Lecture 1 · Week 1 · Özata slides & Gujarati ch. 1 · about 35 min

The Nature and Scope of Econometrics

What econometrics actually is, the three disciplines it draws from, the eight-step methodology every empirical study follows, and the four types of economic data you will meet throughout this course.

By the end you can

Define econometrics and name the three disciplines it combines
List the eight steps of the econometric methodology, using the Keynesian consumption function as a running example
Distinguish time series, cross-sectional, pooled and panel data, and give an example of each
Explain why correlation is not the same as causality

Key idea · Exam recipe (memorise this)

Definition: econometrics = economic theory + mathematics + statistical inference applied to economic phenomena (“economic measurement”).
Three aims: formulate models · estimate and test them with data · use them for prediction and policy.
Eight steps: theory → mathematical model $Y = \beta_0 + \beta_1 X$ → econometric model $Y = \beta_0 + \beta_1 X + u$ → data → estimation $\hat Y$ → hypothesis testing → forecasting → policy.
Why $u$ : omitted factors · measurement error · human indeterminacy. Without $u$ : deterministic; with $u$ : stochastic.
Keynes: MPC $= \beta_1$ , $0 \lt \beta_1 \lt 1$ . Slide estimate $\hat Y = -60{,}356{,}991 + 0.6515X$ ; multiplier $= 1/(1 - MPC)$ .
Data: time series (one unit, many periods) · cross-section (many units, one period) · pooled (different units each period) · panel (same units over time).
Graph: consumption line with points scattered around it, the vertical gap labelled $u$ .

How the exam asks it: three core exam templates

Midterm questions on the nature and scope of econometrics typically test three canonical formats: forecasting from a fitted line, computing the Keynesian multiplier and policy effects, and classifying dataset structures.

Exam template 1: Reading an estimated model and forecasting consumption

Question: An econometrician estimates the relationship between personal consumption expenditure ( $Y$ ) and personal disposable income ( $X$ ) for a sample of households (both in thousand TL): $\hat{Y} = 45 + 0.75X$ a) Identify the intercept and slope coefficient, and interpret the slope in economic terms.
b) If a household has a disposable income of 400 thousand TL, what is its predicted consumption expenditure?
c) What is the economic meaning of the disturbance term $u$ that appears in the theoretical model but not in the fitted line?

Exam template 2: Marginal propensity to consume and the Keynesian multiplier

Question: Using annual time series data for Türkiye, an empirical consumption function is estimated as: $\hat{Y}_t = -60{,}356{,}991 + 0.6515X_t$ where both variables are in thousand TL. a) State the estimated marginal propensity to consume (MPC) and check whether it satisfies Keynes’s theoretical condition.
b) Calculate the Keynesian expenditure multiplier implied by this estimate (using the rounded value $MPC = 0.65$ ).
c) If the government raises public expenditure by 250 million TL, calculate the predicted change in national income.

Exam template 3: Data type identification

Question: Classify each dataset and state the defining criterion: a) The inflation rates and budget deficits of all 27 European Union member states in 2023.
b) Türkiye’s monthly unemployment rate recorded from January 2015 to December 2024.
c) Annual earnings and education levels of the same 1,000 workers tracked every year from 2015 to 2025.
d) 500 consumers surveyed in 2020 and a different sample of 500 consumers surveyed in 2023.

Cheat-sheet & the 4-step method

· Cheat-sheet: core concepts, formulas and data types

Concept / Task	Formula or Criterion	When to use it	How to write the exam answer
Point forecast	$\hat{Y} = \hat{\beta}_0 + \hat{\beta}_1 X$	Given $X$ , predict $Y$	“For $X = c$ , predicted $Y$ is $\hat{Y}$ [units].”
Slope / MPC	$\hat{\beta}_1 = \dfrac{\Delta \hat{Y}}{\Delta X}$	Interpret marginal effect	“A 1 [unit of X] rise leads on average to a $\hat{\beta}_1$ [unit of Y] rise.”
Multiplier	$k = \dfrac{1}{1 - MPC}$	Fiscal policy evaluation	“The Keynesian multiplier is $k$ ; $\Delta Y = k \times \Delta G$ .”
Time series	$Y_t$ , $t = 1, \dots, T$	One entity across time	“Time series: single entity observed over regular time intervals.”
Cross-sectional	$Y_i$ , $i = 1, \dots, N$	Many entities at one time	“Cross-sectional: multiple entities at a single point in time.”
Panel data	$Y_{it}$ , same $i$ over $t$	Tracked units over time	“Panel: the same individual units tracked across multiple periods.”
Pooled data	$Y_{it}$ , different $i$	Independent surveys combined	“Pooled: independent cross-sections from different times combined.”

Key idea · The 4-step method for nature and scope exam problems

Step 1: Identify the question type.
- Numerical forecast: plug $X$ into $\hat{Y} = \hat{\beta}_0 + \hat{\beta}_1 X$ .
- Multiplier problem: read $\hat{\beta}_1 = MPC$ , compute $1/(1-MPC)$ , multiply by $\Delta G$ .
- Data classification: count time periods ( $T$ ) and check whether cross-sectional units ( $N$ ) are the same.
- Methodology question: trace the 8 steps in order (Theory $\to$ Math $\to$ Econometric $\to$ Data $\to$ Estimation $\to$ Testing $\to$ Forecasting $\to$ Policy).
Step 2: Extract and label all given values.
- Write down $\hat{\beta}_0, \hat{\beta}_1$ , the given $X$ , and their exact measurement units.
Step 3: Carry out one calculation per line.
- In forecasting, multiply slope by $X$ first, then add the intercept.
- In multipliers, calculate the denominator $(1 - MPC)$ first, then divide.
Step 4: Write the final bold sentence with units and interpretation.
- Include the units (“billion TL”, “thousand TL”) and verify economic plausibility ( $0 \lt MPC \lt 1$ ).

What is econometrics?

The slides call it, more briefly, the study of economic measurement, and add that it is based on developing statistical methods for estimating economic relationships, testing economic theories and evaluating and implementing government and business policy. A broader description: econometrics involves the mathematical modelling of economic events, the collection of statistical data related to the subject, and the application and analysis of estimation and testing procedures. In short, it is a research method that helps test economic theory, explain economic events, predict the future and support economic decisions.

Econometrics sits at the intersection of three disciplines, and it is the combination — not any one alone — that defines it.

The three disciplinesKey ideaFormula sheet →

\text{Econometrics} = \text{Economic theory} \cap \text{Mathematics} \cap \text{Statistics}

Use it when a question asks you to justify why econometrics is not simply 'applied statistics' or 'mathematical economics.'

Economic theory identifies the variables and predicts the direction of relationships (a demand curve slopes down), but rarely says how much quantity changes for a given price change.
Mathematics supplies the language to express those relationships precisely and manipulate them.
Statistics supplies the tools to collect data and estimate the actual numerical values, plus a way to judge how reliable those estimates are.

Pairwise overlaps of these three disciplines have their own names: economic theory ∩ mathematics is mathematical economics; economic theory ∩ statistics is economic statistics; mathematics ∩ statistics is mathematical statistics. Econometrics is the overlap of all three at once.

The three aims of econometrics

Aims of econometricsKey listFormula sheet →

1.\ \text{Formulate econometric models} \qquad 2.\ \text{Estimate and test them with data} \qquad 3.\ \text{Use them for prediction and policy}

Use it when you are asked to summarise, in one line, what econometric research is ultimately for.

These aims translate directly into real decisions: a restaurant assessing an ad campaign’s effect on sales, a government estimating the price elasticity of demand before a tax change, a private university projecting how a tuition increase of 10,000 TL affects enrolment, or a firm like Arçelik forecasting refrigerator demand a decade out. Each of these is an economic question that theory alone cannot answer numerically — it requires econometrics.

Why do we learn econometrics?

The slides give five reasons: (1) understanding economic relationships, (2) testing economic theories, (3) making informed policy decisions, (4) forecasting economic conditions and (5) enhancing research skills. Knowing econometrics matters to anyone who wants to understand how the economy works, decide on economic policy intelligently, or do research in economics and related subjects.

The slides’ everyday examples of econometric questions:

Decision area	Research question
Restaurant	How much does advertising expenditure affect sales?
Consumer demand	How does demand respond to changes in price and income?
University	How does a tuition increase affect enrolment?
Manufacturing firm	Will future demand justify investment in new capacity?

The methodology of econometrics: eight steps

Steps of an econometric analysisKey procedureFormula sheet →

\begin{aligned} &1.\ \text{Statement of theory or hypothesis} \\ &2.\ \text{Specification of the mathematical model} \\ &3.\ \text{Specification of the statistical (econometric) model} \\ &4.\ \text{Obtaining data} \\ &5.\ \text{Estimation of the econometric model} \\ &6.\ \text{Hypothesis testing} \\ &7.\ \text{Forecasting or prediction} \\ &8.\ \text{Use of the model for policy purposes} \end{aligned}

Use it when an exam asks you to list, or apply, the standard stages of an empirical econometric study — walk through each with the running example below.

Step 2 in detail: variables, signs and functional forms

Step 2 turns the theory into a quantifiable model: the variables are identified, the direction of each relationship is set, and the mathematical pattern is chosen. Another step-1 example on the slides is the Theory of Demand, which leads into the following illustration.

For a demand function, quantity demanded is the dependent (explained) variable, and economic theory says demand is determined by the good’s own price, consumer income, substitute and complement prices, and tastes and preferences:

General demand functionFormula sheet →

Q = f(P,\, I,\, P_s,\, P_c,\, Z)

$Q$: quantity demanded (dependent variable)
$P$: the good's own price (explanatory)
$I$: consumer income (explanatory)
$P_s$: price of a substitute (explanatory)
$P_c$: price of a complement (explanatory)
$Z$: other factors such as tastes and preferences (explanatory)

Use it when step 2 needs the model's variables named before a functional form is selected.

Economic theory also predicts the expected signs: the own price has a negative effect on demand (law of demand), income a positive effect, the price of a substitute a positive effect, and the price of a complement a negative effect.

Economic theory rarely fixes the mathematical structure, so mathematics supplies candidate functional forms:

Form	Example equation	Property
Linear	$Y = \alpha + \beta X$	The slope is constant.
Quadratic	$Y = \alpha + \beta X + \gamma X^2$	The slope varies with X.
Exponential	$Y = A e^{\beta_2 X}$	The change depends on the level.
Cubic	$Y = \alpha + \beta X + \gamma X^2 + \delta X^3$	Allows a more flexible curve.

Theory may indicate the direction of a relationship, while data and economic reasoning help determine its functional form.

Keynesian consumption function (mathematical model)Formula sheet →

Y = \beta_0 + \beta_1 X

$Y$: consumption
$X$: income (here, GDP)
$\beta_0$: the intercept (constant term)
$\beta_1$: the slope, the marginal propensity to consume (MPC), between 0 and 1

Use it when the mathematical (deterministic) stage of the model, step 2, before a disturbance term is added.

Keynesian consumption function (econometric model)Step 3Formula sheet →

Y = \beta_0 + \beta_1 X + u

$u$: the disturbance (error) term, a random variable capturing everything the deterministic line leaves out

Use it when the model is now stochastic rather than exact — the standard starting point for every regression in this course.

The slides’ figure for the econometric consumption function: the points do not lie on the line, and the vertical distance between an observed point and the line is the disturbance $u$ .

Forecasting is one of the most challenging applications of econometrics: the fitted model predicts $Y$ for a known or projected future $X$ , provided the model has not disproved the theory under study. The slide’s own forecast example uses 2,500,000,000 (thousand TL) as projected 2023 GDP and obtains 1,568,393,009 (thousand TL): its text line prints 1,568,393,005, but its table and the arithmetic both give …009.

The estimated slope is the marginal propensity to consume: over the sample period 2010–2022, a 1 TL rise in income was associated with a rise of about 0.65 TL in consumption expenditure on average — ‘on average’ because the relationship is not exact.

Projected GDP X (million × thousand TL)2500

X in thousand TL: 2,500,000,000
Forecast Ŷ (thousand TL): 1,568,393,009

Extra government spending (million TL)100

Multiplier 1/(1 − 0.65): 2.857
Change in income: 286 million TL

Step 7, forecasting: put a projected GDP into the fitted line. At X = 2,500,000,000 thousand TL it gives 1,568,393,009 thousand TL, as on the slide. Step 8, policy: with an MPC of about 0.65 the multiplier is 1/0.35 ≈ 2.86, so extra spending of 100 million TL raises income by about 286 million TL.

Hypothesis testing and model assessment

Step 6 puts the estimated model through diagnostic tests, to make sure the hypotheses and the estimation technique suit the data. The aim is robust results — conclusions that are not affected by model specification — so models are often reformulated and re-estimated by alternative methods. The slides group the checks into four questions:

Economic expectations: are the coefficient’s sign and size consistent with theory?
Statistical inference: does the evidence support the hypothesis about the coefficient?
Diagnostics: are the assumptions of the model and estimation method appropriate?
Sensitivity: do reasonable changes in specification substantially alter the results?

Uses of an estimated model

The slides give three main purposes of econometric research:

Purpose	Main question
Structural analysis	What are the direction and magnitude of economic relationships?
Policy analysis	What might a particular policy change do?
Forecasting	What value of Y is predicted for a future value of X?

Structural analysis tests economic theories against economic realities; policy formulation uses the estimated model to develop alternative policies; forecasting predicts what values the dependent variable might take in the future, based on the estimated model. One caveat from the slides: policy interpretation requires assumptions and a research design that support a causal relationship.

The Keynesian multiplier: a policy application

Keynesian multiplierPolicy use (step 8)Formula sheet →

\text{Multiplier} = \frac{1}{1-\widehat{MPC}}

$\widehat{MPC}$: the estimated marginal propensity to consume (slope from the fitted line)

Use it when a fitted consumption function's slope is used to estimate the impact of a change in government spending on national income — a direct step-8 application.

The slide prints the multiplier as 2.85 because it truncates 2.857 at two decimals; rounded properly it is 2.86. Using the slide’s 2.85 value, a 100 million TL increase in government spending raises national income by $100 \times 2.85 = 285$ million TL.

Experimental and observational data

The slides divide economic data into two main types. Experimental data: the researcher designs an intervention and measures its outcomes; experiments may take place in a laboratory or in a real-world setting. A field experiment takes place in a real-world situation, such as a marketplace, with participants who are unaware that they are taking part in an experiment. Observational data, also called non-experimental data: the researcher observes existing behaviour and conditions — for example observed outcomes for countries, firms or households. Observational data comes in four types, covered next.

Four types of economic data

Types of economic dataKey classificationFormula sheet →

\text{Time series} \quad \text{Cross-sectional} \quad \text{Pooled (repeated cross sections)} \quad \text{Panel (longitudinal)}

Use it when you are given a dataset's description and need to classify it — a very common short-answer question.

Time series: observations on one or more variables over time, at regular intervals (denoted $Y_t$ , $t=1,\dots,T$ ). Example: Türkiye’s quarterly GDP from 2010 to 2023.
Cross-sectional: observations on one or more variables at a single point in time, across different units. Examples: GDP of every OECD member country in 2023, olive production of the Aegean cities in 2022, and the distribution of COVID-19 vaccines on a specific day.
Pooled (repeated cross sections): different cross-sectional samples collected at different points in time and combined — but not necessarily the same units each time. Example: a survey of (a different sample of) 150 houses sold in 2018 and 180 different houses sold in 2021, combined into one dataset. The slides use exactly this type to analyse the effect of the Covid-19 pandemic on the price of single-family houses.
Panel (longitudinal): the same cross-sectional units followed over multiple time periods. Example: GDP per capita for the same three named countries, observed every year from 2017 to 2020.

Widely used cross-sectional data include the incomes of households in different geographic areas such as cities or countries (often used to study income inequality), the expenditure patterns of consumers with different income levels or demographic characteristics (consumer behaviour and preferences), and the performance of firms in different industries or areas (competition and market structure).

Time series: notation and frequency

The slides list typical time series examples: interest rates, money supply, producer price index, GDP deflator, unemployment rates and house sales figures. A series such as GDP from 2010 to 2023 is written $Y_t$ for $t = 1, 2, \ldots, T$ , with $t=1$ for 2010 and $t=T=14$ for 2023 — the slide prints $T=24$ for 2023, an obvious typo for a series that starts in 2010 (24 periods would fit a series from 2000).

Time series data come at different frequencies:

Frequency	Example data
Daily	Exchange rates
Weekly	Money and banking indicators
Monthly	Price indices and inflation
Quarterly	GDP
Annual	Population and demographic indicators

Sources of economic data

Economic data comes from government agencies (TÜİK, the Central Bank of the Republic of Türkiye), international organisations (World Bank, IMF, OECD), private sector organisations (research firms and financial institutions), academic research and surveys (household, firm and consumer studies), and market data (stock prices, commodity prices and exchange rates).

Causality and correlation

Correlation vs regressionKey distinctionFormula sheet →

\text{Correlation: symmetric, measures linear association strength} \qquad \text{Regression: asymmetric, one variable predicts/explains another}

Use it when a question asks you to explain why a regression result does not, by itself, prove causation.

Correlation measures the strength and direction of a linear association between two variables (a coefficient between −1 and +1), treating both variables symmetrically — the correlation between X and Y is identical to the correlation between Y and X. Regression is asymmetric: it treats one variable as dependent (explained, stochastic) and the other as independent (explanatory), and it aims to predict or explain one from the other, not merely measure association.

Exam practice

Summary and review

Econometrics = economic theory ∩ mathematics ∩ statistics; it supplies numerical content that theory alone cannot.
Three aims: formulate models, estimate and test them, use them for prediction and policy.
Eight-step methodology: theory → mathematical model → statistical model (add $u$ ) → data → estimation → hypothesis testing → forecasting → policy use.
$u$ exists because of omitted factors, measurement error, and human indeterminacy.
Four data types: time series (one unit, many periods), cross-sectional (many units, one period), pooled (different units, many periods), panel (same units, many periods).
Regression is asymmetric (explains Y from X) while correlation is symmetric; neither, by itself, proves causation — that requires a theoretical argument from outside the statistics.
Five reasons to learn econometrics: understanding economic relationships, testing theories, informed policy, forecasting, research skills.
Step 2 identifies the variables and their expected signs and chooses a functional form (linear, quadratic, exponential, cubic); a model is a simplified representation of reality based on stated assumptions.
An estimated model is assessed on four dimensions — economic expectations, statistical inference, diagnostics, sensitivity — and is used for structural analysis, policy analysis and forecasting.
Data is experimental or observational (non-experimental); observational data is time series, cross-sectional, pooled or panel, at daily to annual frequencies.

Step-by-step examples

Use this ladder like a staircase: read each setup, write down your own next line, and only then click to reveal the step. The levels get harder as you go, and the ‘your turn’ question at the end of each level checks that you can do it without help.

Easy

Medium

MPC (marginal propensity to consume)0.65Extra government spending (million TL)300

Round 1: the government spends 300; that is someone's income. Round 2: they spend MPC × 300 = 195 and save the rest. Round 3: the next person spends 0.65 of that, 127, and so on. Each round is smaller, so the total stops growing at 300 × 1/(1 − 0.65) = 857 (with the multiplier rounded to 2.86 first, as in the exam: 858). A bigger MPC means less leaks into saving each round, so the multiplier is bigger.