International Finance

Session 4 — Securities Markets: Stocks, Mortgages, Currencies and Derivatives

Laura Contreras-Portela

KEDGE Business School

October 2026

Two bonds, one required return

1. What is the coupon yield on a bond with face value $1,000, current price $1,043.70 and a coupon rate of 8.25%?
2. What is the price of a three-year, 6% coupon bond paying semi-annually, face value $1,000, if the required return is 8%?
3. One of these trades above par and the other below. Which, and why?

Same machinery as the problem we closed with last session, new numbers, and a third part that is the whole point. Ten minutes.

Give them five minutes in pairs before anything is said out loud. Then take the parts in order.

Part one is the easy one. The coupon is eight point two five percent of a thousand, so eighty-two fifty a year. Divide by what you actually paid, which is one thousand and forty-three seventy.

[ask] Before we divide - will the coupon yield be above or below eight point two five percent?

Below, because the denominator is bigger than the face value. Seven point nine zero percent.

Part two. Three years, six percent, paid semi-annually, market wants eight.

[ask] Three changes to the formula. What are they?

Halve the coupon - thirty, not sixty. Halve the rate - four percent per period. Double the periods - six, not three. All three, every time.

Nine forty-seven fifty-eight.

[pause]

Now part three, and this is the one to spend time on.

[ask] The first bond costs more than a thousand. The second costs less. What is different about them?

The first pays a coupon of eight point two five when the market is content with less, so people bid the price up above face value until the return falls into line. The second pays six when the market wants eight, so the price has to fall below face value to make up the difference.

Write this down, because it is a one-line exam answer: the price moves until the yield matches what the market requires. Premium means the coupon is generous; discount means the coupon is mean. Nothing else is going on.

Worked. Coupon yield: 82.50 / 1043.70 = 7.90%. Price: six periods, coupon 30, 4% per period - 30 x [1 - 1.04^-6]/0.04 + 1000 / 1.04^6 = 157.26 + 790.31 = 947.58. First trades at a premium (coupon 8.25% > required), second at a discount (coupon 6% < required 8%).

Where we are

One chapter, two sessions. Last session established the money market and the bond market: the instruments, the rate conventions, and how a price and a yield are two statements of the same thing.

Today, in order:

3. Stock markets — common and preferred shares, ADRs
4. Mortgage markets — fixed-rate, adjustable-rate and the mortgage-backed securities built from them

5. Foreign exchange — spot and forward contracts on bank deposits
6. Derivatives — forwards, futures, options, swaps, credit derivatives
7. Reading discussion — Sartzetakis (2021)

The same two questions apply to every market we open today: what contractual claim does the instrument represent, and what determines its price.

This is a long session, so give them the shape of it before you start.

Last time: money markets and bonds. Short-term paper, then long-term debt, and in both cases a price and a yield that are the same statement said two ways.

Today: everything else in the chapter. Stocks, mortgages, currencies, derivatives, and then the green bonds paper.

Keep putting the same two questions to them on every market. What is the claim? What determines the price? Six markets, two questions. That is the chapter and that is the exam.

Timing. If you are running behind, the slides marked cuttable in the notes go first, in this order: common/preferred/foreign, the mortgage types table, the appreciation slide, and long/short in derivatives. Do not cut the Gordon growth slides, the payoff diagram, or the reading discussion.

If you are resuming mid-chapter, the remaining blocks run about 90 minutes: rate conversion and 2.7b 15 · foreign exchange 18 · derivatives 35 · the Sartzetakis discussion 20. Put the reading discussion before the break rather than last - it is the assessed piece, and leaving it at the end is what puts it at risk when derivatives overruns.

3 · The stock market

The residual claim

A bondholder has a contractual claim to specified payments. A shareholder owns a residual claim: whatever remains after the firm’s obligations to creditors and other senior claimants have been satisfied.

• Equity returns come from dividends and capital appreciation.
• Shareholders are residual claimants on the firm’s assets and cash flows. They also typically hold voting rights, allowing them to influence major corporate decisions.
• Because shareholders have limited liability, their payoff at liquidation cannot fall below zero, while their potential upside is unlimited.
• This payoff structure can create conflicts between shareholders and creditors, particularly over risk-taking.
• Separately, because shareholders delegate day-to-day decisions to managers, an agency problem arises: managers may pursue objectives that differ from those of shareholders.

Here is the difference between what we have been doing for two sessions and what comes next.

A bondholder is owed a number. It is written in the contract. If the firm can pay, they receive exactly that and not a cent more.

A shareholder owns what is left after everyone else has been paid. Which can be a great deal, or can be nothing at all.

[ask] A company is about to be liquidated. Which claim would you rather hold?

[ask] Now a company about to triple in size. Which one now?

[reveal the bullets one at a time]

On the first two: returns come from dividends and from the price rising, and the vote comes attached. Keep the vote in mind - it is the only one of these instruments that carries any control over the firm.

[pause on the third bullet, because it is the one that matters]

Limited liability means the worst case is that your shares are worth zero. You do not owe the firm’s creditors anything beyond what you put in. So the downside stops, and the upside does not.

Draw that asymmetry in the air with your hand, because they will meet it again in an hour as the payoff diagram of a call option. It is the same shape.

[fourth bullet]

Now the consequence, and put it to them as a question.

[ask] You own shares in a firm that is close to default. The managers can take a risky project: it either saves the firm or finishes it. Do you want them to take it?

Yes - because you are nearly wiped out anyway, so you have very little left to lose and everything to gain. And the creditors, who would have recovered something in an orderly liquidation, want the opposite. That is the shareholder-creditor conflict, and it is sharpest exactly when the firm is in trouble.

[fifth bullet]

A separate problem, and do not let them merge the two. Shareholders own the firm but do not run it. Managers do. Managers have their own objectives - size, security, their own pay - and those do not automatically coincide with the shareholders’. That is the agency problem, and you will meet it properly in corporate finance.

3 · The instruments: common, preferred, foreign

Common stock — the right to vote for directors and on certain issues, and to receive dividends if the board declares them.

Preferred stock — usually no vote, but a fixed dividend paid before common. Closer to a bond than most students expect.

Foreign stocks — shares in a firm listed abroad. Useful for diversification, but awkward to buy when they are not listed locally.

• American depository receipts (ADRs) are the usual way to hold one: a bank buys the foreign shares and issues receipts against them that trade in dollars on a U.S. exchange.

Common stock is what you picture: a vote, and a dividend if the board declares one.

Preferred stock is the interesting one. Usually no vote. Fixed dividend. Paid before common shareholders.

[ask] Fixed payment, paid ahead of others, no control. Is that equity, or is that debt?

The honest answer is that it sits between the two, and that is exactly why accountants and rating agencies treat it inconsistently. If you see it in a balance sheet, look at how it has been classified — it is not always obvious.

On the right, the third category: foreign stocks. A share in a firm listed somewhere else. Good for diversification, awkward to buy - different exchange, different currency, sometimes different rules about who may hold it.

[point at the sub-item]

So an ADR is not a fourth kind of instrument. It is a way of holding a foreign stock. If you want a Brazilian company but you can only trade in New York, a bank buys the Brazilian shares, holds them, and issues receipts against them that trade in dollars on a US exchange. What you own is still the Brazilian company.

[pause]

If that sounds familiar, it should. It is the same engineering as the eurodollar deposit from last session: take a claim, repackage it so it settles in the currency and the legal system the buyer is comfortable with.

Close by saying where this goes next: we have the instruments, so the next question is where they actually change hands.

3 · Where the trading happens

Organised exchanges

• NYSE trades roughly 4 to 7 billion shares a day.
• “Organised” once meant a physical floor; electronic networks have largely replaced that idea.
• Listing requirements exclude small firms.
• Auction market with floor specialists; around 25% of trades filled directly by the specialist.

Over the counter

• NASDAQ is the obvious example, with roughly 3,000 securities listed.
• Dealers stand ready to make a market — without one, the equity would be hard to trade at all.
• Multiple market makers post bid and ask prices for the same security.

Dealers quote a bid-ask spread and profit from it. Competition constrains that spread — but a spread of zero would not be sustainable either. What costs must the spread compensate the dealer for, and what happens to liquidity when those costs rise?

Two ways to organise a market.

Exchanges on the left — one central place where orders meet. The New York Stock Exchange still uses the word organised, but the trading floor you have seen in films is mostly theatre now. It is computers.

Over the counter on the right. No central place. Dealers hold inventory and quote you a price to buy and a price to sell.

[ask] The dealer buys from you at the bid and sells to someone else at the ask, and keeps the difference. So why does competition not push that spread to zero?

[wait]

Two reasons, and they are both real costs.

First, the dealer is holding inventory. They bought your shares and now they are exposed until they sell them.

Second — and this is the one worth thinking about — some of the people trading with the dealer know something the dealer does not. If you are selling to me because you know bad news is coming, I lose. I cannot tell who those people are, so I widen the spread against everyone.

[ask] Now, in a crisis, when nobody knows what anything is worth. What happens to that second cost, and what happens to the spread?

It widens - and that is the whole point, because a wider spread is less liquidity. The spread is what a round trip costs you, so when it doubles, getting out has become twice as expensive. Then they cut the size they will quote, and in the limit they stop quoting at all. That is what “the market stopped” means: not that prices fell, but that there were no prices.

Now make the link explicit, because it is the same mechanism as the BNP Paribas story we opened with last session.

Nobody could value the mortgage-related securities those funds held. So the adverse-selection cost went through the roof - if you offer to sell me one, perhaps you know your pool is the bad one. Dealers widened, then withdrew. With no price, the funds could not value their own holdings, and a fund that cannot compute what it is worth cannot let investors redeem. Hence the suspension.

And then it spreads, which is the part to leave them with: the same securities were collateral for short-term borrowing. Collateral that cannot be priced is collateral nobody will lend against, so funding dries up, which forces sales, which pushes prices down further.

The sentence to land: liquidity is not a property of an asset. It is a service somebody chooses to provide, and they stop providing it exactly when everybody needs it.

3 · Price-weighted versus value-weighted

An index is a measuring instrument, not an asset: one number tracking a set of securities over time. The DJIA is price-weighted across 30 stocks; the S&P 500 is value-weighted across 500.

A USD 400 share moves the Dow thirteen times as much as a USD 30 one, whatever the two firms are worth. So what is the Dow measuring?

Start with what an index is for, because they have heard the word for years without ever being told.

An index is not something you buy. It is what you watch. When the news says the market fell two percent, what fell was an index - one number built to monitor a set of securities so you can follow it over time and compare it with other markets.

[ask] If it is a measuring instrument, what is it measuring?

That is the whole slide. Two indices of the same market, two different answers on the same day.

[pause]

Two indices, two completely different ideas.

The S&P 500 weights by market value. If a company is worth two percent of the market, it is two percent of the index. That is what you would own if you bought the whole market.

The Dow weights by price per share. Not by size. By the number printed on the share.

[ask] So a four-hundred-dollar share and a thirty-dollar share. Which one moves the Dow more?

The four-hundred-dollar one - about thirteen times more - regardless of whether the companies are the same size or one is a hundred times bigger.

Now the step they usually miss, so take it slowly.

[ask] What decides whether a share costs four hundred dollars or thirty?

Not how big the company is. Price per share is market value divided by the number of shares the firm chose to issue. Two firms worth exactly the same can trade at four hundred and at four, purely because one issued ten times as many shares. The price per share carries no information about size.

[pause]

Which is why the split is damaging. Two-for-one: each share becomes two, each worth half. Same pie, twice as many slices. Same firm, same assets, same market value - and its weight in the Dow has halved.

Then do the contrast, because it is what proves the point: under value weighting the split changes nothing. Half the price, twice the shares, same market value, same weight. A sound measuring instrument does not move when the thing it measures has not moved.

If someone asks whether the index jumps on the split day: no, the divisor is adjusted so the level is continuous. What that cannot fix is the weight.

[ask] So what is the Dow measuring?

Push them to a real answer rather than “nothing”. It measures the return on a portfolio holding one share of each of the thirty companies. That is a genuine portfolio - it is simply not one anybody would hold on purpose, because there is no economic reason to own equal numbers of shares across firms.

Contrast it with the S&P: when that moves two percent, the combined market value of five hundred firms moved two percent. That is a quantity you can interpret.

[ask] So why does every news programme still quote the Dow?

Because it dates from 1896 and people recognise the number. That is a reason from journalism, not from finance.

3 · What a share is worth

Three formulas, three jobs: value the share from forecasts, estimate one of the inputs, or value it without forecasting at all.

1 · Value it. The Gordon growth model — the dividend discount model with dividends growing at a constant rate g forever, and k_e > g:

P = \sum_{t=1}^{\infty}\frac{Div_t}{(1+k_e)^t} = \frac{Div_1}{k_e - g}

2 · Estimate g. Geometric growth, from two observations n periods apart:

g = \left(\frac{\text{final}}{\text{initial}}\right)^{1/n} - 1

3 · Skip the forecast. Price-to-earnings, a relative measure:

PE = \frac{\text{Price}}{\text{Earnings}}

The denominator k_e - g governs everything — and it is a small difference between two uncertain numbers. As g \to k_e the price explodes, so a half-point revision moves the valuation enormously. Not a defect of the model, but a property of perpetual growth.

Say the framing line first, because otherwise this looks like three unrelated formulas dumped on one slide.

Three formulas, three jobs. The first values a share from your forecasts. The second estimates one of the numbers the first one needs. The third values it without forecasting anything at all.

Then take them in that order.

A share pays you dividends, forever. So its value is the present value of all of them. Write that down and you get the sum on the left.

If the dividend grows at a constant rate g, that infinite sum collapses to the expression on the right. Next year’s dividend, over the required return minus the growth rate. That is the Gordon growth model, and you have met growing perpetuities before in corporate finance.

Two conditions, and they are not fine print. The growth rate has to be constant forever. And k has to be bigger than g.

[ask] Why must k be bigger than g? What is wrong with a company that grows faster than the required return, forever?

[wait]

Eventually it is bigger than the entire economy. That is your sanity check — if you ever get g above k in the exam, you have made an arithmetic error, not discovered a fantastic company.

Underneath: the PE ratio. Price over earnings. A market number on top, an accounting number on the bottom.

[ask] A company has a PE of forty. Is it expensive, or is it expected to grow?

Both answers are right, and you cannot tell which from the ratio. That is the limitation of every multiple.

3 · The denominator does all the work

Before anything, tell them what is held fixed, or the picture means nothing.

Same company throughout. Same dividend of two next year. One base case: the market wants ten percent, the dividend grows at five, so the price is forty. That is the dot, and it is the same dot in both panels.

[point at the left panel]

Here I move only the required return and leave growth at five. Walk left from the dot and the price climbs: ten to nine to eight percent, forty becomes fifty becomes sixty-seven.

[point at the right panel]

Here I move only the growth rate and leave the required return at ten. Walk right from the dot and exactly the same thing happens.

[point at the dashed lines]

Those mark where k equals g. In both panels the curve goes vertical there, because the denominator has gone to zero.

[pause]

This is not a flaw in the model. It is what perpetual growth means. But look at what it implies.

Near that edge, change your growth assumption by half a percentage point and the valuation changes enormously.

[ask] So two analysts, same company, almost identical assumptions. How far apart can their target prices be?

Very far. And that is why high-growth companies reprice so violently on news that sounds small. Nothing happened to this year’s earnings. Something happened to the assumption in the denominator.

You have watched this happen — in 2000, and again in 2021 and 2022.

3 · Problem 2.5

1. A stock is expected to pay a dividend of $0.16 and to be worth $60 in a year. At a 12% discount rate, what is it worth now?
2. If the industry PE ratio is 16 and the firm earns $1.13 per share, what is the price?

Two parts, one line each. But they are two different methods, and that is the point.

Part one. You expect a dividend of sixteen cents and a price of sixty in a year, and you discount at twelve percent. Sixty point one six over one point one two. Fifty-three seventy-one.

Part two. The industry trades at sixteen times earnings, this firm earns a dollar thirteen. Sixteen times one thirteen. Eighteen zero eight.

[pause]

Part one is a discounted cash flow. You forecast, then you discount.

Part two is relative valuation. You assume the market has priced similar companies correctly, and you apply their multiple.

[ask] Which one do you trust more?

That argument is most of what equity analysts do for a living. Notice that the first needs you to forecast, and the second needs the market to be right.

Worked. (0.16 + 60) / 1.12 = 53.71. And 16 x 1.13 = 18.08.

3 · Problem 2.6

The DJIA was 10,000 in 2010 and 20,000 in 2017. What was the annual geometric growth rate?

Ten thousand in 2010, twenty thousand in 2017. What is the annual growth rate?

[ask] Careful. How many years is that?

Seven, not eight. Count the gaps, not the dates — that is the most common error on this question.

So it is two to the power of one seventh, minus one. Ten point four one percent.

Notice the shape: the index doubled in seven years at about ten percent a year. That is a useful number to carry around. Roughly seven years to double at ten percent.

Worked. g = (20000/10000)^(1/7) - 1 = 2^(1/7) - 1 = 10.41%.

4 · Mortgage markets

The market that ate the world

By 2006 an American mortgage was no longer a loan a bank made and held. It was an input: pooled, tranched, rated, and sold on to a German Landesbank that had never seen Ohio.

• A mortgage is a long-term loan secured by real estate.
• It is amortised: one fixed payment covers both interest and principal every month, so the balance falls to zero at maturity.
• The security interest in the property is what makes the loan cheap. It is also what links the credit market to the housing market.

Set the scene before any definitions, because they already half know this story and it will carry the section.

A mortgage is the most boring instrument in the chapter. A household borrows, the house is the collateral, they pay a fixed amount every month for thirty years. Nothing clever.

What made it interesting was what the banks did with it afterwards. By two thousand and six, the bank that made the loan did not keep it. It sold it. It got pooled with hundreds of others, sliced up, given a rating, and ended up on the balance sheet of a regional bank in Germany that had never been to Ohio and could not have found it on a map.

[pause]

Two definitions before we go on. Secured, which means if you stop paying, they take the house - and that is why mortgage rates are lower than credit card rates. And amortised, which means each payment does two jobs: it pays the interest and it pays down some of the principal, so after thirty years you owe nothing.

Hold on to the last line. The collateral is a house. So the credit market and the housing market are not two markets. They are one.

4 · The instruments: eight kinds of mortgage

Type	What it is
Conventional	Not guaranteed; usually needs private mortgage insurance and 5-20% down
Insured	Guaranteed by a government agency; low or zero down payment
Fixed-rate	The rate is fixed for the life of the mortgage
Adjustable-rate (ARM)	Rate tied to another security, adjusted periodically within annual limits
Graduated-payment (GPM)	Low initial payment rising each year; amortises in 30 years
Growing-equity (GEM)	Payment rises each year; amortises in less than 30 years
Second mortgage	Secured by a second lien; used for credit lines and home improvement
Reverse annuity (RAM)	The lender pays the borrower monthly; due when the property is sold

Which of these transfer interest rate risk from the lender to the borrower, and which households tend to be sold those contracts?

[cuttable - if you are behind, show it for thirty seconds and move on]

Do not read this table out. It is a reference slide, and they have it.

Point at two rows only.

Fixed-rate. The lender is locked in for thirty years. If rates rise, the lender is stuck holding a loan that pays less than the market. The lender carries that risk.

Adjustable-rate. The rate resets. If rates rise, the borrower’s payment rises. The borrower carries that risk.

[ask] So which of those two is cheaper on the day you sign, and why?

The adjustable one, because the lender has handed the risk to you and is giving you a discount for taking it.

[ask] Now the uncomfortable half of the question. Who gets sold the adjustable-rate contract with the low opening payment?

The households least able to absorb a payment that doubles. That is not an accident of the product, it is how it was marketed, and it matters for what happened in 2007.

4 · Securitisation, in four steps

1. The secondary mortgage market was created by the federal government after the Second World War, when it set up Fannie Mae. It grew quickly from the mid-1980s.
2. Individual mortgages default unpredictably, so they are pooled — hundreds at a time — on the argument that diversification tames the risk.
3. Rights to the pool’s cash flows are sold as separate securities: pass-throughs, or mortgage-backed securities.
4. The borrower’s monthly payment passes through a trustee before reaching the investors.

Step 2 is load-bearing. Diversification removes risk that is independent across loans. What becomes of a pool of American mortgages if house prices fall everywhere at once?

This is the hinge of the whole session. Slow down here.

Four steps, and reveal them one at a time.

[reveal 1] The government built this market. Fannie Mae, after the war, to get mortgages flowing.

[reveal 2] One mortgage is unpredictable - you cannot say whether this household defaults. But put three hundred together and the law of large numbers is supposed to do its work.

[reveal 3] Then you sell claims on the pool. The pool is the asset, the securities are the claims.

[reveal 4] The household pays its monthly instalment, it goes through a trustee, and it ends up with investors who have never met them.

[pause, then put up the question and wait]

[wait - do not rescue them]

Here is what they have to see. Diversification only removes risk that is independent. If my neighbour defaulting tells you nothing about whether I default, pooling works beautifully.

[ask] So - does my neighbour defaulting tell you nothing about me?

It tells you a great deal, if the reason is that house prices across the country have fallen and we are both in negative equity. In that world the defaults arrive together.

The correlation everyone had assumed was near zero turned out, in 2008, to be near one. Every tranche above the very bottom had been priced on the assumption it was near zero.

Say the general lesson plainly, because it is examinable: a diversification argument is only as good as the independence assumption underneath it, and nobody checks that assumption while it is working.

4 · Mortgage rates track long Treasuries

The two series move together, but the gap between them is not constant. What widens it?

Two lines, and the first thing to make them notice is that they move together.

[ask] Why would a thirty-year mortgage rate follow a ten-year government bond rate at all? They are different instruments with different borrowers.

Because the lender has an alternative. If the government will pay me four percent for lending to it, I am not lending to a household for less. The Treasury yield is the floor, and the mortgage rate sits on top of it.

[point at the gap]

Now the gap. It is a spread, and a spread is a price for something.

[ask] Two things the household offers that the Treasury does not. What are they?

Default risk - the household might not pay, the government will. And prepayment risk - the household can refinance the day rates fall, so the lender loses the good loans and keeps the bad ones.

[point at a recession band]

Look at what the gap does in the shaded periods. It widens, because in a recession both of those risks get worse at the same moment.

4 · Loan and annuity

What you can borrow, given what you can pay each period:

C_0 = a \times \left[\frac{1-(1+i)^{-T}}{i}\right]

What you must pay each period, given what you borrowed:

a = C_0 \times \left[\frac{i}{1-(1+i)^{-T}}\right]

One formula, written twice. Say that explicitly, because they will try to memorise two things.

On the left: you tell me what you can pay every month, I tell you how much house you can buy. On the right: you tell me the loan, I tell you the payment. Same equation, rearranged.

Two warnings about i and T, and they are the same two as the semi-annual bond from the warm-up. Monthly payments means a monthly rate - five percent a year is nought point four one seven percent a month, not five. And T counts periods, not years. Two hundred and forty, not twenty.

[point at any one line]

Each line is straight. Double the payment you can afford and you double the loan. That part is dull.

[point at the three dots on the dashed line]

Now the same household, with two thousand a month to spend, at three different interest rates. At three percent they can borrow three hundred and sixty-one thousand. At five, three hundred and three. At eight, two hundred and thirty-nine.

[ask] Their income has not changed by a cent. What has?

Only the rate. Going from three percent to eight takes a third of their buying power away.

Say the consequence plainly, because it is the link to everything else in this section:

Housing affordability is mostly a story about interest rates, not about prices or incomes. And since this is what each household can bid, it is also what sets house prices - which is why the mortgage-rate chart we just looked at is a housing-market chart as much as a credit chart.

[ask] So when the central bank raises rates, what happens to house prices?

Hold that thought for the mortgage-market story - it is how 2006 begins.

4 · Problem 2.7

You want to borrow $100,000. Choose between a 12% 30-year mortgage and an 11.5% mortgage with 2 discount points. Which?

A discount point is 1% of the loan paid up front: on a $200,000 loan, 1 point costs $2,000, half a point $1,000, a quarter point $500.

This is the one problem in the chapter with a genuinely conditional answer, and that is why it is worth doing.

Set it up: two thousand dollars today, against fifty basis points saved every month for thirty years.

[ask] Before any arithmetic - what is the one piece of information the question has not given you?

How long you are going to stay in the house. Without that the question has no answer, and spotting that is the skill being tested.

Then do the arithmetic. Both monthly payments from the annuity formula, take the difference, and ask how many months of that saving it takes to earn back two thousand dollars.

[ask] Suppose the break-even is around four years. You are a student who will move cities for a job. Points, or no points?

No points. Same numbers, a family settling permanently: points.

The general lesson, and it recurs in the derivatives section: a cost paid up front in exchange for a stream of savings is only worth it if you are around long enough to collect the stream.

Worked. Two points on $100,000 is $2,000 today against 50 basis points saved every month for thirty years. Compute both monthly payments with the annuity formula, then ask how long it takes the saving to repay the $2,000. The answer depends entirely on how long you stay in the house.

4 · Getting the rate and the period right

A rate is quoted per year. Every formula in this chapter wants it per period — and the number of periods has to match. The same €150,000 loan at a quoted 6% over 25 years:

Payments	Rate per period	Periods	Payment	Paid per year
Monthly	6 ÷ 12 = 0.50%	300	966.45	11,597
Quarterly	6 ÷ 4 = 1.50%	100	2,905.59	11,622
Semi-annual	6 ÷ 2 = 3.00%	50	5,829.82	11,660
Annual	6.00%	25	11,734.01	11,734

Same loan, same quoted rate, four different answers. Divide the rate and multiply the periods by the same number — and never mix a monthly rate with a count of years.

This is the single most expensive habit in the chapter, so spend five minutes on it rather than correcting it in the exam.

The rule is one line: a quoted rate is annual, the formula is per period. Divide the rate by the number of payments a year and multiply the term by the same number. Both, every time.

[point at the table]

One loan, one quoted rate of six percent, four perfectly correct answers depending only on how often you pay.

[ask] Look down the last column. Paying monthly costs eleven five ninety-seven a year, paying annually eleven seven thirty-four. Why is paying more often cheaper?

Because every payment knocks down the balance, and interest is charged on the balance. Pay twelve times a year and you are reducing the debt all year long; pay once and you carry the full balance for twelve months first.

[pause]

If someone asks about the effective annual rate, here is the honest answer, and it runs the other way: six percent compounded monthly is an effective 6.17% a year, compounded annually it is exactly 6.00%. Two different questions - how expensive the quoted rate really is, against how fast you are retiring the principal. Do not open this unless they ask; it confuses more than it helps.

4 · Problem 2.7b

You borrow €150,000 over 25 years. The quoted rate is 6% a year.

1. What is the monthly payment? And the quarterly payment?
2. A second lender offers 5.6% with 1.5 discount points, paid monthly. How much is the saving each month?
3. How long must you keep the loan for the points to be worth buying?

This is 2.7 again with the rate conversion built in, so they cannot avoid the step. Ten minutes.

[part one]

Monthly: half a percent per period, three hundred periods, nine hundred and sixty-six forty-five. Quarterly: one and a half percent, one hundred periods, two thousand nine hundred and five fifty-nine.

[ask] Is the quarterly payment three times the monthly one?

Almost, but not quite - it is slightly more than three times, for the reason on the previous slide.

[part two]

Five point six percent is nought point four six six seven percent a month. Payment nine hundred and thirty eleven. So the saving is thirty-six thirty-four a month.

And the points: one and a half percent of a hundred and fifty thousand is two thousand two hundred and fifty, paid today.

[part three]

[ask] Same question as before - what has the problem not told you?

How long they stay. Two thousand two hundred and fifty divided by thirty-six thirty-four is sixty-two months, a bit over five years, ignoring the time value of money. Discount the savings at the loan’s own rate and it is seventy-four months, just over six.

Same shape as 2.7, bigger numbers, and this time they had to convert the rate to get there.

Worked. Monthly: i = 0.5%, T = 300, a = €966.45. Quarterly: i = 1.5%, T = 100, a = €2,905.59. At 5.6%: i = 0.4667%, a = €930.11, saving €36.34 a month against €2,250 today. Break-even 62 months undiscounted, 74 discounted.

5 · The foreign exchange market

The biggest market nobody sees

No building, no opening bell, no ticker on the evening news. Over the counter, bank to bank, with turnover in the trillions of dollars a day.

• What is traded is not currency in the sense of notes, but bank deposits denominated in different currencies.
• Trades are large — typically in excess of USD 1 million.
• We met the price in session 2. Now we meet the market that sets it.

A short section, and the job is to connect it to what they already did.

[ask] Name the largest financial market in the world.

They will say New York, or the stock market. It is this one, and most of them have never seen a picture of it, because there is nothing to photograph. No floor, no bell, no building.

[pause]

The second line matters more than it looks. What changes hands is not banknotes. It is bank deposits. When a French importer buys dollars, nobody counts out dollars - one bank’s dollar deposit is credited, another’s is debited.

Why insist on that? Because it is the reason this market is enormous and invisible at the same time, and the reason it is bank to bank rather than on an exchange.

Last line, and say it directly: in session two we spent a long time on the exchange rate as a price. We did supply and demand for baht, we did the fixed versus floating table. Today is the other half - the market where that price is actually made.

5 · Purchasing power parity

Both versions rest on the law of one price: identical goods priced in a common currency should sell for the same amount, because arbitrage in goods moves the two prices together until the gap closes. PPP is that law applied to the basket rather than to the good.

• Absolute PPP — a proposition about levels. The nominal exchange rate equals the ratio of the two price levels; equivalently, the real exchange rate equals one. Requires identical baskets and frictionless trade in every good, including non-tradables.
• Relative PPP — a proposition about rates of change. The rate of depreciation equals the inflation differential, \pi_{US} - \pi_{EUR}. Requires only that frictions be constant, so they cancel in differences. This is the testable version.
• Absolute PPP is rejected in the data. Deviations are large and persistent: goods are differentiated, so firms hold pricing power, and nominal prices are sticky. Both break the arbitrage.
• What survives is a long-run proposition. Deviations decay slowly rather than not at all.

Start with the intuition. The slide states the two propositions and leaves the algebra off, so put it on the board if they want it: absolute PPP is the exchange rate equal to the ratio of the two price levels, relative PPP is the change in the exchange rate equal to the inflation differential.

The idea is the law of one price. If the identical good costs more in one country than another, someone buys it where it is cheap, sells it where it is dear, and the trade pushes the prices together.

[first item]

Do that for every good at once and you get absolute PPP. The exchange rate is the ratio of the two price levels. Say plainly that it almost never holds exactly - identical baskets and frictionless trade are a lot to ask.

[second item]

Relative PPP is the same statement in rates of change, and the difference is what makes it usable. If American inflation runs two points above European inflation, the dollar should lose about two percent a year against the euro.

The reason this version survives: a constant friction - transport, a tariff, a retail margin - shifts the level but not the rate of change, so it cancels. That is why absolute PPP fails on the data and relative PPP is the one that gets tested.

[third and fourth items]

Two reasons it fails in the short run, and both are failures of the arbitrage.

Goods are not identical. Brands, trademarks and legal protection give firms the power to charge different prices in different countries and keep doing it.

And prices are sticky. Even where arbitrage would work, prices do not adjust quickly, so the gap persists for years.

Then the long run, in one line: sticky prices eventually unstick. PPP is a statement about where things end up, not about next quarter - which is exactly why it is useless for forecasting and still worth knowing.

If they ask what the real exchange rate on the slide is: take the European price level, convert it into dollars at the market rate, and divide by the American price level. Absolute PPP is the claim that this number equals one. In the data it is not one, and the deviations die away with a half-life of something like three to five years - slow, but not never. That is all “long run” means here.

5 · The instruments: spot and forward

Spot rate — the price for near-immediate delivery of bank deposits, settling two business days forward.

Forward rate — the price, agreed today, for delivery at a fixed future date. Quoted at a premium or discount to spot, not as a forecast.

The rate is the price that clears the market for the currency, so it moves with anything that shifts demand or supply. The standard fundamentals:

• relative price levels — the PPP channel
• trade barriers — tariffs and quotas
• preferences over domestic and foreign goods
• relative productivity

Each fundamental enters as an expected change. The spot rate therefore already impounds the forecast, and moves only on news.

Two rates, and the difference is only timing. Spot: we exchange now - two business days, in practice. Forward: we agree the price now and exchange in three months.

The forward is quoted as a premium or a discount on the spot rate. Say once that it is not a forecast - it is a price, pinned by the interest rate difference between the two currencies - and move on unless someone pushes.

[ask] Who needs the forward one? Give me a firm and a reason.

A French exporter who has invoiced an American customer in dollars and will be paid in ninety days. They know the amount in dollars. They do not know what it will be worth in euros. The forward removes that uncertainty - and note that it removes it in both directions, which is the same structure as the hedge we meet in the derivatives section.

[point at the list on the right]

Four things that move the rate. Do not drill the list.

[reveal the key idea]

Drill this instead. Every item on that list is an expected change, not a current one.

[ask] Inflation data comes out this morning, exactly as everyone predicted. What happens to the exchange rate?

Nothing. It was already in the price. The rate moves on surprises, and that is why exchange rates are so hard to forecast: you would be forecasting the part nobody expected.

5 · What an appreciation does

• Domestic goods become more expensive abroad; foreign goods become cheaper at home.
• Domestic producers become less competitive in both markets.
• Domestic consumers gain purchasing power.

So is a strong currency good news? Ask a French exporter and a French holidaymaker on the same afternoon and you will get two answers.

[cuttable - the content is in the previous two slides if you are short]

Short slide, and the point is distributional.

[reveal the three bullets one at a time]

The euro appreciates. French wine costs more in New York. American machinery costs less in Lyon. So French producers lose on both sides, and French consumers win.

[pause]

[ask] The euro strengthens ten percent. Good news or bad news?

Push back on whichever answer you get, because the question is badly posed on purpose.

It is good news if you are flying to New York next week. It is bad news if your job is making the wine. A currency movement is not good or bad for a country - it is good for some people in it and bad for others, and the people it is bad for are usually better organised politically.

[ask] So which group do governments tend to listen to?

That is why “currency manipulation” is a political accusation and not just an economic one.

6 · Derivative markets

A price for a promise

In 1995 Barings Bank, 233 years old, was destroyed in a matter of weeks by one trader’s position in Nikkei futures. The instruments were not exotic. The position was simply unhedged and unsupervised.

• A derivative is a contract whose value comes from the price of something else — an index, a rate, a currency, a tonne of wheat. You never own the thing itself; you own a claim written on its price.
• They exist because risk can be sold: one party wants rid of it, another will carry it for a price.
• Hedging is taking a second position that moves against a risk you already carry, so the two cancel.
• Taken on its own, that same contract is a bet. A hedge and a speculation are the same instrument — the difference is whether you already hold what it is written on.

Open with Barings, because it does the work of three slides of warnings.

A bank that had been trading since seventeen sixty-two. It financed the Louisiana Purchase. One trader in Singapore, positions in Japanese index futures, and in a matter of weeks there was no bank. It was sold for one pound.

[ask] What do you think the instruments were - something exotic that nobody understood?

No. Index futures. The most standardised, most transparent, most boring derivative there is. Nothing about the instrument was the problem.

[pause, then reveal the bullets one at a time]

First: derivative means derived. The contract is worth something only because something else has a price - an index, a rate, a currency, a tonne of wheat. You never own the wheat. You own a piece of paper whose value tracks the wheat.

[ask] Why would anyone write such a contract? What is being traded?

Second line, and this is the one that explains why the market exists. What is being traded is risk. A farmer wants rid of the risk that the March price collapses. Somebody else will carry that risk, for a price. The contract is the transfer, and the price of the contract is the price of the risk.

Third: hedging. You already carry a risk, so you take a second position that moves the other way, and the two cancel. Note the words already carry - they are doing all the work in that sentence.

[pause]

Fourth line, and this is Barings in one sentence. Strip away the risk you already carry, and what is left is the contract on its own - which is a bet.

[ask] So what was Leeson doing - hedging or speculating?

Speculating, and the instrument cannot tell you which. The same futures contract is prudent in the hands of someone holding Japanese shares and ruinous in the hands of someone who is not. Nothing in the contract distinguishes them, which is exactly why supervision rather than the instrument is the subject of that story.

6 · Long, short, and offsetting

A long position: you own the asset and gain if the price rises.

A short position: you do not own it — you borrow it, sell it, and gain if you can buy it back cheaper.

To hedge:

• a long position is offset by taking an additional short position
• a short position is offset by taking an additional long position

Hedging is not the absence of a position. It is two positions, chosen so that what one loses, the other gains.

[cuttable - but only if they already have long and short from another course]

[ask] Long and short. Who can define them?

Long is easy: you own it, you want it to go up.

Short is the one worth pausing on. You borrow the share from someone, sell it today, and you owe them a share back. So you are hoping to buy it back cheaper.

[ask] What is the worst that can happen to someone who is long?

The price goes to zero. They lose what they put in.

[ask] And the worst that can happen to someone who is short?

There is no worst. The price has no ceiling, so the loss has no ceiling. Write that asymmetry down - it is the same asymmetry as the one between the option buyer and the option writer later in this section.

[point at the key idea]

Last line, and it is the definition they should leave with. A hedge is not having no position. It is having two, pointing opposite ways, so the gain on one pays for the loss on the other.

6 · The instruments: five families

• Forwards
• Futures
• Options
• Swaps
• Credit

• An agreement between two parties to transact at a future point in time.
• Specifies the exact assets, the price, and the date of exchange.
• Long = agree to buy later. Short = agree to sell later.
• Private and bilateral, and therefore exposed to the other party defaulting.

• The same agreement as a forward — transact at a future date at a price fixed now — but on an organised exchange rather than privately.
• Standardised: the exchange fixes the contract size, the quality and the delivery dates, so contracts are interchangeable and far more liquid.
• A clearing house steps between buyer and seller and becomes the counterparty to each of them.
• Marked to market daily — the day’s gain or loss is settled every evening against posted margin, which is what removes most of the default risk.
• No delivery needed: positions are closed out and netted in cash.
• At expiry the futures price converges to the spot price of the underlying — otherwise there would be an arbitrage.
• Harder to corner, because the deliverable range of securities is wide. Regulated in the United States by the CFTC.

• The right, not the obligation, to buy (call) or sell (put) an instrument at the strike (exercise) price.
• American: exercisable at any time up to expiry. European: only on the expiry date itself.
• Written on many underlyings — stocks, stock indexes, futures, currencies and interest rates.
• To hedge you buy the contract: a put against a fall, a call against a rise. For that right you pay a premium.
• The premium is higher:
  • the greater the volatility of the underlying, and the longer the time to expiry — both mean more can happen before it ends;
  • for a put, and lower for a call, the higher the strike.

• An interest-rate swap exchanges one stream of interest payments for another, in the same currency.
• The plain vanilla version specifies the rates exchanged, the type of payments, and the notional amount.
• No balance-sheet change, and longer-dated than futures or options — but illiquid and exposed to default risk.

• Credit derivatives pay off on changes in credit events. The notional outstanding is in the trillions.
• Credit options are tied to ratings or spreads.
• Credit swaps exchange payments on loan portfolios, letting lenders specialise locally while diversifying.
• A credit default swap pays off, option-like, when a basket of loans defaults.

Five tabs. Do not read them. Take forwards and futures together, spend real time on options, name swaps, and finish on credit.

[forwards and futures tabs]

Forwards and futures are the same promise. I agree today to buy from you in March at a price we fix now. The difference is only where it is done.

[ask] A forward is a private agreement between us. What is the risk that creates?

That in March, when the price has moved against you, you do not turn up.

The exchange solves it in three steps, and it is worth naming them separately. Standardisation, so every contract is identical and you can sell yours to somebody else instead of being stuck with me. A clearing house, which steps in between us so neither of us faces the other at all - you face the exchange. And daily marking to market against margin, so the day’s loss is paid tonight and nobody is ever owed more than one day’s movement.

One more line they tend to miss: at expiry the futures price and the spot price are the same number.

[ask] Why must they be?

Because if they were not, you could buy in one market and sell in the other for a certain profit. The arbitrage closes the gap, which is exactly why the contract tracks the thing it is written on.

[options tab]

Now the one that is genuinely different. A future obliges you. An option gives you the right and lets you walk away.

[ask] If I am handing you the right to walk away, what do I want in return?

A premium. Two of its three drivers say the same thing: more time to expiry, and more volatility, both mean more can happen before the contract ends, so the right to choose is worth more.

The third one is the strike, and it is worth doing out loud rather than memorising.

[ask] I give you the right to sell at 950 or the right to sell at 1,100. Which do you pay more for?

The 1,100 - you may sell at a higher price. So a higher strike makes a put dearer and a call cheaper, because the call is the right to buy and nobody wants to pay more to buy.

[swaps tab] Name it and move on. Two streams of interest payments exchanged. Enormous market, no balance-sheet change, illiquid, default risk.

[credit tab]

Finish here deliberately. A credit default swap pays you when a basket of loans defaults. It is insurance on someone else’s borrowing.

[ask] And can you buy that insurance on a house that is not yours?

Yes. You can, and in 2007 people did, in volumes far larger than the loans themselves. Hold that thought for the regulation session.

6 · Payoff profiles at expiration

• The strike (here 115) is the price written into the option: the price at which the holder may buy (call) or sell (put). The market price moves; the strike does not.
• The future is a straight line through it. The option follows that line, then goes flat at the premium — you buy the bend by giving up part of the upside.

Do not cut this slide. If they understand one picture from the derivatives section, it is this one.

[name the strike before anything else]

The strike is the number written into the contract - the price at which the holder may buy or sell. It never changes; the market price does. So it is the line in the sand that decides whether exercising is worth anything, and it is exactly where the picture bends. On this chart it is 115.

[ask] The market ends at 120, strike 115. Do you exercise?

Yes - you buy at 115 something worth 120. And at 110? No, you walk away and buy in the market. That switch is the bend.

[point at the straight line first]

The future is a straight line through the strike. Price goes up, you gain. Price goes down, you lose, and there is nothing underneath you.

[now trace the option line with a finger, from the right-hand side leftwards]

The option is the same line on the right - and then it stops falling. It goes flat. Below the strike you simply do not exercise, so your loss stops at the premium you paid.

[ask] So where is the option line below the futures line?

On the right, everywhere. The option buyer always ends up two thousand worse off than the futures buyer when the price rises, because that is what the premium cost.

[pause]

Say the trade-off as one sentence and ask them to write it: you buy the bend by giving up part of the upside.

[ask] When is that worth doing?

When the downside would actually hurt you - when you could not survive the straight line continuing downwards. That is the answer to both of the problems that follow.

6 · Problem 2.8

Rock Solid holds a $100 million stock portfolio that tracks the S&P 500. The manager fears a decline and wants to hedge the downside completely.

1. The S&P is at 1,000. How is the hedge constructed with futures?
2. What happens if a year later the index is at 900?
3. And if it is at 1,100?

An index futures contract has a multiplier of 250, so one contract is worth 250 \times 1000 = \$250{,}000.

Five minutes in pairs, then take it apart on the board.

[ask] The portfolio is long the index. Which way does the hedge point?

Short. Sell futures.

[ask] How many contracts?

One hundred million divided by two hundred and fifty thousand. Four hundred contracts, sold short.

Now the two states of the world, and do them both.

At nine hundred: the index has fallen ten percent, so the portfolio loses ten million. The short futures position gains ten million. Net zero.

At eleven hundred: the portfolio gains ten million, and the futures lose ten million. Net zero.

[pause]

This is where the misunderstanding lives, so put the question to them directly.

[ask] Is that what the manager wanted?

They wanted protection on the way down. What they bought was no exposure at all, in either direction. A futures hedge is symmetric - it does not know which way you were hoping the market would go.

Worked. 400 contracts sold short. At 900 the portfolio loses 10m and the futures gain 10m. At 1100 the portfolio gains 10m and the futures lose 10m. The hedge works in both directions, which is the point students miss.

6 · Problem 2.9

The same $100 million portfolio tracking the S&P 500, the same fear, the same year ahead. The index is at 1,000, and a strike of 950 costs a premium of $200 per contract.

1. How is the hedge constructed with options?
2. What happens if a year later the index is at 900?
3. And if it is at 1,100?

An index option contract has a multiplier of 100, so one contract is worth 100 \times 1000 = \$100{,}000.

Same portfolio, same fear, different instrument - so the only honest way to do this one is side by side with 2.8.

[ask] Long the index, and you want protection on the way down. Call or put?

Puts. You buy the right to sell.

Contracts: one hundred million over one hundred thousand. One thousand contracts, and the premium is two hundred each, so two hundred thousand dollars leaves the fund today whatever happens next.

At nine hundred: the puts are in the money below nine fifty, so they pay. But notice the portfolio fell from a thousand to nine hundred and the protection only starts at nine fifty - the first fifty points are uncovered. That gap is a deductible, and it is there because they chose that strike.

At eleven hundred: the puts expire worthless, you have lost the premium, and you keep the whole ten million gain.

[pause]

[ask] So compare the two problems in one sentence each.

The future gave away the upside and cost nothing up front. The option kept the upside and cost two hundred thousand up front, with a gap at the bottom.

[ask] Which is the better hedge?

Neither. They are different products at different prices, and the right answer depends on whether the fund can live with the gap and whether it needs the upside. That is the bend from the payoff diagram, with numbers on it.

Worked. 1,000 put contracts at strike 950, premium 200 each = $200,000 paid today. At 900 the puts pay but the first 50 index points are uncovered. At 1,100 the puts expire worthless and the portfolio keeps the full gain.

7 · Reading discussion

Does a green label change the price?

A green bond funds an environmental project. In every other respect it is an ordinary bond from an ordinary issuer, with the same claim on the same balance sheet.

• If the cash flows and the credit risk are identical, should it trade at a different yield?
• If it should not, what exactly is the instrument for?

This is the reading discussion, and it is assessed, so protect the time for it. Do not let the derivatives section eat it.

Open with the puzzle rather than the paper.

A green bond is a bond. Same issuer, same balance sheet, same legal claim, same seniority in default. The only difference is a promise about what the money is spent on.

[reveal the first bullet, then ask]

[ask] Identical cash flows, identical credit risk. Should it yield less than the ordinary bond?

Let them argue. The finance answer is no - price the cash flows, and these cash flows are the same. Any yield difference is investors paying for something other than return.

[reveal the second bullet]

[ask] Then what is the instrument for?

Three candidates worth having on the board before you open the paper: a signalling device, a marketing device, or a commitment device. The paper is largely an argument about which of the three it actually is.

7 · Sartzetakis (2021), green bonds and the low-carbon transition

• The six questions
• How to structure it
• The last section

Choose one. One question per section of the paper, so each of you is reading a different part of it closely.

1. Should the central bank’s mandate be enlarged to cover the climate transition — and if it is, who decides which sectors get cheap credit?
2. Who should decide what counts as green, when the issuer writes the definition and also profits from it?
3. If the issuer appoints and pays the verifier, what is an external review actually worth?
4. Why should a green bond yield less than an identical conventional one — and why does the paper report a premium in §5.1 and deny one in §6?
5. Does debt genuinely share the cost of the transition across generations, or does it only move the bill?
6. What would have to change for the green label to bite — and is a bond the right instrument at all?

Sections: 1 → §2 · 2 → §4 · 3 → §4.1 · 4 → §5.1 and §6 · 5 → §3.1 · 6 → §5.2 and §6

Four parts, in this order, with these headings. The first three are about your question. The fourth is the same for everyone.

1. The claim — what the paper argues on your question, in your own words, with the section. Not a summary of the paper. (~150 words)
2. Why, and why not — the argument for it, and the strongest objection to it. This is the core. (~350 words)
3. What could be done next — one concrete proposal, and who would have to act on it. (~200 words)
4. And now AI — your own view. (~200 words)

Two pages maximum. Names and question number on the first line. An answer with no counter-argument is half an answer, and padding is visible.

Whatever your question, the fourth section is the same: does the rise of AI change your answer?

The paper is from 2019 and says nothing about AI. This section is not in the text and cannot be looked up. It is your own judgement, and it is the part I will read most carefully.

Start from what is actually happening now: the capital going into data centres, the electricity they need, and the claims being made about what AI will solve.

This is the assessed piece, so protect the time for it and do not let the derivatives section eat it.

Open the tabs in order: the questions, the structure, then the last section. Spend real time on the third tab - the AI section is the part that separates a good answer from a summary of the paper, and it is the only part they cannot look up.

[on the questions]

Read the six out loud, slowly. There is one per section of the paper, so no two groups are reading the same pages closely - say that, because it is the reason they cannot all take the easy one.

One and six are the open ones. Two and three are where the paper gives them most to work with. Four is the one that connects back to the hook at the start of the session, and it has a trap in it worth flagging: the paper reports greenium estimates in 5.1 - seventeen basis points, eighteen, six, two - and then states in the epilogue that there is no significant premium in the primary market. That is a contradiction in a published paper. A group that spots it has done the exercise.

Tell them to choose the question they disagree with something in, not the one that looks easiest.

[on the structure]

Four parts, four headings, and tell them to actually use the headings - it makes the structure visible to them while they write and to you while you mark. Say that the word counts are a shape, not a rule.

[on the last section]

This is the one that is the same for all six, and it is where they will flounder, so be concrete. If a group is stuck, here is a handle for each question.

Question one: if the central bank is steering credit towards the transition, does it also steer credit towards or away from data centres? Who decided that?

Question two: a data centre running on contracted renewables - green or not? AI used to cut losses on the grid - green or not? Who gets to say?

Question three: verification reports written with AI, at scale. Does that fix the verifier problem or industrialise it?

Question four: would investors accept a lower yield on debt labelled responsible AI, or green data centre? And if they would, is that a price signal or a fashion? Then send them back to 5.1 against the epilogue.

Question five: the data centres being built now are long-lived and debt-financed. Who gets the productivity, and who is left with the electricity bill and the asset? That is the paper’s intergenerational argument, transposed.

Question five is the paper’s own theoretical claim, so it is the one where they have least excuse for not citing it.

Question six: there is no carbon price on compute and no disclosure standard for model training, and the capital expenditure is happening anyway, label or no label. So what would actually have to change - and is a bond any part of it?

[on what a good answer does]

The four sections are the whole marking scheme. Say that an answer missing section three cannot get the top mark however well written the first two are, and an answer missing section four has not done the assignment at all.

[then go to the marking slide and read it out]

7 · Submission and marking

In your groups — maximum five. One of the six questions, one document per group.

Two pages maximum, every name on the first line.

Due 19:15.

Write it yourselves. Submissions are checked with Pangram.

What you hand in	On the exam
Nothing	−2
Something you did not really try at	−1
All four sections, done properly	+2
Written with AI	−1 on top

There is no top band to compete for. Do the work properly and the two points are yours.

Read this slide out. Do not summarise it - read it, because every line is a rule they will hold you to later.

Groups of five, same as the security analysis. One question per group. One document. Every name on the first line, because the mark goes to all of them. The deadline is 19:15.

[on the marks - say the shape of it before the numbers]

Tell them what the table is designed to do, because it is not the usual one. There is no scarce top band and nobody is being ranked against anybody. Every group that does the work properly gets the full two points. If all of them earn it, all of them get it, and that would be the best outcome of the evening.

What has gone is the free ride in the middle. Handing in two pages of nothing used to cost nothing. Now it costs a point, because it wastes their time and mine and it is worse than an honest blank.

Walk down it: nothing, minus two. Something nobody tried at, minus one. The work done properly, plus two. A machine wrote it, one more off whatever it earned.

[be clear about where the bar is, because the jump is large]

The bar is the four sections, each one honestly done: the claim, the argument with a real objection to it, a concrete proposal with somebody named to act on it, and their own view on AI. Four sections present and meant, and that is the two points. Say it in those words so no group can tell you afterwards that they did not know what you wanted.

[ask] Why am I asking you to write two pages yourselves when the machine can do it in four seconds?

The honest answer, and give it to them: because the exam is not open to the machine, and because what is being assessed is whether you can hold an argument, not whether you can obtain one.

Add the practical reason too, because it lands harder than the principle: the fourth section asks what they think, and it is the one the machine cannot write for them. If they have outsourced the first three they will have nothing to say in the fourth, and it will show.

[if someone asks about the checker]

Say that submissions are checked, that no checker is perfect, and that a flag starts a conversation rather than ending one - the conversation being you asking them to talk through their own argument. That is fairer than the detector alone, and more frightening, which is the point.

Next session

Session 5 moves from instruments to institutions: banks, what they do with the maturity mismatch we keep meeting, and why they are regulated differently from everyone else.

Group work: your security should be chosen by now. Not “a stock” — a specific one, with a ticker and available data.

Press O for the slide overview · C to draw · S for speaker notes · F for fullscreen · E then print for a PDF of the slides.

That closes securities markets. Six markets, and in every one of them the same two questions.

Next session we stop looking at instruments and start looking at the institutions that hold them. Banks, mostly - and the maturity mismatch that has turned up in every session so far is going to be the centre of it.

Last thing, and be firm about it. The group security should already be chosen. Not a sector, not “a bank” - a specific listed company with a ticker and data you can actually get. Every year a group chooses in the week before the presentation, and it is visible from the first slide.