The Science of Macroeconomics

1. Classical and Keynesian economics

Adam Smith and the self-correcting market

Macroeconomics grew out of an argument about whether markets fix themselves. In 1776 Adam Smith published An Inquiry into the Nature and Causes of the Wealth of Nations:

• Wealth is a nation’s capacity to produce, not its stock of gold, and the division of labour raises that capacity.
• Invisible hand: self-interest, disciplined by competition, serves society.
• Prices signal scarcity: surpluses push prices down, shortages push them up.
• Efficiency: the market allocates resources without a central planner.

If prices and wages are flexible, every market clears. The economy sits in equilibrium at $P^*$, and unemployment can only be temporary. This is laissez-faire.

The Great Depression: when the market did not fix itself

US real GDP fell by about a quarter between 1929 and 1933, and prices fell about as much. The unemployment rate went from 3.2% in 1929 to 24.9% in 1933 - one worker in four jobless - and was still 14.3% in 1937, about one in six in 1939.

• The classical cure failed: wages and prices did fall, but jobs did not come back.
• Demand was insufficient: firms do not hire to produce what nobody buys.
• Something had to be done: spending and money could fill the demand gap.

The lesson: the economy as a whole can stay below full employment, and policy can change that. This is the birth of modern macroeconomics. (Source: Lebergott 1964, Historical Statistics of the United States, for unemployment; US BEA for real GDP.)

Keynes and the birth of modern macroeconomics

Before the 1930s, economics studied prices and single markets. The Great Depression - mass unemployment - was a shock the old theory could not explain. In 1936 Keynes published The General Theory of Employment, Interest and Money. His central message: aggregate demand sets output and employment in the short run. After Keynes came national accounts (GDP), the IS-LM model (Hicks, 1937) and stabilisation policy.

The two views side by side:

From Smith to the new synthesis

Ideas

The world economy

Five schools in one table

Every school answered a crisis the previous one could not explain: the Depression, stagflation, the 2008 crash.

Markets behind the models

Why macro cares: firms that set their own prices change them rarely. Sticky prices are the core of the New Keynesian model.

Examples: foreign exchange, Borsa İstanbul shares and gold bullion sit close to perfect competition; cafés and restaurants, clothing brands and hairdressers are monopolistic competition. In between: oligopoly (a few large firms - mobile operators, fuel) and monopoly (one seller - the natural-gas grid).

Where macroeconomics stands today

The core is the New Keynesian model: classical in the long run, Keynesian in the short run, expectations at the centre.

• After 2008: banks and credit matter; zero lower bound and quantitative easing.
• After the 2010s: people differ (HANK); who holds debt matters.
• After 2021: supply shocks are back; expectations drive inflation.

2. Long run and short run: growth and business cycles

Trend and cycle

Real GDP, shown in logs, does two things at once: it grows along a trend and it swings around it in a cycle.

Türkiye in the long run: one trend, three breaks

Between 1960 and 2025 Türkiye grew at an average of 4.7% a year (real GDP, constant 2015 US dollars, log scale; World Bank WDI). The trend itself changed at three structural breaks:

• 1980 - opening up to the world economy.
• 2001 - crisis and reforms.
• 2018 - currency shock.

Trend growth per period: 5.1% (1960-1980), 3.9% (1980-2001), 5.8% (2001-2018), 4.6% (2018-2025). Crises (1994, 2009) are dips; breaks are regime changes. The slowest era was 1980-2001, the crisis decades.

Year by year the picture is growth with frequent interruptions: average growth 1961-2025 is 4.8% (the dashed line), with six years of contraction, mostly crises. Growth theory explains the line; cycle theory explains the bars.

The output gap separates the two horizons directly: log real GDP minus its Hodrick-Prescott trend (times 100). Türkiye ran above trend around 1980-83 and in booms, and below trend in 2001, 2009 and 2020. Last years are revised as new data arrive.

3. Micro and macro: is there still a divide?

Then and now

Method: converged. Questions: still distinct; macro asks what happens when everyone acts at once.

Why aggregation still matters

• Fallacy of composition: one family saving more gets richer; if everyone saves more, spending and incomes fall - the paradox of thrift.
• People differ: a rate cut helps borrowers but hurts savers; who holds debt shapes policy effects (HANK).
• Some firms are giants: shocks to large firms do not average out; Gabaix (2011) estimates they explain about a third of US output volatility.

Macro needs micro, but cannot be reduced to it.

4. Globalisation: multinationals and global supply chains

World growth has slowed

• 1960s: world real GDP growth averaged 5.3% a year.
• Since the 1980s: about 3% a year.
• Only two years of world recession: 2009 and 2020.

Bars show annual growth, lines show decade averages (World Bank WDI).

The world became one production line

World trade (exports plus imports of goods and services) rose from 26% of world GDP in 1970 to 60% in 2008. The WTO was founded in 1995 and China joined it in 2001; trade collapsed in COVID-19; since 2008 the ratio has been flat - “slowbalisation”. About half of world trade runs through global value chains (World Bank, 2020).

Multinationals and value chains: who makes an iPhone?

An iPhone 17 Pro Max sold in the US in 2025 cost 1,199 dollars. Parts cost about a third of the price (410 dollars: cameras 110, A19 Pro chip 60, other parts 240). Everything else - 790 dollars, 66% - covers assembly, R&D, marketing, retail and Apple’s profit. The value split: Apple’s profit 58.5%, materials 21.9%, suppliers’ profits 14.5%, other labour 3.5%, assembly labour just 1.8%. Design, software and brand capture the value. (Sources: Counterpoint Research via PhoneArena 2026; TechInsights 2025; Kraemer, Linden & Dedrick 2011.)

When the chain breaks

• Lesson 1: supply shocks push prices up and output down; a hard trade-off for central banks (2021-23).
• Lesson 2: firms trade efficiency for resilience; near-shoring - Türkiye may gain.

Apple and Türkiye, side by side

• Apple: market value 4.9 trillion dollars (2 October 2026); revenue 416 billion dollars (fiscal 2025); net profit 112 billion dollars. Market value = share price x number of shares: what investors pay today for all of Apple’s future profits.
• Türkiye: GDP 1.6 trillion dollars in 2025 (63 trillion lira), 18,040 dollars per person, some 86 million people. GDP = the market value of all final goods and services produced in one year.

The headline: Apple’s market value is about 3 times Türkiye’s annual GDP. The economist’s caveat: a stock is not a flow, so compare like with like - Apple’s yearly revenue is about a quarter of Türkiye’s GDP. Revenue also overstates value added: it includes inputs bought from others. (Sources: companiesmarketcap.com; Apple 10-K; TÜK; Ministry of Treasury and Finance 2026.)

Crises travel: 2008 and the energy shocks

The 2008 US mortgage crisis became a world recession: mortgage losses hit banks in Europe, money fled emerging markets, trade orders collapsed and confidence stopped spending everywhere. In 2009 world exports fell 9.6%, world GDP 1.3% and Türkiye’s GDP 4.9%.

Energy prices spread inflation across the world. Brent crude peaked at 118 dollars per barrel in June 2022 (with another spike in 2026); consumer inflation reached 10.6% in the euro area and 9.0% in the US. The chain: energy prices rise, then transport and production costs, then food and goods prices, then wages and expectations. Oil peaked first; inflation followed months later. Türkiye imports its energy, so a weaker lira adds a second push. (Source: FRED; the 2022 window is the war in Ukraine.)

5. The toolbox: theory, model, data

A loop, not a line

Macroeconomists work in a loop: theory is formalised into a model, the model is estimated and tested against data, and the result revises the theory. Each needs the others:

• Theory without data is an opinion.
• Data without theory is a pattern without a cause.
• The model is the bridge that makes the test possible.

Every step uses positive statements: what is, not what should be. Value judgements come later, in policy.

Theory: a causal story that answers a question

Watch the theory step on the question “Why are prices in Türkiye rising so fast?”

1. Assumptions - what we take as given: real output and the velocity of money change slowly.
2. Mechanism - why one thing causes another: more money chases the same amount of goods.
3. Prediction - what should happen, and can be checked: money grows faster than output, so prices rise.

A good theory is consistent, simple and falsifiable: data could prove it wrong. This one is the quantity theory of money (Hume, 1752; Friedman, 1956).

Model: the theory written as an equation

With velocity stable, the exogenous variables are money growth ($\mu$), output growth ($g$) and $V$; the endogenous variable is inflation ($\pi$). The model makes the theory precise, gives a number we can check, and is deliberately simple: a map, not the territory.

Data: does Türkiye fit the model?

Plotting Türkiye’s inflation against money growth minus output growth, 1970-2025 (World Bank WDI), gives a correlation of 0.87 over 56 years:

• Supported: over decades, inflation follows $\mu - g$.
• Not one for one: 2022 sits far above the line.
• Something the model left out: the lira, energy, expectations.

The loop turns: the revised theory adds exchange-rate pass-through and expectations, giving a new model to test - $\pi = \mu - g + \text{pass-through} + \text{expectations}$.

A century of inflation in one series

Türkiye’s annual inflation, 1924-2025: 1942: 96.0% (war economy, deficit finance), 1994: 106.5%, 2022: 96.1%, and a century average of 23.9%. The 1970s-1990s were chronic high inflation with fiscal dominance; 2004-2016 near single digits was the exception; 2022 was back at the 1942 and 1994 peaks. The latest reading in the lecture is 31.5% y/y, August 2026 (TÜK). (Sources: 1924-2023 lecturer’s long-run series; 2024-2025 GDP deflator, World Bank WDI, July 2026.)

The lecture’s thinking exercise runs the loop yourself: pick a story (money, the lira, energy, wages or expectations), write it as an equation, choose the TÜK or EVDS series that would test it, and decide what result would prove you wrong.

How economists think

The method has two steps: 1. develop models that explain some part of the world; 2. test them with data: how closely does the model match what we observe? Economists observe the world, theorise (equations or diagrams) and test against the data - empiricism - with a bit of art: simple enough to work with, realistic enough to be useful.

Five habits of economic thinking:

1. Opportunity cost - what did you give up to be here this hour?
2. The margin - is one more hour of study worth it tonight?
3. Incentives - who gains, and who changes behaviour, if this rule changes?
4. And then what? - what is seen, and what is not seen (Bastiat, 1850).
5. Ceteris paribus - what am I holding constant, and is it really constant?

Why and how: why asks for the mechanism - what causes what, through which channel; how asks for the method - how would I know, and what evidence would convince a sceptic? “A why without a how is an opinion. A how without a why is just a regression.”

Data labs for the homework - download one series, draw one graph, write down one question it raises:

6. Toolbox at work: how has Türkiye grown?

The case study runs the full loop in seven steps: growth dynamics of the Turkish economy, per capita income growth, comparative analysis, explanatory variables, theoretical model, variables of the model, data set.

Per capita income

GDP per capita (constant 2015 US dollars) rose from 2,449 dollars in 1960 to 15,883 dollars in 2025. The years of decline, circled on the slide, are 1980, 1994, 2001 and 2009.

A comparison changes the question

In 2025 Türkiye produced 15.9 thousand dollars per person, Korea 37.5 thousand and China 13.8 thousand. But in 1960 the average Turk was more than twice as rich as the average Korean, and ten times as rich as the average Chinese. Korea overtook Türkiye in 1979; by 2025 it was 2.4 times richer per person, and China has reached about 87% of Türkiye’s level. The real question is not “did Türkiye grow?” but “why did it grow more slowly than countries that started far behind it?”

Theory and model: growth from inside

Prediction: economies that invest more in people and ideas should grow faster, for longer.

Model prediction: more resources in research ($s_R$) mean faster long-run growth ($g_y$). Now it can be tested.

The empirical model and the test

A simple regression turns the model into data work: $g_i = \beta_0 + \beta_1 RD_i + \varepsilon_i$.

• $g_i$: average growth of income per person in country $i$, 2000-2019.
• $RD_i$: R&D spending, % of GDP - a proxy for $s_R$.
• $\beta_1$: how much growth changes when R&D rises by one point.
• $\varepsilon_i$: everything else - institutions, luck, initial income.

The hypothesis: theory says $\beta_1 \gt 0$; the test is against $H_0: \beta_1 = 0$. First try, 114 countries: $\beta_1 = -0.57$ - rich countries do more R&D but grow more slowly. Control for initial income: $\beta_1 = +0.16$, not significant. (Data: World Bank WDI, 2000-2019, OLS; lecturer’s calculation.)

The data: growth by accumulation, not efficiency

Total factor productivity (TFP) by 2023 with 1974 = 1: China 4.01, Korea 1.69, Germany 1.46, US 1.31, Poland 1.26, Chile 0.98, Türkiye 0.76, Brazil 0.64. Korea’s TFP was about 1.7 times its 1974 level; Türkiye’s was about 0.76. The puzzle: if productivity fell, how did income per person still triple? What had to grow instead?

What grew instead:

• Physical capital per worker by 2023 with 1974 = 1: China 56.4, Korea 16.7, Poland 8.2, Türkiye 7.4, Chile 2.8, Germany 2.2, Brazil 2.2, US 1.8.
• Schooling: Türkiye raised average years of schooling of adults from 2 to 8 years (Korea leads with about 14).

Türkiye multiplied capital per worker by 7.4 and raised schooling from 2 to 8 years, yet TFP fell: growth by accumulation, not by efficiency. (Source: Penn World Table 11.0 via FRED, 1974-2023.)

Quality, not just quantity: PISA 2025 maths scores of 15-year-olds put Türkiye at 462, just below the OECD average of 463 - years in school are not learning. (Source: OECD, PISA 2025 Results Volume I, September 2026; China means Beijing, Shanghai, Jiangsu and Zhejiang only.)

7. Testing theories: correlation and causality

Models and hypotheses

“All models are wrong. Some are useful.” A model is a simplified description of reality; the scientific method puts it to the test. A paper plane is a model of a plane: it really flies, but it carries no passengers. Hypotheses are the predictions the model makes, which data can test; empirical evidence verifies or falsifies them.

Correlation

Correlation is when two things are related: positive (both change in the same direction) or negative (they change in opposite directions). Correlation suggests that causation may be possible, and that further investigation is warranted - it does not prove it.

Causation in demand: from P to Y

The aggregate demand (AD) curve slopes down because the price level is the cause:

• Wealth effect: a lower $P$ raises the real value of money and savings, so consumption rises.
• Interest-rate effect: a lower $P$ means less money is needed for transactions, the interest rate falls and investment rises.
• Exchange-rate effect: a lower $P$ makes our goods cheaper than foreign ones, so net exports rise.

So causation runs from $P$ to $Y$: a higher price level cuts spending. $P$ sits on the vertical axis by convention, but in AD it is the cause.

Causation in supply: from Y to P

The aggregate supply (AS) curve slopes up because output is the cause:

• Labour costs: producing more means more workers and overtime, wages rise and so does $P$.
• Capacity: machines run near the limit, marginal cost rises and so does $P$.
• Pricing power: strong sales let price setters raise mark-ups and so does $P$.

So causation runs from $Y$ to $P$: producing more raises costs and prices. Where the two curves meet, both causal stories hold at once - that is equilibrium.

Three ideas to take home

1. Born from a crisis: Keynes answered the Great Depression; each school since has answered the crisis before it.
2. Two horizons: the long run is about growth and the trend; the short run is about cycles and the output gap.
3. A loop, not a line: theory, model, data, revise - and correlation is not causation.

Also from the earlier lecture notes

The new deck does not cover these topics, which are still correct and useful.

Four sectors, three markets. The national economy can be divided into four groups of participants:

They meet in three markets:

1. Goods and services market: households and government buy from firms; firms buy from each other; the rest of the world buys (exports) and sells (imports).
2. Labour market: households supply labour; firms and government demand it; labour also crosses borders.
3. Money (asset, financial) market: households supply funds and borrow; firms borrow to build new capacity; government borrows by issuing bonds; the rest of the world lends and borrows too. Financial institutions coordinate most of this lending.

The circular flow ties these together: spending on goods becomes firms’ revenue, which is paid out as wages, rent and profit, which becomes household income, which is spent again.

Positive versus normative. Positive economics describes what is: causes and effects. Normative economics says what ought to be and rests on a value judgement. This fits the new deck’s rule: every step of the theory-model-data loop uses positive statements; value judgements come later, in policy.

Exam practice

Summary and review

• Macroeconomics was born from the Great Depression: markets did not self-correct, and aggregate demand sets output and employment in the short run.
• Classical view: flexible prices, vertical AS, Say’s law, money neutral, laissez-faire. Keynesian view: sticky prices, flat AS, demand sets output, the paradox of thrift, active policy. Today: the New Keynesian synthesis - classical in the long run, Keynesian in the short run, expectations at the centre.
• Two horizons: the trend and potential output (long run, growth theory) versus the cycle and the output gap (short run). Türkiye grew 4.7% a year on average over 1960-2025, with breaks in 1980, 2001 and 2018.
• Micro and macro: the Lucas critique (1976), microfoundations, the fallacy of composition, HANK and giant-firm shocks (Gabaix 2011). Macro needs micro but cannot be reduced to it.
• Globalisation: trade rose from 26% of world GDP (1970) to 60% (2008), then “slowbalisation”; half of trade runs through global value chains, and the shocks of 2011, 2020, 2021 and 2024 showed the chain is efficient but fragile.
• The toolbox is a loop: theory, model, data, revise. The quantity model predicts inflation = money growth - output growth; Türkiye’s correlation is 0.87 over 56 years, but 2022 needed the lira, energy and expectations.
• Toolbox at work: Türkiye’s income per person rose from 2,449 to 15,883 dollars (a factor of 6.5) by accumulation - capital per worker times 7.4 and schooling from 2 to 8 years - while TFP fell to 0.76 (1974 = 1).
• Correlation is not causation: in AD the cause runs from P to Y (wealth, interest-rate, exchange-rate effects); in AS from Y to P (labour costs, capacity, pricing power).
• From the earlier notes: four sectors and three markets; positive versus normative statements.