All courses > Basic studies > Economics ::

Economic Fundamentals: Simple Models for Clear Thinking

Capítulo 7

Estimated reading time: 8 minutes

Why Economists Use Models

An economic model is a simplified representation of how a part of the world works. Models help you focus on a small set of relationships so you can reason clearly, make predictions, and test ideas against evidence. The point is not to include everything; the point is to include the right things for the question you are asking.

Think of a model like a map: a subway map leaves out buildings and street names so you can see routes and transfers. Similarly, an economic model leaves out many real-world details so you can see how key variables move together.

What Models Are For

Clarify cause and effect: identify which variable changes first and which responds.
Organize thinking: list what matters, what doesn’t (for now), and why.
Generate predictions: if X changes, what should happen to Y?
Guide measurement: tell you what data you need to collect.
Support decisions: compare scenarios using the same logic each time.

Assumptions: The Model’s “Rules of the Game”

Assumptions are statements you accept as true within the model so you can analyze the relationships you care about. Assumptions are not necessarily claims about reality; they are choices about what to hold fixed, what to ignore, and what to simplify.

Common Types of Assumptions

Behavioral assumptions: how people respond to changes (e.g., “higher price reduces quantity demanded”).
Institutional assumptions: rules and constraints (e.g., “buyers can freely choose among sellers”).
Information assumptions: what people know (e.g., “consumers observe prices”).
Timing assumptions: short run vs. long run (e.g., “in the short run, capacity is fixed”).
Aggregation assumptions: how individual actions combine (e.g., “market demand is the sum of individual demands”).

Good Assumptions Are…

Transparent: clearly stated so others can evaluate them.
Purpose-built: chosen to answer a specific question.
Testable in implications: they lead to predictions you can compare with data.

Ceteris Paribus: “All Else Equal”

Ceteris paribus means “holding other relevant factors constant.” It is the tool that lets you isolate one relationship at a time. When you say, “If the price of coffee rises, quantity demanded falls, ceteris paribus,” you are not claiming nothing else ever changes; you are saying you are examining the effect of price while keeping other drivers (income, tastes, substitutes, time of day, weather, etc.) fixed.

Why It Matters

Prevents confusion: if both price and income change at the same time, it’s hard to tell which caused the change in purchases.
Improves comparisons: you can compare two situations that differ in only one key way.
Sets up empirical tests: in data, you try to control for “other things” using careful design or statistical methods.

What Ceteris Paribus Is Not

It is not a claim that other factors are unimportant.
It is not a promise that the prediction will hold in every situation.
It is not a substitute for checking whether “other things” actually stayed similar in the real world.

Building a Small Model: A Step-by-Step Template

To build a simple model, you can follow a repeatable workflow. The goal is to move from a question to a prediction that could be wrong (and therefore informative).

Continue in our app.

Listen to the audio with the screen off.
Earn a certificate upon completion.
Over 5000 courses for you to explore!

Or continue reading below...

Download the app

Step 1: State the Question

Write a question with a clear outcome variable. Examples: “How does the price of coffee affect cups sold per day?” or “How do membership fees affect gym attendance?”

Step 2: Choose the Outcome (Dependent Variable)

This is what you want to explain or predict. Use a measurable definition.

Coffee example: Q = cups of coffee sold per day at a café.
Gym example: A = number of check-ins per day.

Step 3: Choose Key Drivers (Independent Variables)

Select a small set of variables that plausibly move the outcome. Keep it minimal at first.

Price: P = price per cup (or membership fee).
Income: Y = average customer income or local payroll timing.
Preferences/tastes: T = strength of preference (seasonality, trends, health goals).
Substitutes: Ps = price of substitutes (tea, energy drinks; at-home workouts).
Complements: Pc = price of complements (pastries; personal training sessions).

Step 4: State Assumptions and the Ceteris Paribus Set

List what you are holding constant and why. Example assumptions for a first-pass demand model:

Product quality is constant.
Store hours and location are constant.
No major advertising changes during the period.
Competitors’ offerings are unchanged (or captured by Ps).

Step 5: Specify the Relationship (Words First, Then Symbols)

Start with a verbal statement, then translate it into a simple functional form.

Verbal: Quantity demanded decreases when price increases, holding other factors constant.

Symbolic (general):

Q = f(P, Y, T, Ps, Pc)

Symbolic (simple linear example):

Q = a - bP + cY + dPs + eT

Interpretation of signs (typical expectations):

-bP: higher P lowers Q (so b > 0).
+cY: higher income raises Q for many normal goods.
+dPs: if substitutes become more expensive, demand for your product rises.
+eT: stronger preferences raise demand.

Step 6: Generate a Prediction (Comparative Statics)

Comparative statics means comparing two situations: before and after a change in one variable, holding others constant.

If P increases by 1 unit, the model predicts Q decreases by b units (in the linear example), ceteris paribus.
If Ps increases (substitutes get pricier), the model predicts Q increases by d units, ceteris paribus.

Step 7: Check Units and Measurement

Make sure each variable is measurable and has units.

Q: cups/day, or check-ins/day.
P: dollars per cup, or dollars per month.
Y: dollars per month (or a proxy like local average wage).
T: an index (e.g., 0–10) or a proxy (e.g., month-of-year dummy variables).

Step 8: Identify What the Model Leaves Out (and Why)

Every model omits factors. The key is to name them and judge whether omission is harmless for your question.

Omitted variable risk: if an omitted factor changes alongside P, your predicted effect of P may be misleading.
Scope limits: a short-run model may not apply in the long run (e.g., customers can change habits, new competitors enter).

Worked Mini-Example: A Coffee Demand Model

1) Question

“What happens to daily coffee sales if the café raises the price by $0.50?”

2) Variables

Outcome: Q = cups sold per day.
Driver of interest: P = price per cup.
Controls (held constant or tracked): Y income, Ps price of nearby substitutes, T tastes/seasonality.

3) Relationship

Q = a - bP + dPs + eT

4) Prediction

If the café raises P by $0.50 and nothing else changes, Q should fall by 0.5b cups per day. If you observe sales rising instead, that suggests at least one “all else” condition failed (for example, a competitor raised prices even more, or a new trend increased coffee preference).

5) What Could Break the Prediction?

Income changes: payday effects or local job changes shift demand.
Preference shifts: a health trend, weather changes, or a viral review changes T.
Substitutes change: a nearby café closes, or tea prices rise, changing Ps.
Quality changes: new beans or faster service changes perceived value (effectively changing demand, not just movement along it).
Measurement issues: counting transactions vs. cups (one transaction can include multiple cups).

Practice Activity: Build Your Own Simple Model

Choose one scenario: coffee demand or gym attendance. Use the template below to construct a model and then stress-test it by listing what could break your prediction.

Option A: Coffee Demand

Define the outcome: Let Q be cups sold per day at one café.
Pick 3–5 drivers: choose from P (your price), Y (customer income), Ps (price of substitutes), T (tastes/season), W (weather), H (hours open).
Write assumptions: list at least 3 things you hold constant (e.g., quality, location, staffing).
Write the model: in words, then as a function:
```
Q = f(P, Y, Ps, T, ...)
```
Generate one prediction: “If P increases by ___, then Q will ___, ceteris paribus.”
Identify what could break it: list at least 4 factors, including: income changes, preference changes, and substitutes.

Option B: Gym Attendance

Define the outcome: Let A be daily check-ins at a gym.
Pick 3–5 drivers: choose from F (membership fee), D (distance/travel time), S (schedule flexibility/hours), T (tastes/health goals), Ps (price of substitutes like home equipment or other gyms), Y (income).
Write assumptions: list at least 3 things you hold constant (e.g., equipment availability, class schedule, cleanliness).
Write the model: in words, then as a function:
```
A = f(F, D, S, T, Ps, Y)
```
Generate one prediction: “If F increases by ___, then A will ___, ceteris paribus.”
Identify what could break it: list at least 4 factors, including: income changes, preference changes, and substitutes (e.g., a new low-cost gym opening nearby).

Self-Check Questions (Use After You Write Your Model)

Did you clearly define what is changing and what is held constant?
Are your variables measurable (even with proxies)?
Does your prediction specify direction (increase/decrease) and the condition ceteris paribus?
Did you name at least one omitted factor that could move with your key driver and confuse the result?

Now answer the exercise about the content: