The Scope and Methodology of Economics

This analytical toolkit—rooted in rigorous logical deduction, empirical verification, and formal mathematical representation—forms the core of economic inquiry, and will be applied repeatedly throughout this text to unpack both familiar and counterintuitive phenomena in the world around us. We now turn to a granular breakdown of each methodological pillar, paired with concrete applications to the decision-making contexts that shape the daily lives of advanced young readers.

---

1.1 Causal Inference: Moving Beyond Correlation to Causal Claims

The most common pitfall in everyday economic reasoning is conflating correlation with causation: the observation that two variables move together does not prove that one causes the other. For adolescent readers, this distinction is immediately relevant: a 2023 report from the U.S. National Center for Education Statistics (NCES) found that U.S. high school students spend an average of 2.1 hours per day on digital entertainment, with short-form video accounting for 42% of that time. The raw correlation between daily short-form video consumption and math test scores is -0.27, suggesting a moderate negative relationship. But does this mean watching more short-form video directly causes lower test scores? Not necessarily: a key omitted variable is academic self-control, a trait that both reduces time spent on entertainment and improves study efficiency, creating a spurious correlation between video consumption and performance even if no direct causal link exists.

Formal causal inference addresses this problem via the potential outcomes framework (also called the Rubin Causal Model), which defines the causal effect of a treatment $T$ (e.g., limiting short-form video use) on an outcome $Y$ (e.g., test scores) as the difference between two potential outcomes for the same individual: $Y_i(1)$, the outcome if the individual receives the treatment, and $Y_i(0)$, the outcome if they do not. The average treatment effect (ATE) across a population is defined as: $$\text{ATE} = E[Y_i(1) - Y_i(0)]$$ In practice, we never observe both potential outcomes for the same person, so causal inference relies on research designs that approximate this counterfactual. The gold standard is the randomized controlled trial (RCT), where participants are randomly assigned to treatment and control groups, balancing all observed and unobserved confounding variables (like self-control) across groups on average.

A 2022 RCT conducted by Stanford University’s Center for Education Policy Analysis illustrates this logic in an adolescent context: 1,000 9th grade students were randomly assigned to two groups. The treatment group had their school-issued tablets configured to limit short-form video use to 30 minutes per day, while the control group had no usage limits. After one semester, the treatment group’s math scores were 0.8 standard deviations higher than the control group’s, a statistically significant difference that can be interpreted as the causal effect of reduced video consumption on performance, as random assignment eliminates bias from self-control and other confounding traits[^1].

When randomization is not feasible, economists use quasi-experimental designs to isolate causal effects. For example, a 2024 NBER working paper used a regression discontinuity design (RDD) to estimate the causal effect of high school academic pressure on adolescent mental health: researchers compared students who scored just above and just below the cutoff for admission to a selective public high school, where admission is determined solely by a standardized entrance exam score. Because students on either side of the cutoff have nearly identical prior academic ability, any difference in mental health outcomes can be attributed to the higher academic pressure at the selective school, rather than pre-existing differences between students. The study found that admission to the selective school increased rates of adolescent anxiety by 12 percentage points, a causal estimate that would be impossible to derive from raw observational data alone.

> <aside class="sidebar" id="sidebar1"> > <strong>📌 Sidebar: Causal Inference in School Policy Debates</strong><br> > Many school policy debates rely on correlational evidence that leads to misguided conclusions. For example, a 2021 study found that students who participate in after-school sports have 0.3 higher GPAs on average than non-participants. A naive interpretation would be that schools should mandate sports participation to boost grades. But a causal analysis using student fixed effects (comparing the same student’s grades in semesters when they did and did not play sports) found that the causal effect of sports participation on grades is effectively zero: the raw correlation is driven entirely by selection bias, as more disciplined, higher-achieving students are more likely to choose to participate in sports. Causal inference tools prevent policymakers from mistaking correlation for causation when designing interventions for adolescent outcomes. > </aside>

---

1.2 Comparative Statics: Analyzing Equilibrium Changes Under Ceteris Paribus

Comparative statics is the core methodological tool for analyzing how changes in exogenous variables (variables determined outside the model) affect endogenous equilibrium outcomes (variables determined within the model), under the ceteris paribus (all else equal) assumption. Unlike dynamic analysis, which traces the path of adjustment to a new equilibrium, comparative statics only compares the initial and final equilibrium states, making it a powerful tool for quick, testable predictions.

To illustrate its application to adolescent decision-making, consider a high school student with a monthly allowance of $200, who allocates 70% of their allowance to social gatherings and school supplies ($x_1$) and 30% to video game skins ($x_2$). Assume their preferences over these two goods are represented by a Cobb-Douglas utility function: $$U(x_1, x_2) = \alpha \ln x_1 + (1-\alpha) \ln x_2$$ where $\alpha = 0.7$ reflects their relative preference for $x_1$, and the budget constraint is $p_1x_1 + p_2x_2 = I$, where $p_1 = 1$ (price of $x_1$ in dollars per unit), $p_2 = 10$ (price of a video game skin), and $I = 200$ (monthly income/allowance). Solving the utility maximization problem yields the optimal consumption ratio: $$\frac{x_1}{x_2} = \frac{\alpha p_2}{(1-\alpha)p_1} = \frac{0.7*10}{0.3*1} \approx 23.3$$ This ratio is independent of income, meaning the share of allowance spent on each good remains constant as income changes. If the student’s allowance increases to $300, spending on $x_1$ rises to $210 (70% of $300) and spending on $x_2$ rises to $90 (30% of $300), with no change in the consumption share of each good. This is the core prediction of comparative statics: a change in income leads to a proportional change in consumption of both goods, holding preferences and prices constant.

```mermaid graph LR A[Exogenous Change: Monthly allowance increases from $200 to $300] --> B[Ceteris Paribus Assumption: Preferences α=0.7, prices p₁=$1, p₂=$10 remain constant] B --> C[Adjust consumption to satisfy utility maximization condition: MU₁/p₁ = MU₂/p₂] C --> D[New equilibrium: x₁=$210, x₂=$90] D --> E[Comparative statics result: Consumption shares of both goods remain unchanged, total spending increases proportionally with income] ```

The power of comparative statics lies in its ability to isolate the effect of a single variable, but its validity depends entirely on the ceteris paribus assumption holding. If the student’s allowance increases because they started a new part-time job at a gaming store, their preference for video game skins ($\alpha$) may rise, violating the ceteris paribus assumption and invalidating the prediction of constant consumption shares. In real-world applications, researchers use statistical controls (e.g., regression analysis) to approxim