International: FX, Flows & the Dollar

Phase 7 · Week 62 · ~20 hours

Where it sits. More than 90% of daily foreign-exchange turnover is financial rather than trade-related. The exporters, importers and tourists everybody pictures when they hear "currency market" are a trickle beside the flood of investors moving capital, and it is the flood that sets the price.

Sixth module of the macro phase, and the one that takes the machine outward across borders. M7.01 built the closed-economy machine (spending = money + credit; S = I; the debt cycle). M7.02 put the central bank on it and, in §4.4, gave a first glimpse of two things formalized here: FX intervention and the managed float, and the fact, asserted there and proved here, that a country cannot simultaneously fix its exchange rate, allow free capital flows, and run its own monetary policy. M7.03 taught rates and real yields, which turn out to drive both currencies and gold. M7.04 taught fiscal deficits and sovereign crises, and the twin-deficits idea and sudden stops both live at the border. M7.05 taught cycles and crises; Asia 1997 was a sudden stop, and you'll now see the mechanism.

All of that becomes international here: what a currency price actually is and what moves it; the no-arbitrage engine (covered interest parity) that links spot rates, forward rates, and interest rates; the impossible trinity as a hard constraint every country must choose within; the balance of payments that must, by accounting identity, sum to zero; how trade creates wealth and who wins and loses; how global capital floods in and rushes out (Rey's global financial cycle, the 2013 taper tantrum, EM sudden stops); the US dollar system that sits at the center of all of it (reserve status, exorbitant privilege, the Triffin dilemma, the eurodollar plumbing, Fed swap lines); commodities as macro signals (oil, gold, copper); and, the payoff for a cross-border investor, exactly what happens to your returns when you, an Indian, own American assets. The next module (M7.07) turns the external-sector pieces built here, the current account, FX reserves, the rupee and FPI flows, into live dashboard indicators and the India-macro capstone.

Suggested hour budget. Core content: 8.5h · worked examples (CIP, the carry trade, comparative advantage, hedged-vs-unhedged): 3h · practice set (CIP compute, trinity classification, BoP, tariff incidence, PPP, hedged/unhedged, sudden-stop reasoning): 4.5h · applied mini-project (the rupee, the dollar, and a cross-border portfolio, with live RBI/FRED/IMF data): 2h · flashcards + mastery check: 2h.

App link. This feeds the Macro & Cycles branch and, specifically, its scenario-event engine (competency C8): the "commodity shock," "EM sudden stop / risk-off," and "dollar-funding squeeze" events that fire mid-simulation and ask what do you do? are the mechanisms below made playable, and the portfolio simulator's cross-border positions carry an explicit FX leg wired to the hedged-vs-unhedged logic of §4.11. Five drills seed here: a currency-driver drill (a described shock in, the likely FX direction out), a CIP/forward numeric drill (spot + two rates in, the no-arbitrage forward out, timed), an impossible-trinity classification drill (a country's regime in, which two of the three corners it chose out), a BoP-classification drill (a transaction in, current vs financial account out), and a hedged-vs-unhedged return drill (an asset return + an FX move in, the investor's home-currency return out). It hands the currency-exposure concept forward to the valuation sandbox, so a cross-border thesis must state, and be scored on, its embedded currency bet.

Learning objectives

By the end you can:

1. Say what an exchange rate is: the price of one money in units of another. Read both quote conventions without fumbling the direction of appreciation and depreciation, and distinguish the nominal rate from the real rate and the trade-weighted REER that actually measure competitiveness.
2. Explain the five forces that move a currency: interest-rate differentials (and the carry trade), purchasing-power parity as a long-run anchor and why it fails short-run, terms of trade, safe-haven flows, and relative growth/productivity. Say which dominate over which horizon.
3. Compute a no-arbitrage forward rate with covered interest parity (CIP), explain why the higher-yielding currency trades at a forward discount, show the arbitrage that enforces it, and state why uncovered interest parity fails in practice (the reason the carry trade earns money — until it doesn't).
4. State and apply the impossible trinity: a country may have at most two of {a fixed exchange rate, free capital mobility, an independent monetary policy}. Classify real regimes by it (India's managed float, China, the eurozone, a Gulf peg, Hong Kong's currency board), including Rey's argument that it may really be a dilemma.
5. Read the balance of payments: the current account (trade + income + transfers) versus the capital/financial account, why the two must sum to (near) zero, why FDI is sticky and portfolio "hot money" is flighty, and the twin-deficits identity linking the fiscal and external balances.
6. Explain comparative advantage and its gains from trade from a worked numeric, and its distributional caveat (winners and losers, the China shock) presented fairly. Explain who actually bears the cost of a tariff, and describe global value chains and the reshoring/friend-shoring shift.
7. Explain global capital flows and EM sudden stops: Rey's global financial cycle (US monetary policy and risk appetite as the world's factor), the 2013 taper tantrum and the "Fragile Five" (India among them), and the specific buffers India built afterward.
8. Explain the US dollar system: its ~57–58% reserve share, exorbitant privilege, the Triffin dilemma, the de-dollarization debate presented evenhandedly, the eurodollar (offshore-dollar) system, and Fed swap lines as the crisis backstop. Read oil, gold, and copper as macro signals.
9. Quantify the currency risk of cross-border investing: an Indian buying US assets is implicitly long USD/short INR, the INR's structural depreciation follows from the inflation differential, and FX can dominate the equity return in any given year. Reason about hedged versus unhedged for both an Indian in US assets and a US investor in India.

Prerequisites & connections

Builds on. M7.02 (§4.4) set up the FX-intervention and managed-float material, and this is its formal payoff: the impossible trinity is stated there and proved here, and sterilized intervention, the RBI's reserve management, and the Fed's swap lines all reappear with their full mechanics. From M7.01, S = I and the current account as S − I is the accounting spine of §4.5, and the credit cycle is what floods across borders in §4.7. From M7.03, real versus nominal rates and the real (TIPS) yield: the real yield drives both currencies and, decisively, gold (§4.10), and relative inflation is what drives the INR's structural depreciation (§4.11). From M7.04, fiscal deficits and sovereign crises: the twin-deficits identity (§4.5) is the bridge from the fiscal balance to the external balance, and a sudden stop (§4.7) is the external face of the sovereign crises you studied. From M7.05, cycles and crises: Asia 1997 was the archetypal sudden stop, and the global financial cycle (§4.7) is the cross-border version of the credit cycle. From M3.01–M3.03, discounting, the country-risk premium, and Damodaran's currency-matched risk-free rate: valuing a company in rupees versus dollars is a currency choice, and §4.11 is where cost-of-capital work meets FX. From M0.03, real versus nominal (Fisher): relative PPP is Fisher applied across two countries. And from M5.06, IT services: an Indian exporter's dollar revenue is the single most important company-level FX exposure in the Indian market, and it runs through everything below.

Feeds forward. M7.07 is India macro and the dashboard capstone: the current-account balance, FX reserves and import cover, the rupee and its REER, and FPI flows are core external-sector dashboard indicators, and India's place in global flows (structural CAD, oil sensitivity, bond-index inclusion, the buffers of §4.7) is the capstone's subject. Its raw material is supplied here. Phase 8 is the rapid-analysis system: every teardown of a company with dollar revenue, dollar debt, imported inputs, or foreign listings must state its currency exposure, a line you learn here to fill. Phase 9 is portfolio and risk: a cross-border portfolio, a common real situation of Indian and US assets together, carries a currency bet that must be sized and, optionally, hedged, and §4.11 is the foundation. Competency C8, "understand the macro machine," is completed on its international side here. After this you can explain what moves a currency, why capital floods and flees, and why the whole system pivots on the dollar.

The one-sentence version of this module. A currency is a price like any other, set at the margin by interest-rate differentials, relative inflation, terms of trade, and fear, and linked to interest rates by a hard no-arbitrage identity (CIP); every country must choose two of {fixed FX, open capital, own monetary policy}; the flows that cross borders move together in a global cycle conducted from Washington; and the whole apparatus is denominated in a dollar whose issuer enjoys an exorbitant privilege and bears a structural dilemma, all of which lands, for you, as the simple fact that when an Indian owns American assets, the rupee's fate can matter more than the stock's.

Core teaching content

4.1 What an exchange rate is: two prices, two conventions, and the real rate that actually matters

USD/INR = 84 means one US dollar costs 84 rupees. Equivalently, one rupee costs about 1.19 US cents. That is all an exchange rate is: a price, the price of one money expressed in units of another. Half of all confusion about currencies is confusion about what that number on the screen means. A currency is not "strong" or "weak" in the abstract; it is only ever expensive or cheap relative to another currency, and it has a different price against every other currency in the world at once.

The first thing to master is the quote convention, because getting the direction backward is the single most common beginner error and it inverts every conclusion. There are two ways to write any pair:

• Price (direct, for the home country) quote: units of home currency per one unit of foreign. For an Indian, "USD/INR = 84" read as 84 rupees per dollar is the natural quote. When this number rises (84 → 86), it takes more rupees to buy a dollar, so the rupee has depreciated (weakened) and the dollar has appreciated. When it falls (84 → 82), the rupee has appreciated. Notice the trap: a rising USD/INR number means a weakening rupee. The number and the home currency move in opposite directions.
• Quantity (indirect) quote: units of foreign per one unit of home. "USD 0.0119 per rupee" or, more usefully, the way most majors are quoted: EUR/USD = 1.08 means 1.08 dollars per euro. Here a rising number means the first-named currency (the euro) is strengthening.

The professional convention writes a pair as BASE/QUOTE where the price is "how many QUOTE per one BASE." EUR/USD = 1.08 is 1.08 USD per 1 EUR; USD/JPY = 150 is 150 JPY per 1 USD; USD/INR = 84 is 84 INR per 1 USD. Rule of thumb inside the market convention, and only inside it: the rate is always "how much of the second currency for one of the first," and the first-named currency strengthens when the number goes up. The curriculum convention in the next box reverses the labels, so the rule has to be re-read there rather than reused. Say it out loud with units every single time until it is automatic; a portfolio manager who fumbles this loses money in the most embarrassing way possible.

Two live notations exist for that same price, and they label the currencies in opposite order. This is the single most reliable source of a wrong sign in an exam answer or on a hedging ticket, so learn both once and stop guessing.

The market writes a pair as BASE/QUOTE, and the number says how many units of the quote currency buy one unit of the base. That is the convention used above and everywhere else in this module, because it is what a screen, a dealer and a data vendor hand you. The CFA curriculum writes the same pair as price currency/base currency, and the number says how many units of the price currency buy one unit of the base currency. Read that string left to right and the labels sit the other way round: the currency named first is the one being paid, not the one being bought.

The two conventions never disagree about the economics, only about the string.

The conversion rule is therefore one line: invert the string, never the number. USD/INR = 84.00 in the market and INR/USD = 84.00 in the curriculum are the same rate written twice. Invert the number instead and you get 0.011905 dollars per rupee, which is also true and is a different quote answering a different question.

One test settles which convention a source uses in about five seconds. Find a number whose size you already know. A dollar has cost tens of rupees for decades, so any string containing USD, INR and the number 84 has to mean rupees per dollar whichever way the labels are ordered, and the ordering is then decoration. Run that check before you compute anything and no sign error survives into the arithmetic. Practice item P1 below uses the market convention throughout, and its "which currency strengthened" answers follow the market reading of the string.

Appreciation and depreciation versus revaluation and devaluation. Appreciate/depreciate describe market-driven moves of a floating currency. Revalue/devalue describe a deliberate administrative change of a pegged currency's official level (China "devalued" the yuan in 1994 and again, mildly, in August 2015). Same direction, different mechanism. The word you hear tells you the regime.

Nominal versus real: the rate that actually measures competitiveness. The screen price is the nominal exchange rate. But what determines whether Indian exports are actually competitive is the real exchange rate, which adjusts for the two countries' price levels:

Real exchange rate = Nominal rate x (foreign price level / domestic price level)

Intuition: suppose the rupee's nominal value against the dollar is perfectly stable for a decade, but Indian prices rose 60% while US prices rose 20% over that decade. In real terms the rupee has become much more expensive. Indian goods now cost far more relative to American goods than they did, so exporters are being squeezed even though the screen rate never moved. A stable nominal rate with higher domestic inflation is a real appreciation, and real appreciation quietly kills export competitiveness.

This is why a currency can be "stable" and yet steadily overvalued.

Because a country trades with many partners, the single most useful competitiveness gauge is the REER, the real effective exchange rate: a trade-weighted average of the real bilateral rates against a basket of partner currencies (the RBI publishes a 40-currency REER; the BIS publishes REERs for most economies). Read it as an index. A REER above ~100 (its base) says the currency is expensive in real, trade-weighted terms relative to the base period, which is a headwind for exporters and a flashing light that a correction (depreciation) may be building. When you hear "the rupee looks overvalued on REER," it means precisely this: adjusted for India's higher inflation and its trade basket, the rupee is dearer than its history, so downward pressure is accumulating even if the nominal rate looks calm.

Hold this: nominal moves are the news; the REER is the truth about competitiveness.

4.3 Spot, forward, and covered interest parity: the no-arbitrage engine

A forward is not a forecast. It is a contract, and its price is pinned by pure arbitrage to the spot rate and the two countries' interest rates. That is the mechanical heart of FX, and the one piece of this material you must be able to compute. There are two prices for a currency: the spot rate (S) for delivery now (technically T+2), and the forward rate (F) for delivery at a fixed future date at a price locked in today. The relationship that does the pinning is covered interest parity (CIP), "covered" because the currency risk is fully hedged by the forward.

The intuition first. You have money to invest for one year and two riskless options. Option A: invest at home in rupees at the Indian rate r_INR. Option B: convert to dollars at spot, invest at the US rate r_USD, and simultaneously sell the dollars forward for rupees at rate F, locking in your rupee proceeds today. Both options start and end in rupees with zero risk. If they paid different amounts, you could borrow in the cheaper and lend in the richer and pocket a riskless profit — infinitely. Arbitrage forbids that, so the two must pay the same, which forces the forward rate to a unique value:

F = S x (1 + r_INR) / (1 + r_USD) (with S and F quoted as INR per USD)

A worked no-arbitrage example. Spot S = ₹84.00/$; one-year rupee rate r_INR = 6.5%; one-year dollar rate r_USD = 4.5%. Then:

F = 84.00 x (1.065 / 1.045) = 84.00 x 1.019139 = ₹85.61/$

The one-year forward is ₹85.61, higher than spot, meaning the market's contractual price to buy a dollar in a year is more rupees than today. The rupee trades at a forward discount of about 1.9% (85.61/84.00 − 1 ≈ 1.91%, ≈ the 2% rate gap). This is the single most important and most counterintuitive fact in FX: the higher-yielding currency (the rupee, at 6.5%) trades at a forward discount, priced to weaken. The extra 2% of interest you earn in rupees is exactly given back, in the forward market, as an expected 2% currency loss. CIP is why the interest-rate "free lunch" is, on a fully-hedged basis, no lunch at all.

Show the arbitrage that enforces it. Suppose a dealer mis-quotes the one-year forward at ₹84.80/$, too low, pricing the rupee "too strong" forward. Arbitrage: (1) Borrow $1,000,000 at 4.5% → owe $1,045,000 in a year. (2) Convert to rupees at spot: $1,000,000 × 84 = ₹84,000,000. (3) Invest rupees at 6.5% → ₹84,000,000 × 1.065 = ₹89,460,000 in a year. (4) Today, lock a forward to buy back the $1,045,000 you'll owe, at F = 84.80: cost = 1,045,000 × 84.80 = ₹88,616,000. (5) In a year, collect ₹89,460,000, pay ₹88,616,000 for the dollars to clear your loan → riskless profit of ₹844,000, on borrowed money, with every price locked today. Traders would do this in size instantly, buying dollars forward until F rose back to ₹85.61 and the profit vanished. That pressure is why CIP holds. (In the real world CIP can deviate slightly, a "CIP basis," when dollar funding is scarce and balance-sheet capacity to arbitrage is constrained, a symptom of the dollar-shortage stress in §4.9; but treat CIP as the anchor.)

Uncovered interest parity (UIP), and why it fails: the reason the carry trade exists. CIP is an arbitrage identity and holds tightly. Its sibling, uncovered interest parity (UIP), is a theory of expectations: it says the expected future spot rate equals the forward, so the high-yield currency should on average depreciate by the rate gap, leaving expected returns equal across currencies. If UIP held, chasing yield would earn nothing on average. UIP fails empirically, spectacularly and persistently (the "forward-premium puzzle"): high-yield currencies, on average, depreciate less than UIP predicts, and often appreciate. So investors who borrow low-yield currencies and lend high-yield ones, running the carry trade, earn positive returns for long stretches. Borrow yen or dollars at ~1–4%, buy rupees, reais, or pesos at ~6–8%, and pocket the differential as long as the spot rate cooperates.

The catch is the shape of the returns. Carry earns a small, steady positive drip for months or years. Then, in a risk-off spasm (§4.4, §4.7), the funding currency (classically the yen) spikes as everyone unwinds at once, and the carry trade loses in days what it made in years. The old desk description is exact: carry is "picking up nickels in front of a steamroller." The August 2024 yen-carry unwind, when a small Bank of Japan hike plus a US growth scare triggered a violent global deleveraging, is the most recent textbook case (check the specifics).

The deep lesson binds the forces together. The interest differential (Force 1) and safe-haven fear (Force 4) are two faces of the same trade, and the currency's skew, quiet gains and violent losses, is why FX carry is a short-volatility position dressed up as a yield play.

4.9 The eurodollar system and Fed swap lines: the offshore dollar and the world's lender of last resort

There is a piece of dollar plumbing that most people never learn, including many professionals, and understanding it is a genuine analyst edge, because it explains the otherwise-baffling behavior of the dollar in a crisis and the Fed's most important international tool.

What eurodollars are. A eurodollar is a US-dollar-denominated bank deposit or loan held outside the United States, in the offshore banking system. The name is a Cold-War-era accident (it began with dollar deposits held in European banks, partly by the USSR, which wanted dollars but not in US jurisdiction) and has nothing to do with the euro currency. A eurodollar in Singapore or London is still a US dollar, just booked offshore. Over decades this grew into a vast offshore dollar banking system: non-US banks take dollar deposits and make dollar loans; non-US companies and governments borrow dollars by issuing dollar bonds; global trade is invoiced and financed in dollars. The BIS estimates dollar credit to non-bank borrowers outside the US at well over $13 trillion (the number grows). The crucial structural fact: most of the world's dollar borrowing happens outside the US, among borrowers and banks that have no account at the Federal Reserve.

The world runs on dollars it cannot print and, in a crisis, cannot easily obtain.

Why this makes the dollar spike in a crisis (the global dollar shortage). Now the payoff. When a global risk-off hits, everyone who owes dollars needs dollars at once: a Korean bank that funded long-term dollar loans with short-term dollar borrowing, a Brazilian firm with a dollar bond, a European bank holding dollar assets. They need them to roll over maturing dollar debt, to meet margin calls on dollar positions, to avoid default. But the offshore dollar-funding markets freeze precisely when everyone needs them. The result is a global dollar shortage: a scramble for dollars by the whole world simultaneously, which drives the dollar sharply higher even when the crisis originated in the US. This is why the dollar counterintuitively rallied in 2008 (a US-made crisis) and in March 2020. The world was short dollars and had to buy them. It is the mechanical basis of the dollar's safe-haven status (§4.2) and of the "dollar smile" (the dollar tends to strengthen both when the US booms and when the world panics, and to weaken only in the benign middle). For an EM analyst, a surging dollar in a crisis is not a curiosity. It is the tightening of a noose on every dollar-indebted borrower in the world, India's included (Indian firms with external commercial borrowings feel it directly).

Fed swap lines: the backstop. Because a global dollar shortage can freeze the world economy and boomerang back onto the US, the Federal Reserve has become, de facto, the lender of last resort for the entire dollar system, extending the Bagehot role of M7.02 (§4.1) beyond US borders. Its instrument is the central-bank liquidity swap line: the Fed lends dollars to a foreign central bank in exchange for that central bank's currency (at the market rate, reversed at a set date), and the foreign central bank then on-lends those dollars to the banks in its jurisdiction. This relieves the offshore shortage at its source. The Fed maintains standing, unlimited swap lines with five major central banks, the ECB, the Bank of Japan, the Bank of England, the Swiss National Bank, and the Bank of Canada, and it opened temporary lines to a wider set of central banks in both 2008 and 2020 to stop the panic. In 2020 it also created the FIMA repo facility, letting foreign central banks raise dollars against their US Treasury holdings without dumping those Treasuries on the market. Swap lines are one of the most powerful and least-understood crisis tools in the world; when the Fed opened them in March 2020, the global dollar squeeze eased within days. The pointed lesson for an Indian analyst: the RBI is not among the central banks with a Fed swap line, which is exactly why India (and most EMs) must self-insure with the enormous FX reserve buffer of §4.7. Reserves are India's substitute for a swap line it cannot get.

The dollar system protects its insiders and leaves everyone else to hold their own umbrella.

4.10 Commodities as macro signals: oil, gold, and copper

Three commodities carry so much macro information that an analyst reads them as indicators, not just as inputs. Each tells you something different, and each has a specific transmission into the economy, the currency, and asset prices.

Oil: the stagflation shock (and why supply versus demand is everything). Oil is an input to almost everything, from transport and plastics to fertilizer, power and the price of moving every physical good, so its price ripples through the entire cost structure of the economy. The macro danger is a supply-driven oil spike: an OPEC cut, an embargo (1973), a revolution (Iran 1979), a war that threatens supply. A supply shock does two bad things at once. It raises inflation (cost-push, because everything that uses energy gets dearer) and it lowers growth (it acts like a tax on consumers and firms, draining real income and demand). That simultaneous combination of rising prices and falling output is stagflation, and it is the central bank's nightmare precisely because its two mandate legs (M7.02) pull in opposite directions: fight the inflation (hike) and you deepen the slump; support growth (cut) and you feed the inflation. The 1970s (M7.05) are the archetype.

Distinguish this from a demand-driven oil rise, where oil climbs because global growth is booming and demand is strong. That is a symptom of strength, not a shock, and it does not carry the same stagflation threat. The analyst's key skill is reading why oil moved: supply-driven up = stagflation risk, bad; demand-driven up = growth signal, benign-to-good. For India, oil is the single most important macro variable after the Fed. India imports ~85% of its crude, so an oil spike is a quadruple hit: a wider current-account deficit (§4.5), a weaker rupee (terms-of-trade shock, §4.2), higher CPI inflation (fuel is in the headline basket the RBI targets), and fiscal strain (fuel subsidies/lost taxes). "Oil up" is the most reliable single piece of bad news for the Indian macro picture, which is why it headlines M7.07's dashboard.

Gold: the real-rates and anti-fiat asset (what actually drives it). Gold confuses people because they call it an "inflation hedge," which is only loosely and unreliably true (gold can lag inflation for a decade). The better model has two drivers. First and most mechanical: gold's price is driven substantially inversely by the real interest rate. Gold pays no yield, no coupon and no dividend. So the cost of holding gold is the real yield you give up by not holding an inflation-protected bond (a US TIPS). When real yields fall (or go negative), the opportunity cost of holding gold drops and gold becomes more attractive → gold rises; when real yields rise, holding gold becomes expensive → gold falls. Watch the US 10-year TIPS real yield (FRED: DFII10) and you have explained much of gold's price for long stretches: real yields down, gold up.

Second, gold is the anti-fiat / debasement and geopolitical hedge, a store of value that is no one's liability and cannot be printed, frozen, or sanctioned. This driver has surged in importance since 2022: central banks, especially in emerging markets (China's PBoC, the RBI, and others), have been heavy buyers of gold as a non-sanctionable reserve asset, the de-dollarization impulse of §4.8 expressed in bullion. This official demand has, at times recently, partly decoupled gold from the pure real-yield model (gold rising even as real yields did too), which is itself a signal of reserve diversification and geopolitical anxiety. So read gold as real yields (the mechanical driver) + confidence in fiat money and geopolitics (the tail-risk/reserve driver). It is not a simple inflation hedge but a real-rates-plus-fear asset. (For India, gold is also culturally and financially enormous: household gold holdings are vast, and gold imports are a swing factor in the current-account deficit.)

Copper: "Dr. Copper," the growth barometer. Copper is used everywhere industrial activity happens, in wiring, motors, construction, electrical grids, and now electric vehicles and renewable-energy infrastructure, with few good substitutes. Because copper demand tracks the pulse of real physical economic activity in real time, its price is a live barometer of global growth, earning it the nickname "Dr. Copper — the only metal with a PhD in economics." Rising copper = global industrial demand accelerating (growth signal); falling copper = slowdown or recession ahead. Two refinements. China is roughly half of global copper demand, so copper is heavily a China-growth proxy, and a copper move is partly a China read. And copper now carries a structural energy-transition demand story (electrification and the grid are copper-hungry), which underpins a long-run bull case independent of the cycle. Analysts also watch the copper/gold ratio as a single growth-versus-fear gauge: copper (growth) rising relative to gold (fear) signals a risk-on, reflationary, higher-yield environment, and the ratio tends to track bond yields, a neat two-metal summary of the macro regime.

Together the three metals give a compact macro read: oil for inflation/supply stress, gold for real rates and fear, copper for growth. Three prices, three different questions answered.

Common mistakes & how experts think differently

Mistake 1: Getting the quote direction backward. Beginners see "USD/INR rose to 86" and think the rupee got stronger because the number went up. It is the opposite: a rising USD/INR means more rupees per dollar, so a weaker rupee. Experts never read an FX number without attaching units out loud ("86 rupees per dollar") and checking which currency the number is measuring. Half of all FX errors are this error.

Mistake 2: Treating the higher interest rate as free money. "Indian bonds yield 6.5%, US bonds 4.5% — just hold rupees and earn the extra 2%." The novice forgets covered interest parity: the higher-yielding currency trades at a forward discount and is expected to depreciate by roughly the gap, so on a hedged basis the extra yield is exactly given back. The expert knows the carry trade earns money only because uncovered parity fails, and that it pays a steady premium precisely for bearing the risk of a violent, correlated crash (nickels in front of a steamroller). The yield gap is a risk premium, not a gift.

Mistake 3: Using PPP to predict the next quarter. "The Big Mac Index says the rupee is 50% undervalued, so it must rally." PPP is a 5-to-10-year gravitational anchor, useless for short-horizon timing, because currencies stay far from PPP for years, for real structural reasons (non-traded goods, Balassa–Samuelson). The expert uses PPP to judge long-run fair value and rate differentials, flows, and fear to judge the next quarter, and never confuses the two horizons.

Mistake 4: Thinking a current-account deficit is inherently bad. A CA deficit means a country is importing capital, often because the world wants to invest in it. The US and India both run deficits for this reason. The expert never judges the deficit level in isolation; the question is always the quality of the financing: sticky FDI (safe) or flighty hot money (sudden-stop fuel). The same 3% deficit is benign or lethal depending entirely on what finances it.

Mistake 5: Believing a floating currency fully insulates you from the Fed. The old trilemma implies "float and you control your own monetary destiny." Rey's global financial cycle shows this is only partly true: when the Fed eases or the VIX collapses, capital floods into open economies and inflates their credit and assets regardless of the exchange-rate regime. The expert treats US monetary policy as a global variable that reaches into every open economy, and knows that floating buys some, not total, autonomy.

Mistake 6: Expecting the dollar to fall in a global crisis. Intuition says a US-centered crisis should sink the dollar. The opposite usually happens: because the world is short dollars (the eurodollar system, §4.9), a panic triggers a global scramble for dollars that drives it up. The expert reads a crisis-time dollar rally as a dollar-shortage signal, a tightening noose on every dollar-indebted borrower worldwide, not as strength in the US economy.

Mistake 7: Calling gold a simple inflation hedge. Gold can lag inflation for a decade. The expert models gold as a real-rate asset (inversely tied to the TIPS real yield) plus an anti-fiat/geopolitical reserve asset, and watches real yields and central-bank buying, not the CPI print, to understand its price.

Mistake 8: Ignoring the currency leg of a cross-border return. A retail investor sees "my US fund is up 15%" and feels skilled, without noticing that the rupee fell 6% and did nearly half the work. The expert always decomposes a foreign return into the asset return and the FX return, scores them separately, and knows that over one-to-three years the currency can be the larger driver. Skill is what's left after you strip out what the currency did.

How experts think differently, in one line each. They attach units to every FX number. They treat the yield gap as a risk premium, not a gift. They match each currency force to its horizon. They judge deficits by their financing, not their size. They treat the Fed as the world's central bank. They read a crisis-time dollar spike as a global margin call. They model gold by real yields. And they never let a currency tailwind masquerade as investing skill.

Worked examples

Worked Example 1: Covered interest parity: the no-arbitrage forward and the trade that enforces it (India/US)

Setup. Spot USD/INR S = ₹83.00/$. One-year interest rates: rupee r_INR = 7.0%, dollar r_USD = 4.0%. A dealer quotes you a one-year forward of ₹84.20/$. Is it fairly priced? If not, construct the arbitrage.

Step 1: Compute the no-arbitrage forward. F* = S x (1 + r_INR)/(1 + r_USD) = 83.00 x (1.07/1.04) = 83.00 x 1.028846 = ₹85.39/$ (to 2 dp). The fair forward is ₹85.39, so the rupee should trade at a forward discount of 85.39/83.00 − 1 ≈ +2.88% ≈ the 3% rate gap. The higher-yielding rupee is priced to weaken; the extra 3% of rupee interest is offset by ~3% of expected currency loss.

Step 2: Diagnose the quote. The dealer's ₹84.20 is below the fair ₹85.39, so the dealer is pricing the rupee too strong forward (too few rupees per forward dollar). A forward dollar is cheap at the dealer's quote, so the arbitrage is to buy dollars forward cheap from the dealer while doing the offsetting money-market legs.

Step 3: Build the riskless arbitrage (per $1,000,000 notional):

• Borrow $1,000,000 at 4.0% → repay $1,040,000 in one year.
• Convert to rupees at spot: 1,000,000 × 83.00 = ₹83,000,000.
• Invest rupees at 7.0% → 83,000,000 × 1.07 = ₹88,810,000 in one year.
• Today, lock a forward with the dealer to buy back $1,040,000 at ₹84.20 → cost in one year = 1,040,000 × 84.20 = ₹87,568,000.

Step 4: Settle in one year. Collect ₹88,810,000; pay ₹87,568,000 for the dollars; use those dollars to clear the $1,040,000 loan. Riskless profit = 88,810,000 − 87,568,000 = ₹1,242,000, earned on borrowed money, every price locked at the start.

Step 5: The lesson. Traders would do this in size, buying forward dollars from the dealer until F rose to ₹85.39 and the profit vanished, which is why CIP holds in liquid markets. The takeaway to carry: the forward is not a forecast; it is spot dragged by the interest-rate ratio, and the high-yield currency is always at a forward discount.

Worked Example 2: The carry trade: steady nickels, then the steamroller

Setup. A fund runs a classic carry trade: borrow Japanese yen at r_JPY = 0.5% and invest in Indian rupees at r_INR = 7.0%, unhedged. Notional ¥ equivalent of ₹100 crore. Trace two scenarios over one year.

Scenario A: the calm year (spot barely moves). The INR/JPY rate is roughly flat (say the rupee weakens a token 0.5% against the yen).

• Interest earned in rupees: 7.0%. Interest cost in yen: 0.5%. Gross carry = +6.5%.
• Currency: rupee down 0.5% vs yen → −0.5% on the position when converted back to yen.
• Net return ≈ +6.5% − 0.5% = +6.0%. A fat, quiet gain. Do this for three calm years and you compound ~19%, and it feels like free money, which is why carry trades attract enormous leverage.

Scenario B: the risk-off spasm (the funding currency spikes). Late in the year, a global growth scare hits; everyone unwinds carry trades at once; the safe-haven yen surges and the risk-sensitive rupee drops. Say the rupee falls 15% against the yen in weeks.

• Interest still nets +6.5% for the year.
• Currency: −15% (the rupee position, converted back to the now-much-stronger yen, is worth far less).
• Net return ≈ +6.5% − 15% = −8.5%, and levered 5×, that is a −42% loss on capital, in weeks, wiping out years of accumulated carry.

The lesson. Carry pays a steady premium (Scenario A) precisely because it carries a fat, correlated tail risk (Scenario B): the funding currency (yen, or dollars) spikes exactly when everyone needs to unwind, so the losses are violent, simultaneous, and levered. The return distribution is negatively skewed: many small gains, rare huge losses. This is why "picking up nickels in front of a steamroller" is the exact right image, and why carry is really a short-volatility position, not a yield play. The August 2024 yen-carry unwind was Scenario B in the real world.

Worked Example 3: Comparative advantage: gains from trade even when one country is better at everything

Setup. Two countries, two goods (software units and wheat tonnes), one worker's annual output each:

The US is absolutely better at both (8 > 4 software; 8 > 1 wheat). The naive conclusion, "the US should make everything," is wrong. Prove it.

Step 1: Compute opportunity costs (the only thing that matters).

• US: producing 1 software means giving up 1 wheat (both are 8/worker) → opp. cost of software = 1 wheat. Opp. cost of wheat = 1 software.
• India: 4 software = 1 wheat per worker, so 1 software costs 1/4 wheat → opp. cost of software = 0.25 wheat. Opp. cost of wheat = 4 software.

Step 2: Identify comparative advantage (lowest opportunity cost).

• Software: India gives up 0.25 wheat; US gives up 1 wheat. India has the comparative advantage in software (0.25 < 1).
• Wheat: US gives up 1 software; India gives up 4 software. The US has the comparative advantage in wheat (1 < 4).

Each country is relatively better at one good, even though the US is absolutely better at both.

Step 3: Show the gain with a concrete reallocation. Start with 100 workers in each country splitting effort 50/50, then have each specialize by comparative advantage and trade. Illustrate the world-output gain from shifting one worker in each country toward its comparative-advantage good:

• Move 1 US worker from software to wheat: US software −8, US wheat +8.
• Move 1 India worker from wheat to software: India wheat −1, India software +4.
• Net world change: software = −8 + 4 = −4; wheat = +8 − 1 = +7. Wheat rose by 7 for a cost of only 4 software.

Step 4: Confirm both can end up better off via trade. The world now has 7 more wheat and 4 less software. Trade at a price between the two opportunity costs, say 1 software = 0.5 wheat (between India's 0.25 and the US's 1). At that price the US sells wheat for software cheaper than making software itself (it pays 0.5 wheat per software vs its home cost of 1 wheat), and India buys wheat for software cheaper than growing it (0.5 wheat per software received vs its home cost of 4 software per wheat). Both consume beyond their own production frontier. The nations gain in aggregate, the strongest result in economics.

Step 5: State the honest caveat (do not omit this). These are national gains. Within the US, the software workers displaced by Indian imports lose, and if they are not retrained or compensated (as the China shock showed they often are not), their losses are real, concentrated, and durable, even as US consumers gain from cheaper software. Comparative advantage proves the pie grows; it does not promise everyone gets a bigger slice. That is the whole political economy of trade in one sentence.

Worked Example 4: Hedged versus unhedged: when the rupee outweighs the S&P

Setup. The learner, based in India, invests ₹10,00,000 in an S&P 500 index fund at the start of the year. Spot USD/INR = ₹83.00/$. Over the year the S&P returns +12% in dollars. Consider three currency outcomes and compare unhedged versus hedged results. (For hedging, assume the one-year forward locks USD/INR at ₹85.40, the CIP-fair rate for a ~2.9% rate gap, so the hedge locks in ~2.9% of forward points and gives up any rupee move beyond them.)

Convert the starting stake to dollars: ₹10,00,000 ÷ 83.00 = $12,048.19 invested. After +12%: $13,493.98 at year-end.

Outcome 1: rupee depreciates to ₹86.00 (a normal ~3.6% weak-rupee year).

• Unhedged: $13,493.98 × 86.00 = ₹11,60,482 → +16.0% in INR. (Equity +12% plus a currency tailwind of ~+3.6% → ~+16%.)
• Hedged: convert at the locked ₹85.40: $13,493.98 × 85.40 = ₹11,52,386 → +15.2%. Hedging cost the ~0.8% difference here (you gave up the extra depreciation beyond the forward). Unhedged won.

Outcome 2: rupee appreciates to ₹79.00 (a risk-on year, heavy inflows into India).

• Unhedged: $13,493.98 × 79.00 = ₹10,66,024 → +6.6% in INR. The +12% equity gain was cut nearly in half by a strong rupee (−4.8% currency drag).
• Hedged: $13,493.98 × 85.40 = ₹11,52,386 → +15.2%. Hedging won decisively, locking in ₹85.40 and dodging the rupee's rally. Here the currency, not the stock, decided the outcome.

Outcome 3, a crisis year: S&P −18% in dollars, rupee depreciates to ₹92.00 (risk-off, dollar surges).

• Year-end dollars: $12,048.19 × (1 − 0.18) = $9,879.52.
• Unhedged: $9,879.52 × 92.00 = ₹9,08,916 → −9.1% in INR. A −18% US equity year became only −9%, because the depreciating rupee cushioned more than half the loss, exactly the 2022-style dynamic.
• Hedged: $9,879.52 × 85.40 = ₹8,43,711 → −15.6%. Hedging removed the cushion and left the learner with nearly the full dollar loss.

Read the table across. The equity return (+12%, +12%, −18%) was the same within each outcome, yet the INR result swung from +6.6% to +16.0% (Outcomes 1–2, same +12% stock) and the crisis loss was more than halved by the currency (Outcome 3). The FX leg was, in every case, a first-order driver, and sometimes the dominant one. Note too the asymmetry the structural-depreciation logic predicts: unhedged won in the two rupee-weakening outcomes (1 and 3, the more common case for India), and the crisis-year rupee weakness helpfully offset the equity loss. That is why, for a long-horizon Indian investor whose base case is a steadily depreciating rupee, unhedged is the sensible default, accepting Outcome 2's occasional disappointment as the price of the structural tailwind and the crisis-year cushion. Hedging buys certainty: the locked forward already includes the ~2.9%/yr of points (the expected depreciation), so what you surrender is any weakness beyond the forward, above all the crisis-year cushion, plus the rolling costs. It is worth it only when the horizon is short or the money will be spent in rupees soon.

Worked Example 5: The rupee through an exporter's P&L: why a weak rupee flatters Indian IT (India)

Setup. Sahyadri Systems (synthetic) is an Indian IT-services exporter. It bills clients $500m/year (all revenue in dollars), pays $150m of onsite/US delivery costs in dollars, and books ₹2,000 crore of offshore costs (Indian salaries, offices, power) in rupees. Volumes are unchanged year to year, and only the exchange rate moves. Compare the rupee P&L at USD/INR = ₹83 and at ₹87 (a 4.8% rupee depreciation). Throughout, ₹1 crore = 10 million.

Step 1: the strong-rupee year (₹83.00).

• Revenue = $500m × 83 = ₹41,500m = ₹4,150 cr.
• Dollar costs in rupees = $150m × 83 = ₹1,245 cr; total costs = 1,245 + 2,000 = ₹3,245 cr.
• EBIT = 4,150 − 3,245 = ₹905 cr; EBIT margin = 905 ÷ 4,150 = 21.8%.

Step 2: the weak-rupee year (₹87.00).

• Revenue = $500m × 87 = ₹4,350 cr.
• Dollar costs in rupees = $150m × 87 = ₹1,305 cr; total costs = 1,305 + 2,000 = ₹3,305 cr.
• EBIT = 4,350 − 3,305 = ₹1,045 cr; EBIT margin = 1,045 ÷ 4,350 = 24.0%.

Step 3: read the sensitivity. The 4.8% depreciation lifted EBIT by 1,045 − 905 = ₹140 cr (+15.5%) and the margin by 24.0 − 21.8 = 2.2 percentage points, about 46 bps of margin per 1% of depreciation. The engine is the firm's net dollar exposure: revenue of $500m minus dollar costs of $150m = $350m net long USD, so EBIT = 35 × (USD/INR) − 2,000 in ₹ crore, rising ₹35 cr for every ₹1 the dollar gains. (Sahyadri's ~46 bps runs a little hotter than the ~30–40 bps rule of thumb quoted for the large caps in §4.11 precisely because it offshores a larger share of delivery: the more of the cost base sits in rupees, the more a weak rupee drops straight to the bottom line.)

The lesson. An exporter is structurally long the dollar: rupee weakness inflates reported revenue and margin even when nothing operational improved. So when an Indian IT firm posts a fat-margin quarter, the analyst's first move is to strip out the currency. The rupee did some of the work, and only the constant-currency residual belongs to management. This is §4.11 rendered as a P&L, and the exact mirror of Worked Example 6 and of the importer in Practice P17.

Worked Example 6: The strong dollar as a revenue "headwind": reported versus constant-currency (US)

Setup. Beacon Global (synthetic) is a US multinational. Its domestic (US) revenue is $4,000m last year, $4,200m this year (+5%). Its foreign business earns euros: €3,000m last year, €3,150m this year (+5% in euro terms), the same real growth as at home. The euro averaged EUR/USD = 1.10 last year and 1.00 this year (the euro fell about 9 percent, so the dollar rose about 10 percent). Reading only the reported top line, did the business slow?

Step 1: reported (translated) revenue.

• Last year: foreign = €3,000m × 1.10 = $3,300m; total reported = 4,000 + 3,300 = $7,300m.
• This year: foreign = €3,150m × 1.00 = $3,150m; total reported = 4,200 + 3,150 = $7,350m.
• Reported growth = 7,350 ÷ 7,300 − 1 = +0.7%. On the tape, Beacon looks nearly flat.

Step 2: constant-currency revenue (hold FX fixed). Re-translate this year's euro sales at last year's 1.10 rate:

• This-year foreign, constant-currency = €3,150m × 1.10 = $3,465m; total = 4,200 + 3,465 = $7,665m.
• Constant-currency growth = 7,665 ÷ 7,300 − 1 = +5.0%, exactly the underlying rate, since both halves grew 5% in local terms.

Step 3: size the FX drag. The gap is 5.0% − 0.7% ≈ 4.3 percentage points of growth the strong dollar erased in translation. Not a lost sale, just arithmetic: euros converting into fewer dollars. This is why US multinationals (Apple, Coca-Cola, §4.11) headline constant-currency growth every strong-dollar quarter.

The lesson. A US firm earning abroad is effectively short the dollar in its reported results: a strong dollar is a translation headwind, a weak dollar a tailwind, and neither says anything about whether the business is winning customers. Read constant-currency to judge operations; read reported to see the dollars a US shareholder actually pockets. Same FX force as Worked Example 5, opposite sign, because the two firms' currency exposures point opposite ways.

Practice set

Problems 1–4 are guided (framework applied with scaffolding), 5–9 independent, 10–12 exam-condition (12 minutes each, closed book). Rates and levels are "as of mid-2026 — verify"; use the numbers given. Round money to a sensible unit and returns to 0.1%.

P1 (guided: quote direction). Both strings below are market-convention BASE/QUOTE quotes (§4.1), so read each as units of the second currency for one of the first. Over one month, USD/INR moves from ₹83.00 to ₹85.50, and EUR/USD moves from 1.08 to 1.11. For each: (a) which currency strengthened and which weakened? (b) An Indian exporter will receive $1,000,000 next month. Is the USD/INR move good or bad news for the rupee value of that receipt?

Solution. (a) USD/INR 83.00 → 85.50: more rupees per dollar, so the dollar strengthened and the rupee weakened (depreciated ~3.0%). EUR/USD 1.08 → 1.11: more dollars per euro, so the euro strengthened and the dollar weakened vs the euro. (Note the dollar rose vs the rupee but fell vs the euro, because a currency is only ever strong/weak relative to a specific other currency.) (b) Good news for the exporter: $1,000,000 now converts to 1,000,000 × 85.50 = ₹8,55,00,000 instead of ₹8,30,00,000, a ₹25 lakh gain purely from the weaker rupee. A depreciating home currency helps exporters (and, symmetrically, hurts importers).

P2 (guided: covered interest parity). Spot USD/INR = ₹82.00/$. One-year rates: r_INR = 6.0%, r_USD = 4.5%. (a) Compute the one-year no-arbitrage forward. (b) Is the rupee at a forward premium or discount, and by roughly how much? (c) In one sentence, why must the higher-yield currency be at a forward discount?

Solution. (a) F = 82.00 x (1.060/1.045) = 82.00 x 1.014354 = ₹83.18/$ (2 dp). (b) The forward (83.18) is above spot (82.00), so the rupee is at a forward discount of 83.18/82.00 − 1 ≈ +1.44% ≈ the 1.5% rate gap. (c) Because if the higher-yielding rupee were not expected to weaken by ~the yield gap, you could borrow dollars, invest in rupees, hedge with the forward, and earn a riskless profit. Arbitrage forces the forward discount that exactly offsets the extra interest.

P3 (guided: impossible-trinity classification). For each regime, state which two of the three trilemma goals {fixed FX, free capital mobility, independent monetary policy} it chose, and hence which one it gave up: (a) the Eurozone; (b) China during its 2000s rise; (c) the US; (d) Hong Kong's currency board; (e) India's managed float.

Solution. (a) Eurozone: fixed FX (shared currency = permanently fixed among members) + free capital mobility → gave up independent monetary policy (the ECB sets one rate for all; no national monetary policy). (b) China: fixed/managed FX + independent monetary policy → gave up free capital mobility (extensive capital controls make it possible). (c) US: free capital mobility + independent monetary policy → gave up a fixed rate (the dollar floats). (d) Hong Kong currency board: fixed FX (pegged to USD) + free capital mobility → gave up independent monetary policy (imports US monetary policy). (e) India: independent monetary policy + open-ish capital → gave up a fixed rate (managed float: the rupee moves, smoothed by reserves; India straddles the corner by retaining some capital management, but its primary sacrifice is the fixed rate).

P4 (guided: balance-of-payments classification). Use the sixth-edition three-account split of §4.5. Classify each transaction into the current account (CA), the capital account (KA) or the financial account (FA), and note whether it is a credit (inflow) or debit (outflow) for India. (a) India imports $5bn of crude oil; (b) an Indian techie in the US wires $2,000 home to family; (c) a US pension fund buys ₹500 crore of Indian listed equities; (d) Toyota builds a ₹2,000 crore car plant in India; (e) Infosys earns $1bn of software-export revenue.

Solution. (a) CA (trade/goods), debit, because importing goods is a payment out. (b) CA (secondary income / remittance), credit, a transfer into India. (c) FA (portfolio investment / "hot money"), credit, foreign capital buying tradable securities (and note: flighty, can reverse fast). (d) FA (direct investment), credit, foreign direct investment in a controlling operating asset (and note: sticky, cannot flee). (e) CA (services export), credit, earning foreign revenue for a service. Note what is absent: nothing here touches the capital account, which is the small account for capital transfers and non-produced non-financial assets, and P19 supplies the transactions that do. Reflection: (a) is a big CA debit India offsets partly with (e) services credits and (b) remittances; (c) and (d) are how any residual CA deficit gets financed, but (c) could reverse in a sudden stop while (d) cannot.

P5 (independent: relative PPP and structural INR depreciation). India's long-run inflation is ~5.0%; US inflation ~2.5%. Spot USD/INR = ₹84.00. (a) Using relative PPP, what annual rupee depreciation does the inflation gap predict? (b) Roughly where would USD/INR be in 5 years if that held? (c) What does this imply for an Indian holding unhedged US assets, and for a US investor in Indian assets?

Solution. (a) Expected INR depreciation ≈ India inflation − US inflation = 5.0% − 2.5% = ~2.5%/yr (the rupee should weaken ~2.5% a year vs the dollar). (b) Compounding: 84.00 × (1.025)^5 = 84.00 × 1.1314 = ~₹95.0/$ in 5 years (rupee weaker). (c) For the Indian holding unhedged US assets, the ~2.5%/yr rupee depreciation is a structural tailwind: each dollar of US assets translates into more rupees over time, adding to returns. For the US investor in Indian assets, the same 2.5%/yr is a structural headwind: Indian assets must clear a ~2.5%/yr currency drag before the dollar-based investor earns anything. Same force, opposite sign by direction.

P6 (independent: comparative advantage). Two countries, output per worker: Country X makes 6 phones or 12 shirts; Country Y makes 1 phone or 4 shirts. (a) Who has the absolute advantage in each good? (b) Compute opportunity costs and identify each country's comparative advantage. (c) Give a trading price (phones per shirt) at which both gain, and name who exports what.

Solution. (a) X has absolute advantage in both (6 > 1 phones; 12 > 4 shirts). (b) Opportunity costs. X: 1 phone costs 12/6 = 2 shirts; 1 shirt costs 0.5 phones. Y: 1 phone costs 4/1 = 4 shirts; 1 shirt costs 0.25 phones. Compare: for phones, X gives up 2 shirts and Y gives up 4 shirts → X has comparative advantage in phones (2 < 4). For shirts, X gives up 0.5 phones and Y gives up 0.25 phones → Y has comparative advantage in shirts (0.25 < 0.5). (c) A price between the two opportunity costs for phones (between X's 2 shirts and Y's 4 shirts), say 1 phone = 3 shirts, lets both gain: X exports phones, Y exports shirts. X gets 3 shirts per phone (better than its home 2); Y gets a phone for 3 shirts (better than its home 4). Both consume beyond their own frontier despite X being absolutely better at everything.

P7 (independent: tariff incidence). The US imposes a 25% tariff on a $100 imported component. Empirically, pass-through to US prices is near-complete. (a) Who most likely bears the $25 tariff, the foreign exporter or US buyers? (b) Name two second-order costs of a tariff war. (c) Give one legitimate argument for the tariff, stated fairly.

Solution. (a) With near-complete pass-through, US buyers (firms and consumers) bear most of the $25: the imported component now costs ~$125 to American purchasers, so the tariff is largely a tax the US levies on itself; the foreign exporter's price is little changed (some burden can fall on the exporter via lost volume, but the evidence from 2018–19 says most fell on US prices). (b) Two of: retaliation (the other country tariffs US exports, hurting US exporters), supply-chain distortion/rerouting cost, lost specialization (the economy makes things it is relatively bad at), higher input costs for domestic manufacturers using the imported component. (c) Any one, fairly stated: protecting a strategic/national-security industry; infant-industry support until it reaches competitive scale; countering a partner's genuinely unfair practices (dumping, subsidies, forced tech transfer); negotiating leverage; or resilience over pure efficiency after 2020. The honest verdict: tariffs are mostly a tax paid at home but can serve real strategic goals at a measurable cost.

P8 (independent: hedged vs unhedged). An Indian invests $10,000 of US stock at USD/INR = 80.00. Over the year the stock returns −10% in USD and the rupee depreciates to 88.00. The one-year hedge would have locked USD/INR at 82.00. (a) Compute the unhedged return in INR. (b) Compute the hedged return in INR. (c) Which won, and what does it show?

Solution. Start: $10,000 = ₹8,00,000 at 80.00. Year-end dollars: 10,000 × (1 − 0.10) = $9,000. (a) Unhedged: $9,000 × 88.00 = ₹7,92,000 → (792000 − 800000)/800000 = −1.0% in INR. The −10% stock loss was nearly erased by the +10% rupee depreciation. (b) Hedged: $9,000 × 82.00 = ₹7,38,000 → (738000 − 800000)/800000 = −7.75% in INR. (c) Unhedged won (−1.0% vs −7.75%): the depreciating rupee cushioned the equity loss, while hedging removed that cushion. It shows the FX leg can dominate the outcome: here a −10% stock year was almost neutralized by currency for the unhedged Indian investor, exactly the structural-tailwind/crisis-cushion logic.

P9 (independent: sudden stop and the taper tantrum). In 2013, two EMs each ran a ~4% current-account deficit. EM-A financed it mostly with FDI and reserves; EM-B (call it "Fragile") with short-term portfolio inflows. The Fed hints at tapering QE. (a) Which is more exposed, and through what mechanism? (b) Name the "Fragile Five" and India's rough rupee move in 2013. (c) Name three buffers India built afterward.

Solution. (a) EM-B is far more exposed. Mechanism (the sudden stop): the taper hint → global risk-off and higher US yields → portfolio "hot money" reverses and flees EM-B → its currency collapses → any dollar debt balloons in local terms → balance sheets blow up and the central bank is forced to hike into the downturn. EM-A's FDI financing cannot flee overnight and its reserves absorb the shock, so the same deficit is far safer, because the quality of financing, not the deficit size, decides. (b) The "Fragile Five": Brazil, India, Indonesia, South Africa, Turkey. India's rupee fell roughly 20%, from ~₹55 to ~₹68/$, in mid-2013. (c) Any three: FX reserves rebuilt from ~$275bn to ~$690bn (~10–11 months' import cover); CAD narrowed from ~4.8% of GDP to ~1–1.5%; formal flexible inflation targeting (2016) anchored inflation; a higher-FDI/remittance financing mix; JPMorgan bond-index inclusion (2024) broadening the investor base; a more credible, better-communicating RBI.

P10 (exam, 12 min: the dollar in a crisis). A global risk-off event hits: the VIX triples, EM currencies fall, and, puzzlingly to a beginner, the US dollar rises sharply even though the shock began in the US. (a) Explain, using the eurodollar system, why the dollar rises in a global panic. (b) What is the Fed's tool for relieving the global dollar shortage, and how does it work? (c) Why is India not directly protected by that tool, and what does India rely on instead?

Solution. (a) Most of the world's dollar borrowing happens outside the US, among banks and firms with no Fed account, in the vast eurodollar (offshore-dollar) system, with $13tn+ of dollar credit to non-US borrowers. In a panic, all of them need dollars at once, to roll maturing dollar debt, meet margin calls and avoid default, while offshore dollar-funding markets freeze. This global dollar shortage is a worldwide scramble for dollars, which drives the dollar up even though the crisis is US-made (the world is short dollars). A crisis-time dollar spike is a dollar-shortage signal, not US-economy strength. (b) Fed central-bank liquidity swap lines: the Fed lends dollars to a foreign central bank against its currency (reversed later at a set rate), and that central bank on-lends the dollars to banks in its jurisdiction, relieving the offshore shortage at its source. Standing unlimited lines exist with 5 major central banks (ECB, BoJ, BoE, SNB, BoC); temporary lines and the FIMA repo facility extended the backstop in 2008 and 2020. (c) The RBI is not among the central banks with a Fed swap line, so India has no direct access to Fed dollars in a squeeze; it therefore self-insures with its large FX reserve buffer (~$690bn), and reserves are India's substitute for the swap line it cannot get. This is the structural inequality of the dollar system: it protects its insiders and leaves EMs to hold their own umbrella.

P11 (exam, 12 min: reserve status, privilege, Triffin, de-dollarization). (a) State the dollar's approximate reserve share and two elements of "exorbitant privilege." (b) State the Triffin dilemma precisely. (c) Give one point for and one against the view that the dollar will be dethroned within a decade, and a balanced verdict.

Solution. (a) The dollar is roughly 57–58% of allocated official reserves (IMF COFER). Two of "exorbitant privilege": the US borrows cheaply (permanent world demand for Treasuries holds down its yields); it borrows in its own currency (no "original sin": it cannot be forced into dollar default like an EM); seigniorage on foreign-held cash; the banker's return (borrow cheap via safe Treasuries, invest in higher-return risky foreign assets, earning a positive net return despite being a net debtor); sanctions power over the dollar clearing system. (b) Triffin dilemma: the world needs an ever-growing supply of the reserve asset (dollars/Treasuries), which the US can supply only by running persistent deficits to pump dollars out, but the larger and longer those deficits accumulate, the more the world may eventually doubt the dollar's stability and value. Short-run global need (more dollars) conflicts with long-run confidence (sound dollars), a built-in structural tension rather than a policy error. (c) For dethroning: weaponization (2022 reserve freeze) pushes diversification; alternatives are being built (RMB/CIPS, BRICS local-currency trade, central-bank gold buying); the US fiscal path invites doubt. Against: network effects and unmatched market depth/trust; no convertible alternative at scale (RMB has capital controls; euro area lacks a unified safe asset; gold isn't a payment system); the reserve share has drifted down only slowly (~71% in 2000 → ~58%), and mostly into small currencies and gold, not the RMB (~2%). Balanced verdict: gradual erosion at the margin (a slowly more multipolar, more gold-diversified reserve system) is plausible and arguably underway; a sudden dethroning is unlikely absent a catastrophic, self-inflicted US failure.

P12 (exam, 12 min: commodities as signals). (a) Distinguish a supply-driven from a demand-driven oil spike and their opposite implications, and explain why a supply-driven spike is a central bank's nightmare. (b) The US 10-year real (TIPS) yield falls from +1.5% to −0.5%; what does the real-rate model predict for gold, and why? (c) Copper falls 20% while gold rises; what does the copper/gold move signal about the macro regime?

Solution. (a) A supply-driven oil spike (OPEC cut, war, embargo) raises inflation (cost-push, since energy feeds everything) and lowers growth (a tax on real incomes) simultaneously → stagflation. A demand-driven oil rise (booming global growth pulling demand up) is a symptom of strength, not a shock, and does not threaten stagflation. A supply-driven spike is the central bank's nightmare because its two mandate legs conflict: hike to fight inflation → deepen the slump; cut to support growth → feed the inflation. There is no good move. (For India, a supply-driven spike is worse still: wider CAD, weaker rupee, higher CPI, fiscal strain, all at once.) (b) The real-rate model predicts gold rises. Gold pays no yield, so its holding cost is the real yield forgone; when the real yield falls from +1.5% to −0.5%, the opportunity cost of holding gold drops sharply (you now lose purchasing power in a real bond), making no-yield gold relatively far more attractive → gold up. (c) Copper down + gold up = a risk-off, slowing-growth, fear-rising regime: copper (the growth barometer, "Dr. Copper") falling says global/China industrial demand is weakening, while gold rising says real yields are falling and/or fear and reserve-diversification demand is rising. A falling copper/gold ratio signals deceleration and risk aversion, typically consistent with falling bond yields and a defensive market, the opposite of a reflationary, risk-on regime.

P13 (guided: PPP, absolute and relative). A representative consumption basket costs $80 in the US and ₹5,600 in India (synthetic). Spot USD/INR = ₹84.00. India's long-run inflation is 5.5%, the US's 2.5%. (a) Compute the absolute-PPP-implied exchange rate. (b) Is the rupee cheap or dear versus that rate, by how much, and why is the gap not a free-arbitrage signal? (c) Using relative PPP, project USD/INR in 5 years.

Solution. (a) Absolute PPP equalizes the basket's price: ₹5,600 ÷ $80 = ₹70/$. (b) Spot ₹84 is above the PPP rate ₹70, since it takes more rupees to buy a dollar than PPP says, so the rupee looks ~20% cheap (undervalued) versus absolute PPP (84 ÷ 70 − 1 = +20%; equivalently ~16.7% below on the 1 − 70/84 convention). It is not an arbitrage: most of the basket (housing, haircuts, healthcare) is non-traded, and the Balassa–Samuelson effect means a poorer, faster-growing economy structurally carries a lower price level, so its currency looks "undervalued" on PPP more or less permanently, with no tradable profit. (c) Relative-PPP drift = 5.5% − 2.5% = 3.0%/yr; in 5 years 84 × 1.03^5 = 84 × 1.15927 = ₹97.4/$. PPP anchors the multi-year trend, not a level you can trade this quarter.

P14 (guided: the real rate and the REER). Over 5 years USD/INR moved from ₹74 to ₹84 (the rupee fell nominally). Over the same window India's price level rose to an index of 132 (from 100) and the US's to 112. (a) Compute the nominal depreciation. (b) Compute the real exchange rate at each date (real = nominal × foreign-price-level ÷ domestic-price-level, in ₹/$) and say what happened in real terms. (c) Express it as a REER index (base 100, up = dearer) and give the one-line competitiveness verdict.

Solution. (a) Nominal: 84 ÷ 74 − 1 = +13.5%, so the rupee depreciated 13.5% on the screen. (b) Real rate q = S × P_US ÷ P_India: q₀ = 74 × 100 ÷ 100 = 74.00; q₅ = 84 × 112 ÷ 132 = 71.27. The real rate fell from 74.00 to 71.27, a 3.7% real appreciation of the rupee (a lower ₹/$ real rate means the rupee got dearer in real terms), the opposite of the nominal move. (c) Inverting to a "value of the rupee" index: 74.00 ÷ 71.27 × 100 = 103.8, so REER ≈ 103.8, ~3.8% above base, i.e., the rupee is ~4% overvalued versus 5 years ago. Verdict: a 13.5% nominal fall was more than swallowed by the 20-point cumulative inflation gap, leaving exporters less competitive, not more; nominal moves are the news, and the REER is the truth.

P15 (independent: import cover and reserve adequacy). Use these illustrative figures (check against the latest RBI release): FX reserves $640bn, annual merchandise imports $720bn, short-term external debt (residual maturity) $280bn. (a) Compute import cover in months. (b) Compute the reserves-to-short-term-debt ratio and judge it against the Guidotti–Greenspan rule. (c) One-line verdict on adequacy, and connect it to why India holds such a buffer.

Solution. (a) Import cover = reserves ÷ (annual imports ÷ 12) = 640 ÷ (720 ÷ 12) = 640 ÷ 60 = 10.7 months, far above the traditional ~3-month floor. (b) 640 ÷ 280 = 2.29×, so reserves cover short-term external debt 2.3 times over, comfortably clearing the Guidotti–Greenspan benchmark of ≥1.0× (able to repay a year of short-term debt with no new borrowing). (c) Robust: ~10.7 months of cover and >2× short-term debt is the post-2013 war chest of §4.7, and it is India's self-insurance substitute for the Fed swap line it does not have (§4.9). A large buffer is how an EM without a swap line survives a sudden stop.

P16 (independent: uncovered interest parity and the carry P&L). One-year rates: r_INR = 7.5%, r_USD = 4.5%; spot USD/INR = ₹80.00. A fund borrows dollars and invests in rupees, unhedged, on $1,000,000 of notional. (a) What end-of-year spot does uncovered interest parity predict (the break-even)? (b) If the rupee actually ends at ₹81.00, compute the trade's dollar P&L and return. (c) If instead a risk-off spasm sends it to ₹90.00, compute the P&L, and state the general lesson.

Solution. Convert and invest: $1,000,000 × 80 = ₹8,00,00,000, grown at 7.5% → ₹8,60,00,000 at year-end; the dollar loan owes $1,000,000 × 1.045 = $1,045,000. (a) UIP predicts the rupee depreciates by ~the rate gap, to E[S] = 80 × 1.075 ÷ 1.045 = ₹82.30/$, at which the carry exactly breaks even. (b) At ₹81.00: convert back ₹8,60,00,000 ÷ 81 = $1,061,728; minus the $1,045,000 repayment = +$16,728, or +1.67%. The trade made money because the rupee fell less (to 81) than UIP predicted (82.30), the empirical failure of UIP that funds all carry. (c) At ₹90.00: ₹8,60,00,000 ÷ 90 = $955,556; minus $1,045,000 = −$89,444, or −8.94% in a single move. Lesson: carry earns a steady positive drip precisely because it wears a fat, sudden tail: picking up nickels in front of a steamroller (§4.3).

P17 (independent: the importer / dollar-borrower P&L). Garuda Air (synthetic) is an Indian airline. Annual dollar costs (jet fuel + USD aircraft leases) are $1,200m; domestic ticket revenue is ₹36,000 crore; other rupee costs are ₹18,000 crore. Compare EBIT at USD/INR = ₹80 and ₹88 (a 10% rupee depreciation). (a) Compute EBIT and margin at each. (b) Give EBIT's sensitivity per ₹1 of USD/INR. (c) Contrast with the exporter of Worked Example 5.

Solution. (a) At ₹80: dollar costs = $1,200m × 80 = ₹9,600 cr; total costs = 9,600 + 18,000 = ₹27,600 cr; EBIT = 36,000 − 27,600 = ₹8,400 cr (23.3% margin). At ₹88: dollar costs = $1,200m × 88 = ₹10,560 cr; total = ₹28,560 cr; EBIT = ₹7,440 cr (20.7% margin). The 10% rupee fall cut EBIT by ₹960 cr (−11.4%) and margin by 2.7 points. (b) EBIT = 18,000 − 120 × (USD/INR) in ₹ crore, so EBIT falls ₹120 cr for every ₹1 the dollar gains. (c) Garuda is short the dollar (dollar costs, rupee revenue): the exact rupee weakness that helped Sahyadri (WE5) hurts Garuda, and the dollar leases also swell its balance-sheet lease liability, a second hit the P&L alone misses. Same currency move, opposite sign, set by the direction of the exposure.

P18 (stretch, 12 min: the full cross-border decomposition). The learner invests ₹20,00,000 in an S&P 500 fund at USD/INR = ₹80.00. Over the year the fund returns +15% in USD and USD/INR ends at ₹84.00. One-year rates: r_INR = 7.0%, r_USD = 4.0%; a one-year hedge was available at the CIP-fair forward. (a) Compute the CIP-fair forward. (b) Compute the unhedged rupee return and decompose it into asset, currency, and cross terms. (c) Compute the hedged rupee return. (d) Which won, by how much, and reconcile it with the structural-tailwind logic.

Solution. (a) F = 80 × 1.07 ÷ 1.04 = ₹82.31/$. (b) Unhedged: (1.15) × (84 ÷ 80) − 1 = 1.15 × 1.05 − 1 = +20.75%. Decompose: asset +15%, currency +5% (84/80 − 1), cross term 0.15 × 0.05 = +0.75% → 15 + 5 + 0.75 = +20.75%. (c) Hedged locks the forward: (1.15) × (82.31 ÷ 80) − 1 = 1.15 × 1.02885 − 1 = +18.32% (≈ the +15% asset plus the ~2.89% forward points). (d) Unhedged won, by ~2.4 points (20.75% vs 18.32%). Why: the rupee's actual 5.0% depreciation beat the ~2.89% of forward points CIP had already priced in, so the unhedged investor pocketed the extra ~2.1% of beyond-forward weakness. This is §4.11's default: over the long run the rupee's structural slide is a tailwind, and in the years it falls hardest the unhedged investor is most rewarded; hedge only when the horizon is short or the money will be spent soon in rupees.

P19 (guided: the three accounts, and the identity that ties them). Use the sixth-edition split of §4.5. Place each item in the current account, the capital account or the financial account, and say which side of the identity it lands on. (a) A creditor government writes off $1.2bn of the country's sovereign debt. (b) A domestic firm sells a patent outright to a foreign buyer for $300m, with no continuing royalty. (c) Foreign investors buy $9bn of the country's listed equities and bonds. (d) The central bank's reserve assets rise by $6.5bn over the year. Now take a year with a current-account deficit of $25.0bn, a capital-account surplus of $0.5bn and errors and omissions of zero. Net liabilities incurred are $31.0bn (FDI $18.0bn, portfolio $9.0bn, external commercial borrowing $4.0bn). (e) Compute the change in reserve assets and show that the identity closes.

Solution. (a) Capital account, credit. Debt forgiveness is a capital transfer, and it is the textbook case for the account most people forget exists. (b) Capital account, credit. An outright disposal of a non-produced, non-financial asset is a capital-account item; a royalty stream on the same patent would instead be current-account services, which is the distinction being tested. (c) Financial account, an incurrence of liabilities to foreigners, so it reduces net lending. (d) Financial account, an acquisition of foreign assets by the central bank, so it raises net lending; reserves sit inside the account, not beside it. (e) Net lending must equal CA + KA + EO = −25.0 + 0.5 + 0 = −$24.5bn. Net lending also equals assets acquired minus liabilities incurred, so with reserves as the only asset line, R − 31.0 = −24.5, giving R = +$6.5bn: reserves rise by $6.5bn. Read it: the country borrowed $31.0bn from the world, needed $24.5bn of it to cover the external gap, and parked the remaining $6.5bn in reserves. That residual is exactly the buffer of §4.7, and it is why a country can run a deficit and build reserves in the same year without any contradiction.

Applied mini-project: the rupee, the dollar, and a cross-border portfolio

Task. Assemble the real data behind all of this and write a three-part "external-sector and currency" brief on India and on your own cross-border exposure. This is the exact material M7.07's dashboard formalizes, and the currency-exposure skill every Phase-8 teardown of an exporter, importer, or dollar-borrower will assume you own. Do it on live data.

Data to pull (all free, all primary):

• The rupee and its REER: USD/INR from the RBI (rbi.org.in → Database on the Indian Economy, DBIE) or FRED (DEXINUS); the RBI 40-currency REER index (RBI DBIE / RBI Bulletin). Pull ~15 years of monthly USD/INR.
• Inflation for the PPP test: India CPI (MOSPI / RBI) and US CPI (FRED CPIAUCSL) or PCE, enough to compute the 15-year cumulative inflation gap.
• India's balance of payments: the latest RBI BoP release, for the current-account balance (as % of GDP), the trade/services split, remittances, and the FDI-vs-FPI financing mix. Note the CAD %, the FX reserves level and import cover.
• The dollar and reserve status: the broad dollar index (FRED DTWEXBGS); the IMF COFER table for the dollar's reserve share; the US 10-year real yield (FRED DFII10) for the gold linkage.
• Commodities: spot Brent oil, gold, and copper (any free source), plus the copper/gold ratio you compute.
• Your own (or a model) cross-border position: pick a US index or stock you hold or would hold (e.g., an S&P 500 fund) and one Indian company with clear FX exposure (an IT exporter like Infosys/TCS, or an importer/dollar-borrower like IndiGo), then pull the FX-exposure/hedging note from its latest annual report.

Required contents.

1. The PPP test (a table + one chart). Over your 15-year window, compute the actual cumulative USD/INR depreciation and compare it to the cumulative India-minus-US inflation gap. Does relative PPP hold as a long-run anchor (the rupee's depreciation ≈ the inflation gap)? State the fit and the largest deviations, and say what drove them (oil shocks, taper tantrum, risk cycles).
2. India's external health scorecard. Report the current CAD %, FX reserves and import cover, the FDI-vs-FPI financing mix, and the REER level (over/undervalued vs 100). Give a one-line verdict: is India's external position robust (2020s buffers) or fragile (2013 profile), and why? Then name its single largest residual vulnerability (hint: §4.10).
3. The cross-border return decomposition. For your chosen US holding over the last 1 and 3 years, split the total INR return into the asset return in USD and the currency return (USD/INR move), showing FX's contribution. Then compute what a hedged version would have returned (the USD asset return plus ~the rate-gap forward points, per CIP). State whether unhedged or hedged won, and reconcile it with the structural-tailwind logic of §4.11. Separately, for your chosen Indian company, quote one line from its annual report on its FX exposure/hedging and classify it as long or short the dollar.

Scoring rubric (self-score honestly; pass ≥11/14):

Case Lab: run case CL-7.1 in the app's Cases tab for this module: a cross-border case that walks the external-sector scorecard and the hedged-versus-unhedged decision end to end on a rupee/dollar portfolio.

Reading & resources

• Barry Eichengreen, Exorbitant Privilege: The Rise and Fall of the Dollar (2011). The definitive, readable history of how the dollar became the world's currency, what the privilege is, and the case for and against its decline. The single best book for §4.8. [Paid] [Intermediate]
• Hélène Rey, "Dilemma not Trilemma: The Global Financial Cycle and Monetary Policy Independence" (Jackson Hole, 2013). The paper behind §4.4 and §4.7; free PDF from the Kansas City Fed / NBER. Read the introduction and conclusions even if the econometrics is heavy, because the argument is what matters. [Free] [Advanced]
• Krugman, Obstfeld & Melitz, International Economics: Theory and Policy. The standard, excellent textbook for comparative advantage, the BoP, exchange-rate determination, and the trilemma (§4.1–4.6). Read the trade chapters and the exchange-rate/BoP chapters. [Paid] [Intermediate]
• Aswath Damodaran: his posts and classes on currencies, country risk, and "the value of a currency" (free on his NYU page and blog). The bridge from FX and inflation differentials to valuation and the cost of capital, from the valuation master; the perfect complement to §4.11 and M3.02. [Free] [Intermediate]
• RBI: Database on the Indian Economy (DBIE), the Monetary Policy Report, and the BoP releases (rbi.org.in). Your permanent primary source for USD/INR, the 40-currency REER, FX reserves, the current-account balance, and FPI/FDI flows. The raw material for the mini-project and M7.07's dashboard. [Free] [Intermediate]
• IMF COFER (Currency Composition of Official Foreign Exchange Reserves). The primary source for the dollar's ~57–58% reserve share and its slow drift; updated quarterly. Verify the number here rather than quoting from memory. [Free] [Intermediate]
• BIS: the Triennial FX Survey, the "US dollar funding" and "global liquidity" work, and the quarterly review (bis.org). The authority on FX market size, the eurodollar/offshore-dollar system, and dollar credit to non-US borrowers (§4.9). [Free] [Advanced]
• FRED (fred.stlouisfed.org). Pull and plot the series yourself: DTWEXBGS (broad dollar index), DEXINUS (USD/INR), DFII10 (10-yr real/TIPS yield, the gold driver), DCOILBRENTEU (Brent), VIXCLS (the fear gauge behind the global financial cycle). Plotting the dollar against the real yield and against oil teaches more than any summary. [Free] [Beginner]
• Perry Mehrling: "The New Lombard Street" and his free "Economics of Money and Banking" course (Coursera / YouTube). The "money view" of the eurodollar system and the dealer-of-last-resort role (§4.9); the deepest free treatment of dollar plumbing. [Free] [Advanced]
• Michael Pettis, The Great Rebalancing and his blog (China Financial Markets). The sharpest evenhanded treatment of trade imbalances, savings gluts, and the twin-deficits logic (§4.5–4.6); it argues the causation runs in ways that challenge naive trade narratives. [Paid book / Free blog] [Advanced]
• Brad Setser, "Follow the Money" (Council on Foreign Relations blog). The best free running commentary on global capital flows, reserve accumulation, and hidden dollar exposures; read a few posts to see how a top flow-analyst actually reasons. [Free] [Advanced]
• The Economist's Big Mac Index (interactive, free). The friendly on-ramp to PPP (§4.2); play with it, then read the methodology note on why it is a rough long-run guide, not a trade signal. [Free] [Beginner]
• Zerodha Varsity, "Currency, Commodity, and Government Securities" (India). Free, India-specific treatment of the currency and commodity markets, the rupee, and how FX is quoted and traded on Indian exchanges. [Free] [Beginner]
• Odd Lots (Bloomberg podcast). Search episodes on the eurodollar system, the dollar-funding squeeze, de-dollarization, and the yen carry trade: practitioners arguing about these live questions in plain language. [Free] [Intermediate]
• Going deeper: Gita Gopinath, "The International Price System" (Jackson Hole, 2015; free PDF from the Kansas City Fed / NBER). The dominant-currency-pricing paradigm: why roughly half of world trade is invoiced in dollars regardless of who is trading with whom, and what that does to how exchange-rate moves pass into trade prices and inflation. The rigorous upgrade to §4.2, §4.6, and §4.8. [Free] [Advanced]
• Going deeper: Maurice Obstfeld & Alan M. Taylor, Global Capital Markets: Integration, Crisis, and Growth. The definitive data-rich history of the trilemma: how the fixed-rate / open-capital / monetary-autonomy choice actually played out across the gold standard, Bretton Woods, and the modern float, and why rising capital mobility keeps forcing the trilemma's hand. The deep backing for §4.4. [Paid] [Advanced]

The Modern Analyst's Addendum

Everything above teaches this skill from first principles, by hand. That is how you learn it, and the mastery check still tests it that way. This addendum shows how a working analyst amplifies the same skill today. It adds; it never replaces. (R1/R10)

AI-Augment this skill

{
  "skill": "Quote conventions and the nominal-versus-real distinction, the five forces and their horizons, covered and uncovered interest parity and the carry trade, the impossible trinity, the balance of payments and the quality of its financing, sudden stops, the dollar system and the eurodollar market, and currency risk for a cross-border investor",
  "use": "FX is the module where the arithmetic is unforgiving and the conventions are where people actually lose money, so the split is sharp: use a tool for the conventions and the plumbing, never for the level. (1) UNIT-AND-DIRECTION CHECKING, which sounds trivial and is Mistake 1 for a reason. Paste a sentence and ask it to restate every rate with explicit units ('84 rupees per dollar') and to say which currency strengthened; a rising USD/INR is a weakening rupee, and a tool that restates it in units catches an inversion you will otherwise carry through an entire analysis. (2) EXPOSURE MAPPING FROM FILINGS — 'read this annual report and list every currency exposure: revenue by currency, cost by currency, debt by currency, hedges outstanding and their tenor'. That list is what §4.11's company-level section asks for, it is tedious, it is in the notes, and it is exactly the work a document tool shortens. Then check each line against the note. (3) BOP AND PARITY BOOKKEEPING — give it current-account and capital-account components and ask it to close the balance of payments and name the residual, or give it two rates and a spot and ask for the CIP forward WITH the derivation shown. Both are checkable in one line. (4) REGIME CLASSIFICATION — 'which corner of the trilemma does this country's regime occupy, and what does that force about its monetary autonomy'. What it must never supply: the spot rate, the forward premium, the level of reserves, the REER index level, or this quarter's current-account balance. Those move daily to quarterly, a stale one is invisible, and every single one is published free by the RBI, FBIL or FRED.",
  "tools": ["Chat assistants — Claude, ChatGPT for unit-and-direction checking, BOP and parity bookkeeping, and trilemma regime classification", "Document/PDF AI — for annual-report currency-exposure notes, hedge schedules, the RBI Bulletin's BOP and NEER/REER tables, and IMF Article IV external-sector assessments", "Coding assistants — for the CIP engine, the ex-post carry decomposition and the REER arithmetic below"],
  "prompt": "For each rate I paste, restate it with explicit units ('X rupees per dollar') and say which currency strengthened between the two dates. Then, from this annual report, list every currency exposure the company carries — revenue by currency, cost by currency, borrowings by currency, and any hedges with their notional and tenor — with the note number for each line. Do not tell me the current spot rate, forward premium, reserve level or REER index; I pull those from FBIL, the RBI Weekly Statistical Supplement and FRED myself.",
  "verify": "Three checks, and the first is the one that pays for itself. (1) Restate every rate with units and re-derive the direction: half of all FX errors are the quote inversion of Mistake 1, and it survives every downstream check because the arithmetic is still internally consistent — just backwards. (2) Any parity number must be computed exactly, not by the shortcut: the CIP forward premium is (r_home − r_foreign)/(1 + r_foreign), which at Indian rate levels differs from the plain rate gap by more than ten basis points, and a desk quoting to the basis point cannot use the shortcut. Compute the fair forward, then back out the interest rate the quoted forward implies, and express the gap as a basis. (3) Reconcile any long-run currency claim against its own endpoints: a stated depreciation rate and a pair of dated levels are two expressions of the same fact, and if the compound annual rate implied by the endpoints does not match the stated rate, one of them is wrong. That check takes one line and catches an error that otherwise propagates into every long-horizon return estimate you build. Then the judgement: any hedged-versus-unhedged conclusion must be stated jointly with the horizon, because §4.11's answer flips on it.",
  "diy": "The gate is unaided: you read a quote with its units and name which currency strengthened, compute a CIP forward and the arbitrage that enforces it, convert a foreign-asset return into home currency and separate the asset leg from the currency leg, distinguish a nominal move from a real one and say what a REER above its base means, place a country in the trilemma, and read an exporter's or importer's net currency exposure off its P&L. Nothing in this callout is admissible in that room."
}

Modern Data Analysis

By hand first. You computed a CIP forward and then built the arbitrage that enforces it, rupee by rupee; you traced a carry trade through a calm year and a spasm; you converted an S&P return into rupees under three currency outcomes and watched the FX leg dominate the equity leg; you ran an exporter's P&L at two exchange rates and a US multinational's revenue at two. Keep the arbitrage construction especially. Anyone can quote the formula, and only the person who has walked the four legs understands why the forward is a price rather than a forecast.

Today's workflow. Three changes. First, every parity gets computed exactly and then inverted. The forward is S × (1 + r_home)/(1 + r_foreign), and the useful modern move is to run it backwards: take the quoted forward and back out the interest rate it implies, then compare that to the cash rate. The difference is a basis, expressed in basis points, and it is a real quantity that widens when dollar funding is scarce. §4.3's parenthetical about the CIP basis is a series you can construct rather than a caveat you nod at. Second, every cross-border return is decomposed before it is discussed. Return_INR = (1 + asset return) × (1 + currency move) − 1 is a product, so the two legs are separable and must be separated: the asset leg is skill or luck about the asset, the currency leg is a position you took whether you meant to or not. Attributing a good year to stock selection without stripping the FX leg is the most common self-deception in cross-border investing, and it is a two-line calculation to avoid. Third, nominal and real are held as separate series, always. The nominal rate is what the screen shows; the REER is what competitiveness does; and the third quantity, the nominal depreciation that would exactly offset the inflation gap, is what tells you whether a "weak rupee" headline describes a gain in competitiveness or merely keeping pace. All three are published monthly.

Tools & sources (IN + US). numpy and pandas for the parity engine and the return decomposition, matplotlib for the NEER/REER pair; India: FBIL (fbil.org.in) for the daily USD/INR reference rate and the published forward premia, which are the two inputs CIP needs; CCIL for FX and forward-market data; the RBI's NEER and REER indices, published monthly in the RBI Bulletin and DBIE in both a 40-currency and a 6-currency version (trade-weighted and export-weighted), which is §4.1's competitiveness gauge, already computed; the RBI Weekly Statistical Supplement for foreign exchange reserves week by week, and the RBI Bulletin for the net forward position (the forward book), which is the disclosure that turns a headline reserve number into a usable one, because reserves committed forward are not reserves available today; the RBI's quarterly Balance of Payments release for the current account, the capital and financial account and the errors-and-omissions residual; DGCI&S monthly merchandise trade data via the Ministry of Commerce release; NSDL and CDSL for daily FPI flows into equity and debt, free, daily, and genuinely better than the US equivalent; and the RBI's half-yearly Report on Management of Foreign Exchange Reserves for composition and adequacy; US: FRED by id for DEXINUS (the USD/INR daily rate), DTWEXBGS (the nominal broad dollar index) with its advanced-economy and emerging-market sub-indices, DEXUSEU and DEXJPUS, DGS1 and SOFR for the dollar leg of parity; the BIS effective exchange rate indices (nominal and real, broad and narrow, monthly, free) for a REER computed on a methodology comparable across countries, worth holding alongside the RBI's because they will not agree; BEA International Transactions for the US current account; Treasury International Capital (TIC) data for cross-border securities flows; the IMF's COFER database for the reserve-currency shares §4.8 discusses; and the New York Fed's weekly central-bank liquidity swap data for §4.9's swap lines. The free-versus-paid line, and here India is unusually well served. FBIL's reference rate and forward premia, the RBI's NEER/REER, the weekly reserves and the forward book, and NSDL's daily FPI flows are all free, and the daily FPI series in particular has no free US counterpart, since TIC is monthly and arrives with a long lag. What is genuinely paid on both sides is depth and history: intraday FX, a long tick history, the option-implied volatility surface, and any ready-made cross-currency basis series. Those live on Bloomberg or Refinitiv. So the Indian analyst can build the parity and flow work from free sources and will have to construct the basis series by hand from FBIL forwards and cash rates, which is exactly the exercise below.

# CIP exactly, the implied basis, and the ex-post carry — recomputed in-session (R3)
fwd  = lambda S, rh, rf: S * (1 + rh) / (1 + rf)          # S = home per foreign
prem = lambda rh, rf: (rh - rf) / (1 + rf)                # NOT rh - rf
S, rh, rf, quoted = 83.00, 0.070, 0.040, 84.20            # WE1
F = fwd(S, rh, rf)
implied = (quoted / S) * (1 + rf) - 1                     # invert the quote for r_home
print(f"{F:.4f} {prem(rh,rf):.6%} {(rh-rf-prem(rh,rf))*1e4:.1f}bp "
      f"| basis {(rh-implied)*1e4:.1f}bp")
#      85.3942   2.884615%   11.5bp | basis 149.6bp
carry = lambda rh, rf, dep: (1 + rh) / (1 + rf) / (1 + dep) - 1   # ex-post, in the funding ccy
print(" ".join(f"{carry(rh,rf,d):+.3%}" for d in (0.00, 0.03, 0.035)))
#      +2.885% -0.112% -0.595%   <- breakeven depreciation IS the forward premium
reer = lambda nom_dep, pi_d, pi_f: (1 + pi_d) / ((1 + nom_dep) * (1 + pi_f)) - 1
print(f"{reer(0.0, .045, .025):+.4%} {reer(0.019512, .045, .025):+.4%} {(1.045/1.025)**10-1:+.1%}")
#      +1.9512%              +0.0000%              +21.3%

Verify. Prove the engine on the module's own worked figures. Worked Example 1: fair forward 83.00 × 1.07/1.04 = ₹85.3942, forward premium 2.884615%, arbitrage legs $1,040,000 owed, ₹83,000,000 → ₹88,810,000, forward cost ₹87,568,000, profit ₹12,42,000. Section 4.3: 84.00 × 1.065/1.045 = ₹85.6077, premium 1.913876%. Worked Example 2: net +6.0% in the calm year (compounding to +19.1% over three) and −8.5% in the spasm, −42.5% at five times leverage. Worked Example 3: opportunity costs of one software unit, 1.00 wheat in the US against 0.25 in India. Worked Example 4: $12,048.19 invested becoming $13,493.98; unhedged ₹11,60,482 (+16.0%), ₹10,66,024 (+6.6%) and ₹9,08,916 (−9.1%) against a hedged ₹11,52,386 (+15.2%), ₹11,52,386 (+15.2%) and ₹8,43,711 (−15.6%). Worked Example 5: EBIT ₹905 cr and ₹1,045 cr, margins 21.807% and 24.023%, +15.5% on EBIT and 46.0 bp of margin per 1% of depreciation. Worked Example 6: reported $7,300m → $7,350m (+0.685%), constant-currency $7,665m (+5.000%), drag 4.32 pp. Section 4.11: (1 − 0.18) × 83/74 − 1 = −8.03% and 1.10 × 79.8/84 − 1 = +4.50%. All reproduced here. Then run the check the module does not: invert the dealer's quoted forward for the interest rate it implies. (84.20/83.00) × 1.04 − 1 = 5.5036% against an actual rupee rate of 7.00%, a 149.6 basis point CIP basis, which is the mispricing expressed as the quantity a rates desk would actually quote, and it maps exactly onto the arbitrage profit of 1.496% of the rupee leg.

Verify, second pass: the sections added after the first release. Section 4.1's notation box: 1/84.00 = 0.011905 dollars per rupee, which is 1.1905 US cents, and EUR/USD = 1.08 inverts to 0.925926 euros per dollar; the two conventions differ in the string only, so the number is never inverted in the conversion. Worked Example 6's currency move: the euro goes 1.00/1.10 − 1 = −9.090909% against the dollar, and the dollar goes 1.000000/0.909091 − 1 = +10.000000% against the euro, which is why the printed pair reads "about 9 percent" one way and "about 10 percent" the other. Section 4.5's sixth-edition identity on the stylised year: −47 + 1 + 2 = −44, so net lending is −$44bn and net borrowing from the world is $44bn; the residual check CA + KA − FA + EO = 0 closes exactly. Practice item P19: net lending −25.0 + 0.5 + 0 = −$24.5bn, liabilities incurred 18.0 + 9.0 + 4.0 = $31.0bn, so reserve assets change by −24.5 + 31.0 = +$6.5bn. Then run the check none of those sections asks for: take the same year and shift $5bn of the financing from portfolio to direct investment. Every number in the identity is unchanged, the reserve build is still $6.5bn, and the country's vulnerability has fallen sharply. An identity that does not move when the risk moves is the reason §4.5 insists the quality of the financing, and not the arithmetic, is what you are paid to judge.

Quantitative lens

FX is the corner of macro where the identities are tight enough to check to a basis point and the theories are loose enough that the data reject them. Both halves are worth computing. Everything below uses the module's own stated rates and levels, which are illustrative; nothing here is a claim about the current spot rate, forward premium or REER.

Covered interest parity is an identity, and the shortcut everyone uses is wrong by more than a desk can tolerate. The forward premium is not the interest-rate gap. It is (r_home − r_foreign)/(1 + r_foreign), exactly. On Worked Example 1's numbers the fair one-year forward is ₹85.3942 and the premium is 2.884615% against a naive rate gap of 3.0000%, a difference of 11.5 basis points. On §4.3's numbers the premium is 1.913876% against a naive 2.0000%, a difference of 8.6 basis points. Those gaps are small in percentage terms and enormous in the units this market trades in, and they grow with the level of the foreign rate, which is why the shortcut is more dangerous in rupees than in yen, where the same 6.5-point gap produces a discrepancy of only 3.2 bp. Then invert the identity, which is the move that turns it into a diagnostic: the dealer's ₹84.20 quote implies a rupee interest rate of (84.20/83.00) × 1.04 − 1 = 5.5036%, against an actual 7.00%, a 149.6 bp basis, which is exactly the arbitrage profit expressed as a rate, and exactly the quantity that widens in a dollar squeeze. A CIP basis series built from FBIL forwards and cash rates is a dollar-funding-stress indicator you can maintain yourself.

Uncovered interest parity is a hypothesis, and its breakeven is exact even where its verdict is not. The carry trade earns (1 + r_high)/(1 + r_low)/(1 + realised depreciation) − 1, so the depreciation at which it exactly breaks even is the forward premium, not the rate gap: 2.885%, not 3.00%. Now run the module's own long-run figure through it. §4.11 states the rupee's structural depreciation at roughly 3–3.5% a year. At 3.0% the Worked Example 1 carry returns −0.112% a year; at 3.5% it returns −0.595%. In other words, on the module's own stated numbers, the rupee's interest advantage has been approximately consumed by its depreciation. The forward-premium puzzle that makes carry profitable for certain major-currency pairs is not visible in this particular pair at these particular parameters, and the "steady nickels" of Worked Example 2 depend on a specific pair and a specific sample. That is a useful corrective: carry is not a property of high-yield currencies in general, it is an empirical regularity in some pairs over some periods, and the honest way to hold it is that UIP is rejected on average in the literature and can hold perfectly well for the currency you are actually looking at.

Which brings up a discipline worth making a habit: a depreciation rate and a pair of dated levels are two statements of one fact, and they must reconcile. §4.11 offers both: "~3–3.5%/yr" and "~₹18/$ in 1990 to ~₹85+/$ today." Compute the compound annual rate the endpoints imply: over 36 years it is 4.406%, over 35 years 4.535%, over 34 years 4.671%. Run it the other way and a 3.5%/yr rate would take 45.1 years to carry ₹18 to ₹85, and 3.0%/yr would take 52.5 years. The two statements differ by roughly a percentage point a year, which over three decades is the difference between ₹63 and ₹85. Whichever figure is right, the check is one line and the habit is the point: whenever you are handed a rate and a pair of levels, compute the CAGR from the levels and see whether it matches. The economics is interesting too, and it is exactly what §4.2 warns about, since the realised depreciation over that window ran ahead of what the inflation differential alone predicts, which is a demonstration that PPP is a gravitational anchor rather than a law, not a refutation of either number.

Nominal stability and competitiveness are different things, and the gap compounds. §4.1's decade, with the nominal rate flat, Indian prices up 60% and US prices up 20%, implies a real exchange rate falling from 84 to 63, a 25% real appreciation of the rupee measured as the fall in the real rupees-per-dollar rate. Stated as the actionable number: the nominal rate would have had to move from ₹84 to ₹112, a +33.3% nominal depreciation, merely to leave competitiveness where it started. Annually, the arithmetic is %ΔREER = %ΔNEER − (domestic inflation − foreign inflation) and the exact offsetting depreciation for a 4.5%-versus-2.5% inflation gap is 1.9512%/yr, not 2.0%. So a rupee that is nominally flat is appreciating 1.951% a year in real terms, which compounds to +21.3% over a decade of quietly worsening competitiveness with no headline attached. And a rupee falling 2% a year against a 2-point inflation gap is not weakening at all in the sense that matters. It is standing still. "The rupee hit a record low" and "the rupee has become more competitive" are independent claims, and only one of them is about anything real.

One closed form the module leaves on the table, and it changes how you model an exporter. Worked Example 5's firm has revenue 50S/10 and costs 15S/10 + 2000 in ₹ crore, so EBIT = 35S − 2000 and, dividing, margin = 0.7 − 40/S exactly. Two things fall straight out. The margin asymptotes to 70% as the rupee falls, and 70% is precisely the net-dollar share of revenue, 350/500, so the ceiling on the FX benefit is a structural property of the cost base rather than anything about the currency. And the sensitivity decays as 1/S: 48.2 bp of margin per 1% of depreciation at ₹83, 46.0 bp at ₹87, 40.0 bp at ₹100, 33.3 bp at ₹120. So the currency tailwind on an Indian IT margin is self-limiting, since each further percent of rupee weakness is worth less than the last. A rule of thumb calibrated at one exchange rate is therefore wrong at another, and the closed form costs nothing to carry instead.

Honest limits. Covered interest parity is an arbitrage identity only where the arbitrage is actually available: it needs balance sheet, credit lines, no counterparty or regulatory limits, and a liquid forward at the tenor you want. §4.3 flags the CIP basis as a caveat; the honest version is stronger. A persistent basis is not a free profit anyone is failing to take, it is the price of the balance sheet required to take it, which is why it widens in a squeeze. The rupee's forward market is liquid at the short end and much less so further out, so a one-year computation is cleaner than a five-year one. UIP's rejection is a sample-and-pair-dependent empirical regularity, not a law, and I have not run any regression here. The carry arithmetic above uses the module's own stated parameters and shows only that the result is parameter-sensitive, which is a different and weaker claim than "carry works" or "carry does not." The exporter's closed form is a translation identity holding volumes and pricing fixed, and that is precisely what a sustained currency move does not do: a persistently weak rupee invites price competition among exporters, so the margin gain is competed away over horizons longer than a few quarters, and the model cannot see that. REER indices depend on the basket, the weights, the base year and the price index chosen, so the RBI's 40-currency and 6-currency versions and the BIS's broad and narrow versions will not agree; quote which one you used and its base, or the number means nothing. And every level here is the module's own illustrative figure. Pull today's spot from FBIL, today's forward premia from FBIL, the reserves and the forward book from the RBI's Weekly Statistical Supplement and Bulletin, and the REER from the RBI or BIS, each with its date, before you compute anything you intend to act on.

Do it in code: compute parities exactly and then invert them for the implied rate, so a mispricing becomes a basis in basis points; decompose every cross-border return into its asset leg and its currency leg before you discuss either; hold NEER and REER as separate series and report the nominal depreciation that would offset the inflation gap alongside both; and reconcile every long-run currency claim against its own endpoints. See G1/DV1.01, G1/QM1.02 and G2/QD1.03.

Where this goes next: galaxy cross-links

• G1 · EC1.04 Parity Conditions & Exchange Rate Models owns the full parity system, the equilibrium exchange-rate models, cross rates quoted with bid and offer, and the welfare arithmetic of trade policy that §4.6 states qualitatively.
• G1 · DV1.01 Forwards, Futures & Swaps covers forwards and FX swaps as instruments rather than formulas: pricing, settlement, roll, margin, and why the cross-currency basis exists as a traded quantity rather than an anomaly.
• G1 · MS1.04 Global Dollar & Digital Money takes the eurodollar system, offshore dollar creation and the swap-line backstop of §4.9 to the depth the topic deserves, including why the world's dollar shortage is a balance-sheet fact rather than a sentiment.
• G1 · QM1.02 Hypothesis Testing & Regression covers the Fama regression that is the actual test of uncovered interest parity, what its coefficient means, and why a rejection in one sample is not a strategy.
• G1 · AL1.02 Commodities & Real Assets puts the commodity mechanics underneath the terms-of-trade force of §4.2, which is the channel through which oil reaches the rupee.
• G2 · QD1.03 Backtesting Engine is about testing a carry strategy without lying to yourself: negative skew, correlated tails, funding costs and the difference between a Sharpe ratio and a survivable position.

Teach it back & journal

Feynman prompt. Write a one-page explainer for a smart 15-year-old titled "Why the rupee falls a little every year, and why that's good news for your uncle in America." You must convey, without jargon: that an exchange rate is just the price of one country's money in another's; that because India's prices rise faster than America's, the rupee tends to buy fewer dollars each year (a small, steady slide); and the payoff idea, which is that when someone in India owns American investments, that slow slide quietly adds to their returns when they bring the money home, but that in any single year the currency's jump or drop can matter more than whether the American stocks went up or down. If your page needs the words "purchasing-power parity," "covered interest parity," or "depreciation," you don't own it yet. Say it in plain words with a rupees-per-dollar example.

Journal prompt. Three entries. (1) Pull today's USD/INR, India's latest FX reserves, and the current-account balance (rbi.org.in), and the dollar index and 10-year real yield (FRED), and write the date beside each. That is the first entry of an external-sector habit you'll keep for years and verify against later. In one line each, note whether the rupee is near, above, or below where the last few years' inflation gap "should" have put it, and whether India's external position looks robust (2020s buffers) or fragile (2013 profile). (2) Take one company you've studied (from the starter bank or your own list) and write two sentences on its currency exposure: is it long the dollar (an exporter like Infosys, where rupee weakness helps) or short the dollar (an importer/dollar-borrower like IndiGo, where rupee weakness hurts), and by roughly how much would a 5% rupee depreciation move its earnings? Notice whether the FX effect is larger than a typical quarter's operating swing. Often it is. (3) Reflect honestly: before this week, when you saw your US holdings had risen, did you count the whole gain as your own skill? Write down, in your own words, how you will now separate the asset return from the currency return, and why confusing the two is the most common way a cross-border investor fools themselves.

This module's flashcards and mastery quiz are wired into the app: see the node's Quiz and Reviews.