e < 0):
i. Consumers do not notice changes in price
ii. Searching for alternatives is too costly for consumers
iii. Consumers are not able to substitute products in the short run
c. Exceptions (e > 0):
i. Giffen effect (decrease in real income for inferior good)
ii. Snob effect
iii. Price as an indicator of quality
iv. Expectation effect

Question 6

Cross price elasticity

Answer 6

a. For substitutes > 0

b. For complements < 0

Question 7

Linear response function

Answer 7

a. Q = b0 − b1 ∙ P und ε = −b1 ∙ P/Q
b. Easy to visualize and understand
c. Parameters can be easily estimated (standard regression methods)
d. Within small intervals, a linear approximation of more complicated functions is possible („window argument“)
e. Constant marginal returns are not realistic - Q can become negative
f. Increasing |elasticities| (with increasing price) are not realistic

Question 8

Multiplicate response function

Answer 8

a. Q = b0 ∙ P hoch b1und ε = b1
b. A simple transformation (taking the logarithm of both sides) can turn the multiplicative response function into a linear function, and the parameters of the transformed function can be estimated through linear regression analysis
c. ln(Q) = ln(b0) + b1 ∙ ln(x)
d. With appropriate parameter values (exponents) decreasing, constant, and increasing marginal returns can be modeled
e. Decreasing |marginal returns| are more realistic than constant ones
f. Constant elasticity (b1) is practical
g. No saturation level (at very low prices)
h. Not defined for X = 0 (problem for budget instruments, e.g., advertising)

Question 9

Models with Competition

Answer 9

1. Dependent variables:
   a. Unit Sales
   b. Market share - relevant in case of competition
   c. Market volume - relevant in case of competition

Question 10

Simple market share models:

Answer 10

a. Linear: SA= b0 + b1 ∙ PA + b2 ∙ PB
with: S = market share, P = Price, A,B = Products
b. Multiplicative: SA = b0 ∙ PA hoch b1 ∙ PB hoch

Question 11

Models with Competition - Problem

Answer 11

Models do not necessarily meet constraints:

a. Market shares should range from 0 to 1 (range constraint)
b. Market shares should sum up to 1 across brands (sum constraint)

Question 12

Attraction Function

Answer 12

= Model which satisfies the range and the sum constraints: attraction function

• Market share of brand i = attraction of brand i / Sum of attraction s of all brands
• Our share = Us / Us + them

Question 13

Data for Estimating Response Functions

Answer 13

1. Data sources:
   a. Market Data, Experiments, Analogies and Meta-Analyses, Expert Judgment, Surveys, Auction
2. Data Requirements:
   a. Availability, Quality, Variability, Quantity, Costs
3. Analysis of Market Data
   a. Data on past prices and sales
   b. Analysis: estimate response function –> sales = f(price, controls)
   c. Data sources: Company records, Panel data, Household panel, Store panel (scanner data), Single-source panel, Loyalty programs …

Question 14

Data for Estimating Response Functions - Problems

Answer 14

1. Enough variation in price?
2. Multicollinearity (–> difficult to separate price effect from other effects)
3. Endogeneity (–> difficult to establish causality)

Question 15

Experiments

Answer 15

1. Definition:
   a. Empirical test of causal hypotheses
   b. Manipulation of treatment variables whilst keeping all other variables constant
2. Types: Field vs. laboratory experiments

Question 16

Managers‘ view on field experiments

Answer 16

a. Expensive and time consuming
b. Results are informative for competitors
c. It is difficult to control for all outside influences
d. If marketing efforts are reduced in the experiment, there is a risk of losing customers
e. If marketing efforts are increased in the experiment, there is a risk of upsetting customers when marketing returns to regular level

Question 17

Analogies

Answer 17

1. Basis: elasticities –> invariant to scaling –> comparable
2. Analogies
   a. Can known elasticity estimates be applied to other products, regions, time periods?
   b. Problems: Reliability, Generalizability

Question 18

Meta-Analyses

Answer 18

a. Systematic analysis of various estimated elasticity values
b. Advantage: broad data pool
c. Collection of numerous elasticity results
i. Empirical results from different sources
ii. Estimates typically show a large variance
d. Calculation of mean elasticity values
i. Price: -1.76 (Tellis 1988), 2.62 (Bijmolt/van Heerde/Pieters 2005)
e. Variance of elasticity values can by explained by
i. Methodology (e. g. functional form, data), Objects (e. g., products), Regions (e. g., countries), Time
f. Problem: publication bias (non-significant and non-plausible values are often not published)

Question 19

Expert Judgment

Answer 19

1. Typical problems of objective data:
   a. Market data typically has low variance
   b. Experiments are costly
   c. Markets may not be comparable
2. Basic idea of expert judgment (= subjective estimation):
   a. Managers have a good market knowledge, which is the basis for their (intuitive) decisions
   b. Managers are asked to make point estimates
   c. A response function is estimated based on these point estimates

Question 20

Evaluation of expert judgment

Answer 20

a. At least the same quality as intuitive decisions
b. Knowledge is revealed –> discussions get more objective
c. Reflection on problem is encouraged

Question 21

Considering Uncertainty

Answer 21

1. Environmental uncertainty (e. g. competitive actions) and uncertainty regarding response parameters
2. Sensitivity analysis:
   a. „What if“ analysis
   b. How does variation in the different parameter values change the results (e. g. profit contribution)?
   c. What is the optimal marketing mix for different parameter values?
3. Risk analysis:
   a. Method: Monte Carlo simulation
   b. Input: Probability distributions of uncertain parameters
   c. Output: Probability distribution of results (e. g. profit contribution)