Why Best Responses Matter
In a strategic setting, a decision maker’s payoff can depend on both their own choice and another person’s choice. A firm choosing a service level, for example, may earn different profits depending on whether a rival invests in promotion. The relevant question is therefore not simply, “Which action has the largest payoff?” It is, “Which action gives this player the largest payoff given the other player’s action?”
A best-response analysis answers that question one opponent action at a time. It makes strategic interdependence explicit: a player’s most attractive choice may change when the opponent changes theirs. This method is aligned with the SOA VEE Economics (Validation by Educational Experience) syllabus.
How to Read a Payoff Matrix
In a two-player matrix, the row player chooses a row and the column player chooses a column. Each cell contains an ordered pair \((\text{row payoff},\text{column payoff})\). The first number is used when comparing the row player’s actions; the second number is used when comparing the column player’s actions. A payoff may represent profit, cost-adjusted benefit, or another measure of preference, as long as higher numbers mean that player prefers the outcome.
To find the row player’s best response to one column, compare the first numbers down that column. To find the column player’s best response to one row, compare the second numbers across that row. Keep the opponent’s action fixed during each comparison. The two players’ payoffs need not be compared with each other: each player’s ranking is determined by that player’s own payoff.
Here \(R\) and \(C\) are the row and column players’ available actions, while \(u_R(r,c)\) and \(u_C(r,c)\) are their payoffs at the action pair \((r,c)\). The notation \(\operatorname{arg\,max}\) means the action, or set of tied actions, that achieves the maximum payoff. A best response is always stated relative to a particular action by the other player; it is not a description of an action in isolation.
Select one column when analyzing the row player, or one row when analyzing the column player.
Use first coordinates for the row player and second coordinates for the column player.
The largest payoff identifies the best response; keep every action if the maximum is tied.
A complete analysis records the best-response set for each column and each row.
Worked Examples
Worked Example: A Retailer Chooses a Service Plan
Two retailers choose service plans for a local delivery market. Retailer R chooses Basic or Extended service; Retailer C chooses a Standard or Premium offer. The entries are hypothetical quarterly profit units, with R’s payoff first.
| R’s action | C: Standard | C: Premium |
|---|---|---|
| Basic | (7, 5) | (4, 3) |
| Extended | (6, 4) | (8, 6) |
Method: compare one player’s own payoffs while holding the rival’s choice fixed. If C selects Standard, R compares 7 from Basic with 6 from Extended. Since \(7>6\), Basic is R’s best response to Standard. If C selects Premium, R compares 4 with 8. Since \(8>4\), Extended is R’s best response to Premium.
Now compare C’s payoffs across each row. If R chooses Basic, C compares 5 from Standard with 3 from Premium, so Standard is C’s best response. If R chooses Extended, C compares 4 with 6, so Premium is C’s best response.
The complete result is: R responds with Basic to Standard and Extended to Premium; C responds with Standard to Basic and Premium to Extended. The row player’s preferred service plan changes with C’s offer, which is one clear sign of strategic interdependence. The method avoids choosing the largest number in the whole matrix: each comparison must be restricted to one player’s payoffs and one fixed opponent action.
Worked Example: Three Possible Capacity Choices and a Tie
A small producer chooses Low, Medium, or High capacity. A distributor chooses a Narrow or Broad sales network. The producer’s payoff is listed first.
| Producer’s action | Distributor: Narrow | Distributor: Broad |
|---|---|---|
| Low | (7, 5) | (4, 3) |
| Medium | (6, 4) | (6, 6) |
| High | (3, 5) | (8, 2) |
Producer’s responses. Against Narrow, the producer’s payoffs are 7, 6, and 3 for Low, Medium, and High. The maximum is 7, so Low is the best response. Against Broad, the payoffs are 4, 6, and 8; the maximum is 8, so High is the best response.
Distributor’s responses. If the producer chooses Low, the distributor compares 5 for Narrow with 3 for Broad, so Narrow is best. If the producer chooses Medium, the distributor compares 4 with 6, so Broad is best. If the producer chooses High, the distributor compares 5 with 2, so Narrow is best.
There is no tie in this matrix, but the procedure also handles ties: if two actions gave the same maximum payoff, both would belong in the best-response set. The complete answer here is Low against Narrow and High against Broad for the producer; Narrow against Low and High, and Broad against Medium, for the distributor. Including all three producer actions is essential: omitting Medium would make the comparison against Broad incomplete.
Worked Example: A Best Response That Switches
A new supplier considers a Low or High introductory price. An established rival chooses whether to Accommodate or to Counter with a promotion. The hypothetical payoffs are profit units, with the supplier’s payoff first.
| Supplier’s action | Rival: Accommodate | Rival: Counter |
|---|---|---|
| Low price | (5, 7) | (−2, 3) |
| High price | (2, 8) | (1, 4) |
Supplier’s responses. If the rival accommodates, the supplier compares 5 from Low price with 2 from High price; Low price is best. If the rival counters, the comparison is −2 versus 1; High price is best. Thus the supplier’s best response switches when the rival’s action changes.
Rival’s responses. Against Low price, the rival compares 7 from Accommodate with 3 from Counter, so Accommodate is best. Against High price, the rival compares 8 with 4, so Accommodate is also best.
This example shows why “best response” is conditional. The supplier does not have one best action regardless of circumstances: the larger supplier payoff depends on what the rival does. The rival’s comparison, in turn, uses the second number in each cell—not the supplier’s payoff. This method identifies incentives for each possible action pair without assuming what either player will actually choose.
Why the Comparison Works
The logic is a direct application of maximizing a player’s payoff over their available actions. Fix the column player’s action at \(c\). Once \(c\) is fixed, each possible row action \(r\) has a corresponding row payoff \(u_R(r,c)\). The row player’s best response is the action that gives the largest value in that list. The column player’s calculation follows the same reasoning after fixing a row action.
The inequality states precisely what “best” means: no alternative row action gives the row player a higher payoff against that fixed column. Equality is allowed, so tied actions qualify. The same condition applies to the column player using \(u_C\) and fixing the row.
The reasoning assumes that the matrix lists the players’ available actions and the relevant payoff for every action pair, and that each player prefers a higher own payoff. If an action or payoff is missing, a complete best-response comparison cannot be made. If the payoffs are costs rather than benefits, the player instead prefers a lower cost; either convert costs to an equivalent payoff measure or identify the minimizing action explicitly.
Strategic Interdependence and Common Traps
Strategic interdependence means that outcomes depend on the combination of choices, so a player’s payoff from an action can vary with the opponent’s action. Best-response analysis captures this dependence by making a separate comparison for each possible opponent choice. The examples show that the row player’s preferred action can switch across columns even when the available actions remain unchanged.
A common mistake is to select a cell with the largest payoff in the entire matrix, or to compare both numbers in a cell as if they belonged to one player. For example, in the supplier matrix, against Counter the supplier compares −2 and 1, so High price is best. It would be wrong to compare the rival’s payoff of 3 or 4 in that column to make the supplier’s decision. Another common error is to report just one action when two actions tie for the maximum; a complete answer gives the entire set of best responses.
Check Your Understanding
For each matrix, identify the best response to every opponent action. The first payoff in each cell belongs to the row player.
- A firm chooses Route A or Route B, and a competitor chooses Local or Regional. Payoffs are A–Local \((5,2)\), A–Regional \((1,4)\), B–Local \((3,3)\), and B–Regional \((6,1)\). Give both players’ best responses for every opponent action. (Answer: Row: Route A against Local; Route B against Regional. Column: Regional against Route A; Local against Route B.)
- A producer chooses Small, Medium, or Large output; a buyer chooses Contract X or Contract Y. Payoffs are Small–X \((4,7)\), Small–Y \((6,2)\), Medium–X \((5,3)\), Medium–Y \((5,5)\), Large–X \((2,4)\), and Large–Y \((3,6)\). Find all best responses. (Answer: Row: Medium against X; Small against Y. Column: X against Small; Y against Medium and Large.)
- A platform chooses Open or Restricted access, and a partner chooses Integrate or Remain Separate. Payoffs are Open–Integrate \((8,3)\), Open–Separate \((4,6)\), Restricted–Integrate \((8,5)\), and Restricted–Separate \((7,2)\). Identify any tie and all best responses. (Answer: Row: Open and Restricted against Integrate; Restricted against Separate. Column: Separate against Open; Integrate against Restricted.)
- A supplier chooses Invest or Wait; a buyer chooses Expand or Maintain. Payoffs are Invest–Expand \((2,1)\), Invest–Maintain \((9,5)\), Wait–Expand \((4,6)\), and Wait–Maintain \((6,3)\). Find the best responses and state whether the supplier’s response changes with the buyer’s choice. (Answer: Row: Wait against Expand; Invest against Maintain. Column: Maintain against Invest; Expand against Wait. Yes, the supplier’s best response changes.)
- A firm chooses Design A or Design B, while a distributor chooses Channel 1, 2, or 3. Payoffs are A–1 \((3,7)\), A–2 \((8,2)\), A–3 \((4,5)\), B–1 \((6,4)\), B–2 \((5,6)\), and B–3 \((7,3)\). Find both players’ best responses for every action of the opponent. (Answer: Row: Design B against Channel 1; Design A against Channel 2; Design B against Channel 3. Column: Channel 1 against Design A; Channel 2 against Design B.)