Economics studies choices under constraints. Its models can expose a tradeoff or likely response; they cannot turn one statistic into a certain forecast. Ask what changed, for whom, over what period, and compared with what alternative.
🎙️ Published & recorded: ·
An incentive changes the cost or benefit of an action. People do not always respond, and different people respond differently, but ignoring incentives is usually a bad starting point. The useful question is marginal: what does one more hour, dollar, worker, or unit change?
Saturday choice:
paid shift: $120 after travel costs
certification: expected future benefit, uncertain
family time: valuable, not priced
Opportunity cost of choosing the shift
= value of the best forgone alternative
≠ sum of every alternative
≠ merely the $15 bus fare
A demand curve holds other relevant factors fixed and asks how much buyers would choose at each price. A supply curve asks how much sellers would offer. Their intersection is a model of market clearing, not a command that reality must obey. Search costs, contracts, inventories, queues, bargaining, and regulation can delay or reshape adjustment.
Change Curve movement Typical pressure
buyers' income rises demand shifts price and quantity may rise
input cost rises supply shifts price up, quantity down
own price falls along demand quantity demanded rises
technology lowers cost supply shifts price down, quantity up
# “Typical” means other relevant conditions held fixed.
Price elasticity of demand compares the percentage change in quantity demanded with the percentage change in price. Absolute elasticity above one is called elastic; below one, inelastic. Time horizon, available substitutes, budget share, necessity, and how narrowly the market is defined can all change the estimate.
Price rises from $10 to $11
Quantity falls from 1,000 to 850
Simple percentage estimate:
%ΔQ = -15%
%ΔP = +10%
elasticity ≈ -1.5
Midpoint method reduces dependence on direction:
%ΔQ = (850-1000) / ((850+1000)/2)
%ΔP = (11-10) / ((11+10)/2)
Do not infer elasticity from two observations unless you can address what else changed. A shop raises coffee prices during a festival and sells more cups. That does not show an upward-sloping demand curve; the crowd shifted demand while price moved.
Inflation means a broad price index is rising, not that every price rises. Disinflation means inflation slows while the price level still increases. Deflation means the broad index falls. A consumer price index follows a specified basket and method; your personal cost change can differ because your spending weights differ.
Index in June 2025: 120.0
Index in June 2026: 123.6
Year-over-year inflation = (123.6 / 120.0 - 1) × 100 = 3.0%
Next year inflation falls to 1.0%:
prices are still rising, just more slowly
the index does not return automatically to 120.0
An interest rate connects money now with money later. Borrowing rates can include expected inflation, credit risk, term risk, liquidity, administrative cost, and profit. A central bank controls or targets some short-term rates and influences broader financial conditions, but it does not mechanically set every mortgage, bond, or card rate.
Real rate approximation:
real ≈ nominal - expected inflation
Nominal loan rate: 6%
Expected inflation: 2%
Approximate real rate: 4%
Exact Fisher relation:
1 + nominal = (1 + real) × (1 + expected inflation)
Higher rates often restrain rate-sensitive spending, but the timing and size depend on debt structure, expectations, banks, fiscal policy, exchange rates, and shocks. “Rates rose, therefore recession next quarter” is not analysis. It is a forecast pretending to be an identity. Nothing here is investment advice.
The unemployment rate is unemployed people divided by the labor force. Under standard definitions, unemployed people are without work, available, and actively seeking. People outside the labor force are not counted in that denominator. That is why the unemployment rate can fall for good reasons, bad reasons, or a mixture.
Employed: 96 million
Unemployed and seeking: 4 million
Labor force: 100 million
Unemployment rate: 4.0%
If 1 million stop seeking and no one finds work:
Employed: 96 million
Unemployed: 3 million
Labor force: 99 million
Rate: about 3.0%
Labor productivity is commonly output per hour worked. It can rise through better tools, skills, organization, infrastructure, scale, or by shifting activity between sectors. It can also jump temporarily when low-output hours disappear during a downturn. Measure the numerator and denominator before telling a heroic technology story.
Before: 500 units / 100 labor hours = 5.0 units per hour
After: 630 units / 105 labor hours = 6.0 units per hour
Productivity growth = (6.0 / 5.0 - 1) × 100 = 20%
Possible sources:
better machine + less rework + product mix change
Not established by the ratio alone:
which source caused how much
Productivity growth expands the room for higher wages, profits, lower prices, taxes, or shorter hours, but institutions and bargaining affect who receives the gains. It is a capacity story, not an automatic distribution rule.
An externality exists when an action affects a third party and that effect is not fully reflected in the decision-maker's price. Factory smoke can impose health costs; vaccination can reduce transmission risk. The policy question is not merely whether an effect exists, but how large it is and which remedy has the lowest total cost.
Factory's private marginal cost: $40 per unit
Estimated external health cost: $15 per unit
Social marginal cost: $55 per unit
Possible tools:
emissions price | cap | standard | liability | bargaining
Compare:
measurement + enforcement + avoidance + distribution effects
Markets can coordinate badly when there is market power, asymmetric information, externalities, public goods, missing markets, or severe transaction costs. Government responses can also fail through weak information, capture, poor incentives, or administrative cost. Name the mechanism; “the market failed” is too vague to choose a repair.
Comparative advantage example, output per day:
wheat cloth
Region A 10 5
Region B 6 2
A gives up 2 wheat per cloth.
B gives up 3 wheat per cloth.
A has comparative advantage in cloth,
although A also produces more wheat per day.
Trade can raise combined possibilities;
it does not guarantee every person gains.
Trade changes prices, job locations, bargaining power, and supply-chain risk. Aggregate gains can coexist with concentrated losses. Compensation is a separate policy choice, not something comparative advantage delivers by itself. Likewise, tariffs may protect a sector while raising input costs elsewhere and inviting retaliation.
Before interpreting a chart, identify the source, unit, nominal or real basis, seasonal adjustment, frequency, population, revision status, and start date. Levels, percentage changes, percentage points, and annualized rates are not interchangeable. Then ask what comparison would have happened without the policy or shock.
Raw CSV field:
month,value
2026-01,104.2
2026-02,N/A
ValueError: could not convert string to float: 'N/A'
Fix in order:
1. Preserve the downloaded file and source metadata.
2. Read the series notes: is N/A missing, suppressed, or not applicable?
3. Parse N/A as missing, not zero.
4. Choose and disclose drop, imputation, or no calculation.
5. Recalculate and check whether the conclusion changes.
N/A with zero makes the parser quiet and invents an economic collapse. The technically successful chart is now false. Keep missingness visible, check revisions, and state whether growth is month over month, year over year, or annualized.Data-reading checklist:
□ source, release date, vintage, and revision policy
□ unit, price basis, seasonal adjustment, and frequency
□ level versus rate; percent versus percentage points
□ total versus per-capita; mean versus median
□ nominal versus inflation-adjusted
□ correlation versus a credible causal comparison
□ uncertainty interval and plausible alternative explanations
Do not turn a macro relationship into a certain forecast.
Do not turn this page into investment advice.
The habit worth keeping is disciplined modesty. A model strips reality down so one mechanism becomes visible. Put the omitted conditions back before making a claim about next year, a country, a household, or an investment.